From 3c79e379a00cb5ff4ce8231e0dc8c34a5a328da6 Mon Sep 17 00:00:00 2001 From: KovalenkoAE Date: Wed, 14 Aug 2024 23:47:37 +0300 Subject: [PATCH] update --- docs/.nojekyll | 0 docs/_images/HI_deviation.png | Bin 112570 -> 0 bytes .../ice/anomaly_detection/datasets.html | 24 + .../ice/anomaly_detection/metrics.html | 45 +- .../anomaly_detection/models/autoencoder.html | 75 +- .../ice/anomaly_detection/models/base.html | 180 +- .../ice/anomaly_detection/models/gnn.html | 607 ++ .../ice/anomaly_detection/models/stgat.html | 733 ++ .../anomaly_detection/models/transformer.html | 778 ++ docs/_modules/ice/base.html | 726 +- .../ice/fault_diagnosis/datasets.html | 14 + .../ice/fault_diagnosis/models/base.html | 104 +- .../ice/fault_diagnosis/models/mlp.html | 20 +- .../ice/fault_diagnosis/models/tcn.html | 20 +- .../ice/health_index_estimation/datasets.html | 22 +- .../ice/health_index_estimation/metrics.html | 13 + .../health_index_estimation/models/base.html | 123 +- .../health_index_estimation/models/mlp.html | 16 +- .../datasets.html | 51 +- .../metrics.html | 2 +- .../models/base.html | 121 +- .../models/lstm.html | 556 ++ .../models/mlp.html | 25 +- docs/_modules/index.html | 4 + docs/_sources/advanced/index.rst.txt | 5 +- .../advanced/optimization_tutorial.ipynb.txt | 881 +++ .../ad_benchmark_autorncodermlp_256.ipynb.txt | 6745 ----------------- .../benchmark/ad_benchmark_gnn.ipynb.txt | 240 + .../benchmark/ad_benchmark_stgat.ipynb.txt | 242 + .../ad_benchmark_transformer.ipynb.txt | 245 + .../benchmark/fd_benchmark_tcn.ipynb.txt | 461 ++ docs/_sources/benchmark/hi_sota.ipynb.txt | 338 + docs/_sources/benchmark/index.rst.txt | 8 +- docs/_sources/benchmark/rul_sota.ipynb.txt | 187 + .../ice.anomaly_detection.models.rst.txt | 23 +- .../ice.fault_diagnosis.models.rst.txt | 3 - ...ice.health_index_estimation.models.rst.txt | 5 +- ...ning_useful_life_estimation.models.rst.txt | 10 +- docs/_sources/reference/ice.rst.txt | 27 - docs/_sources/reference/modules.rst.txt | 7 - docs/_sources/start/datasets.rst.txt | 5 - docs/_sources/start/tasks/task_hi.rst.txt | 12 +- docs/_static/hi/HI_deviation.png | Bin 112570 -> 0 bytes docs/advanced/index.html | 25 +- docs/advanced/optimization_tutorial.html | 995 +++ .../ad_benchmark_autoencodermlp_256.html | 6 +- .../ad_benchmark_gnn.html} | 144 +- .../ad_benchmark_stgat.html} | 164 +- ...256.html => ad_benchmark_transformer.html} | 131 +- docs/benchmark/fd_benchmark_mlp_256.html | 6 +- docs/benchmark/fd_benchmark_tcn.html | 753 ++ docs/benchmark/hi_sota.html | 665 ++ docs/benchmark/index.html | 12 + docs/benchmark/rul_sota.html | 593 ++ docs/genindex.html | 130 +- docs/objects.inv | Bin 1790 -> 2013 bytes docs/py-modindex.html | 15 - .../ice.anomaly_detection.datasets.html | 31 + docs/reference/ice.anomaly_detection.html | 8 +- .../ice.anomaly_detection.metrics.html | 52 +- .../ice.anomaly_detection.models.html | 208 +- docs/reference/ice.base.html | 98 +- .../ice.fault_diagnosis.datasets.html | 31 + docs/reference/ice.fault_diagnosis.html | 1 + .../reference/ice.fault_diagnosis.models.html | 51 +- .../ice.health_index_estimation.html | 3 +- .../ice.health_index_estimation.metrics.html | 21 + .../ice.health_index_estimation.models.html | 51 +- docs/reference/ice.html | 653 -- ...ining_useful_life_estimation.datasets.html | 28 + .../ice.remaining_useful_life_estimation.html | 6 +- ...aining_useful_life_estimation.metrics.html | 38 +- ...maining_useful_life_estimation.models.html | 87 +- docs/reference/index.html | 1 + docs/searchindex.js | 2 +- docs/start/tasks/task_hi.html | 27 - 76 files changed, 10245 insertions(+), 8494 deletions(-) delete mode 100644 docs/.nojekyll delete mode 100644 docs/_images/HI_deviation.png create mode 100644 docs/_modules/ice/anomaly_detection/models/gnn.html create mode 100644 docs/_modules/ice/anomaly_detection/models/stgat.html create mode 100644 docs/_modules/ice/anomaly_detection/models/transformer.html create mode 100644 docs/_modules/ice/remaining_useful_life_estimation/models/lstm.html create mode 100644 docs/_sources/advanced/optimization_tutorial.ipynb.txt delete mode 100644 docs/_sources/benchmark/ad_benchmark_autorncodermlp_256.ipynb.txt create mode 100644 docs/_sources/benchmark/ad_benchmark_gnn.ipynb.txt create mode 100644 docs/_sources/benchmark/ad_benchmark_stgat.ipynb.txt create mode 100644 docs/_sources/benchmark/ad_benchmark_transformer.ipynb.txt create mode 100644 docs/_sources/benchmark/fd_benchmark_tcn.ipynb.txt create mode 100644 docs/_sources/benchmark/hi_sota.ipynb.txt create mode 100644 docs/_sources/benchmark/rul_sota.ipynb.txt delete mode 100644 docs/_sources/reference/ice.rst.txt delete mode 100644 docs/_sources/reference/modules.rst.txt delete mode 100644 docs/_sources/start/datasets.rst.txt delete mode 100644 docs/_static/hi/HI_deviation.png create mode 100644 docs/advanced/optimization_tutorial.html rename docs/{start/datasets.html => benchmark/ad_benchmark_gnn.html} (61%) rename docs/{reference/modules.html => benchmark/ad_benchmark_stgat.html} (57%) rename docs/benchmark/{ad_benchmark_autorncodermlp_256.html => ad_benchmark_transformer.html} (60%) create mode 100644 docs/benchmark/fd_benchmark_tcn.html create mode 100644 docs/benchmark/hi_sota.html create mode 100644 docs/benchmark/rul_sota.html delete mode 100644 docs/reference/ice.html diff --git a/docs/.nojekyll b/docs/.nojekyll deleted file mode 100644 index e69de29..0000000 diff --git a/docs/_images/HI_deviation.png b/docs/_images/HI_deviation.png deleted file mode 100644 index aed36e9256ed17c68933447ea038ce208e6d043f..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 112570 zcmZs@1zeNs`#;V`gER=zAV^9aq-zq=AyU$aN_PwxsdPw(lF|(#5*slIQIJr&OIk)V zn%~3s|2gNV=da_n4ffRi-1l`|@3LY^6_JlD{`;09jfVPJ#EF)*(`0{q7S(PCi#-)jsERS?}j zuXRA2|N9IW10&oP1M7dEF#?{i|D^&yz}EkJ#>@r%yT@Gc-=D_Z$;JHpHJIpnvl`vH z0^kYXMftHi1_m|x^$*B8=R0r-5DXQC`?}tspIP{Rx+5o-Ij_I8ZnJ>EES7T#{?u3x zUF0jzbSm_X53lUYM~tS7HsmTtCr2kWDn|=Z_7Cr4PoSkC<52JfTKzBg?JH@HkT6oY zzh0|9zca8a)c-XLHsC*iCKed*7Y%xer5yC12XYhNFgdiy_W74bvegZ#IEfhe@qauZ z-E4z%lY(P?dP#r$1$Z0eOqLsjuH00#SNYG*B{)F@$Hcx1O%CMDpg$gStPs$)@mTce ze-G%P4W`^-IB0jDV1t1Fc(f`8foB`4s4Ln3+#&f|E*9s!2`ZuIUv2gp;UpSEAze@I z>*-P5)@Yi>$DUN7%Nsre&G zujB#!Dw9a_%e^$iwg72z3JS8DH^VYB=@k?e7Z7t7e2f9d3(wz*`5g;fp0CO-RW{6i zXO$Kge=J+xG~9@SZ|=NgI>&hZ`gPx%ILUY;>n))b9po57H=3INQ*ll}R5TwP zukB*XnT(FET$VxFx9;X%59SqceN~Uaml@8j-@0dE+bxYpllEFUilGBj4k8tX@4)5y zl@cu%e*Bj^;j+kekB!0e{jJ`CKW`{92}}2MC-?qK0sB#2t?c@atZ4A%FgrgdCo+tP z?m%)`xSC?yb1p)SzlBmoMa8hei9z?3pOKN#j8d$Oapi|Y&}R1Lwr+z)=&~qOYX7TH z-kkAGwYWi{b z7tzkD)?{iEmta2C;=QE-j(NQV3B6|2a+>)A3ya)Y^U$Q9b8Y6-pEoDob5ylmo`w~R zubw&0z!!GAnZj#d{%6uOM?xMv2pg51eVQaYDF+JNC`(OFX6EPDf;Cc*llQ|%d+e(^ zTgVqwf95z(`=e)E#OS>jLa9AUnos4*twuCn?CiLF&5~?T0aYK(G`VxszJKre+5h1~ zvgFv<2`?GrWAQ(ic^!LjR8&-bX103{urvb+6f`u5%F4^hNJ(?|XXpatb<0KdJT&n) zdwP1pm~swfBqV5;KE)1Mo}3+R$=leJ%qsbyMy%~#Ep^A;W@lfnuo}VX5U2~f`uTpq zV;1)K@#Fk6*V}&{G#=8;3><7jf`| zFBUq&H|0kw9>45ngxx^T1iBDH_bo4;2I{w-Dx!>Kt^a78I z!U*0OIIilB94j&U{HXIb=#HTmCnu3UARVnTjplDVCpWHl_%=eD@kYzPsDI97e9MdB zsJi)tTMkJb4&Tjnh>MOMI_FyS-{t=P{EzUbMVcEVtN2(i-5_*PcL1tqQ?;ONhGLNP zvf*#pz@8h{9{H))51*`lr@FuK;Qsb(i|fw(eSf|Yr;jPFdp}f6nfaxqO{bQ7=x1Hj zeSE~-+}u1ioPlJ=!6lQ!*mD|OQSiebodN%bp2Lie~iie7tlZ~5(L?gOib zIIYcmlbU~qQxqEnWFEJYE@(AaZotoL)=cBvdU8iM0$x^Ba(e4>MYQePEK1ycB{QMC z#cf&D?`*5VcMe{v+lttW+Malin-g$aT4hwd@{nXi{&Fc!b~GFP^{aXejbI+C=BDhK z-MC*gbr0sBR~Upfj@x<(wP+8)Kb{x}z!?FfvSFx?xJ-Vz?(YpM4BvLF7OTI`+aKFx z67t!5YI>j<+A($*W+1-TFc*M0s+B$a`TqEPe?ZV~nA6lo6F*>t)(IH|YTWOa{D9kX z>DPK}wEV8zCaA28+uvQDs^HN>Jo-Np6Yv&w8$#kIo!XP{zggMvWT8DYUCcG_sG{v~ z)YvFS=Hy3Gx@@4#ix)387(kt$K5d|hqO1MUh`CS+{TV#%ka(~ zm9HuO%qbAu>s-x2Ex9Re(YqPcqI`hJu$z3Xlfe(C6F3o!ytnpEtfakNE>p}Ed~t>z z>pGBtqGDW^J}tZ-kQw3v$1-(Q{&%$QvO_>~I#0~tZR;4BdO`AxDDs^!JnwK*sWa<_ zTv?5`1qBM!J8GJmqf@$&MaBfNtQ@U)Q#xjw?UcPft3hdT!m5fOu- z7khhq2c>$O(#t6ZDaYdfopRGkL3(R5-L>-l$?@^IJiE>_aJ}lb++2vkcH8k{v?Ln| z-lrj*^W+o1mxF( zqN0uPwu;Ej-^b9IOxA7to>B-zW+`15RF~na3|o3)JMF}Xifs}xfZXw%-=^1sfADTKsc(WKCjV%u8&=~3#~N)n4eiCn{=_Qo#Le0P+Z^<1GZUkRngKQD zs$wU@8&-=qMIorn*RY}S`l6M2>p!k_F#Bk+mN!|~A>2}oltLWq+X5yjZKA3aTQ0sf zazA>I8?4&QBU}wDyAx?^W zMH9lg3wMu)_yebMXuI=44NEgyDiWwNEcRFW4B;MJ*v)E`KBT{7E{M|Y`FlS(tEc%F z_KcT(azc7Jn7Nz?o+va=Ij}Gg;f_t+0QS`v^n(!bV->@v8&oV`AuW_Qp|d`#(~^4>e#X&oFYH=n(5iNtkt;(XuHDiSs!e{dYIi0@4&idXjZT<8q+i*uI@%-;o zja-Z)c;gM&7QcZ|u`ok>i}QhkO_383HRJcK%QHftU>k0s0t{W2sN_?a5Wg$jt`$o=*8R{iE}B{K41e6C%gLhYHL z_t2G_RN(oOn)|TJh3y zlv&~lGkIS3#w^OwUB7>08(0|^7X&a1e=h_Bhh`>3QBnF9ugoK9r%>R_a z{{6HlK#%O?CmtF#5jKUYJ-jWxyV)Ca1&L%zB+0Bv}U>Pqr?(>B?Djhewic*CK+IlM;P(leqRPw>g(&YqU? z$;!FSz8n}&b5_n31}@fEKCb(>*o1-aVF4!Ra~@WHb!LR0W`4&9<+Z*lLmBxliIj$P zB)G|C-4Vcdt(BkM%zis(W>6cuh3dewk%0?^j|L@GUzEI+B@&nT|AwnBuVn_Cw>5&>Pu4>zCXfF?lY_# z`Lx&Z)Vxk-Z@L_a8hDZ4C_2v=HL78~+HTwWzzT9^hcNz8Jk94|sShQW?LGd?W5()( z!p0-i9Kzm^d-=$Ms#+2k-}?;!VT&Wr7EW$-AXjq&Us-4S;80OfDXXY_1OPyI>7(M) z1-mA;UoJoR6(qNLk9jXDHMe;q7qy!G#zma9{Vy^L!j$Fr%BD)k#NJD^T=OlvWyk3n zBO@b5b+S81%ryX_h&YS~0RYkRb398#Tpaq{$e_t@`&CDpW$`egsW`c&_{@$hfSZW_ zjSRE!t_fSmvg96hhuWwnXq%ln@5A$oIa>?8)L>P~g4_r8*5%T(0>O=(gnV?)DQ%7Y z(ze>#q^IcZxce`0SeThHL^6n~+b$)Jwx*Oo+u3R#AOD(lC)0rlsv9*_dEa&Y`4@L6 z^PMO0Kj0*aF52LZ<*9V(VrLm(p|8B(+mP0}`DPqGmTZ<>bdJ5m_=x>4#!~#7>U7@@ zLiEKgU0in7ld`3J$gfM=+2bL)lNZ0`jpq0uZfyfyICf|uo%U9|6gnQY*hzcm1ZCPO zNVO7rbEx2zIzU2bRbB41>KAy{A{RW&AG{KqCnL8>kYiXqY8CBlfNq0=o zNKa1>W7?PgePI&?C+=k#H5%wMl54?O6PJo^v);qJuVb415~~@=+0CM6RMZYSEGUi8 zCWOGVXcBSX15Ti%6)z|(TsWJ%3W$zd1^t$(=JCN)df|$V&I-0Ce7m*btbEV!<+YzN ziwQj5o6gv7-V3Re!o?rv00L|djyXb|nS%qjx_OU4PF`LRiZZ$Qx%X5m{C>kk3T!67M(iMJG8Zu(ZPq-~Lyrc79MX@o8r;3;w0ak5%lBd4 z35EAUvEJMP;@Ul?fXN$RRiyRv=ZnClf7|+@%YR%`BfwQO=CO--uHI+*>+DcQJ_8r6 zC$kP9K6o1pumr0pN4zfSf>}V4d-wMs&|Z*@$;v`Sn(<-}1gb0Gvi#3?c;{9P+D264QbCj)8S zUmTII(7_{tSncdV7JF8Zv3GehnEpO`bu(1J1y^Y=3I1_*X7`b{x{(&L+zWeC9#z2j zRP?*Y#W61IV{dR1c22co86MKRI7r6}GNkX|#Y;@{3=TzjU8fo1K_As$VgQg0f=9^& zG;qtxw4!Y0-ooI<7K$Fn7M~eA%8cf$IFU?9mG?};Me*n+;|NvyDVxB?rI$!d zu`~4gW(45d{6Stp-0{4_)2#{CwVCWAFB# zpnxox)iYPJpQeLc0ko_gy27G!jQw`=PjA+s6fKt2K2WV#UtrRBzjUsS z9z~=9s}knpH9nzVZ5Eqzc~Fc;O#A={UwlnJ$feGAIv7%Mseg^=^L*etH}W6sO_SWN z_Ga%XZ1;1~X4Rp1&T_}C#EZFUQ+?~NIQ--+EH8hspColn8V2>WDcxU8a)E{j6LM!6 zp!gFt&$>DO9NNR=6^?(hzddT1@*>MJcu`MQYoVX(QIQ2ejhs$TOiZK;Srb3+dlL%6 z#AzI1)O_&EydHGLOD}q~+O!(o-BKV$2pSb&>r#JU^3{e*oBHQI^N-dLKFtPQxgXE~ zoPyX?7oRVlT5W@?X`is`$beWX0^VG=UETJa-}*^QX+XzXKagLaVVgN5@J;x8n9>8> zx3IboU-B#Tc``Lm2So%^#SRr8Q$OtAhsl_vMD1WojtQ9$SGuk9t|1mjd;9v(PsMsgK z*~eY0V1*z_JcYVd+-jsuFI^fY0>;GL9h4_|Ydw(NCmwV5+X_P3f;Br{wI?2Gn?b|h zV?A8B@HB25w6G0Ak;uzENBiOX?Gw(gN#t%nTl5ShP~O-kiC&_e1ahsPF~!l~`TM`X zxr|%lpJM64DL6DO`rZhqOG5eh+RoiPHb*O^H5QwGtsDG8gKYC^90-Ao3na*Byi(pd zaoc(+C_05IbF6<+9!b7i=N!N2QC4>T(;)bM1Gw)XEejfen!m!DG}t|nog;2)uQEkE z8O*(JSYrV~HShNme+i1d^GoSvBNGf@F0sdr%G70q_^S8EG)0x4ov24*rMPIz-YTm- zPwWm(>_A2^4TtqAK@XF4(tV zZrxqDsOgg)%-4qh)>eNsz+zXc#=V!g}dTL-;RA1c1KBW$D##t_?p|K}}8i0ZQS zS3kPIj&rClf%jEQJ)`ks8yili#s!=i5zQ=Jy@I%yn1k^XpWmw9IcAG=;m`a=eApYK z?Y6qAOlY94p=D25Uiu+ga23+KZHmF0$m*#ys+<5X4|*qytrHHaXP#@+tosfF!aQvI z!|ttK-tYB%tgLdfexGb;RexDz^Mc?a$`TD{JaeN|W^SV1`eCihh=1dClo7P|g+ z<*jS&=gv)H>0!DzHg%5_D->A%7f-u_=Koc ze%U<)yd|Go;DGR+HptW2n|b+?d{7u=7ZO&W(%5W$TcpiotCNoKMpLB4lUZqy6ZKA6 z@mbFtCN?72axN)+*iDlbQVl!v+dM~CQ0`p99Gm|7EsEQSf|payJo(X4W%A^VOi`My zj{qp}CVKM!v{nk|naO;Jqqci$PFmhzke_I5E zr)4L4AoY%k0w|`JqI(N}qmY7%@pUKT@^hcw?cqLJd(^EM3wi1-i=P%py?~mKLdT_w z?JkgGcDoUoh<^#Qz3uS%=aYmZkIG`5N1zICobf$c?T6Y^H>y`Iku>ge7tF8e0O;OG)K~m`mGQN!Qz4LFooW*@feaJIO9*jA=Mg^9N)km9fSF zsq0?qo~GVTQICc#iFz%%!!$aLs6(NR+%_ zEW^vJjtkqgIi4LGQ#$)~64P>MFLB;?ow(`q^zEmIF*Wfh9n*kZsbUpb^v)*|yZlPM zkvlTyqmldu8Q;xkJAZ!BHxc_9WoC=A)i7w(3OUOug> z6RfjaPo3tdsOMQlS)wNAHjVeieE=we*HYD&Q~gDs(j6lo=*vXmlTHm&fRYpq zFtF3E-3+l>^H#>ay}i3#)aLUu&^bkbdckgvGnIYwi@?{%Zb37`nR0%VxaX*iy#aPe z_)hM`%i7_)$-Uo~)opsVOr4S5)*cyZU#MsLI?|r6ZXqv0zo58L=`oWTsDxl2v5J{c zimWTPhati_5cb%a;aG390MN%+4$z!dv&}xoYx!~4y4rd;`3{|^Q>XiYbojNdw>DBv zU8I#`wq5ecoIByNTlSJH@TAvV;7KRmbl}y6vX)jFL8ktXl@+^$seGllPb6NJo3Gq( zYkydZPD3fQi{w0luqmE^$tVUr#uJ{@J?SDafo**_eI81vHqY{FL1(6pyqCpUEydaH zkH9-N-9w1r36i}eqx#15_G6xdv=m#79&G9g(T<$V!ohiFKh)Y++UQA&_SLP*#l zjh=o9sETEe!DgZJZNLuyzC_pZ4POQto4`KQ7vKcS8bz{(<7dE} zMA~h^>u>cBh@%~JkbvNz9F6|?5_i+%B_I!c`k=VO`D;+BX4AHny|b_<$!-!rr%ofG zE_3GzbG!U^MP9^uw>6=Dz`ZfbI7BdZIg##2dZSkql~DI&Jo%ijF})tLwc5U8nS>W7 z5f@|?$ZEMimynIhzMD4(W}W_!e=WLr7Dh)!<+iVmq~Eh8-{?FmjT3|&IY^zZFP5drYu;*7j45?6DDSvzO5pDo&MzgSb=g@+#|&SfsHAMKRXWf(SebXxajjW>38)|ly=2!0F`#O%L-ePF+{ zvQ=>icP6Vw3=x~`_*;ylgAALrKHdkNl6*WJM@`!%8vUb0MAcjmDxvm1i&oUak#Xh zV3y!W<){|z$#$Z|L7#!E72QOygHnH7I9D z|AX8qHj7S@pqPE@{%kr{S(adtYk82s&eio@zeZeCltk}YmDj_jO;fnXH@zPzgm0f9v5E_^4WNBFJI`3pWqVH2lYK4nuUGn zn+jXZBiQ5mc-g~KJ^wl3_9PvEh3_`qUvS*alm2&s4U*w=^1})*#G+J1>;?d^bc5XL z(`^yE^aD2%M{2f0syn0G_jZHp!l!(Yoe=^0CV>f5m_kO|o=|;Ej`u;a{Gi9%;J5F{ zW6yXj));=S&+<`ca7gn5Vf|4=GVWul>_-c8s@|gXJ(!HNv;^#VJlx#00e+v(YHRy=G& zCXkJ|!JPqHwI!*)~IBb`r!xl}J zkIVIaHi|-NiDaFZ*S*F5T}<&o8TcisE+fLBK(Pv7#*>TZK*VqTe$&YhQ5RBfY-ZaO zIWk4m+=N83knJXg8@le_;DHyPZ$rIiTJQSTr>XLz`;bv{D&yy@%x^_qPI^DI%os~h zqd7oYKs6()XhN0tXz|gZlfl9PQ8@V#cerV!4izz=(6TIhvbF(`!+LA7^btoEZrjr@ z#22f?pYAs<4@T`AR`ojClUs(xXq zucCP3#5&Ev%v&&eYlzXojy|YN;A#cT(K=(p%+>!Q+&Q%sb-VwSp<}8410wA=m=#xu zAMiFq*umNr=J4%%RfX?e1IcZefGk_w$idFOZnw*wX1tswO3&RXBPnSNTL)zNshuUE zS%8{n2C7)X@lOgBdxJc|)e2R<$4fR2Gnb8;RgwQLXurrHTUz)VzsFr>WGEhmPHN_R zB|_v+tXcm0>Cl~2JX#b$W}y4#_<2k4m<%*87pu{sc@I7nPdalt>f-fU>qFSt%IY;t zUoAw2x(62oXB)8$$b(TVgQql<=<#o;_a*wyFL8+EY;8-4otUf~r3cXE>GPnyVZhhYve50P^;?PTbjL#VjZJ7s&vxfq zimQ*`n&Ubrp31LwPAG3n6`OV;_TbynPv#A-Jmx1FG9zN^=Ap7TkXNSEgLb=BzVKC5 z`$jDk#%z)ds)sOeWD^OUE%t`)Q^ySXL;!-_Ata7b+KA9OCGBz{oM{n=8N(+#-EmA~ zp_9pD=SNe{vJ8yX#CoB?%`Shd(88@_AbO2<45`0hBHFB5N1di0^8B?-A{ymx2Gx z+{I+%mEW_7eZt12zj&&$9g+ktM{nJ^1uo&k991qkujRlv44Bp`tQwYwC@1&c*$3#a zYsTe}FPMCSMD|<&aP%;KeJzB$-FjBl4>j=bPtj(Q4Uh`BbdN-oiUL@-c&dm&%A2%j z-%RuxSZ%!vygXTbYgS(G$_)(wK=*i~YyQo1*^6~8GJ1LwC|=I5wXibM5lgqx@7W01 zkDGTDZM*oea6aWeYG;o~+St=qJL|#Fo3$Ns!W?MZLe5$tkMHYG3r&SNWf^fyQ%Myw zh&56kF9tAcHE&B7fvD9APdba8zcVOt@>Wcl}%kuwVwo53a@PYHF!qo3G#Ojp zin{|slg%46fz+t(R8PXWoc&CL#r5JfBF|2@7B+~1Vs?msv#fv2u^OhYHvzay_Jmf- zGOC@J{+J_@cqL{iAI08>CaoHn#L4f_IY&jW%F+^>3ZHOY370LYm3 ziX;+D+~<>RFc!AsE~q`93vy<{)SX3atjgRE@v8~xRkM~N25MJecEp;fveyi-f2Kn* z($I{%4Bgr`{%h@DSuou6=hQ*@d5b ziu=`Vjwh@*(NGxM3rq^SSjxi#ZW4OY-E&FeiUE;t@1wD`rk}}gJLzKyUp#_8evZ1r zG2Ee8-)hBP6H#ej4g{pli>^&nSX0I`3@9sz1Q1Km%npGvdVtNv7mcdi#L46_`j|5| z@Zn#O34xIUlyZW3vyRjx-NHR?QmR>Vxn(65w#Fnp=(4XYP>F-9~KW_>fULLoS<4eKCDbw8Np0N(J5H}2>;Ff_A%o<Pmq#HFQBY=#p6@a@m_vJxf@8q3+Mdj9gLIt!+!$Uif7U8O2>%u>t) z8{y^}wJ(b*H=frP+w=6rowEo$klG>3>fF3+E7Go%#0RDnKl@-E_r6ZP?dSxSD!ao) zOi`0?ZEIpC`s);k+;r;+`QRl*0InCctC z&6-j67S&&PW-rUb^^x)Z3(^uO*Si&Px zC38N~W1UB+Y!z`PY;EyR+uNysHMgCQ4b`MnBRNuI)4inS(oK6~;piE9a&C%r+ikQ~ zi7c#Pvv|{tf)bHte|0%|?&sMU&2%mPB}Lz2S5esbW+v}5Ga?`R`LlAxHJDE0D-3e3 zCT!pT2O=L6fkp_ba-UWnc*vHOe<|)97|cfJA{!t1ptjnE9nWqwUxrf9K{J%K-pyay zJJ2NCl2yYtop^PrFgMvv%%q1mgGE8q(|S`w(#&-iV;Y3Y$RPk~No!1#46J&TIrw}L zmsl<}*d}3=ig+z#{fC;jb0heBl(a;5OM3u~h?)TTlWssVqF}kN!a6;`#IlahL#V3m zcU&2KRTV$?GH;|!Xe4a?Ht*V-Y2WVeuSkME@M9>AjfVA-v?-N+)uJTvd%>aq$vr*@ zdcD5-Hy0o2m)5%y-sDBvSGR<@;4fE~mB|4JLJ-JYIYzj#zu0Y^nYGP)AOjCCAzt@uivPZBfYpq&K2nk`koYAVMZ0mhrcpKi& zb8WdFg#t5dU);Cd*-6vZA1sMe;U5EDVQ;1um)WQFd(D94^91nvCBCG}E&oQsAfX%W z<6N6m|1Yr{BuY_I!*neKChMV+fbuVbc)K_hW~fYU+jR4%a1qbP%P%?{H=~2%B5USH z)g03A4M!y(kK&_iDFBM1;O=x=a1S`Q`bQ&4^TX!t{KuxV_Ixr#w*S*5{N^x5(58n$ z1|-D4q=w*BvXZ)0fQjQ#V-xT0p0)=YJrp1jpfz1nt@k6M3C zBQzMLrexf;=|m-IT)%P(D-Tnen!^-nBcM0pe&K&Z;L;j0pcbs{#aR^>RjqIBG+ygDey06ndt()B=UP(WG?(diY$0)}ZDioeBw5nLk{tjpL>-b*^fP=wKU2+y-Q+GKjy8-_ z@H^lHh(QHyi;9ea(bM&MCV|~UBMuq!XUtrmUSt$axlDMA0;y55@n>8xAWg{q2jtd z=571u0eQ$ycoxp=$t2`HzJPBK_L!PE+RV1{Mw%?v&og+vu*s|@^U|VGH4Lr7*g`LL z3;!4!7{j*4nAvv`-@ofPOw?jOd-e<-Iz9jKC16%wXZZAu;NSVVLkcQ>57ryucv!*R ztCo%EB4?gG0!75aF@PAsSetiJTv{gbPQDqHqw0C$EK2iv?a9ZMv2mAOPAcwg-$p-A zE*Eb-N6!-moo6z(7y=iM=U+%>IMbm4--lQd2KM=DspEb{@bdEVwTsaVwj~Tu9c=9E z3*VHO3WJ~t93=m^rgx;ln)w7HBPXoW4Ys>gbh<-~praT*l0ghQ1x_>3AEc+luOLPp(lJVV2B%e6JR(L1{&^sjoGY-w@N(CT*fs2BcoX{|f%gxI$Q4 zHs@sh^|pScQ7DPbF-F_j%=Tclx$K3dqGH!;$z6tjSHdw1sMt~|Hs6pC-JVEPD~Cz; zoq&(e_eQ&%1Z+7fx>x<_mgp`V56!SARhNeC{(PdK7_n&vJ$!Gbmpq52s@XU(BFnkf zO(PDxn6^@AY$)-M=|OP>-t40!^kP8+GG{mJM=S0J1jx`6cRyw|i#!7is-*9&2ad#Z zx1Q}7Sqru%OSUevE>=1v?%=4i8NU?dXcW>}ZYp(M8)PY=6PiIlyIcLwNEsMnkB^Uw zh!=h_tIZExyH0>rt1N}WX18A!brf!YiC~d=o_XN5y^>;I?urracEgW#WLehCj#<3_ z#aYz{%Of`9@}D2n8k;xPTa7}__)7WaZ1G&W)R4#wBjLLwK!sofLFR)207;i-;)YKQ zB&LW6JS=Q%70tnk%f-Ar)=Vj%+Tc>ZDav4%@%Sf1a)r0%@0?Vj7%+Ex&SpOgmpMn1 zqc&U@JMpJI$4pO8UjYW{w$qVnQmSRs7WK`kzOjGFmU#k@JpCPFrKSVd0l9!()sn%J zn)xc_h%Dx8w1+{F!@gBAv^jUUlK?+0@X)<@ocN1t!{V92Bq_D$(e~~NZ9)6;J`}Wh zh2!A&xU9DuBZ&#~(>WF%fisGC8X&52o3Nzr%%`9)tYyz`B^MvnPdRlkK&OKNGaMFf z&enw9Sy=eKmY+L+-lTKE0kK}(5siL1RZ~+_Nm*HYG=D2z3ixf+A=A~dQWowUqsVUN zFK6xqE-K)aD1ewo0_VjWCTnV?yh`w$e_Ltj*_*8*;-ssI zyCNooO8C&RP6StWc9h@x>ZL@@9vaegJc~2Uys#G+g~>VOSmX|fSiv}0Qrz=typLbJ zv+7=~4!p3>Z*K{9^xM3R1$Z|SQd7gSrC}Rz?39$0YerM%bnQ0NV^H6}cCBD{3hn+v zqXj-jC>ct)pPUR;svMgY`ne{1Rc4S-M*NNa;*4|Vu}vh))}~!~Z=jM<$Arnx8yuV# z$~2d#fgtucOeMBu*(Euy01mlq$A#wk7eiQW9P@&pYUQ##Pe6qpVl-;wnEH6L1gn{R z&&S87W@^i@#aCFDPW+erx$_K&3gqn1rV@(MnSLn4v0Dm7 zb>O-}>w!m(j1E;PA*9%QObX>0-8X^fcSaL`QR+Pj0~xd#ZTd7|k8XVg2rZy5WBwsX z)&hyaAXhC{A$}f>JlH(4-fiC5CI$NucvWCh4Y zQtNZsQ9pjz7q%G5Qorn%4tJW-#N-cL>HK7RCrrijM9C@1mFRFrJRkR zEP5Y-4)|UjME0HbmE(dG)YRt=x%Tzk(t9VtzUI_&s1Nvh?^B5{?J$9R&H{SxHp~O9 z48zenKD#2un`N(0Nf@d^{^nVb2G{4)RVJ1d(lcf6rr21VtCmQP@~6Ao`cq?5BHMZ~ zNPhcK`0M z+ZgTAP?FkhE%C<4Urto3=83&y=YxjPauYQ z)1>h=CuV71HnEyw7f$P(%)|k~<=p*%((lX1Ma%DsRri@ z4L^;)=gqOHieb9fmyc?uXzw6MnImeFOG*Cr%8xyb9-osVJ@lGmHakigz&Jq!%NT&I7%>t%b8ga`I&-uF zQwHFN7%6t_3g5kZx7BJv`>T0-P!PXF`{YHUXGw)bRBo=AZK>Bg?`a{xHhQ~#CSNw% z6po=La?4wOvVBYH=@VVqlvT!v)*0SfoUx@zG+#2MwV&loPYXX&!VQWJump#(v)|Y)9{Cfhv$L3fb8XOivetDC4AtGqL?xt7eJFl zNb!^W8-PJ+_{vR%i+h!vyfX#TsNmp-$}w3$ivc5EYK>31qFyQ*^-#}ou|*N5)zEhS z>D%DmJL4|jTzGuFKEUbu%u_KdPK0W!i?tTb*|yzZhI2t4j!a+!GMMi}40fmaf+UAC6H zfmv-w{d4@CJoJ_~pdt9T2hjdwJD8WnQcveh@Ny}Gm;yu|8Jz4pg+Fh*{E%m;kNHkx zdHFk=mn=Rg&8J&Jg3KK+5LI}{`E31NBx2=%&-`_f9!3o_JZ7g_Mu4dMF#jr2Y4jr@F`>x@45P@(>$Gg{f2ap& zE_89yM~3-}e#0dXKY#i1@oc*-j)+d=9z4$2U|T6^IepnMj-n(4Q(siYTOMW6eVg)i zge}l4+3kI>_VBs|s&OqZdc*lwBGQHe9JUUz^#L#$1M4(NyB*O3hfdNjt^*2&$_CZd z+~&CK4TIMVBZgGHt5|E>FrGEl;V1-7s zoVT*54}a7M?gPb?$Y&Wueodc9w-J%8Gj|(Z-vDv}oB7-aY|5Vr^yr3QULV~bpw~Ky zHb^1}ny@ejfiAv8)UVy7e>R%jK~4%bG_0(~9&ap9Bi$(pGO*Wwdq9uki0A!?`jRSA zBK7?R45V31fq<&*Yv1DL0DQFnxI?EkkhbHBHQP!6#Fg4kTE81{V>`B_MkaA}Rz3tq z7|`4})p+x(HO-y+pMvioA!lX0@=g+lE-iTI(nX)`%(Y3{9<8Vb|5iY6AWNM^F1k&XM`saW8;+xTQU1*`cbkc0hlWc)q6J-xg{0FQIv zO%OXu@A!kSrlzLMrn}RQ6F8iN=@Y=VUt4CkMUqGVNx%SVD1I0k)gy1z2&lv(M@k{j zS-G3}+Y8eW2v$;Q(8?e_X0#~nv2P`p-kyd!jYAiEex*)D|A!KgUM^}7sPLS@X9d+T z1rCd^afR*Yv5EV$BNRM)D0iA=B}+ggD6X0?eO))E8+iW$~1B}-SUf|W)T+0x| zNb&C!Iq!~@+0U)3dgZn3%3nFo{)$l^?Gi_9u(0=9<^v=9=-8HkfH{yYBpmXfZ_Z^} zJHx)nJH{8u+-2mhIXw2|@CI+u}BwL*7g<( z@OM5r$afb*xg|*n#;Q3bd~aRkoY$12Q9&Efvv-Y9!-yH80HaKPULJU=-mwFKO@l2k z=r#TM6tA8n;4M%)ANHl0wfLUo9^QNX_hdn4;EBbm>Q1<`b(@ADel%scZ0NfPCcI=5 zoo)z$6*7!_ejZfZ(033dWv`sq%|0u%s*H&BxN9O~UV9TAh+W}N#wiQ(0K!UJaXq|Q z)Kn>MidED*X3y$F`$Z@1#U$;U7-ZdgjVCRUnoCT##v;B7GVA|f$z+K1y}!Q;;9J?V zOm3&qN@G!8Cd6Klj&aj>xRq559e#1(*`C-OslyY?68ursBL4Mzdi7Z&L*;RQ2a zN-?;8$Qj{ZXglV2Id!3Dl_9-z!|w=(Ux#5n@4_cT!4^J0#(t#Gyw`~9g}|y$7OcT~ zP}d6$;$R0sDX!P$Ck{5Yd34)VOVwY1PoPkK_2r8f@!8q0f-!O7F#pV79Ro;QykP>g z9~tU(NniCS1kZUu2h zpX|0<%k))Hp^?t8sSlE;;4mjNRTMppja@&zY4!+2srv5X?o}1lwrU*lYs{E4F4B>% z)t$xx3v1z+GQU#pHi}nEy_Whb5iZ+NL0|1H)uVQ0a z_-%d18xV+QMT=uICpn~~=s$k^7@Pi8<6pE10k>*G^iryUZn}r-VMYs8(L>KzWqj|o z%lTD%_{L&><`>+VTfQ@E{&P5Obq`-ZYmRcO`s+nkW+|3`iIsTB6q!Il9PC3r z-bt6eHY+1k7;s_TmU+%A_tL+bzu~w-Ti^Ccmp(>SMA2HNw&D7_s_$W%_XLu;sXoP_>)+_P*Uh_&h z#2%H4HK~qFOG{h2?!nj5AxE`c1>F5{byywvaUp&xY4w(#o*q1w9{bz}vTOBE%|O0` z57@TJERNpDQ6^W&C4}@ETgZDT$nG}A$t!E;X;tLCO(=t(qdIUdkKR#C9lcr0Cir@x ziSdcVjb*@;V0NMxAeV70)|Zfsr2;w3vHpC4_%wd@9YWA#G*}T8&&iLtZT*2bMadJt zLa>E&a1&>I5!Zu~)Pw^+^UmONu-S4$HSjeEK-pzyS4LcTwUXidhDxrw`Jk+XA>5f% zd!1ZZSom}$CaMY#4{va6Mzo?h<)5kHs00Z?=q5B6tln<35s&J9&qh8JmL9>P9Kjep z&ph|DtFnDri!$BYvz5ZUN|2O+cp4TF$_Y?qL@l;Uu!o$LF_jcVe|j$6H{c*#W1v*A3y{zI{uV@t0WXOA7Q9`zJ|D@PdXs z^2#J|eH3{LYMaSX_jS^p8i`X#2zL6+rge(&*{KqjT0bK*2(tv(MlS5gwi^0$McIw` zrbp89*O+TgeQXKQ)Xb31)3vK4efKyF%hp@TgFV{_F8B=iQQ{SYQrowMJ%KVy4Vo80 zRn#AEYT%mlfc9NvnSoJnWuencM;(rtcL#wm0roV5#u_`L26s!aBzikSolrIEen~E#A2SdX(76Q})Ux4rbe> zIbQLW{EnnQI`5K?id!&*FVeyeJVrpo%$n-1`)A*QfC&YBZu>rUv{*8Jx`m zupw71+x{dZQac!@M_VvWP@CtbzUes%h1f0%a{dUli1Wdcbj2GfaW&r3@FclMzCdHU z9&Y0!Gp5Z&;)P%f{-V8bu+dWHkNl@Q252ED$v9os!@XW*&!}K*$x>kacS&_)(gMsn zkj9m_PR=jFID0h^MCbB*(+vR4{+xO~m<}vZmV>{*@}T8?CX$wlSb?65h+~&;=2bo=EuhSe)ytkh zBEOb?$P0ywt2JLbJ_f6lX@5eKLDcx+Yu(g9da-8OHQvynwtdDyHSkTX)h5I$N2X}5 zWHpgdV|mnGxI$wZ0}i-@z3t-3xX!0`THczt zw+c^VxrGKTftq%*f-u(U5)(Dx6`FpKYx``pXDB%(WnrxCQW)sN@I3T6q;4rw0I;0D zukpnr%9}SA0f3y`#(8&b@`L?F-0`UtmP3uI(9MCl1sUKgZoIZ8vGv|VLi@_?MFD#( z&}|w4zLXk^VfvplCWxP0ps>u(bwUK-er~25S>X}HC>1wGlTpnYG;fpb8Ru#Qn$>gF zZqUxZr_yZR|G-j;ap#kA_F3$Z zt;w(IvtUqVQp!{;D&kEBd@)YW)%mIg$uW*a^MRwGA+GEwG#-!^P))NQdd*LuDH{G* zqgHfnsVl$%%9>m2`7j@(v}-5veXIS;Q9D4bK}bkA1|*=xCIZb_;(+;rJtYiaQyK%c zq9C3=33Kx*YD3GP|Gg0_XUJ3|uZfdvr^;3i-guz{x)ke9xsWh-vQ?TFoh|YK3UvZ) z)Mz-j7CAvA3@bz?Y*!t&l#IAlGFje{w_oh$ch$|V)kmjEf3ggHi6%902gHXkYId`z68HV5)itbs zC1|z*-Z4(|X347&EahYGDp~_IgqR6NSACGBqbUmD?c1B1yOi7m0|SCg%dk)HBLCOK z+e23q!mM@VDCLD%D$r=*sI(-c#mBpI2Tk6(7^?g3ujTV^!@ZNF+lU4`A zKgm%VNz@Z#ebynDTPgl73gtjddjNzb)j@wEgCeE+J(5ye2{M{ZJ7`{ay%Z_#+Q^u| z9C2MRirx)wSICnDc?7-S|Kxw{Bf2>$A-iq$q5T>1f?E48lBsa#i`$KOMFv4Z?MT>7 zmX~RP@W(U)B`Z9?n{%t`YXPqRUH)x!HT7+olv_ri7f!IMA5uBSb0Qr(jVSt(Q^6m6 zUH62AneSxaH4@e=1-}*m$s(1ehz>^F?|%A-W+|{OUm+s~OAcD!4@|%c9J4la3)JZ;-`y5CfYP8sdwS$VF4)omE>Q)FAS+>KA9*&96$ma zInf9dgbxoDG*n8moKrpjS)qKY9nz>Ahq66z;aEZgJ1_THE{ zeoh_nh>bJo05YihWpAus8WmqkR4QIgun<%%4E+rA)>Bgp0j`F@_7MDRT8WQ+kAe8s zD?^v^e;*r0_%fN)5ojPVEW#okIpTBeb{Q7?J{J3u{p38N&74dGWBXPOwl7eB$w$!S zv?wGoL6rgd7A0Z9f@Q5GHeu|iLgf`Woa=uVNW zPRMC3#Ga|h`zskSv1J@!MJ5=ly{TG1%kO{7|H#}vSZq**7y`8L9+1o;BvJC*5q zvHL3%Ux3-y|L4zC?GsdN@|8VT;h6tw;Xq_*{QMdSxL%-nEV+^S14e2f)5N{Vd%!qh=i(;~ zti(YO@Z5O$f1VqGk$@(jx#1A5{HaP-j8ZyLEne|`NX@wlqNw-{2;Rl-Jkgztqu`ZN zbhgN3VAxaa{|ULK@>0J;|3tVA>);l>AA)k8f+|ONk|`kJ-6Tui5P(C}v?1MdSm z9H2eMirw=kXbAD}5X=10_q&*Gw7v!`muke>Srcf!v9+-6q4oo8_jZQ>xXvR#90X_? zE}~J)@(3?R#U6zHfub#2qeA}?cGPyQd-S~<4Y=n9_>ytozU8>Ukr0Kyv`W=E{~u@k z;gK+|tn4N(3~f@@ucG~;RR(UC`f~Bzf&j|n5uNQ2t49;KJwUqDp&Rfc;#rU zfkDUp2c3eP#C8e#MN?R5u z`(K6+dA{QM>BHAsqZ7~(X)Ihc#c^-VRoSS8fFOxk0R{LbpVX1|Jt#DrJKA3p9A~^3 zy(@gm`-j|0N=0`_(tIAn5HibPE(N01*Vf6lKL~CQ+`p~(efzQDW0gbNV!oyYgG~mC zLKw=AJ8Mp}g}2x=3DF^OY>o-rLzl#wghdQ=Mw|fH*m@ln^eH|t>n|jVvIlfrUszA0 zV_{9M>Dg6K4oZ{>PXn}Rw>05XUBG7KVt=zP58Lf5TcLS2Lstspujr-0`30aWD}08& zCu*~I-GwI>;;rj8tq$5swK?85cxW6ify)oC9%hvRW$CNNT<8LHzWROjse6+Z?Z1!W z6$^s8RO^KQc|fP0%F}1>nx4JN1^jt#d|5z^`&$WF31C>HCB%4*#R|)!Dg~q|*)-@j zdSF}zF4Gz%xn`Kt==VJeFHS^s>-g3lC(xdq7L<=LoZ89ct|HbHE92VP%4#?y6aIS6 z`mJRMiXUo{$#L8>Z4jDY=kI@S zh48C$JJCX0J+FeY#53b`$Gb5!9n(c)*@U^T^7Mx zrI}gk(Dy5bG2zo%s9Umb*SdhCMso#8vLevM%1Xk3z54k(vs&4?^~I{&_AV7Q?9pp?82kW6`fM zz8UUTws00aUweD@+}zx_loZdIpHn!Y?0StmI|wf>NBPvoTLH6>RX|5`b$1j{;j*nB z)Ov<-(&#mP0-%>6g2~nuMfTRkcH0yUG0*rltEt}<>fKk_YZcEmZ1KDpV_&2QzwcNl zD33_wd979xq8*Q!p&{XfiHFw*NJE%cC|7&_gTGb)riIl9c)jtV69+q(u>l(9YUyk0 zY0xc+J#)CS=r?kieq@>h0Ed#GiUeVpALET%JBE_HL@BW3=Aq?nd0g&nL2NzJ&jnB{A+!G5HV!#_s#ZRxJ3u zjqX{~v+V5;Y>P`4C>B6p){zMz7Q@r*JW)eG98kr{_$->3n7GwmbF^N)3)6W3?$5-s z*R=Jc8{PP_y$AcbuM6`isIe+LVkk5u-LFVFTupsUA>49$=-zi7@^D{oFIDvFa5RN$ z-M1<0qCtim(4-~WHx!Agh2Eu7X-L#_nvZ~9 zRAP*rk}|eVy}cQ(uPRw&*1sKvKuYwKQwRUrJT)x9%xainC03TJ(Xm?+SITxu2^ozM zw92~7E#>rM!|I5SDj-$tmUU;y(!wP0^&Li-D6xwK7K4i@LXTSC5p={D$c4yvH-J_N5&hFrh^Xa zz{d~wkOzw+k@oeH^4bTFkz$y&PTd9|9HlnlB+_H#UknT@KKh31 zqB!R>JZ4??I3+@Fr=ymSNo+$&tn<#wM{Xn6%e8&4e=F6pSv@bi6JCEhoGxII1^Lm( zeWY*GvQO{~Q)u;;gX$eaAK`A2*PGITwn_xr{xfHqGJ$?st#-iGF7nact`!dDM)mwn z8xRn1c3*cOvBjWpv-C_B~4|MurV_)vBiwV{rmW|%)xKS6YWpx8!Q#)Yzd5) z2rsgc69NfAr^vQ~gR@?PT9C|`&WV=dpTl8o2FO03(E`t`!9F96`)oZ>U&O1GIc7=+ z?zPrTMmPGXYG3QBP@N$4Cz!cknPlF?;rxYAIOM2$P?^B;OAF+3yS9}^9LLwC3=_k6l`f=$j z#BD|OK+3t#i{_X&sRrJFgigQLi=PZV z_@jA8t5QFP7BNdh%%b5!IwH4s$gE}@FmRF<`CP@}?x{O99jcqLT>YlSvOd(0$=P7^ ziHYd&shCGv(MCL_x2J!i+=E!!q2abQyMiw=F&QYXpXWLa`m;CvR|Gzh7qVMleYUd! zsqWhoM<#uFm&WVL&&@C5M7@j!nSq^4eLnURv31~Sdcbd%o+i1sd&>nES~4=2`Lw?4 z^ttfk)h_8cfXC2yGDLRAChbee7I3v1Hhh>%Hu^x%wo4Om>*Ra4Z_o$0@Gea;E^P^X zJQDM{EY*7QP0d>?MYGCSs{Titg4%zlgoEf96ZV=o5&GU}facP%!hFzEthZxu+z7TBXQj7(1Cr&Bl!p7@je z8RM_QN!-1)qew!3Sgmhgav+V~eYfDjxugyJ$SnFXqX)N--I}?>7lr1|5H-^#WNe z`kxtW4Pg+%V6==DbIF~PJ?e%QqS)9E?s$cF$Ft}}|Z z4}2X&eUp$^;dU#+GUd z;<=8a3T8V?%sdwZe5B+Lo2GX7N`uIhImWLC;+YII9YGKG@U-;88cia0m7|9O4nK@p zJrBKjuvVnJA>~1%y@g`n=c#r42`X2AfXj#&mxK_O>5@gom3#G4c6uU*mm!WNY zL3~o(iizxMV)JBxk@F6`+t`+w>$eY=ODl@JU*T%ki(N=@bEAIut=AGRs0Ni=XX^ee z)d9@{5kvdwMGlDait>=X`GRnXj)87QZ?x`0Hj^jN{AQ8wvB@tpUeRh9!eROiIf4dujHu{JxU<78(6cp*`D1b-*!6}Wx!268hn!a zJko77yor@H4Jxf9#IQ*-)3-h8lL7Ui?m+KgV<0AsOJWdK&? zO_uA3#iyXOw&fyr)Fi*&j$U1fc(7yQ$+Y0UIYi%J>YplD6&pR=rd7o`sltLd?>WnX#IxBcI&>| z3KPaBGUwX<_GQlv(l6H1^nGbbO>oBMrbXhpb20^NyCBB)%mq*5>)p;;+VYMSMBtOjx(StbS7^AwPVp?}*4k z{UuaH;Yw{U@|$pn7gpkQUMuwV|Y8CT$ck#%&+u+*ZUk7(akj>dEv72`%TS93AMTB_gjV) zHPBK~9Z<-1Da8NYF4ieId05mZF?5vG{aF&*YNAG@$Fwl2PVSK70DVJDV+n`es#Q>}0$%MslaHN?$>j!TRR4S7i5 z=;-emQtmJCwe+}ql&<3Ce!OiaX*ZSz6NM=!krlNI{6=+?vf|-(U5UxUZR_r-d`ILt zaceg(A;xX@pth4FlMTbS>q|wlN0`|09p(`pXYRjt?k~$0)5qKhM^Wj#DRls+t@;a6~N%1T^6kl|CXF zn~{O7AKUF>L7aE5lAeUjcIegMxSyiVnHBvo$H(>{l;6D>G2(!|G=FN@-4abpy(EWm z{t_{RaplDgYx%i~273(j(Sn}KY>?P@Pc^F(V7Eko_}?t-FlVu9mG0th5{iV@XNLU~ zNUqos_T&=OskjJW&NKBoFFk_Lx{sR=S2QwM3H`IF?KvaI4J&K&f`Nf&cZc#A} zcLsw}U7dffWrWgBQRs=eg)^5i>HcMfhY|lM^Vq8|O@_~end+v;x0 z`!z9F?1Y5x(E~y6_#r{@P*gVxpOS0M{NU}t`TG9`mAeWzzS(f$Vu2)eV>(@<;}vvDI&nB|Y}_m$ajDSM811nhj~n$v%O4X(cPhaK48QzRn2p ztv7}7?*bk}+qw7SPKN?)alb$ERC|Dth&e#KUd`}W<}$tqd||o)+XOqR6*5*$!<+nA zO1|=0gKX#%y4SBKbVmUtB;bae%1lN=vdR%~jRm-;egW#YLE-zOhkFNMfDdPPNw$N# zUGAdx@0lNq6Svdjir&4f4uI2NI*&>iiVNTL>f23hBHiZ7*?_))VKseem)z?9WoTjC z)POJ;&u<}GU|sjX$YETs&CWt69p}J1rt^*@`<9CHfWjfe=V5ob6}rT{rr*aJa+hXV z2g7wt7{6Sg$Q<>=gpD?F8o$OTP>#$lxvQ*xYG+Y`z!yd$M1$aG(TTZ_mmGkcW%&$L zw>n)0oQa(+WI^@z*=kH2oX|zdep%B?pyAs)p7*{AFo zIT~W&R{G=>HF90U(qas;2G>P|%zI`urKO84hi_ zI?H;%c7i0g!eWI_(Sd**WPX{e$nsyt-ilmeJu)WmFE+?BNzWk4|8ksvFzk34B9{8| z>`y$j!vtmoo@zLfua@_|&6w3r5gT~-z#l1c0Crz15NOX7Qvp2?y%a^j*20v#5su@h zL^la(EMtb*32DE#-&|oOqwkjTKayUV;u^_%io~7V!0AN8X-J-NDiN`w@f*o)7p|=8 z|3&JSzVc=`SjZj`UJnbBRl)w};Z3C*dx6-xDr{_QOji%UH1vFYWG`O)`L|W@E){V z@H#m4jV!}A|nEA-K^3KBAnyz1*qMEGde z6Q^B_@~xl!8aA8wSWS^9H{L7NT;|1@K|S$kt+eX6iove!MweN^Qt@o8Kg^9b#K{C| z4n4j~F5)#mq4E!YvpNtZH;P<-%`x+b!_@DpABs5kmVWZ~#e}~hw@!Q2ccs_M?-da~ z`FDf{nP^sLDPcXNN*t%phgn~?y9mE=tZrS(@Fdv>hjFga9ND4F>M3Y#{{#YbGU1LZ z0*Bkt1fD-2f|zUa;J7@PH;EeF8p((R8a)vHNm)QVfap@-TJs~Y&ER?7W^25?nU<~> z9&Rh=1AqY&6|^0Pi2i@i;#8G|kJU9n!l(T46lj9h1}@%HsSZKxw?TK;B2+WOBgJ=< z!@_p!!PLCG2QK{`9W}}fi%Ti|e*ncz9R}>Xs%t_Km)=5KAa|4+lhT3y#WTZW!kZVh z>F2Ze7#?#%RQXO&IP=2fC8EHO0|`hN+i7{;{IVJuYBZ`b16EKj@f3~hTQetAF%JKz zn%K`m=GdWdL+QnJL0k7m6Cf+}W)D&zIwt()2bi$}iXmK=c#7SR@r+NCPoHv|8?JSS zGY)HmW8Et7(sVT%%U3&p3QtEr zHPQPTZJFmNhLPW@&;e&eIZ|+-1W7>H%;$P|yG#Uhp$ zF)tl|Y{$p+2r%suP5!(W|K2d!IVbGmz~8Ts^2A;W*DTf762`BUQ2@kk(T~}xD9tr- zMC*dPXh0Qh`#6b*VGo$H;7b^zP~wKjuv~DyI;~;wW)ok9^aKFp7IYy(N6kcSx0@VQ z=ED^E`S~k3VKgdeH#6V%NdL)iR*;aN%eUqVrim=ZR=0`-n;Bj6ooW-?Ree&JfK8xter)zAb7^u@Clwa8%(Ie+b*2GE&i^I;hqn zX6U#(VROY^r7pBF*-y4U!aje~I1o?YN zov*(wuJ84O(^H-(voDHUTyJtOk`M0q=zKwkVb>`Ut@;~__++S_D+D|thd+goK_`mz z7FwyrHslk<6kJ43P-)(AH(s;ok@qFJ-=Uofj55!8{U{y~!d7%@9O$iN`0_gp&JUXJ zTiS%2-Qh2sYK__bgYF5+xAUcYTU)=vr4@|=l%BaD5cwltWE^NyJ+fwiLhox#OfT?{ zyq;SZY}W+xEISjW)kEG@`k&>E@gf$Q$k>VE>?%;02!=ODj%!*T(_OxXrA}&n5(m4_ z4A;M6E;1ahta56rpV7nKEzdMsO`W%odO9V&>OVngMFy0H$MkZaQqEBf*S38PKxCVU z^S0qky{^pQAYF<>scyu0G7zVgD5p6EK7RkF7Lp>R@WN#(|-2%%$eWzCrC<37!tLKFZ#nE< zxf>Xbx64kubJTeo&R&{+)O(e~Agx1|e{R(^V0W;7uP9wEhWI&`dEqGE^FcTizEK#v zk2N}@6ubR9QSVb~eSpuNJ?hN-39l2s<6#E7oKEL8)1X}B3h`$4r74B*HQI_ASw@9} z0k+32v8It6*FW)W=lNrL-P3;%BLvOHh!PWpHH@w~?1ShwSIaXd?QX?ay-YQ=p zMS8U&v@4Mvru<)uNQv$#Oj7)l%{29vi#VC~9rfha-CeF+sxMGQM9iiv2X>^JnAHZa z3+M+odelcG<8YegWBnp+^n&oZv`$R6@9MR8Bx~;9i|%)H1)bC*sUE@XG_Kct>0-D2 zJ!63dDWF!P@WZMmgPv+|$@`OLY~1Y!E+Bf2WSG|6JrNEKTQ$}Sjl4MY|2Xt=k*RBw zY{EvA!#fu^Nmsa!&&Kz{%gL;J;z!oN^=7hD7Y^SF&>^M-`drjsF|_aSEpBRUW`PJn zbfLK0^1m(1MrQv3*Xktd?q8OeL8+)DUWQlbQT-<%mFx!NB`CgIuKe zKuMgtv-7aoH%4Y=3tCeZQh+MF=GS)z6Gha>$PsYzn}Wiiv0;ivv)UH-I#%Y2m1!RkvjgF(LMqh zPq%+Fyx7@%aac`*YW(Gm-(tCMUF!H{gbF*a>FdcAB*RhEOeCcJ2}I}@pJw}0u0l#f zw)6#=#{S6!=z;qrDz037JwsSWz|JdRI~CKXe~KkG5vI}TV7v^nf0wQLpz(dDF~65j z?gyCy!|`5NGkd^$Y3UiU0+^_RA>c>id3(}F51Tl;13GSR&CFit1B8g%HL|wHgtiLM zHOc*W-P1QJSyc27fV31Jh!jfzHIs{1l%{_Mb;~rqYIsIK+YH zDqhWt@hSPl63YBJI9K!8J-$yvKqBs8@8YmW z;DRvZRm3@=IqAWsZoMi(?c zd<)t&`tbqfU_Em~h^dhK@58~vFlL3QzYzWCQ{~WdLHe4To6CL?gr!==>adrTKm6D` zV2ogC>2a$Gev;h(LQknRxqO4uZjCh{b7aJei=M+7Im}Q5u8mJ zt$ce=W28Z%q3STb{Xek$H9}vNyoisR9XqD1yJ|D-+H7`ob|CV3xYp~Ni<9T5~;=9vq1<%L#7AfAy>tyipSGq<`Z z&$37MhuglcEYZZ$^-(#u$`dw22;-3pMycJy!LxK2COfa>kOTfJ-xA+6lv_88Im zgwG2qg8fXL4s%6vsm-z~SpSr3zUkscm{(ym-sq2`!|WS2NpJ6AZSSR|LJ8czd$Tdm zDT-VWi(aiD317~cq;LiId=y-e%VziB5Voj8bi18tF_r?PyZFR56aA$tTWwYAYio4# zBxtymwS(8Tm>-kLj11J&Ol26<+Uu`X_P+cZXI~*mFN&zU#*bz@LI&XLwn$_l$7ADP zqf>%b=xDL<>b*si+k)h|bfhe8v37BZIz9nZ-sst z8XzIJTHA*tM8y%txt{mKR+?4otH~gAXt6RIDzooqx*N^o4-N80EW7p4Z<|kSNVYrx zWHx@j+d_~IK%XU-@Ckdr`r+@aH*No`W7{|Ji;ec{*CvqU;vqvxj@P$r1U;5;b)Les&Lqdx z{bkXeYj^nvWit`(KDqs_m+aS2`zP)0Ur(gbGfbV5c<_BQN12LwId*$K0Q9Lg#75@`I zlOzTPQ=Zu}p}XxUMY?j6xx21IT>aa?&bxd*^?M(fmL}(bG3jUY`J}M?L8dlr+Z@MH zlNQf&CBseHz;F*Q>Mb3A4~y14y2zk2oQJ}1%i6oE3bXM0_S*W$cwtV*{69TDtugrn z)fPTWKcg$La9`Ukn`42Lk16k;ewbR?#69>EzF+nCf@-9TC;P4gbKfX|mGGIn6XqnI zTQ@W;*}F8YS^O0G4~hbkFKUF{S$%DPWRxBuCFKm1Nc&TQ920W8Ea@P@_FMO2GW#Xy z2D?coZz$XcOm&MF>LVSKKZpcrgG37uvDWq3=z(U(((NB0fp!l)xNqe|V$G1@h1)?+ z^}maRF3Nsgs9X=F_<B;dbV!&L^S++y~vD8A;V_1yK#jK{XwGXK7cepP{jCE?m+l%he$lZ`a zec~!yq*i#w;Cu`6^1%MKyFkP!hGvK|sKwAq-Ky;uJyT{L*6q@NhwY0On1 z)LZA6h>k(=Rn#wufya{*+>W9OGHCvAU~_g5N zq;@{>J7P-|>82T2y-1jN7{--qk%AGBC1!Ev{ZdRM3o-Wn#4)iyxwF{-%mBOvcz1n( zLUGpR18jwx;NdfwzGp`Y{-D*(_#W@8SA)9X>$B~&58=87PK6FsIY%9u=TrX#r)(T# z+q$=G?22!dcWD}nv$j0G&`!3$r+?+kfQWRXl8$<*y=t!tYO7B4Tb##*c0WavAPjWy z>xj0>W0h0}!Rt?1*dNfPP8=*RFY&SqLu%Ks<}sl|;uF4wS=A<~@Mb#uF0;Kjd25># zhPIe3uOm*Wdweo78RnV3C-nitX$-W(sQiGJng`_Aopk*|6fc&$Y{gTtP+i>0<9eFnK5$@yJ+QZ+XX_uA~Ybov<@tfYB(hWgl= z?stkY76fh?&Sc^=rJ&HlQf=D{oRo{hM0F-*4^eqhc58%ph;Z&)styQ_>#O~OygG$HySBGJ@fskQ z1@g@|otKM0UeVGH%hi4IuQe-0aDNDDS^N>f{&XYH9u-56Zy0_$WaNgP#>`X=#mF}VYi)>g7vHsbQ1%E+55D3-((O(JgL^3l{a z<&Dh@Q@77uPF`M~)9{l&4Nxu&e!D+(2|D_xda^`Ci*7z!6*Y595j(NCKc>NYLD_si8ffS}*} z6_D>Mn(_Et)e=MF$(>>kUsd8xFwel8xmL#%==l25aK+38$wg%X#TsLQ{G~xQH7%J*kH2yZjNzIZNHzw{+T*m}po!hi z-SO#)#0NZ?tk-(_75>W#fIn3;Y(GNAh zfxg9=7!!(jq5&@)LJmj{wQ2-QO0%_=(N`Lq`0J}m?P&Wgg=7);!c@@H4)%_dIf7%n?GzeS;EM*@|YVP<;Y(}_VEYliwNX#MHqXYh| z#>j56LkOR$OUlgY_Jcy{{R(`)7R+e&=fFW zd1rL7H_bS#zqi`6qY!pP6t|!H?OTwBz86Z7VixZULSTlk5HOWmZWIQz^}fp9#nDSR zj*8T@Pyb6268RE%IF*z83TyKKtgrtgJXNr8050`R&=>RFXEb{LY});_CULDB@>KxA z<~^*h+6}Dl72#SA1t&LfnUsbS5JTC19Bxyi#FXUg+AN`+e(SrNQ10{Sr`$|!Tlp@ z`hv5jQTVZJBj%@W{F!<@3c^J(-`PGvF$OBD?57%A>K%d}1dX?xN zXdPrh2;4fpHet=4g7J~)12J5AvWloOmLJH5&hLI6u!()Mug;0aPy5T*8 zGA`(+4vQWlR5FBh8P_aDP+9MRg-n0{8*1?wAM0dtTwr*K*!6&7lq00Zib%3jHviM& zPQl?~iTW?Ow1z|$4c;_kNN`2ZW`_3a$81h8`1nEw!?Z8w4Eq_=-PJGp${?^@-j6{y zKR~!E=hY2p6K(ZzfVmcr{y9NNKZN>Dz>xbH$z5P-bcdx}u&&@)$$j2GU_z@xyOVIK zm6*h8ru?cumHM;=NSw0$(jlBtUX;i4<k6w&Lvf}^hHF|$|da^_`x^-1cQz!i$8u*3|^u2)Vf9K?bpG?nMnDWcUSa< zgGWYizNC`K>qYUII*!KiG)6JU{79@b+EE$p){hB#QM=FemW|GrFn7BaF6Vg)D6Np2 z(Y^lVd^uV@LSjS2FW%@c4A_C|aCUchUotZXwY7->%09W9T1hK6UBsa3_kjr>L3_@q zBB>8fW91@eeH5Jy%`_jI=17r!f#ED0pEUXmXtv*ri?^ekuqzI@#Q+n{cYH+XS~MOr z7FQy~TDJn}QVliPl3S<30kSR#b@+$yx-jhbmiRb*x{aJH$!Gkb@MRv#{mVIgf~@|p zHIj>DinVm*r^C^EgqW~HV(7s-`|hb(QP91soxv6LrE`%|49&!%P{r*TNkM7VDE=xm zh39%bEO5b^VgqN*mp^%2m3i^06AbV6%9l#&I!>1-(Pg-8JX-6;;ycqV~Tn@b>btKo*S~ zPiRLCwH-cC62#)+15`xOz?gYjG5Gu_!0hFI$7DD=9)s$s7f<4GzrS1*8=`=jakGA)g>6X-JX;j7t!2D8 z-2IKJ7M%Mo(o1`PjAkNt1xyz57-&Av{uk4EjSyyTQtKKESY;Sufq3dZ3(2T9 zTG#nSO6zV=Nrpt59J!O2UJ(DXrZUY!xhw*EI#9wvlg0K*pRo);BhmX?nyhS=+8{9T z?M3I^m{>D#p{K%hQRMCskCJC-b4-q8pKgwdJ%gU;kNYQ}dc8>`-m59qx?vs(9q{d9 z?5CngN7EqBPg*MH2V$fx()LtBCkZi*N0yaipW++46+!o{Mvuo_GZ1h$dbM)l;y90- z-zox>53b~oHc|k8Bl_(2(Q&dF5VU#%Jv@4=b6}_hIoG>fJ{*5y-OF(|w11ZB$c!xP zP7>;a@#>di4u};c(KX(qDHxx4EQ`TanBiStO8;KLK@(GJQEo)-sx4amS!J|Gy|*J{ z)>iYK-qmRlxl@rN9;W3@Q(4lQru9NKF9MqW@ueeP%%wEAXjw77ae!L!R|(?F5sxU0 zr4GCHw=-ZerEb}X+bga1s22xIqRI2nnVxMrhNYT7lTmlYW&atF#khIjDEP*1<&9i? zR<|gWr|G|doFOfa;#-shZ9x0Za}6~4&jNPvt-qXW7(OZkXyz|LoA}CA(q)f)&BidL99hz$E!33d;G(Vf@a< ze4=XL_J+Wm^H{J)-CTds>SF(MGHCNs=c$%GyEBhSj&*8BBDV@h*~;bp(|*|kCf3Pu z3qO*ScdIl~s-PGuV_}UEfemg6x6zxkE)uK8EL)OUCfpl`0fyan2e% zhAcR05;uZ3zQAgPxIzX!a&Dk7K@tv4L|HTd<{rYS|t}%^${-bP+kt7Tsj)u zNH~=EDyiKmm*4tP+wA^oSW(hlaf^f0e!)VYf=p!EKiOY!Xigq?FWiGszU;f#tO+g* zRnQ}u+>lLS2*A&DY1(}iKFJ+`-MH!oU%^i9u{OwFbmWiFu^W0~J^sj6Q0p9T|wLTO?UKGza zHj9d5=}!$| z!=JmBbAL(5c#dYKlG{>WYm3fKtTI0L4nO0tfI5AuR(8LBe&`2OOd)1jye61w-^i}1 z=(`_SF&k(@%y}^{_VThp7Q5`t{e(33;8fVY#Jyzaw+JM(@N{#K{nxHaVmH1YSf@!U)8?0Iy%oSF&iVB!v zmw#APpuoBwDEe?k8qZoAr26nV$R>U9-RiAQ<6ztVVJfdViRvcpWnGBP#Y%D|0tc5Y zEx!djZ$9mL#HAml<#&z#sat+_epbY-itM~QVs{gVbWQqkpp~P39aVEAxzy02_F$nk zM1$JqPkeA^-W5UR{%xu)gn3Wr-^@A~-L*I{`Ft;xVpn##mHy}q25ajdZ}#&UD+6%3 zzo^4GBQP0@da|h{pY=zfYgFCrR#lzIpyWaxLm}tFavL9l)ntQA_fl8DC9T`>>3f<3 zQ=!^8x|P%)r1o0YKu$B+smB0)6ZIGNJ+WEgffu6ndBW>QZ)N(X>SeXtSb64jE&Gl0 z1GcSc;LlRWt42fvT{1Z-Q{IoW-z^{oQ%Wy{zgzy(-^fLvG0tO?1T%ftA(3zHVP zo_yjW12tKwQhw3(*+1{(jvylSs}*2?mm^>w$nf$jg!Pd8v7LE3z~nT_b|c8phyCw`P)*(t2qB%Ma4g5^5-)0dU@_neHY)_ zB0IvT?H6m>-?hz2BVGv;=Ss~BIHcY}A=RxJwNS3EK${0K2R|P=m{UQRr9(IlI?%Tou zvg$gKBnDSfhAyVx1zN%cJZwAONNGtURlf&2;6k>yCS_=SiU>&X&YI6UBgHgAUTp%U4Rv&=j+D@^#AMt+`-sO@B`lP z&YYO&s~F}}c_xhZLX3>6TW(G=vBZu&~7FsFQr?DSi3sX`EV`C7^ zo@M?VV_aB9kok=k$M(|cxEGbu!T$8pk(O^N^k~_}XaqU3^+`G7Mg0`Wq7*1XTxaUx zTtAHdSOosSL%4f-=dLU~bSM)Op@P;ldzKPaM{A7T-Bld$(9tdrt1iE{rps*EHtWb{ zUbWXC;ga&_*c-78R!z0TJT1sva_EnC?28E<;yAy37wj)xzVT+#@2?g~FkUKum7;f>;Q~T2*}C5 z=I1}ZJ&6sFw6(SEQ~VJ7H_-1GA-^h7%qcddl^Il(qlGD|z z)-`%DFF|ge@y)U+I(Fx!!C>RXk9A$o0|mZbyjr=Irj+S7GzW!WHj?Lr?(hx}MsF@T zrFXZ*26(Q0=7#*}oiSZG;M80=?CReA8qAz#D_gyk<>hz?Uc z&#=&MV-kDnCZN;ocIcMq7>XMRQHr<(ePq;IHv2?D(KP=?)aDLo%Rb_MjYvVbB*%#D zhcg^e(BrqDbOA=d=%CVwE#DbyQsN)&_Q?94_1rC@_SlX~4w^P2 zypHtaQSE330#tA0HWhO;AtS_I4IC_6?+|Quia1v7s`%7J1J6uZ^9NqG74ZkNpYr4l ziL~DzUdw%pH3!%4ZkdPTWYbeaKEwhl)(Nlw3a&lsWf@8d$Tpzh zH(587H^MDJV@V-7(|=gOv^@|382`ep?4h&VwYt5H_W!85%CM-`wN0neNJt|kEgb_$ ziBc*G(j_HbL#K3ybcdjV!r7M_NTs zPwzI%{ZCo`8fB1xwNw-k@I_^1mqdwH>z#uN|9ZihPrt}2Ki_es--rcco-pHs@9C`n^)MO`Y;x7MQzCmt8J|0E(eM6?ShxCb0- zPO(2RT*{90?hQr)+5$g(ZShpphE@g&Z>fbWL?vvm7pWiM&aTVfAE}sbHdv zj)@^aXbJ;lKXin7rqxUtG9be+Ssd1teUA7EcP8YtoTZ8|{?xIr6}R))(905e11Vay zd74N3U&5COBk0t2bj$C@J!m$iDVv+BWw{*UtnX73o_-(PiS2#@K z=~CyE?NaG%S9iR+FIeIh?-W{zu0yG|9{w9AB8XHn#nfrdF!0>)42C5?U#jp)YOQ1M zpKAASA_xu+Uc;2UY?C8&t zMe43bVjx%aGC|2{95bE(oB<(e{IeNL!YR_}DNu+xN%#RYgfCydW*0yE91COM`h-CT z;EdJQU%YMY1_lqNW9wo*ri)u46e2|SFoSVqOt((FAeQ>$1ST6Ls!int{7(#;zsa@B z5AMZcno@c{p-YWX89=A~kd^eKbM5eBs^a=7@y76v$+a7EZwIgs2Riw02wT&+9sGP( zK_ywZu@UERAzn#M441Iz@t8v-arAvnjn0?PNcFUuH9s@+><-mKTUn>-`L_PucARS= za=RP^blSs6)|N15Zb8gqF$u_zu*gSIVegPFyM{Q{;VrYgK2oDI0`c>o=e|RD#L3FK zl92*1klBOn_&4s$k}?@Nh?-{Q6k8$8)1vc@{KB|0T&VlI@h4CJoms9RC*-@)i0WY7 zJ3sehnsu>{(}tKuJIpOxA)g&mTa~lpZ*wP1>7pTXP=1d#0AmUxA0WBR$X5_P9WB#*eK zw2z>@v`-lJO?oUDCf*mjEDvYs7+cW7;LG<*`8|LE-9qg!H$eCU0uB9B_SvDJc@lt~ z(aVxGL3w%T@!fe7xCD@R-~1lMPe7my`(eeUN5+TB_T^$GPjgvfMs^Gfb^r5u*?G07d7TFs1yL0|F8%-^ww}lL`*vh^9mz`1V_(_DVe0jxNM!!d z27L0ykajO3%OPUQmOPM;SdaL??U9MHn5_tf@65`ONSDV`s}>^a{y18d>pCUlJY17)X8O?v#aKKOp zz+FEZ^Nz>`;l39rREEB>Gw_iORF$=p^5N2g<*;Q;an*ohzZQ7)BR`&hYefaxNzOdL zWzkNS=xR5|fz9-U2R3wk61R_)yvB>E3dK7jnesA!$aiF4&_8Va&hu&B%DH<3pNrLM&AS`&!DLo#q`MT{s_0JZ9sKJJmavm5~1w9QjNkg); zj490AY{0~`^kZqgqL`*v(%(A#lE;sW@K|C%rf^CHC2nk{#m8sCTTFXUfRdQ`I4S-Q zk5P)7?)4U%oSTWDoJT=u^|X5*!3tSu$VfRB-Yb7Bz_aC>p4958Dlcb$n|c_OI@s!C zmh*^ESK!_Zh7D!jp>gt4ZqUMGyq#(9eVI~M$*K&teNrigJ;dbVx5RYX(`7q&wM4Zw zCQFpva>ORBDu&9~aS1Dah=E{3^5{c~j@xJhj8+0$_SypIp6ibtDj$^0Tw{5N*oFv) zQ_b((53ukwGV^Xo#R%nDy?V_)R%te5DU8b@BarRQDr zE3%ZYU888q&m$l6Qzj43UA5)uDo_nY_?DSA-|vggn6Uh2%Yi9cFb8@(O0UQH(54hy z7~V+A`)dK%nI2?h+Wb?`yg~wi9k;i)53i>_E$=-X6Fn1UVqsZBIIJK5T56@7jtQbR z7CulwU!0QcVrnx!EiF6?(Az;cBS+Ygtw}_88;|vVJ-ol(EA768Ek|377gSZVd+GkI zJ3S+$-Vcfx%WAywifadjR~>8=n?*mDo%s-e@$i{YO+(768Le1SNoEwpPcg@C*?C5A zwaLHBi7}4cK5MUQ-+qc^cb@%}0(-fO;NpWCs_pE$nAMYzDUbDt^DN zO(VJ5W%yuqVJgSdN5e~!I8hy3{`%jz&+6At_oryusIi_0$J(b%$P|A4niBFWkyBgg zC!}^vi0Tv;LddA}3EZ(Ic5M#^R}(HF$|C!+`Nik+30cY|Zg4TGk;IWu`wnuXiJ z!l1@L|K7b1r7LMrcy;s9doVhEAVgw(9rf|oeIfCv+OEgH^-|?z>htWLI@b9UV6pQf z1C)*qPR^Bf%yFr^wKI!L@Bc7=vzrhFmg3`bK!k?u`E#bP7RAroETjf@$IAnE2bj^bAs;EP2gtl1)}yasUQkW`sQ&~lh>Mq^nTMDZ7wo~crFJan+_C0c z7Ztl<+T?CoV|IIqv>IQYx%(u-*V_G(N2!-O+<|Wu?KK2u5_uDsRyUjyv0BOE+0psA z%ftCcxM6_D6oU zxjE}xyu?wzseNZ>XIEKiW8=aXPy|IsddHc}uf{S`K1t2mZ!pPkKDtML?v$=_tKqFf zMq#>iQ;+N{@Ssw^HIR?J4D;kyTkoK-#y>8+t&I{S?g3P)?@12lffCNJB~n^`B6uYE z@NBxye$oloD_1hgVfu^Ow=5&s^fxCR`^#M4PYoROP#9PFTKak}3wO?G=Bs6rNGk;; zmf7dV$*W3_Cb?pv!im}3hWTDMRt6+$>l8w^;6fE^3aY8{`3*DHIeaw;pIoK|vXO+D zT(mF1o;;eBn2Z*i`d4EkJ03AcBrBiNYI0@CFe?fsBz#)6SFt_vMW&HM8la1&4EXkk zrk85h-~s+-ZC}{b=wk0SgRqtnc+7HcW>=&9ts)umhxzM<%gtV9 zWK2E)UalrMwJ)K+Jfw6wT5Nki_VCdoRcF4j+mpM#onyT|2sA=KYFCnN1-}9h6tsFR zZ?KxdA7xzcY57?{iC|Q(uX&%yLh_fETPS4fqif{$s^-I*8$pA14g+Dv1I{pTeW#cJ z$zicIEX15!$wb#fLT6ynC6KAjb_C}Gk#(U+#(yLjbf0GfyjqR)k*_prW6-aYX&j>b zC9_Fhuh4WJ9LRo9%)ffu1qe8N{Ith;=N|jzCMb5d3m#A8KUt*lcwEd)`By1b!|a=` z5qUOS!SDy#A*}rHJ7)crlbd8ba^hkC*3EI0&itd6ySy<0>EePSALs^jh5 z%zJva!Qqb^KZRHXotvXn>!R9PdmsZ!7p`LMJsxZ~3-nldgt{MRT;myz9K3X{C-eMt za^j1=t-?sOTMkrfPL>4IFr4v>7z?G{dNWc03q+jv>PHdl84irE*g|!w58ZdMIDYuF+H+5gpms+h3@YMOK+6XI zBrz6M|7xaDq}Z}1oOSL@%|IjbbGWHfNtrAHDGESAIIQ7Qr*fkje?xAz8rC;o$!Ja4 zss-66LK;56c1AX!y5@-#9=!z&;IBu7zV$(%on`Pr6RYTn#v!5lO|e<$QMPZ8Y;Oo$ z*W|)ngXk$;@gS}-Q+n%HxspEUpi`e&#bET1!3VsT@$-NKI=9wGVdvd1nq`;0s^0#V zC3OWUE_>sMIlo^N6P@z`!-6HqgtbWF#mg&0(Y~yld`Eg=R&4v&GXfgXxsc1KjF-c6 zV$&&ao(o;3>lE}o>j-I`^@JTkwPBV64@}o)FkAiZbBb;R zOW797Hd^z!%kXaJ60cmz7fW+5lj^}1SJghmb|JYS*b{J1HuX z8&!GiktW1Ahx_(-ucSfl;kyH?3qd6B&!sS^RUz;H-@j&ADfFAGBEX6};`rE(cqb8Z z)FmV9WoVc~I0_g#Me5s}*z$JAKXQ63WWz7-%CIsoma6@(wk&JJ?h!6C3gHduIFIB> zJ4Vv6lr)vswMX9Z&qm!xLX>!HFI@q7O?(B@g3O08faE^XhDS;=NLO!p6N=|^A(KeMZHm^*VWyl6BzHBPqVf%W*63)HfZM| zYrbKVd z2WzImFBYo1K&y7dnEXldBfnVZf-<64fsD>=kv)(omnor}sRB?&I2SsuIC$;x4YYWB z^H-fi>b5WK@N>+g)0B+so3mVNB>s(S2=6stNgt+P0(zeo011+Xe6dFIJ!+8qgzR$X z%jTalZ+j3!FXD{5(P-}gVA`9zq2R9Vvf$bf_`tB|XZQ=RL@9asAYPsupuD9kxn(#8 z783*%6yK$r!%Jgn^svu$PzJIc#DS&C+hVgfkiujVw<~R(^*nQVGEzamH!hFl_RPtU zzpnt6&?cmK)y(s*#-Ae@HSze!d^^=R5Heq}FSp(|SoQZb%Fd53V?sek2a7c17!Qn} zO}#6q#@8P`|GgnhWRbM(&1SjtHPH(uT1u!(VY$G#)_6N>tGz0aiyQvs7V>0H#NtY+ zOtOLH2uThOe4kVjTdyJbz?SLqdzW(a4h7(5{Rf)~B=c~fsX3cA9?5>5ENstk2Ve$L zonZHu(f766uR^gFn>=bPXoUdeBW~&B^w(+JXtNi(s!c%%TkbO;3lhz>IT>l_o(X2cK7jJ~EB3 z6}ERivinbgjmt7#+L{|_;BxPw6rI++?X&f?Vh2+7d@I`Fo|yywnDSkU7at>G7MzA`h=-fA?e06%0;lKg`m zh$kc0YX*%B1!ZuP^QMFP)ht@aUFHHvzIOgA%K)?s&3 zEr(1?4~VDfY+>O0Dd<*aG_x3cgT56n~xGc=i zPX@cR?|3Z3X=H!(6TH#qKK3#h62%8<)_QKgxj7>e4XRu_Re6)imv=t9!7IMV3L&~L zMD%(lplfcFm%YGuCS-xi79`{?aD_P9WRjW< zaVPT3>o~s@4Fjj`oPb|4M|X>ac{CCps2PM3QfH0-8>F&X2oOqqC@Q%Vh&7Wp&?TlN zS~X+5*6Q&BWPE=CUce`nS+@_kTW+4^H>Ka)0!)O}wB?%u-g--6TMH0Np5aH`M^ABZ zE{lR?7)~54KOyGMAnDfO6uxc_haQ1M`f%d0hGy>&k(F4+7X;)ppRm5(asJ&{)P_i9 z6#8FGTd+O_8SKt$9t>gGQC)X&7`H%qv5xcdC;}M{dbr%P?nc#7@Ss%S<1S|WYAU-A zQLT`RST1qb%D@z3PE&!HeK!bGVdKsT{H`4s!;o8>DLUE5WjS!etGRQJ*zSv3vXh3x zU3hcmyL)aVaG>kfOB1d;Z$(Y7-H7Y!NN!HXHag*Y(^oJ7A&-XrFQJfmprYtkNfoM- zCR!0X$Dka215tEuRBGSZkYJ!dm8YOy4_Cy03&}FTabdbyj137*H(@wBy?) zt6uaDS?y)Ia?Z|FaRL!;kP2x|`e)+K_2{M%3@aVu4%NZzDY3S+Ajyd7xkHORQo#e=Cv8U z^mmyT2bMjq$O<|184!P5Riv)PzC?SMvi@yXsJ&U0KfGOwU%Tn6>%(olXMNtM$!n^%+zYHme$$1z)Roh6h9JV~dd5qO!m=z>u%CGah z`~l>wl1}JW+|$$V0zs)oRL37sP9vw;WMtuh@S-(vZh>1sCFk(V+V1neaEaC$Y=r_f z%?*2C$-py~8Jb$p5C(l6fQJ8Ew#PgSie@T}t_D$S)a@CB9yp$M&b}{)dLGa`KH^)= zjQbtK#AD_kXlYi*^C`PC(2VTnDm^&CkYVntEqn*WpTLfZF@(AadX7AFeU268dVeN} znh>#4XtM#;{(#wn-zBG)_#Pt)t)u~$Tr58J`UwtL{uK2y8}za4cAC+8*+!PGYe24- zWTyvZ`cr7GBPJ!x73M2v_af4lJtoD|NagaR@3~sT*nfzg=L!f=^#-OEpjvJ3p5%TC zC_+&K#2q<+L#y26)IsU+U5xCk1^^vE6}H{q{!RZ<0uuFpHP{0DtXgv2|gdr#CM_H$QY(Z=*jH%5%S-;;9w`ysIO$HTl6S4*@9}m_^RIk0iy;TvwMt2 zE=e4`9i|zO6?AK^at8;qV}RQX#|sy^M7;GNj&00FJPR*1+NgO0ax+Xm>HM<)Zk`nW zuaqjlUr-@EHbT)*S2vw2X+ogo&IW~)51iK2KLL*Ae?*N@n#p{4E+&9hV><<{uc*RY z-sMY;)5Kc{AB>{y`7I$a@Av||j*)dihV@ztdlIT7uCZVIih^RfY`Co+L_^h=uO?R1M~-d0ayC*|1XVJxh#B&wRIT#CDXRV?7B@`i30QYBW!9-^ z%qy5H;qfG?C+Co*v`}i*{>zROJ=zUv@}j;cV@q>(pI$Y_LT^YE8iWAmBiB8F{2XS- zE1f>LAIZT2h0HBixQ)wyiVjv}ne1`2#*lLa_Q0PlU3Q_fs?l)0=`J;-5*|Tuvyt7g zh23*u*@!eg;V9w5Ao?6gcA^ndCX;Mmp7jCL%c`otvJBA>WL=#JPvD0}o=e32TyDSe zd0FVi5=rw9vRb=QlT~Qfa_!9Q z8XqPNl?YpotuoEoByF{R=gK$7=i50^CzA8nMzVGHL}Ic2G}n+?K%@U(k*Y)EFd+pJ z(g<*ZzbT6mBY!K|u$G?Oc--r(3>X?5dBem$nwlETR+zW~T`nR{ksfT#^!+TMQb>NF7cvMlAV_N9t6s@^y8MU<9`AiT#WNdoRz>iHb6XoIC$QY{!ZbsDZQ{ z>eJ>FO*UHcr^oz*mN6$oeAfr56hc$)_Q3@6og;nK)*R-==bIwcYd;5V`iL-pb)g6y zYv_o}a1BNPVhL&6*TBFy7$Apiw5wMbFcP{D9wWYV(1OJiJb}_TvV!6ga_*9=4!(>fjnb0RNdPik7WSa zGl&hKF8P;4Hchxqgv@O*=>gs6CDaa`n8I}qU3V9!2?r%Y-k zZX|FeCKSGO&8)XsURbn;R?@xcy#SwC=gOI_s4Y0LV&AsG6;AATRG z(KA{h@gTZ)3+L+yirQttPx5wLGEnN)xzvN2C^UebU5JnGvryBHvg@FwUgbK!HB^o?JBs8? zPKbRPBh0G>u->4C>?PkrZSR9ry-(v^_q3N^VIs5Rzb}Csbn~L?{XjuS(PkDq)SgE_ zfK;0~^5Kn`=$01ENt+)T9j(LC3v?|DW&PX|eLSFL3v1Yi9u@+HRiL5rSx!sz17sZt zb4b@mq&25{_w(@Ljyt-=6$|Suru1Pm(3G;-M%APezkXZ2y;YT`Ep1MdM+WV5Nq|$r zB$GL}aqC2|t7jw=pd|x#;2!P_m${m|3#xczBs7dCgd(`~q_jV`_*J9c{zm zD}L~8^q06vPUR1UXkeVkGEj3>G-HQwP{8{yFB&1q@DBsCzlx{2-YwsRIyr&i_ zhTTrbOY(v*gEw9lz4I!WA7U;X+lj+k|XdP$LJ z&TNN4QWfSpWrZKGCIP9jUAK4Ft#JkG^T92F|7C$Mn*q)jHNOhvMvmh!51+{*y+FVL zyVi0-w6nE;m&>;W5cxjNqrJt_zv|N{^MZpFz14?Q+xvy5dU~)B)|?B@)4HyJzuGb2 zAVHgvkzu1}bB2>q3+(0WfElDD=B|Dh{lQP*ckIGQUc1#;fm25$b{5tWo=c>qt+)tC zo7>9D9GHCj@!_C#aUoYVcDr*?X~0!j_}7>%CP=)o(ib)0@~LUmas^iiySa5sHg8D% z6;e=g-gz(ta%5r!zY7LEA#b7C&S$!%tfkzqig5a5DBkd_yYwq#Vk?M9m${5@Ha#iz zV^+4u&1+slYD)+og(02#3v*k!+5-o)-DFU6K~fj20wkEORc>wj=YasvrBx(XO7mVd z{nb~xes?xKtw-9LCW`%IPJk!zKqpkvew#q~;nWZgI@)ocRH=o&epbKi{iaY01w!>2 zk4ku@_|!7x!I6;9nGl)o?r<1xWjsIn&>QsAD>Qnq#HkB5)0iU{&2_$6Hg8ErKyD*C zC{+qDYy*|lSF-4z(zsEZE)`!91M256yfV*;cHFJM6zk-4WTZQ**2A~w_r~7 zfZD83n8x`)U(jq~9pu<#@A@zn3sp=!&S?r3k)I}L-BE=6*@{TTa zruo-~&8|=La)DAGuC&UuCN&S`JHtjnnTGqM5t^mmoIm%Ri#t^kAC;B1gd~Ec#ZVIx zT{2%uB{4yY7PO7N2+YAm6dU43dUI*$Ai>(z#B)0|`Yln1rvv_GgNRh{HzTO$qnMz& zcFNc;)Dkv;4;;8A->&^ADrq;9IHTAS@FgN||C1v<$ZLzO7c2@D4Yapc| z_nDVuSMV&>m0naDe*UHcKHnN$>UJhby~_mp$UXlttv*Vo_lQe$>26-&t$bJccwaM^ zuPpXrc3OYCp`UxcFVmR7OYryVH0~K{&+tAoWW17-@3U0?TwPbNdJ|=iyR_Ki+Rv1K z-(W%OYz`urIW*o~l)8lOtqbaFYwvw(V&`kpDJnJmc``csM*Lzzt@dg&&Ayk()2RHP z*I#CZEo-EZwsm84+@EgyBneuIb&t6jSwLb?xiq}bt8 zMKdDHhw1c*mL|g$4?M!k_x73C;pa4_F|0i)I5~z($`Nj=HNWg_LSVUzDtvB$w9E+cY0jcp+xe>3 z{{dR^bAY4;uXf*g0NJtQWo_s9ucMP@#;hEVo5n0pn_bQN?jvV)m+Q7OawVy0w}<|b zWIQEa4^imZVJ>~P%?Bg@9+R~aaXEF97FL%pN?br&M79kFf{kYxM^3#%pFVvW88fQ- zn8i4P{pg*g?$U!;tV%gP`jy$acY#tHTd$b5ho<(*&IHl>@{ktGun#J#x02a(%JmJlsZuS z$j6^z4Sip&;dQduz{LYdp=F~Kh{M2k!Ac#f<~?m4{GTwJqLE#U)N<~SaHBM3tc-4B zIdvS`HTRNj1n`h=$HTxNUsm88?s9cnyf^3H#c%;I%qqLr*2Zcpo^O1s)iKQjr87gf zn6dXgc4%6HtL8Kh!&e?*PmcCAhmJJ&GUci*f-MX6knePOsr*({!r_&{IJJ9b=nF7b z-SsVsN|o^Is1iD@xaSWfiCXiPk!e<;ym$K5sjpiK61dKlPt zU}PHI)ibqWI>{@J+e5S7TP0(S$v`{ULo5Ls!|y^`vp)yspLJFZ{->m<1TrgXqpz}a z8*G1eZ8w1Z@YdSa<1t#EJ7;`v?MJZxbtR!f{}peHgtfJGNzz#@m;<<@tidS1z|6Vq z1WR!{jLC}o;KP%rxAM6%a*&70RT?=#BYK2;Qal74-7hm-y6L3H-)CH{nBUMCT-6-i z4|FWzlZ~j7U9sNd&uLRg=|2rwUqg;qM%y_-5__}4C(<#zNEN8K^l3PZC zIs!J`Z}uixQW_VoLc(lek_B365g~#R50CcId}4rfQzn+H&HwuMY69_{6ul4cSXJ_t z-}`pjhZ-;~ZZ#6TR^)D)K=$g~spbtaj(QNva?7&1XYLug&zNTx6R(fQ2#E354kfofdck+$t zo67_AHv)uc!Ks9dj@`i@P9u%kn+7ObHj+?gu2r&|li*s1oN);tf#-YfgfnJPy9wVs zB&uZWcg~EByUb{5VdeX#CX(*)Rzv@uLU@>jmA#c#*bYf((TNyn#E<)1Zu*nc(|>6+ zpOTunB^?&2|7=Gr$bcQ8=w4$&oY=5LB)G1i)vZ{HSxotZ&!bA*h7^K>7{^F_jHg(8 zd&#~wO^VWFdJ$&&T%?&cKG$#l)dJ}DZ;nvRN@iQZgjD#zSZKiGVblb z!olYevYrF>3ZN^X*F$OhCoB3nzDKU>W)E7d>*B-KeJbN>93HSY^4gKD%?q?j3UZ3v z8ip8USoqa|>q(Azo~`~afs5a)oS@)Uf;|$|FOYsW8q$X}e?ZdJLz-no9=K13g}CHJ z_HGH@3dWB%k($rSnwzoJO7$%MAQOLCDplh;WkgoiKX%CZpqAqlOM~Z%iWm>vh(eFz zkzGIDvfOx7(fVK2IkH-IzLX~N9bze@2dkg{4Y)tbBC> zf4aRPCUgfk(@51r76@E(t+GzD!VQZw77XFn*(yF5RS2gY=9;$n3>|lZr`QJC-AypI z;wOh3j{2c=InWUKgQx6|!qCbSOqR9ykOw5FTi1t>dQQw+uxdAlljZKne7i0rAi-#d zEC&1S{#l$7s^&0L{EGF*$MzO}m~cH|>dHQ#qEa ziKjs7-gI^t({lT7O3c%tIh$iJrA=M%la5}PB(nG!>v2zI@p$G54=I!<31=aZFE!TP7xxJomG0OmL#TB#M`QP{r9A)V#&{vn1mUfR2@ZF|o zKg7e!ea$K*@cZw9?W);gxbw|TG_0Xs#f09zsBx8jY(x2u%HN9WLRy=HDAu(Jak{oc zWzDSWdbGtlaa=zbBvg5!K|#@|4Oz!_0YOK-=E(7c_>}=ra(YjA7ZO(qgX}#=NRAJ4 zEiH!W2fUI4>6ITYs9r+Xq1z0_%QPF;Ymq;Da`lo;bK5MW;r#s#UklAmR4)Z)Ej00+ zbv_&1ilVVoQr}{}TOE!Y$ykM3H~gmw9HK?ghV91d3TAfMOk zUly14pzB%|udL0SOFK05PA$93$3K8)p)e7>^l1&*S- zk~hRDG2$v9?QBUHQZuHOIWc?*-lFmO#61K3ugZSwJEE-Efj7gN@bv+%m;JnT&F>+x zqhH6zW$|9#-ZX$e*f0J23w=a2kKw&nc+P%?7%HX3X@urg6eT=N6g#KEt=z15|cS#@Jy~M~uSlZ+L z*;&paQ)Ze->$)Qx`v9K8<~@Vl@OB)NeV)*~KIRcr01YbW-k~9Oe=QZjVlVD^3;ZzH z|4PA8@tZ@tEc{ykdXmGmrYcKih-%jca zkX-}f+vXb9IK>~(-|iSDy6mw)U=0x`b>g3Y%o@d-PQ2GwV7C!MJI0&CJ_It0b4`U* z670?k&Ds4MEnOMytfE4e0fkT^I9&aiRL*RNYk07XBhAM5jD`4qJci2{_H zS4h1nNR&IBOWm?gMJ2{oaImz8P;0BINp`9?IK|9W0lR@zL1ijoJ^nAr$bh5TJH+vw~23KhB z7}`Op2)Hx{>8@BZIqP;f*lK&iF6-Yvc%iW=I{ge% zf=@65pT)m=1nw-~N^#}V3eV=)vlb&LHnOPs(4m>wx0vsdCo~~crKk#=m z!!Ns&?y8GV!5=0mi!V9FMLQPfJQ@^0ez$L;;&+deEl4e^oVu~REx7E?Ibx4KA}i!( z5?u-a@Rx~>H-lpG@VyP2Y4sdP^gfMYf;)=wHAYxdRKvPQv=Q z)nfpqcm{g0H*E~yiDqi;836;FfpwL{NVh;Q=6xq6KSy;xj>T2JS>7Zzi44Gr06LnR zkU=lu6j&MsBb?FRVOZYpYqrR!O3FjxYl|Mwm5?&v2BL2__C8;Pq1h*rcRb5!zFzYi zPL4+#qCxShe4AcCSaBuZI$ z^4u$(>jVt|t(sL3N0R%IWZf>Tt-vULm{qI0B z(GU%3vB8?^>P(9F``>_q!A||@YXi3$C2$z_RzVB{#0=};`v8_7sB_OSm(Mh`wD`k@ zZxh2@S29$6i_1q{Zy!f^$3km#<65H4^}^mC)fcgsICDM4`H7L2jRMUwsst5(11_8KEvz0@r~zRY!hYS=CD)AV%~F4Rv?=%48W}W zbDjI2q`I>J6DXVG0js3PkL7ofkR^s6xL6ILH&;iuE*-AUe3moUM|PNH`_EhqIKy!T zDbO|`$P6wl=}OEe=i{nk$r`sXFE})(rG+*N7#r(|)RmNQd9E|kwsF=hoK|p;fDtiv zs{|#p`@wph1(X8w&>>W{;6=<@g@GQ$T7pxFdKrx+OIgk6`%KNz;tU81c2HdZfiBmH znv;b_Ze7l2G_W*$f&VI{{NRL=c7{7fOx$qMS!>pEJEugjW8Lu zca-ZP{&Rnl@3~X_rC6P=SvvgprS*6L+8lSLG(2z|=-b=W@_Ll<=o^CTa4IY)M@o^%Gw44x69RQ9(jY(7tWIyQzl3G zUzEyl_er;F1hl%Rn600IArb})?;I*yuy=4c6BM4{s(hh zsdC*>nh@JGDgG{U==bOAo7-=Aha&x7iB9`U4N@lRA2C2&t~M6mO%49d82^;ETYe}e zn3pFf_7M>g!XK6^&G2d5x9QCyB^}nWET6ux-$>BN&CBaZND;$7uTkU&J`7^k6*sqb zcYnKPUn@QAZDyU68vgh%*y}qy2NfONc^M%ZPQ-H!!c1?T@zqsru+eG2&^Nu*$RNW? zu0s_iqWfz=n!R8SkNpmL_-XTFQ@(IHM^%yIRAh~5gl9-1G}pPQVh$6U$h9J!XiEIe zWxRfAwlI$>@M4VDjEo#6(M|4(ha|3u;q!&}V0ffY!d}={l-u|HBt>Ho@2)lffzVdm zcAY7P*Yv|&=MKUpv99vraV803*-zPFttv^E*VG{s#mr@}5vC=FoG(t*-~SXn|GZ}) zU~AhpO|o&B`v8oihMD=fxhM}74%AzFlkx;$4Up%Hzi%<6rz&EuI$DvbKuceSKR&0j zY>h1Tvusp>@l(3Cr_d^-B3SxCTs2r4{Zu3Jnrd6Vk9LlCO z%}53RmW$~Q61j@gps6auo$KJ31~(?INws>mcW}VGKI*80SAMhXJZ-`4VLCOBEG%#9 zCz!!L7=LNCPuZ5522Y8wK`MV5B*rn|m|cKUH~VURRY`EVmKx}pYrz8XpG^-zkqH3r&Ln%GeN{C4KJ9j=i1{xJORM}3?CtkdII=3;{tzP&(>0lfmh}% zOFQ3d=x*%(>VH-G(FM_Puk63imNI}PKU`v8C$G%RHSvn!-G8C8DpJ>3gpP^^{lHGS zSMo8mhGse;VrcG!l%SZM8B{oQrCHZTJcfV7TuhtooZbA3G`0{hWycGavagi~iFxC@m^ z&ytibD)H%33=1yL$>XET_ciGD`|1QHbvHg$XuKb>7VRlEoyyGfE#7eWtiO-8Yf)bW z8%Mw6X;___G{f2}E(qdbviitBmdc6dtTjblOgxj|de~{83sfQld{Rjs(jLrvhWD{l z#moE8<8-hN}hx6=} z{!f?!=J!=NUr1MyGTk$0zi`#ApiJAg+=3SX;p)qoZi&r`F?@XdhWca0ni@BtEc>8| z;~Jrwq*`4-v$#4Qp{e==|B82i=lRDEo18uMOMX%VCc&ka3%lAgJ4CctndMoT2~0*O zRybcOq@=i&9nrQ=o@Zxq9hcxvi43prXXig)L(*!{Y(98f-!`7x{KH(c^oNM|z%{O} zQ)ag_Dw_B)mGLy~VnTE8>FVfx*CZMJYF3!Aini7ueECTW)l1JGlD3D8t$Cp(@u zP(0)9tL`Pt4*{Lu$`jI$R|zG9!i0?SeYZq3T)F{{llQ=eHLxJe z5^G;cv|DRG|1w^Ds;;hX-I3@~Aqq55Q)?1^8}g%pZ7UdWglNdsaZ(HaM5EqIpt4Q> zfF(KJMEC2$)2FI>YWJ6SZIyK>P%5q!Kg{?6+2!ceu#TA#SF|H7M>J4UM4homV_u3U zqqK7q|50hKNXt47uzmiY8=H(`qLFQdREl|z?|u9kZwA5Dz})x0`oIy?&(-h%VS#B= z`x`?;zN9)4Js9z5T5(>t27^rkR0}Y%_MCzXwLOD27I7l13uj43?NE1rd>D%yp)q^V z@T@CKT&u)a5%jQd#UKMdc`3(yU})9q;*<2akz=po2YX$5YW16+&Yea2_v(`6>6`DH zIf* zVFarB<&)`$4ue-vi?F|prn3k`X1b;on=+gKwN2CvOV0a8pHv3iB(`lIEH+D9Q8!uw zM@0EBJJbh~Z`4T#MtdkFEUEqoO9v+L9rO8Q4XUgJjB>Ft z)KM4!=ddDr3M{U(fA7$fkH+YQYI^wr#j4PSOj0l{0Mml#U~Pe&so0&MtoM=elwz& zYLL$qrSxPDSo%zd*DP#cfU&ZpkcZ@x2(ZV)A+yiFw_lN`=5z^?TMR$gxN3U z>W^Co`>4O_jTK<%cu@W#Kbz{AKri`6$Jxt0X8dJ8))#fICIOJ`ZQ3?%L@PI3?6ba8 z;_IQeSQj!%Jp`su>2c#0)PA2&96->IpQ`y1*EWHA^?JOeG9p{~Sy56>Loo^CIeG`y z-r1C=)+MCHU>41Vh+@T4a(qG!>gy90;G*2!U&no&rJ;xPrOo(B2#x1kM?{%cKsrNb};|Bt^66Tm_u^J zIX3wJO$75r3MP&VtM~gi?sM}wir794rOlz{_iFN|Vdg0pVkz4Tb(r7_)79jfpGTr! z>SFNwrxP3{TRTdOJ9+!DfSzBBCR@y`mrbBs0}Qbuh2=@JNC!?Es5F^1Pk;aoM*-pK zUGOm9*meK!_UL@&AGZP&Lmphttxgpc`p~{;6F%ExeS#&SSV z{md*2#96gl5f*Ey`05B5))BTf%KD~PYGckIpkw;Q>TB`r_|zPZN|MPZ!E=fMp?Qpd z;(#>l2rJpNCsrToX>ZjQYq{&F0_+#-+>$UQg++Jb)I54_o-d!?JU5kQ#hZH484mxM zp!Qks|GQ)1DHu4%@Y(yu3dv`bx(Bs`09;fv5BM<{RWJ1F8{6Mc*m*7K$X-4y*S06% zQ}w*mZVQdeE4m9Oz#Q<^R6sD%?)*2F+Nt`STwB`aM&Kus=<;;@PTJ%XIaBE3Qeay3 zaf~?$R^h{W(Up7dlLfwa=F~0s-!0>qsY0)2IuvA~GvcHRHh)MD zh0@=c&X9aVswR%cfkBAtb}Lzp%N?%PPH3T2=dx{df@+2t!-duJC{p18T>O2{XnCV| za&YS`dYRFa1bpp5oO?cZ%REu@yz1(?GdgusHOE`ATiVrk_qYP8`_RLnF`OigR&XTBh1N=Xi*cPnK7zF+x!wWLUKq)NUi?PmY0!9`PPkmjP;&JUUVM<4P_ zF4}94iFaRK+0H!0Q{Pk-t>&fWIJk#RaLaMjbD$?|1IwWwLXm2wLxr9xJXf&SBg ztXc2T2GRiy=lYZqtX(zh4uuLSMAJlNE*5t=MeP~hStxqWk3pVeU<6{ks{wm#->EyS zMP1W9)UKcagMyGMErYohUbs{CWc7W{<*yNRui1c7+xttLl%Oez1Pz+U$#t7*Xqk?vJP8{Qg(Kt@mwG^$YQ(3` z0@CAgC+6M~?Ii@!oa;1eXb>8$J7up=fSkDZVq^*i%Y7M7y1A>ZWc@&xjlk6~96u3? z*~|~tmTW%6cgFvB1@>PGduewJj-Ty&xf@dFO9-VzAAu%03~Of(!$%AIID&0NW@Et_ z-z}T1E;AT?uJLILUoVJ|7coNhM{I(jlO^`!g@ADS7;|clZEt;CteTOKq{*!FI>N!g zZ8h)-b4LvR&v{= zM&7~F%jOe93|ERl6^Ejy?S$(a8J^F2Zdi_THV1!0_^!Te{?Hr^ld?9SB2pvve7)ms zZ(tM^Tnr>>h{oDWZiilTXaXwx&-FypBy_)oeN&Hnx=y1;ACWGA(-J%3Xc1Bxs`_=S za(;8IzKPMW+0S9iN|~ajs3{#fP$sIkYWBveKCz#P!ujh{q0;wwV^r2(wN3u#{}rZJ z(CvtMrO(yBXC%)On$@1*^c9)$kETryt(^HVLMcH(pqVH{!ieGmj}?Q)ELQP&5fMc7 zIm|AG-{=ttM+45%6~8B*OYwhxL75f}<29YBJV%oMVC!Ka-(X0amz$``q~S>-{C@}{ z3{DtV${ADM$y7Cu%67&GBG{k})z_hX8rSr)e`YlDvNAuL@eWfl@@wfMoMXS%aHPY-!qsXg(qZZwzT3}{3Nwc=R_(IK zwmuWte2qPCYd|vF$Yk8~mGd>)(6rUIJC&V}5zutV?^ZqTXSslCO{lfd1F5gD%&v7t zgk*jzhAGs(?&iJWErK8~6CZAhzv~jO*hn_#%>87}@^44uWu*)@M|!MH8oo$0J_7R- zrD{772Uwdl*JfwT2Sh^KT@!aZu>OPSsBR@E*+YDv=ikHVw@Z*E7Mh!p(OYx(w;?#xOGHHrpc%(cI! z;zvwN=mDa>Cf>6$sxRKaFnZnvS!CzQ>6p;h0xgEt)!nOI$RkanKDYh$P-iUhbQVR% zsT?tZEfgd;^PoS_vdzr26sWWGIC|R;|DuurK$?4SrmRFhCV2uxqz^CQq>1tp#Qi)} z7d>?@A)%gUk1-BP4h~qFj;DO-oM>5UJq!}vzA&^2u8kSmR-(!LDQccL>P)}wiaSV{ zHoW}@i9KDtjWqA|1TrXkk&Q1pMeI0Ot+&_>)B`0)jeVR^6kw%74yN@6)|m&!4UKhB zWRT|BzG?1uYL2C^{%PCH-_3MHU$TSsR;&Thb)C-*)t?i_5ABro=YfA%wziupbL=oi z@ZLL+T;|KYG~)R00mHrtqvXjd>o+`9fM^r}hojW|uWnCMl%t39R%~&Z(oYfFSdi#! z_%0_2BCYYO-|=+~BhAakkkI4c-_c0_x3mCt!pN$R?|6j}j#_R<%C6qe2r8s^$aNf! zN>#MO{$97AA&LF12lD+&|BNRwepLQh5H)Ly5s;>b(Vk~6c)(qAjq+zYw~==j%~9DO z!scE?C@e+$jD@vE_=ZVAxrwt=D<93ix3!^}F)hD*1#OHvHtl2k`1g*_uZAnUm3%lG z%mDCW$P?a%Z7|`(ez+dR$*T=Sd+R*)QEWxC5ID*Ri9!#@!rSWEBSw9C=AB8cd}|M9 zm_$_|Hx5spUZ~L{0duI*(U_@&l-T5t@c96%7qS z4WQ-cSpLqdpJO{HGcwQRJ?WEH%yDmTFR#^l2w&d~5}^aKyojPU>JF0#49+~Cihsz*3oSyU@_uNJbvv>g^F`yG7dkUJXbl)v|<|khN@2Yi<9&X zJ$yfgzaQ?^UrIPSFN@yxi`nYReD<7)cbES8mvQQ@R*IhIvh+KrjCr<8VXaoOx)Dm% zIA4ZH^=Dkd607CLPO^sH%zm-ge2}o28Qj5;7vT{1bDCadhVJKQRNs3;1*_1lJ5c+g z10-tQY1`4(-~0<*3G0A9`pLdIgNpThWRUcWy>U<5g;E{u21| z*MV^+-a}u^kh!Ue@k|yWumnXZNwX0Z8M_2Jr3$quebAWI zR`0ZF!5-^uog~_yN%)571t z)aOl2ZmN=9StR5g?#)fbFyiS2D1ii)dp~$K;d!F58!PJ8tiG<8+6XgR<%t-25v(c^ zwIQ`;Hs!CXQi`=igB}!KlMqKbSF&SGx}|_qGlt0H?Dru_W*ElP(RC}}81edaRA3O~ zt$`+6h3@6yG>a5Ic~WcBX@*@-|2&iZYJ_~+9%~;PJpiX~?`~8J*a2I`-uK4p#p#h! z4wAy#JJ4*L*Vaw_)Ui#^V{q-+B)fb|<;ExA8>(#}T67fNRgJe3jwH{rZ)eLlcOM-_ zZzs8dd%PAzScm&Z-Vjc5r@{2y8G)hIWlZgn?fl&*jOBM2`-x5{DAMs#$^~o}{mjLC0;tSl>u=ITD@BjQyz1u+xK_dSW zoTZ}PX+Q6a#P4=I!3#W!KUXIJ#jo?|)%ESi_wQRS+v@Ecz;F98n^MzhTgfxY0&k{PqroGBkr?nRIP!8`nQ{S7k{mdXcrWp{BhM8oN5ep&wZz^s(C{Bw+xDNHdKviGB#T4gczeRAk*2S) znqr(2|JX*Eaa)KUwz^0WAJ}NI#ay4vh@;iWGa0@-qAU(29nTCm*3zY}ycE!HtY6>6 z;Yq2;b+ISN;={)J3cR!6t##%iycJpV_>m{~Qa4q)ok>nHp;oCISZks4oVwDNjIu55 zRfBo9H^|>vw08%{<&7vgxuq7;Gx?^es_qqU%2=v3U?_$`p-Mv8fD-U!3qm-9MLE@40txyS3VjlbMAzvb`x zI7#bZ?1I(=1D@Oqkf z``Q!N#C25Jxrp3Z`WQI4mmyo3e{w#ZM3k|9i0$o0s$$x zUq1CPUnqW%C|!>yVwMq%EZ-HJ9v)#*MZEmJKJHwxHUnjDE=qRqBsOX7922mRg(}D1 zaP2MWj?tWRwRTL^TST7Ynd8kw1d@VY#E-QTC)tkpWOuhqSehDT{hc(&@Y2HGVmI>X z&w#`)`8r;fn3zmsZ@^s1R;L=f5}6A>n0cFwEQ#!4wveoLdYe7jEt zsHsK5o~RWR8G5*ly%!9hmpl6P2ckf=WbCh_3i{NdjK4`$A=i|UxICs!*tRza>CIW) z<&5Yg4D5_fZHlKtb1mSFwTY-iizZg`$86|^Jn5|A4P7R_@NrW4jN*r~cnO~WO7AMz zm#ZY4gU=^_a7|oMvHg_0c}LVW=x7SG@nW!EC2_(>yu*uG(p;URDp%lC?1(_~?-aj( zEMsDV%lBucDoE>SErGOcRmQ+r9vDEAu^@%ZwRQ&=xELd@s?xO7%PQ>c4AoNqCLZ|R z;{;2o^SsHJqavJJ%*?Mdgy7)jUs}yL2Ll3Xv|O&X2;^ojxvl8B;fC6Jq{eo%t`GYj zEgzVTbRnzUTA=@U&pNM&b=IDqHJ;1e9rr~x$W(Vd+^G2RGbquXJf8BSQsm?uNR>QJ z_@^65OqOR{JFmA@{GjzbH~(ZBF?T~xR8>_|sGLYi`>PUDK4{hUUNNz-jbP}-RS3kx zPwrkdhtczvw?Clmik4ny&3ef7MFW&t$%oigRlm_aQwsh_({u6$%aJ8sOJ@>0m>_uy z#mG6R)O;=|6VhH~H6WWltDsA40Wk=7i(7}WFM2ZYEDof3H~Nx0;qQfenc5dxu)FLv zmIn-<=4`ikSgLPtNR9FLZT*-PJkjatQ0?CMp4n|C83GIRD7S6SsoAOzE^GW@HWO=p zvQ^}-)wBCiNSmN*+$;H=gF$dqOn~~@1u^rb}nAnf2TlWMU6<&}Jh$~Y*-f3ySk|KU!`;i8#hXRSP8%hz z$UeOA8Oc?SPu*zChftY+TfciUO?Q9TVicpt2bqz5SD!+SzGMq5=`8x~5u=ehV{UlH zR44N)e(6UD_Sy_RjgQCnXpHu&5-Qm}o}FZRk=dQRU}4FLBds#&mZCN(HTR|Isz&*L zSFk)f=(OtR^mYgBZ$lr|Ug%IL1ffs@pnLFy3iaC3QLuP*+o6<~udlMz<3AxkmJeH& z$2E^PHGHmS)q5Qs9b1x;j4FIbDE9VcK)|iX(n7M9WgQ~M-sr8A;uWE&c`Q>1sxxK3U?ev_QY;R`hRt#OvY*YqRR!qSYhP`LTc6tIS|SHwKa7l9e4 z%;rV$)NxKJPF5Y!94nYAE!_s24}xBVW+r$(YiViY(#Z0a=97i+X2&!my9dwEs)lJM zgxWG@5m_*+NIu=Vtzf9k2SNMFmM-lxQ#h&zF#F6Eu<&JJf{<`r3N=_t(*#s&9-`0I z_@0@KvFyq6Bo16n?i6i_;qucorzMZ5_2=*R6qfoVe#X>|j$b`SmC-KK)o! zhm%MwC(BF!$X@=p>jAi)j<`0=cZiB{>F?G{VcN*}nSBf?p^x%zyifM%Dmcat6K z>?x@AIcA`6SHijIdhFflk4}f!m42Iq2J+N|lEIT%UfIhcdA`<<4-eAK-tx3rm79F; zq#w8G({2&id$1(8#H1n{fx|3$84nQWb3xd9&J?v$Z{w%lRd6BmFLT#&PqKs{XwlT( z?mnVg8cS8R|2eT)QrI6NcRR8eOGI*Le&QkNhY{)+eN4kH1VOji+uqzoK1s+UtR z3TBzzwNLO8+;qHrnZKh8yI)INcgDdRUCXHUYSe{2o$a4o!rUu+7$#Lq{`dfM=) z4goYb635t@q44#}I}NPyzNgVouJWjKKQLLY(J@*AZwmEJpw|#n&mSYoH_v#}_e7iT z$>W#mEyleaO+Ut7=7X!=SqL53U8^BPkBYO_ngp4W$9(sTy!%s^e2VV*wOf|nZ{ULo zO{He~t_2~1RwR4*OR`|<5?2#y-Mx&h&%D`UyAERyN>yHg^9|`wmPM-z19sK}9U0sn z3kHKc@*;5*L(o(r07cF2hW2D}<8BcE62Yj!gcwj(iAD`*yKnU-{vk>VDPTFJ_E)Qg z{ykjfbj%~;YvHJk&1ZAV8ur2MB$Rj%$CbAtv@qm|w3YalnHnL6%Wy5p8>EjUBRRXR zUipO=vSli{QuK}I@?c_9A~)o|7;KpFl-lU`rUbmr&*Bi8t&uUwHHtWXyiNgDoxvq; zw)wPbqPt()3t zUo!lc<#lgG$wbyAUzbafjjrRue-^I)1wxbL?$4+EZc)}qrBsjpo__;B#}jipA<)f% zT?y|x;#7cyO!B{$81X2Ao${!A%^Dt?7v54dn;=Gkw z`hOy&O*%6pK0Pr!LpR-~a;EVTj1G)s-NIJJ3Y5+Dkn6rCJ+l5hNTdSJfjkHwCq*k> zqE$x!UhgwGQJl1#mFgHcyxSACVs=6<2~a~3Z}d`Vpx2Bp{LO|fIc+a?4SymkMFU4< z^*z_MRXHn4hTij!0LwM^b9mP4y|`0gEt(#ZzY)G?^fVLtC)=Z3qi@D8Mu3H19W_?GkG5r_GN2(K3h%Rtz$X~ z>KgfFv-j>%D?gdE?Wo zW1SQ^*Ok3+$Dtz4BOKlWn`2y1rhd@!^`3DgOEEBKhyBpkUZe_{&{d&vS!2osaaZkW zxzc%@ATXwdp%(jdav+5G+4bXjfRP!(f;!FlSm_YKlRZEt-=7LGKn*eVX{kbB=V!c- z`z657WP%&Ga=_=V2kR(;KP-%P32*srT@&h4i`luQT;TTFa?sC@4jQVW{t>FlKz10baj#rI!bZ6aDwIDOkF#-Q7@#< zrd}9u@%k3r?>2cNJPOWs4~-Lo9xpB5+p7_Z-S^3_PuN!+2Z)EDQBB~SlddWsG!9?> z)J-)};8ez+1*H8RXc6?{RC|+4qtgFdg5@6gIq|M{TIAIVJWz`g+jO_Xbf5|%W%*34 zwm6*j5B`?OSX*=xp5r$loAtwBfFd68-u+?lZcd&2(X7WUiwPZzBU=yB}hL<&kE+;)*T5 z3W+_fQEoQ-qBDsm7Y`Yw4Wh|1i5aL4#Iy1hX*c}qxMX8)P-ym9d9Iue;)18m!j~E6 z#+0<3R%`9uWWe=fvK*hK)+@K|UvZ1MYt9?5XH5b2rI`_UUOuCeo?1Lxf_NBlocoFlg(_f4j7t1M_)CE{;) zr`H~5;MC1qP05n}>&cqR%u{h=c@_*HyLW8ec6(B1f9%k*I{wQ0INnQ*5|dML$N#t# zQK-`$>Lj3bxh_eirU#{Lb$_+rN;&_h?KO-2?&a%s?Lpqj>*qbN-@mc8w<+6hVH}}3 z??*fobIcxerhnw=V3M_Jn;;WhV(e3?EF)+aElY7PUkwbef!Hd`l)g%0CLQp;Y7~FpS%}}jEBY$&Y7$+RXwnSP{Z67P zT0X3Ua>hoeYBmc_XTRi`1B1PUfL8gAgZx{}$=hvP-pDS|^kjy->A9ZN_mAL)fx3R^ zjH9>Fzw?%1fVySyW%F0_twci05#E@T7{`;kL#@+#2rOw zM!k6Do5%Nx^9RB77Tu9eDNadvHBTnwPFvyVX(fe%be1<*Cf6ST_AV-tG*4cE(cOz3 zudLHOSb@v8uspVESOzs&hb9?Eoml6{^5|i zri}SPcU66T=%_c*8wcz8>?pQkJmllb7FnRYCOMB6Q|l$&Z$QOM8=)9c&Cp+R(lOP3 z@^qNEy5jZJClNaNQKG*Wt`o~D;bAx#UTw4WHHw{uxVxj;KTR_diBCPglT4@^viS}- z@+ki0ierNtxDWB4Y&ch|d8Vg3Vi+oN^ag$n${@+BN5ZcrC^wz6-E6>(p}t1IpUA)y&YmScur6Fek&sa-!JT3Ji&(;UcJodab>7f0UM<|~i`Dl>U5ywy8x{vRSDN(<4 zV{3w)0QOL1$-@wp1V-`XM^@s$_mdy_5Yi#?|jY>$~8RNiw~V!rgG^Zg|bIpiIhj)KFUqv~S$KVAx#3 zgTNKkSgN>k!&B{&clGy)RRgBGQ4K5%Y}Gy;MA*`s`832{UAq2%5##M_*nbI*yUwJ$ zHk?XBwzw_B{`TWfVxQFe$)f@_`0_I=SAAg~xO^Swlc`>c{75Z~T~v?vn*76tyjhCB z?U3cdhjgP0|F8#ClvhgUdHXfW2-VhuW?Q(@6(aA7Dh!5yEM)UT#mGcLdMytTfumpA zhp}bbE$H6G2P~hs27bfe3*N1HpnbfM4|&#w%#a_r?Z9;Lx~{E?A&CgR&(7-e{9W+*u{||*$VoX zW_4Q`#g0STb7q7_C}7P>x76~V`JSq1msN2=9q}{nP*E`H3!QB~MhB4@dux@wTW6*ddGM-~~_W^z@V+pg=UJ}d_u z^~+wu8t|6u(@@zA0H0-y<|Us+;QAG+{0QETJ6mb8@wX#c`zqcokd+aI90#ruTDnJ& zEjGLP@(=aa44-m6;jA?hYHMie|1t)RG~DR2cqG1avEdCy_C3ah+$XH`a;gP0gaj{z z^jU6FyIae|cLnM32G?+e&qqct-81?ye2}N_e&Fn*0jIj2vRL1Asy*{`FdMtN7S#-g zep5PaWgD;*hWvG8ydYyN+?5iEH;;&7tzns-TIdiCon5kVpqKxAvm+-mbExv}jMX8d zlTacr2tyBOPLnQ6Is*tAF{G>*b=G&*PA06gjCNSwqtr&oh zjXUerAA&iNLZgwN@8Wp$StYdGqIdx16$Pc0goSeS zDvd&sewb@nq~_u0c&TTN;Ux~|8Jt*}Km{w$6pt2qN{QiJO6H*ORr!|8@rSeZA@^Rr z?`EqmUSBp^-$l%Ik~y&dsDRve#Cq>JSES+9{^kl)_Tbr=rlIu@L@oBq^c9+>)bMs# z`@vop`?dzpdKUd)`JUG*d@L+(8V^J|TmzHPu7>&T}r8+Hy$SayP z|J&D#B2~n*(yKm0i!q^$Yb}sZOFb=_PT=oGw=5CF52Br`$&3L?zdrtFf$$GHY^&RS zpPT-ACTF{j?|I(|j9Hy_Wh+TePVO+&Saj;yvuEb}SDqAY*?VUO?KdUz_m1t4_eZ`= zA*cx0!m=`aR0$!;@dn^=ZYG)<@(#OBlWgLhE>{x>{(YjJZ3xpY<=4Q5piI;xbro z$ru+S{!BAQP7$5k@nQCNmtc&f5mxs#Hj_=w$EOW{?A0u%FgeT^y9}&+X5iZHfXuTzGLr_!f;DFlwzHrT8!RWDqGe?MDWMg_M2^T~R7<{Rxo^1I)aj z+#w64so~VTe8^K3@{C!&w7K^GXe5_Vee1eMYyqm|xN{~AHMs`12-K%Pe@mqJisQ~v zib9ctK{19ccdIdRBSH>vQHYfkBY=3RR!|DJQqc3J8KF0Gbp9;hNdH}J*VGHo$N!ur!nd@~9gkS(|O{sqXpU1Bld@bH>8@e!azMW(b zi6{!W_nou+Z}@R_`UMl_flY#J6e@9kqr|o3SBVYP-1V^hx6LK0 zC}KM77tS*f2xNBmfGc+#2A}y1wTh_!3*ilRbO|v3u7rtooO=s{*8<_%;fCnynUT?8m;9_FevCc_;C4r>vLHAmID!3@6y~ zYYN}pvhTq1u*zXL@xCMyP7Bn+yudSWiJprtxy~D<+oL$!!`2jNTM*%0`lQ;r)@(~^ zoj`C->!wreLwkrB7~&@sIcG)_|DCpymSLdvOP{f55FW!Bu5PBx`fctRvm`}X&&MUY zVKJA1K!_brql?n7I>Ha;ttmH4COQ7~Qu0saN-K8nG|pOa49Q+*ASZ?PO-Aj3SxF-e zFoC`&jdJ~)?n5FQA5HAs!Xj{952SE5YUb#8W>ahA|MF*?(Zbhh{lzug5zaAVZ0a$&Fy{VG753D}PUmm%cBG1E>+!(94@JjP1 z)E`&Z>v`!Y*=yr5l)Sgmp6O&#_wXS0fG~Y(TCsKWk80HP2##X?XezG5?LDa|f_wFOQXNFgHar#G|7lwP7zo%vNoms2wcu>h9F--o zTQI9X0o~|^ZRz&5S_LL)W9sPaP515I3e(=q*8avtyC^1e^~ zY(E;k2Uk?L{!LEgh7{$lGNo(U2}ZQ9nt3m7J=9#?*(|QE<^l7MK9DzUB&X^P_=mM$ zloba#$|onsELa3fKSQ%HV*UI`Ol$?;4{l!@ymdbqe84|DPguGAb-8+<`6zv;Jy!T_ zXk1Jva&;rK+{2%B_ev!rG>!O4ta}&Ub-Or-$TAN*1MrZq<~*R6Yuq&b#is3lAbN`m9_W$fO^k_{tl#b;#Nm@K)IcjC7W$<`WK zo?2bYdfQcOr>SN^1WiviDjAK2z+U~CMuwo$Lc6AYq^r8Pk2X?{zZ);86@yEo#XE7;<8&GlV5P$h$%qGQwPx>SL$JbuG-N2UY{ckrVSfT!V zC5~UM*bzVPXr}|mpRYv~f0ern!yKgV_l{lMlQ~aNW7{E&{Hi1mhoLEXpmjgNa2UP$(z`T*^dn`813p3eC4EIgn0m(W54#S>_?HH}0ME|-jZ+@We zCg*dCa~k2hFMrAG<&531@U!s~sup)iD9k^b1}tMd|~-syI+<;7+G<;~F) z-2?aG;7ZA($%MFJF0)pXPH+;q5_;#2;gp?wyTf}J{|@*lwLZam{3;e+CCvgh)H4&W zOV0&dCgfskA^o|u5{TJ$8;t0zCZ~@~nRN|IUN)$ni}cSEtCdrNAKX%j)nF>c6e<&vpy zaRv@$iZcT>IJdM+W{EI)SbE}ibQWrwR(TM=Q&=PsS<@jERktkPJCb`%$M=x>7{-6# zhbR4*AV27qFtZWX;@`ei=D3eHcwB-gukAp7LcMQ6`~zn4YBZ0`8P@a0MPN6%TZm8& z_w(TPq}N+l<6{;~O`v_p-a{I5bE?~Nnz=R5tW$A+@MDyeOty=%LUW=(Q=_RYaOfwJU!0cRy#>y(sD?TSGHu@0j@@B2Itj%Cc++3R)e9#x`Mx0Kx-UG;GLpOJnr_{{nTdnc z*w?HtsM~EAC6Fh1Q7U$<2V#G?fp>S|+4lzX5-!ujcD*rjNPOi|T9AR}M@ro_uTmw_z7$=KVN*&GfgUW?f6qGVhKCl34lF|4SBlK0qCMBND{B&Shb&)Ecw9g#cMVh6bdl zhfM4K-l_C6GozWINX%rxadEAazm>4^UtYX`h;=A4mrb&G(qh9;yVf5%R;z$MyJd@ z$tv(+`9Kq`S) z(4i%A{byT&fq~F2)nko($#o$@0~lC^hg?WlgcfZ1zBo_)*bFO>a9?#h0$7HcN-a;= zIBA`%to!rIaz0_BY_Uocz|Bk4AVg$=8JJ=*l4BmL!#!wG6+$;@gBkyGY{`#L98}Qkv^oeiCrEm3b z7<;Ojrv)q$c*271DqfIzIIkGkydki?;_+g+ZjHj>W`D&Bx$=lF6%q{KyE*ELJu>j! z;P)9@4`HY9^YVSVm>a*2WO&o=VNRo!{8d}KOa>#I07C&=aaA%tdO=}WcVLgG3QlV} zHsxim1I%vSQyjA=f5m1mn3mPoInDF-Ob;)CQFUpYxiOBYAYkQw2XuF<`I|w!FpL z6$kW+9?eRY%G_jhJ?vPHBn!ZDQs>Q2I8rg+qOw~36ZRS{c zCiroX{u=G^ACL>YUk}M5ysyJw`DupYzH33#8Z`#NHSu%$j6Y2 zWz6>K3hm3_1g{v7TMD~Iwdc|6{ogMl6BAW2KKwxw7HC}CfCvBqhr{k`nUJmJ7LqX{ zBG@qzW-1-IP-J8x)?_wgiIjHMy2U!nyA&Ri5ht~Sz@s+ke3BWfbW$MOeQHCakM!;S z{UPht-E5clIL8aHn{4=UYloFoFxw13;qT2s+Jy4->(^>JhKw16C3(e-7|%2y9`5Rr zUpJ{IfuXan2CnQWK6pAzk1+4xg$g+q4R0bpxc@-D4gDds$|wVJlBlTQkkCGbGe6LL z5cYj@?CCH6i@Uldb_n&7Ok54F^{tPz;BMyfZB9CpM0CVT*rBTP_~K_lE8#~iyTJ1P zmgBZ>`{Mgzc&!l{%dumt^nJk%z`(p>)*ta55gp&h3M0Bln6TWYH>chxfl0PaXT{K4 zI4aW^!VB|Y=&cWi(btkwY(|lVXtJMjJ23cc$;KP5{LiK20bCWuHdtGrc=5fK)61C% zuO6vXQ4hQS?R`cqW$wA%liYTW9)FfdMUX>-9?o(vVQ?MqVSjjbwDcjkPs7!>9F=c& z^@b_g@^p;k$9TNd=UdD+kFiPtx#RQ%tC2hQZidIO#@u^u!_I4#;-I@&@W=kQckRWP z2?oAaI`;fNGoNa!IU;F!Tzstkli(JXJzKz;AFeu~1#k7LCz2TKXcfF4N<=?(=BTF$AZkR~=9zC1nDWX%RfH^p#qNkgi z+|PrIBQC}k=^geZ!?_^A{)x9@Sv{~jEtWmbR;!bM*{S}a$Rjfpf>|RGi9dj4%sN8} zuVgXIalRYUR{!#$g?d=J`?6r80~O)obD~XhjD1afZrkrxaBwLQ0vZ~*0^weBl^0={ zsyA{P-ET1GdQ7x0inw0N5Zw&|>k;bXE^opcIU=!8loFD!8Zx;=7VU}NiVH`D%rQc> z8Jboij=;7sOi!1MIs*Ff^G|ZZAd7IvqZW_hf*jh{)8BKc!oYIMrrPl#hB)0Y%S+Wq zWQ-IbT8VS#tQjKo->KKAH;fgSAnrvsw`7)3eQxz=n^Ro{Y+*>KmE^K`AN>2R0$;rb zhKj|nv%xxQ65Hm2hln$GgKOW!Ul(MnJuCSRgLW{#ttV-h`SOEQkdjq-lrho9g>q!c zYvTaV`gU;$?E4Pzl#;`GWSXtg&gFXj{2S)vH{H02<4D%~SH7~oufqFA#X@JVr)O1v zHcc0T^9K~~*D$-KkMdmlxE-wbu8a3&8aJOZNfjk^U>vAyIq-6+;8vY8G&=d73(=bA zsyliMGiQCASI6RT1_PZhmUBP=H}o0dakkLKYElgQkyMox+#ufD>Jhp9YeW5>3^!|r z*2;SnD#+WeKh&d7bkN*!>tA>w1WX8}m*n1JTq-W>$? z>sDOo3S72}(Nw-mGipYYv|y&W@vQ$#|-tGaPz)!f)ZaSeDwFALyFL$YN&j%8n- zp@X3<`&ygF#p3fmDuBHJstm{ zJyZlnIK`p%5P~saUvSJEv^QPGx!?R`<6T&}0%KL6b!99=cWjuG$jFQSQI+3NyiPTGAqg6nl6;m~GKPt0|*oI7$c~{?MB*zDKVt4M@hE z?0IR6p>~9Xl@w+NiGOlO)K3!b%R*#(%rALrkJd|Vwal6F7`L1Q@Qcn(R|6Wz0qBMl zlw1VcX!5;+3T}yVu80GjZSIJjwK*-CuczjP^i9K`adcSukqo&8U0ihT=bRd#vIds> zrj^W>U*5GfRTk?MwsE@6EIp&{_=`c;QH@U}9#$X}Y%={;p~~5}JBY=BB-|H9J}Ueo z7^VdVzosJ@xIPT77x%@!0RA7YzB8<;t?QN+y7VGSv(P(;NDZK30clEyPy;H`J0YPe z(h;P06%gsYh6L#yq}R|pp%VfDZanv%?|a|-E5GtQd+)W@Tyu^&<`_-v<8C{FG6j@V8&{=@>{krjPQ6uM;Yno|EV_1F#`uuw( zsl0lHvIkO9!`Y)uFzJH>25%8$<}wjvSn7u0B_=_nk~s5kC(ebJjlW`OtM%Q^zgPs_ zVe(X`i2$6voA0fXMtV_~J;^SU5BqLVC;S4?Gf7tC*mEKyoBO3asC#ZCmM$lvJ)q8Kon^lo+qg=0{hfObNG6R{3>vt5G z(lNo6FxnfTbVz%%(u1T+>+1gLHoX8YAUpRDu1B|8VX7A}RmjOOjQvBzB3*50_wMyx zMHwNk?UH;l;~&D*KMI?<$e%=hCMLI=@q+xzCaCj7vVE@0jb}O3$H4R9nEj;)v2VYA zQR2-GIgxv_^^Ag2e7!gnt{;_MQE^!0*aiA}MoAt}Lj8t23Ei*Ne!8sE6g*u)X?m8F z^$Z;ZM&3h^WkT6~oE(t(C3VqRVTI0GQyQ}J(;3DNd5=r)U1uPy zRo5MZCP3^^uEm#?iL>dMiuSUTiSGd`#2q>+hA#Ad9e~pt7MbHd)VpId-8oh9`E)jb z?T!!u60tBZns^7{@4 z6~}q=iOkYvemU&f#pI}EWWxfoL9d*FZ70#3Yk~`8a`%{&ee?;AQjl)Go(wE2X(v!u znAJ|nzoMyuuZ6^a5tJS}s@@rN*IPND5B8CZ6KRkyw{~)?mDtp8 zN{PizBrUQ{Rk;ej^;VD`PYU&2CGr}Bx-;WtMd7Bq+o$DjUvpt2R0PY*^j5Mp>36e# z1SxC% z+_U<2!sc)LCMj=Ys!1OSaO^Sw`Ur>#5EkBfGuc@6g#HQvZs!XMlVU~Yc_daliLt)@pe-z&SBynjb!`J7^^ zJ?&qA3jF<~1Mo&1PYnI=NJ&XK!;c|tjRhL$pC_`|Zqj_z*4O{EIg%gaB)z*es6nW@ zK%M6#mM34u%F0z}$$Su1Cl0-pXa>C}IdHEH=*e|)iPUP2pfV_F?GbWpjv90z?z2X# zW?-cl7lflVZAEc$FDEXOgc3nUXj*u0bEnZ(trNk#?*JiGzqxX2+XF!2O9*>thOVqP zTTk1G%6e=xsaKvB@2dX~EwuMWJ2wU(Njj>;Ov&ACdCOZ&YKAL(6a?9 zU+d{wnyAOCm0d~~w32*x_he87k}4A@PL8GTn3WbL9^MDAJE#epu$`A9_ib z1cs*|jQhJ9m*%Bj~`VPNLH{rK_@A=3(4SE&|AnhJK9Kf&=TIdTmmrBdwF#;VBob z)Sca7BnlSh`stBVIC)7S1N{>M+ZBHLZ~p}sseuUh#>Fv+ z)j-j+^?MZiuJx=dc=TvF(ZsQ(dIBWx$|^N5SoMBPPys&&1RB3`feE!X;AE^kIJ;!% z99eP*Sjl=MK5i`aKZ2hUIqz{G(@KfHh P9n5~~J!mz&1Ny=2|GtdNU%mD`!tHl* zitpil?Vi)7=1<&O$tc%0Fbr+;;XS8_U3iLl#~=5K>;#+p!{@>v=3mA3)@ik$VV)|FVQszc1wji`&QeKa#AEzN?Ze-c-T|oXmZpIDKq%|( z2?ueuSgjx8Yo^sH03BYCh?UKfWq~BVH8m z1K!uDcHXzJTf&!8_~w|?5{w#$=i+|?2O1PL5j0mw z4WZZWq&Xn#{mGH&to4LAnm!|RgZ~)WrmI*osu!zpa>G7w-H@I;VWC9VV^vR?XC3>M z$2`#|<8j6mx)`nnP@rlS-b(2L19NT=`nAA`FKD72>l3p5Lsd* z;8uG(tNG4Y7}se8?^$4cKN0ZQ#W7lX!DWZ1qt;sVko=ID|Aq~n=h885b{Uxg1bBY1 zA`q0SL7lOlT}7%N^kxdPljkk3#gSk6zNB2s1AHl1n%$eS@}6<4^`ba<$hdL)T^J*^ z%?N|}Q>xh@Z)$j=)J^$Pu`7XenvvuO>9yX2_Rj-|-$a185upvu7c^3yE@mv}+!W8i z+j+yP=JiAMb3&g$B^Q>wfv;s1?5zt4EcM2*IV=#BOP#BMw`oP^qSc)01is!}-=Xtb zv+E4oV1^Sgd@%146n1I@XI&`VMh&@qa6%vNjX!?0$6NeNwn1}S&U(C%?fX@UGQ4}t zpmgzECCFDyWJ};2n(L;i)C&Hpp$U?yl-0?1PEEmOXx<)2g^ide?MnX(OYo)<*#>gO z^EF#&5d7ouadowK4bwos{np%VL+Y(Ka%xEXbpZ`vzyFft=56!v2yy1kB81|?{p-H( zNVBRgCrFy(g3(69)$6mhK72u2Zu-PX?DASby)!@j_s?t-=t?BN{HKgpmy=zoM^0s? z84SxDKgKlCwQ~I$y!olG2$BbzV=YRBF`IX6IoCf5Xg>UX5IHZZ18mzB!QWW}5FSy_I%!A0Ek%JLx_odoPqC->L<=arh53_n)ir3*9R*K{MXQNWjhm5r*TaQU z(yc}@BPV9)y&h6A>v}=$F05L;h|N>h?7{s869WO6|EL$mB`I$rVxu;U{6f0f=Y;RTe9bh9YqzU)F7(g-SUB}8@(JD~-4@tXfp-VYl^40->-zcrrOipl zqEeGe?M3aa^Mk_{C53?>-ey%qJr#V0$wJw7?3Y%mmk*4wUzxN%!MWclrD)&5T)2Fs zbxkw(&e>*AnL-!V4$BQvF!Hp`U`fjk-sLUvnt{cmLDk#&m4k+D6<#A8LqaeR3+_kY zJJyWu>9;~i+Ej+QRr=O3t+%c+b}&Dq9l&u(;}j2YBJX{kPxOT9LCR;(+<)u z&D?5%(_(uj#k|}?PdlTH<(s?b%rZ2hl|;SoB{kL$DJOZ=H&TsaDxa80^9|C_{Q>JZ z*|9hB_B=9QGa6f6gXswP8(xIrje$TnE!T{5C zV9ZBvdHtBsvfGFbTVds9vcso=4+d|29!qN3E=BAF;^PydEn`t$8`^zxcF``g#AVXE zUpVWIo9Ll7&EQgbSaAO2$0InYk2@Ux?e2!h_49=IY4;z4ZV3N~#PfU%W|1(l#ClRh zpqiPh-T`Tq}l-Ia7_;$`{L|xj_8OUNhQ@?lN%rsnkT}pToTJ{yhx9&wt-%&`h zc`=FUiW|@=6V;x8z4G3~JTBSIt;!p`lWi_-=Cb)Bld7qw%1E+)IV<%)zvh!S_#;2B zWh+AusDj*r09JlAKF|)4#cN@A%bqyAZF^c?aK@cCT&%QGe`MAE-Wz(PCN^b;mGW}-F_q+VXW2Ty(lJ9*l$ZI-3k(zr!!XMtEr>g*NK=p{0 znKy=pb<}C$e&c*OBl~#zget4|mV}M~S%12-;vV5PQMJX!LvZcwZmA#lJlIZ+4T6p&zwW)x@ zX~UU{J`Sq>fo{vhisLQIirdVUA!7qE+^^`otq`x1!l?EPB6uq%FS|s5S){n&t%Lg0 zqv1F1UU}c{OF!TCkyJ8F*-<%n*Mt>3nEC21x4E5!ABg>bdYpfM;@{vasI$@0H-31n z?vtLbSGv&MnrAwJ4}#96-3x)hhk%$r#aBi!*mK86? zJ^J0F(aiPL&--R`de0PZqSYMr$to%vV$-Gq1+ZrPU1((0Z_hq{r-h~)@CTFbj;~B| zhXLk^zU{NUu3}U$VCVN8w(>WhnkUzPGxov$q0RpD?T|lSBQLIi#zq^tws=JhMEbDp z79&~h*!cL;{$lHV^X#mtM%u%EdD}k){-iPrQ}L7Vq&MNcb6oP_z~N#TZ3cvDvjM_k z>npuNrVf-G%yCFe8-!NwkVf%7OsQ7ojX&ipAS)rZ*_zwuvL9Z!5l4F704Vn-yGx835Ft1B9Pw4-DF}0l$Nfd571l`Y!t9SiO0? zj<1u32I;84-3GJph!!r5627EMnuc|u;>_!UL`ZXQV%DNcGXXFSFLi!N{{51ShEu9| zC+c516yA3xZZfLd&ceNNh!zeQ^Ss8J<;FDbHQ$z&mga1_xTvhG{41C5o3h0!YQti0 z&-Yh$FX%qs2f8)S&W)0R5EEL)Nt5z zm1|)(_I|d0&GtuW^UZ5hIRsy{>R>8F5 zDMoIQjE>G50M{DAczHD6byuydITc8B<3$B zY5%1)BAIT!08lSH)9rtqX@5)z>W;zJrJ=W-n*P$YpBbo=@7nwCiYJm0a|epVi^WKeD`XTlN;~-@x2Fgd z3A2>Nx_YC}IJ#^VV0q&ya>_H6ldnC#(K|HNb81!`AKv|*f&6ldNKtB5sQ2<}w`jA| zj^&P#AjQ19?|#=D=*5AcVGJz&!K(`P-RPVIfF>gwt) z2korOGpMVFF#KG^((p5`_HbsQ*c2ajo2Z@snJL~nqMoN&x1YDH_Guu}PnH3P+f%iY z)|X3$grbjSd;DVaWyp0$lG&i7W|ivn%{%WyygsO{F%k2u8S?r1hSSjR;o`z#w~4 z)nfEMQmJG0LRny8bVhcAi$p<&FgtZNsFUE&*_4pbgx8K+A$)_X$)RQy86K% zUr?;yl7Cd=(inX}kiM>D8%ZoHX02TNVDtGfVD*EH)DT} zG2KX_Uoq!tLxi=0if}kXrSWX)1VUq{OI~}&1E%aU@-J9*?X%6xw+|d#d7JC=KVWkAVij&U$7Xkw;h%C2b@>|0%|v)|<0kU}f2`wcsd6BcbY3GX zCTjC&Uu{}plxJNMYfRnkK1C}&e(SK!BA$iF8#QLS{na9y*37HZg)3$g-&vC0GnS^H zU;Oy&`4Ih8?v10=Ttsc3!_#yz!S&!5>&+U*J<@}q>`T#Qk3-Yd_%bjE{qAf z@MY7|;zgD&h&r6=>q!@-<6rpM1Xp8(F}mN-rWV z-JggxQ!WaXJhs?+aGkL1%4FWP1U(c!<;$m4DW*M{_nvLA828#9uizyjt#L78B&<*_ zq}l>6q8BGe_P9~S^6GU(vjaqE-kZnsAa-}*u%U$RRbkKHQeh2+PD&3?YAk6X`c?Gp zuNt+d%Wl+HL&~d8Jo@gro3Zw&LB}7AG7mg6J0M~gO%FYF?-;A(VO)7ka;_b4Pnteb zJ*fp;0-5==v+R58)$}2MEMFkm?AKR`j#n}S3&AthA~Z|$E=rM&4ymJ$>bK^3reo-J zkvvE$L9fel5mA^J&}{2GU&j%1iy0zjfNSxFtmD&HUMtww!RV}{M@#m28Rhzw#*z?U zwt2~;mgJKw!(E5}fW|co{&pH(cI~1NTwE8Eoti#`oqIoj{W=u3x^d%1=U#=!Yl|mL z&(|lmQb1~Zliiq9nK+*+H#H3qNWSgNy^gTt7oyr~VRI9A8h%-|iwc#` zgb)Hsl|Is5Z-tU8;N;H>q`pjV^=?jIs-$Ygs3)#xy$y;YdX1X9oHUVlJ7$`MXloMe z#Q-Ly&l8<%W2m`mBjttaHCAy>NUIVaX7=qP(>ym>S1Q`i?qbi3y@dXJ?|;7i!Rl^+ z7aI_99K)iUu5&h5f&>!#IzPyQTry$g%bhZr;-g`w8Td2rA z0lzQdG*H|slNsIQC(25c87`>_;GDP>VA}*IPwmEJpn!4H7~ea9c%(uDMmgA~qU0TjM1e`W3I?poj6?h_7QS3CE%j?n`-4_UFIO zOUMa^g2(Q%*B=V{?M1!%T;$I=QG^PREiJY-YA`2=7$h4&kQn;VZ+<~nj?cd)L`f<> z7qWXSlJ9%NdgT7rp)vZ}x-f86;^Y9!jL_-2h7@ zHKTXHl|JEja#+wBr)~^F+`T@N+nrN;%Uq8ao1D!oN{6>YASz9i3-va>K^LOu#L_+i zrPg40l6rGP&l0%JrkJa0GT%)$!`Z0>8-LbnSL^tH$H&CqG2he+F!)eVF~pTsg?4{W z7OX`oOBC&kM~m_Lnx$eNWgGa?eW#cI#ZB0c7iHu_ZL7%FWfoLVE=U&8K8 za*fqC2OdlDXHBfVdCY!)qGfZ)I?JXp&Ik*@4lkY8wzi|$6S1+rG7ykW8M7-xZ25p;RAH$-U2B1;QdBi=?dh1#pHJSRuIH!>oT)v^??*8K7p1D?t_?sX7*&sxfVZ{kA%>>iX$$bB1A48JaGBE1E=Vv9cZ?} zZ-w@W!kvuZ3sEN{wxhJ0fu)b2G8~*oOK8ABflD20*M+i}m!F9J=*7AJOtg=aiNwEX zINk!Y1o>09!s)4j(W4WocP|6%@aahsWtZy1;BK^Sna0%Jj#j7dCUE5|S+!q+5IBK9 z%Y}Q8UuyGyMeafGNoXa}nX(%7;E(mCn5RMqL$e8n6Q(ihTQ$pvk4>Ck^t1eMnj>$* z3FpL!&bqgBD*X0iECl2v3UbDy4Px=^bqI_v4 z&mI8D(`NN&(=LTT)?0qW;+^71Y%0M^KIrc=*6qyWAG<9^wN@;8)K13B z^UE z)eV}kwlqi2_8bP#eH9Brob$KJ-ZL-mErx!{%5rkOA3&rtPL7G6+f@-lFkvii5yCi^ zSELh3zb?b}BZ{`NfWwnf5ZY%4v%^Xxt(K++>7VT5rpx^HOuD8T0?{X>l~q%X)Ue%e zR1Ymdcy8qTLF1$3muJ1zc3A{5u(KrYY}P)Z#>*VVfy+#zJ-Dyi6!rJmLg=QjU5bGu zoZV3>o2$9mbEQjAxy~Tbk8300iPPJ=iY;)ic+|5Qxa8^u&nd~?A-VRFg3-b1H?9K$ z^OWOO-1=`q)i%h#{R~Efz!@h5$88zMw&L1(ITxqHS=3Ze2K24Di3!-L}= z^nG-WweSy-IC1C~y#o?u+z_A#u|4Le#QIfI1qf?SCih;R`F1d#yvKTJ4vDJ6w|iny zY^5%!7M|QGw-=ZGRi;nrhlxt1M_P+V%bbRnBZ%j zRWjW(@I%h(-^*O`4DB2w*d+2p70p~|)BPJMqxW-oW(YF~2)*vwJZNt22`(Q`JR>UPL&~D@Huf zTXGzW?90s9}T~8)Pv*{kX>Ue$TA(! zD)>ir5{zCL;SoqqVNtd0;`8C0oLs<%ykg>q!!OOBO?(&=F-XT*jVGJegir0f%V6!Y z8?v$=_t8SRU>=KNAP)%}QLfr3-w<2PPj75iB_e|;X!50iK@CgmfWra}@0$`@kuI2Y zz9Ybq(+?z+ZbQE5zHboZMAXSC3nQggmt@k+Wha(|Z*a2IzBQ7-1o`p|yM5<27dRtz z`=Vi6usi=xvcLLF+xsA)r zP_~ycE2-RkuT_XU922dC7abA#oQU@-8{PJ&{An}ih?jwMxIcur!S3+mU1cyM-dd2~ zZU-0nl*IQ`fe4!Ha`V?K{GE#zQ1E^e;EV^BXx6~hPM8Id(P8zg?$h(Pl@k&bvlq@{ z$zINbhyoWRW^sP-r&t#zsq^~-&;EY^!27@OUOLu?HY|@uP{~b6@Gt@Kdgh~(2C)-jFZ*N*%3(_p|jAkdBKS=7J#9t4F$q7N`1A}cfr%E_^m8+}Pxw^&@Dj62F z{fXDFC6c(aZm&#eq4-PX4!8V0sD_>h?QXkQ`>)E5-#6F#=^yy*D6Onw8ZwThN~VkH z6#NcXR#HV$yQajm z60VJ4OYXLHj-|K#OgCxqyE;vH)2FJukPG#dt|~zjOS1v4ce3VWP=VdPu6XEUSlR~p zeeqoQ`A;v$GuQ6HwU7c~%#vN1M#6W593{&OtZ&}Vw&$Lg+t<&4Z6QmsG zdc!tQ>noo`#l$dsgrt5mjnkj@D*^|N^N;CISs`O-<~O_Y6O=xiWn7-Qc(C443eTHY zGBu%VL4D1&q_AF*iZ}W`bs?hE?y+m5apK!BT{w=j{XIX-;5i?~)x=Cdp>AE^M=rO~pGL@|JV` zGMh6JK*qq8L5aSE&Aydn7Y?{7S&M4cd7)n*76_~TU!=NJ2Yud6cooI-$6m69VKZZF zq>L3$u0!6}MGE8BzO4V_f1G4S9FuhU!w=aJmuZ=o{98QlbRgr!(71hNsVTxjWLy$-HE@n_e{e;!itzI=Cq5LuvA0 zwubjCcKhy`9?sm{8GV_5!DJf5Lk{SZe;4P+wDS-jcou}2#RGuqivv4g(tW$UiF)~I zf~(ng{-UV*q{PvT5Fevez!66kO)lLUI3;Vw{PRh5!C5eSiUk;$CG!%qVr03N?h|Iaxg{5A?LasR%zfXQ`m!f6vS)f=$o&+ z$Z-gk!@bc!x5-h_zNVm(eQ3}~g-S6{>E4X;ynr?E-;5Ujln?NCUY1(=ml^y=b~TPh zr{HgR1Jo0x##vcC)$UNAgxnd8WTztwrOzql?C!O;%lKNV$?tx^Z)XOj_Cq&wtKW@m z1$N|zk}3tgVO7B_=NuZ*r@Zlt6k9S5?rdmqf`|?Ki9*A)*pD%N`1`1vL` zfa`Q=!6qvd%et#rHmb4P!5SylV@n)_6+C70TiZ11%FZ zS7M4~tc>IPNL#WW@(J*|lmVDOm{0@?m_yC5Wxcfm?fjsW^Omj+u?D-dnYz5l{h< z;~=c8-?*Sz{;lKxdI@nr$CE%ySRq%ueYp(2va>j_m+Lu$`J_m)!V_-V`) zV0jf1*g@LyMoK01{jk&b>Z{FXfxy|t0PV1Sn-tceZD}`_KdB~*17IMd7NOBh!n!is zmY%Dl=wlux*3u^O@TAiDwb@;-10$dtZP}?z2u8!A$DLC>IVqtzZqP8E@c(-ewWonn zxQ9T&2@5_vV=AyxTUoqx#j~PsP)+oW^;K~BySLA`kOM2WA1FCnO1}CsF>|RE@r9UI z*)5zx6?l2L{B>eK4zc-8K$12Hq26Vg-lTH{>b-#xqFyZLB{bP3pOmi?iO0u!3lf34^(FitOFLUU>NDFz$&wn$ayOPr{@Fi|xfQ6E3atz(t zNZ841!?Cj4la z=zl2H|J?oQ)%cTPCz1glVtTFtXo|IMQGn=k2i?7U`TK|JEFD_KtcZB~+q1Ife72mg z@2fuAAF#*tZTGiTtSIz7p&8CvD2h*ac>ST;YH93U?1?d;xT05|apN6c6=d%%8I7LQ zg=}$MBgN_1MF8+mc;U{5ubH4o^G6;Mr63$N5p-1}JBa(|BZz%i4*t_caWZeoepG(J z!*~m{k$EtzrZloE+6Y%0`$k*&sV1N~DE>hZb~A!ax5e)Le*m_B2LS)(-n1ry@(Kqzja)2TYb`>`3ed8q}PNk zBEBy2BT@vCNs%A9){50oq`FhquSZ@^|vTBnSWa7c)LYA>_n2{G==0M=jIghuOWTQ^^}*!|SNn8?1Uv=jkSRSm2& zyII+m1Y=)*db2W2)%yP(j{yAkBG#P@!l#r>9t@u2*`gdlAW~0Hfvb&F%-gXo|S2d2*$K5PGbFaz@r|y`RTFP|6+?!reDg`Mg-+n6* zq!4)dHex&IJnx>U<$MmY@~+AR;I!iGeVod;Zy;2(xQQnOTM&Y^GX4}n{!y;B<4bDf z%{on!N4Z2jUfG#iP82Qm6|=R6Y&Ao2DCBzn4{x$bbAKYkYyP=iaQC{K&}(u)@i7s^ zXy?BOvf^R*^~q|ZMLEtf?KAUh-N;su4q8pIEYj>ns*5Ok=hZU~?(Fg30pgblhe-rw=rLWxpUL0gIi$O83i{t5TW z`IlCb^IJqpEoFh@^HBHRz4u-zha-P{IE^75B|8ds_@P`@24QgfV1Bz{K&hdh{rCKU zt-mZrzRyO4PgR%dZ8`VZfzuS75J=6Zh zMQFzn*Y$i2Y>o0KNrvy#2h6m-Qsuq$kS`w1j$A@)JFfcY98Pt=$N-0V7Ih)YAo!0z z?PFP3`5B>qaV9*Ghv}xw-?}{Q8JP7wWn>1CN=JTaAEobQE8tF%{Z_~8C`Io@X96eb zka9u!5WRYZPuy8)e}aA8mk6Y0xrpqCMpVVJX*&Xyvh>f1j5qOUJQjp^?YP=x6Zwe& z_jLXEn0R^CcXD~@QpP%4DO1?F*c|38)&~^kwat++Ll%h2*Iv0Jly-W?ja2x z**9+uQLTZel6sr}gB33h;ElyE9pi@ObTNLCHb+^{m~Lx=`E(J>+c3_!X8v~b_q3Y+ zr(WvCo+fwGxcG?MBg7k&?u!I>nm>t~50(d7AZFa^fvV|R-@ZdXb7ysA1}UgQVcZ{I z98R*CaB1r{mcM-w83Rlh1MP`<;cVyAl|fMc6^P?`Kh0T!0C4(6*}4aQqUd|p8!a@J zZRhG|7)c=XJa$-Xp)(fXHo?^qJP7s?vHVwN zb@Lg1vv+bO5?o-{vpQ-0&mK4J|E9u~aD*r;w2Ao}n0YWapO2MVO;vzfeXE=hmk(j* zBU!%d_D*=rK=D@8A;eF{wm)#j(@2x7FbM6m&Dm?K&=!u1EK7*WIs!Hun{*2=1*bdO=$JL{yrit)=cSdl}YvHTp7=5g&@s_l0<&GI?AZYwTBigkenD0QaFQf%ygDZwWj5qP@u{|Q_VBDZi zZ0U;t8dca0v)}S~CEhqOFQ}_hkR@3Lk(N>9W7EOx1P}|fe(fUBk7qeP0+y^S){E#_ zeW#FfbYhtiZ?=9k(CzZAnnAPfrx+sr4lQ*&U02~=z8-$ixL=IeN#Rxh&)lMeU$4bv z@XsoG&9W};v-nnzXXu2#Gf3(7Ey$ySWsj1J-qej@SFceIxGia^%XDPe>5m=Q|Av?a zC_V+`V%x1P_k(y?mwic_B;VOg~8er2UDz3Y5%OA+|Jv7{0U_Q?>`# z{=;|%&SkBmT1#NkDnCvMIe$O|-&R3WMEgF!EXxWZFiP1O)0!ap@Um=J5F5QCc-t``n}<3JHof&rb;)*ZXMEObH&D*PJ8_5*xrlgLKL0R6Zuot z3MrY6;|R;vef*F~wE<-3>>ex=~! zwdI=;4J19p3Rfu&Enl2m5jjSsej0Ypi(*s(RjBZ=kN)_i4B`J^-{|ECn)$jKa->^= zb~W0qby~3LHnwBa%JEQgqLjz1i_e4N*%gD19iuA}sRowkLLH6z6${T}p;s=c1p5z0 ze_rH@635>&woL*Z{))9zmgA#YG0M@V^zrRW@y~lXdGPP*1PF+?FfYhZ+E*ka&Xywh z#l$-ClW~wpZdqB`vqr>)yuv{nT?qPU>!RU&zj0c0)*SWm`oK}1k^(&;0lPBp=+WZqN>Ypa;zw!tS_I2R;_HF% zX!ljDg%HAL_Y&@Fji0}-lBj4->cHKsNT$}3I1Vplon&?G`&O-_8q3;8>-lscYpMTI zMx#ncJ0GNQ(pRF92qT1BAmUlf2hFnd&Hd}uXKW~Jlb%kRpPsszC0eWe8V8hevX9)F z<9i-$l6yH^`KEKxh`_d$H~*Imrx!^kO+_c+5x|+eWbgXH{2$^+o{s1;Eh$^JhEV8J&ujr_Ux1))+`&nY zyPtz`s@)ev#KA%MmutOOWTO2P%WF$a!eEw*xaSO#T2Buj&)?;td%|;I3eEmfO>&4Q zKk)WZ?TmSeznGVgER6YMEXL{w6v&+2*BjS#7Tbmw9p@0tU2IO5mYkitaaseN-R@N_ z7`0RU>}dkV^+${)f>Y*cSLS9 zDa{)Q7LsdZP`nWhF__mJ_f24-W*GSI%!Cww!@jn`n+0ueyDWZW{r2`4qT z9|Aw0ae&uV)-d}2+Uu}3F=hj1MW%)8I^QTh1e08`H!RTxuabeeYh~3Tw^+_hy4}Mw zFxFcfvlotMyR#F|WNZ%9OyH~1M=>8iX#OAe-ZCJn?t25JLn&#bK?J2i>24Gl z3`(S1y1PrdOJYDoL^`A!=>Z1moIyHg=!Sdn{k`$O-|x5kk(ptieb!!koqg7N*7HOr zYwb_7s0j7u$PW61OuFhU$(rkdLKLm0{^`(Ad4~BIqy5rsE?L>Ho>QebiUa<@-0oya zd+zPx?%CsQcM%g3%nMsvmm;x2w_aLzIIHA&gvILZ{JKOhRPmj++NcZY6z1PX;m(~d z>9RqDuimTsRc+sR&gR~uZ)i~&(7mQ=zWv=z-m9F+GlQ}2o6Ad4r2k3X^7kto34RvL zsEif~QA(-Ebv0jdL78u9&s(C45aZYn{e8V~_g=)aPF?01mcKjB#61~EEH}Z}6?W;> zdhZKB&QrZ%UggK?Gq0~#eAg$~?qmn;(v~`_ z*jJj0^AO`!XAtK&imEqfeSRf{U1}fkf6iGR;Cm`K0ck1k_7IA3bhNcJ&TEkU7cF`K z6dEtd#h?MR4{bf^x$Tsra%7TNlpCVu^tUMNJl5T>d)&bHc%o1|>iuaOGL+jx4|_ng zy={1`73%g0ZCNVt*{2yv-A`M`$0bj0w`H?E(p^^(!t1N$fbh9g!Nn4i#M?4<@)s_dhbZ=bGwx< z3{%zhdI5;fJD8_V7jE(8ZO>-yY_r&N`CPUr7T|Vr&)40X_bzIuw9EFMP|)S>oK&as zQ7zodO zF9x>cE_SpSY^hydh}UauPG<3CIw1XP`PAG?D(RcJ z<5yRWU!(f>g3j3Atx9%soV>A0p?Gu&72SHhAM*=$ev3@*YFU@tI$Cf{0VY>rZ$8_c zm(OwTG@`XB`dqKBu>@&VX(-m9KbkAv;;&4M!T5wRbvOGT zX2MK^R`sto6eT;sx5B0TcP<4NC#(Ocvo%3jMY?buat7nkyeSrq{XNH_=+2_Ief#|x zW?JQ2Jts6NwfKNik)thE>w80+hNiwGORX7Fu+GX9C71}0+cJ^u6<13R@@D!g$3@k2 zA<4!9<$(fcA->Ihr_8x>o9nf&{v~1SE~?7AYsAnA`({j`S)Areg20@zleroC3rq`F zzaHYinjWCJWDCgESx>MF>!bO!jug3_to?gjaR>sQpUGbUf}1owH!3>Ml?Q)FAoqKy zvEQ#2>kyt9+#VYwBqRim_}?7%;;_}17ZvR%j;dzxg=$1&ahgd2vaIjqyiUc=&x?P zt>Su>nT9bIyi4?zxH>!k8a^Rn8kaYMeocqwCm36GnF~{(xOOYE2Eg{xdlOYpVWM-F z>h>Whi6$y2ea+`q(PtKo;`S=5RongZYpG-PBLF4i+cg@=;=cb-WEw@E)`P2JN?8LJ^F;$@k8Zc8e^*c2 zNtGCY%K0*(tl5Hq-TcR2aDn!!e$r35PRG0EW(V9(}lk;f*d*@d(*5eC3V-5 zGPee zZ+pFW*Aj5iL6Cg~cH;J{DOcfo&~GmX!%UlNKf-+*zlN3$+ov6Z9x6OHtp`NvQ<8N% zWtI+?5!=06x$G_<>MJn(AIi;580RR;IDPI(A*}BkTaYLZnb|GJ(0t#)hbzvZqa~P5 z1d=MJ^y(g2Cl`$e%cgltx4f9v`A^JIE(ExduM)I0HN8^ZSn8VY^oF?s#n_zV_cOb9 z9Xm@=23Nmv=?PN6Z%R*?0sfz)7Qgl3>>M+FZYV~;#lGtPQ5BTBoxVI8Xj^5~3i#F~ zmx#*Ee_i-8aeuUv$IbO5o==;YpYnVK}2U$Py)UADxO_g>oqDQH8e-6oL6I!_CL=i!_Vw+ z-7p;ZPgoRITZkEF775rLt^uqGfX!(UniLz=>0V%A3>8VltS~Pnv@*k7UKP0Iq+@WqPX|a>2C1+4Bol>L2rH{XsJ27Y77z23O=H zJp9>f<=d8pyZ61xYhTJ&aL*#^igRubDS2A6+%LLRw}IQN?_N59-@E;^tf;51sxZN)sC{Xl5&hu3VPyGa|C(>*=>i3 zTSyaM%B0tSUZ8lmfb=ANTHFxV9mZJ=$eHs5%p3BVe|c=AYKMa@a=Fg|IEL33YSyr8&^YrXuRAD-DZ)S~-aLZqNVbiR#E%uVj`0M4P297)Ubun5;g>!Y9 ztKR^gg`>f~bncG_mL)9Eu~pyzaBc?w(HmGI_HH9%sWs8c0ppPDuwJeGZ@}B#n*nYS z5Q}6a$eY;iGVNBFd{pd`+~OO@&xMVyh2zyU2k&PvO<`81`MPGFN$%tGv~aKPA>^3N1xta zUuU1G+9qFO92%A6X|w`-Dgan2L|tliZCs;_poA4Rx$|L!UU=ojWWS8RC1KLA23bJ zonFK!dheqxuY&EiI!vN7(pYYMI-B{X!HV#aT95Mwql;>Z4EIb|>PwDSHTHr?1rZ%r zsJV9Z4h|F*nacYDEq#2P82&ozg(~jK`)?c7>@L0={H+w&4$PoO4p?vAgqg*kz#qhV z0DOKL43g5iWm{e{Fv0Sk^0rDq98^tno6xjy98#EuS7r_NlXHVN;wH)vDrD+cO&1WB zy%eWv$@ja4t!~`gPz2VZ5BW4QBnr%3mirtmW$+?CO~U}-M|uMLPcbh#5;*bXGZkl# z-_4;K&8f02b!KzI-Ti2Ztn-T9r4r^*(_1}E5uUHDI_0Ia0>J6l`D^deGK^&E=H9N| z!>7k0f|s)1o94G%>Mymt8|D!68zo3Ie*npHz=GjzRyR_3|G(7@N?yzz-Sv;kR;&Qm zYhc8;c9q7dV*kcRX6-FE@|{=j%&TtS_)@G9EpkWdu2AsdF4mA%OQ`)E30xo{Q|}ki znVU%SMcIfD2}sLqX$YVn)qX5-x`%Zw3r$Fl#|nue7Ipk7rJ zRH6SV1v}jP*4oTBUR-x{vaAzywXEV%h~W`*VI0)-VnQ#OR*sDGD5Yj#Jq&Mc_mIDX z?Bf^IYJe}N>?_{sWnJaom1gWtBp7zUm;1}@ySQWw4|i^wnM5H}XZ_8u1?%0ye_IQ) z-U4@a5)vJ8e>U^e{J{4@ zCy|Stl5(+smkr2au;=2PMrLbiS${Y2Y~L8 zi#Hkm*PHO}P!kNHoPPp4@`jX+xx~Nh*~IpoPU0S~OtbN-zwT710sYny&~;p<#=~9Q zoVZ%Re+lri8@^_Xf61ezb=Cao{m!YKli8MOf&YG-OBcc=)uVeDWU#U|Y%wMR9tO^d z6T1B(h1u87{ueQ7?jpvlqP^f=mLl7q*yB} z@JYOn|J@A%6=+DLX`a6gJ|FdY3;0cpA*h;+p@7e8f0@{gxtMXCBt|nk|_&i%h0I-GHUS-4z z>O}6fUa;jbKyDG@yCp1yCB_F$*7rK=lgmzC_@wp6JW+`RIwsQJ-b)_J>(0LZmH8g< z@Kue^cD-1J+g_771}_Z>8~9wuCEce^Kw;q`9aEYCl*@^Iufkn8xtyY&oA|F4PboF=pvn_NKZ zSk@>2m-XN58HqCBZxKh&b^(#iH-_(@IQY5YY~70Rhq-dRbVn?)(M2@P4x-(5G!vB# zfjU+j7)|-3090cb?#Wr|f!$lT8_!QPJFlpRMY!3oN!n29ZhttJQb0!Z2_QyV z*Ph)&)-nv=%EeCZKRW*#=@M;lhu}q2SF;0;jzHYcQURKsFD*T_Zaa4H#AMUkz*IXx zQc_ZB+b84_^Kgnn&cV!&QW)Y>{@N`S;uqwDvLi4IU?G0!dSz^TxjKti>^{dnnzPE& z3cI|~PkRQElC)q^{}F_H0G!-;xV0|7tnm9Ac!&PI&m8`gGJ6Tz7Z(Aus|4oV!>}^D zRo9>1Zn@?agU!cWQURox{^{Z>=${9EEvQo(5sURe!V|^0jiRn-zklWgiD14}^dn1J zDk`4XbLQdxrsaRF+%s|7aOSDAUW{{8z3BwbPy}Lj&cghz_~uUdUsskgW;a9uPmlDL z4FlJGT!|KRKB(iihrT13PTLg~opH--q_4-a)u}SeS$qw&qic=vNruk+!8@Rku@yO7 zC&rBtgB;2i32ZF*#9UdryqEke0>L6lJpOb%X?hX+Zp$fdIIVes?LCo>m^`xz z@#~%iyR)E5IuqlV)!jWKfLX+E+}-*GN!9kBTZcco$7Um_!eErgQ8G+mf?0|%C|BxQ zJJA;#cz75`yms5>K8M^DxBbS{ZlD|AN!R0)S^?aUiWeqY7J&($Quz}-i{l{>;|DZ zRTK7#YnbMZfYCTwL-tN+r8uy@MQ8s5H()>|1+Ym0;f85{5zi>Qkv+KWz5t9YrFLE=(`-k3%sR1vt!i_*A9k`RRKR(OV=oHSS3?`_&MFp7Uj1C$Q}a)5!B0 z-CmV2lpa2q%p%NJ(%JuexBu-K$#%E)od$msNePleCpE9x(jBrvhLE_&mEIY+(4Bzk>ZwU!Ng_OewD% z2u^xb+p97LvVxbpv@P0-THa>0{;BF=4K)Bln@Nw%`BFngS=N_jjkb zwE0EopP&E$XZlp!Bp?Ruzg&XOw*Y3My`Ao`$;I2YmcxL${BJjBIJiV!xOxl>&yN

!YW_n}=dI))t0#yrZ@@nM*~dquR;Z#FgyG;MgjS}N(kFI;o2@dg zqUSPFI5?<`3>YB+M3#@xoZ!^TnyxSIKH}NOf$$$EM@VL!tD1XDil0n+uR+?<_I?zc z)Xni-MuELW^-M3<`m4u}fZpAm@ko*?$aKARa96eb)UrG%pL8S49`*l-&tr}sp z>R|-N)$WbC;JlmWpRtb!fw^8{Yl&;z%@rzyF@4!uKU8&C*S*uPuX>sze1N!sT#2ti zoGBMwYwi7dp1@u(4+R7x?<&mdgY$*sVwwOeU7=s4RiR3his+-E zl9_Xb7))1%f5uH0!|4kIiwPBz-u83;L{n%)q>mEg5DZL(s1}n zzSjOmaHKWGyiRMJ(_(NxEoV=u@`+(X{&d-*=A_s2mI3 z@WFN4N#EXv*IrB6r?zcZqMHPwqdC5Zbu9;KjSw0yZ+5+R#*KNk7K^E?b0t?e&MWJ6 zZ}u!|4FeMYvs@uktbk!fc}L0q@`>~i9Edsgd?h4DlGuCIUNHP7HLbS*8|EXKeRR+^ zrY?%G^K*tfu4z@Yg@)lPJA=gKKh2r%gm1MRCdEJWO8#ePbqkD@_Nff74^4L;<%Z!n zRM7f2jfiu!m{uP)f7*MTnbJ07zQOJ5_x_@L%)USv;%G2=P$IIdo>_1k$1mB0iH3tU zZM~iHj|~79WqS~hF3Hwkrgk3{m~JC_pbmr)=22U89Y-#1=)G?5bDr%|Iy8K^cwVee zlO}|)Gs;Xb=r>>U6ukcIMnt#j(Xw##W*`$m@Zvk!-wqqnz9J<>aJ3?(9EvVH3vs^C+?C|)qsr_UKY*l^8pFY;wEPP0lUKcq89T6N#w!%6*u&CM2 z@+$w|>fDVT_?3wPyR|4;qwUWfJXbXI5{1~I&k2suZ$9xxT)9Gq=bF~+f3TcF#^cx{ zYXXFk#So!KrQ(Ng1zt)%8!{e5glEc@w;VpFNBPkDk6odGfFlqnxk&%zZ;cNE8+Nk~ zMIrsLXNzJ4T}?e%o-`Ni29#-B9&YcS?!X4teu5j%@T?d`XY>#E7nGELDF5rJ+Q1rt zIGmoc{P~&ezPc}R|H22H^HbW6%cWLN!IHpz?sZ?O{``T^|7}Q|KtI52 zK9T|774O|G24F_{fYpQTx!2w|$^Th1=v$1hmLGb@x&QL7e*?vz->Av}VUf ze~rziif1!y;vG2DRW1s64;PfjtSYgRwqBS9-~z4*|$ z_R0DeM>QmI5m0E8NZE~pV2~=wzdGt&K8Em5`CU9 zE=?c!leLSv+d2t##qXPp%U|Cv=g-uPsb%gpZkHXM%-NqhzdZ5~Mn0!PkK$^y6uCYc z8%utpCJg7K<1SOFyr6`nqG0!Yz4=P)Tn{BTp`ou`RR0)vm zzW1^)xD-bHWvVflhGZPb9|6U+^i*x* z2z^!(1JBiz8ld(Bg7T+W{eFP7ddpALJI$=1c4hL`se%!iPk-H)vg&uaD>&V#zkMTd z(8p{A_B~O)tFL@mi0ha<)6mpmxjrB9r?XiobM>#A{ZQ=k^MT?QK*_9X@|sd5Msh9P z>9&H18Tk%fjN|U>@l5Z2ZrEgmK z*q+wFV1g)%K8Hg`!=i}s@}=PBZYW;MSF)_*hd92Y)=rC;X|?ZUU~OjtC!23npKEAw ztlM+6Uyt+F8dfMO7wO}u2rjrpnodQTN+iwU>*jpGnhOx1Xz2mJ_9co}D<&)*;FK?a zGI1YuN6~h5o8Z%V*w(9jg+zZ`*MeO74c4P4KP z=qA@KQ7F}XQzR5{!AVH?_l(A!*-K0fT(`y01iDT}qhqS$RH|Oi8;#KrF$nY_MS1hxv{&yrDy#vhm zv;e?2u*g#_^0wUGTo$r|eX4r-xSNqOzf=4NIqTkk0xmL^v3Rso*-N(xFkEd^s$|r*>dT=rXh{n{alRF|XT6d+T&-unYkgMfo5#+Qz*nYdk{&Vq_hTh4HjV%} zl+@~4zguAF7vVHcAKTn%ycn_FW~~Zcfa*-$!CT++Zf?JXNdtf1w2o>BU)11=!R^IV zjPNmCMkS*4@+)Ap*-N3R_s-{+pA*cQPn)u!s#^MWNG8p_e}tDqZ|bC zeldQSQtrdT2`sM!7x_C&#`4n+Z2FxdC9mz$x?fKUvl2Cd|C9uop3GRBy3;Z9dlzrd zT9>N=Tm`|EeC5m^0B^a6>=uEPjwc*ylNnrw!m+XfJbSMeOG5k1k=i ztuX?9&v|z6%P$daZNUNtW@zOoC@2Eq%5u_gueY?u?i9;&zJLER)GKkqTi#aJc*iMT zv*0%O!?LSB*xfsSXb^zq5{MzCBtckSPIOQ}Gkx=!IV4^Ga>8W!a2h-mfBWG$PE)r1!tt&r!;b zX9R8+v_L0Q>%7*jKyABX@}*ZMcr^^SmiMIa5r&ssJLNZ?$6drvqxbKTWOWmW?x;^2 zNruwOy}p9oUK_-jxh}Wiq<3hicRJ9hE|H2Nnw^b?-y4Nl7_oYjn@3Z?uGR!^b>YGG zQ+n=tMu1nOo9LyZaH!ytxdniXP*XaC&}hw6GDNa-GlaRiDca7uJMqdplckIl>>RK5 zd))zRQEDmENHP2qGG=LE4W6G7s1wuMlh&2sUmN4{vYC%;`ynA-9Sn=~^*=v^VI_a2|3q$ywHq~BT z@tD1v_WmYL`0NfWw(jz*dvz)|u0^CMKZ{B_lu9L($}*Iyv^%Z!sKB_JOzWsob%xBm z&o#Lc1!NSKBssOHwP(iB{wCdMZ-f_sD;B_Bm8nt0il0oQ?VJUlLX6nmfY{X;^e(ie zxP?cyVj{Z`>^I9BkQ43MBPwRtlXUnK*d6l&nf|=tvj`9nA%{&22}ky6Jcgrxi#7ST z0a8?1YAd|q9lco1_@K-{y>!R%S7G;~@ja%aL22|y-9cFTQ_|S9Z?)TVtKREz-2mb* zPCN18NIdO#@L$IS+B!T1JnLl>v>tmaNgy{J^cYT*-(vhQpF=V?c;ao@!>+X&GICSL zc?m!`N#v&Q_0|u#>VCN7Ri1{TPoP13_)4I{hMZ|V)o1-FT-Gf0>EaD0@9i-(`5drM zrfqG;;O}qVQJl4G@e{5}8@@fPE{2f2OR9C?NaAUBxcj3Z6w^=KABvSD!zx8xQEh0Z zsAGu^QU^hM(ClJGz19F7c@ssd`#*)np~BMCJu*d@ z&%GK(!wpq6^=RXW-{eDEEWkVI&5x6qeHGR^D&ax#siSE_WJB$m9EFo3VDyz;E|)e# zUoVUAXWt_VjwPy0oQL++fO1@=@p6oyn~DdPkJj@Wu*dI+&LIBSDAjxC+6opO;Ejxz z+$}Xu9J{D)%E z%sn45=jQ3KC}w$;#+15-;*-!Nl{O`p7HU71+!slHqK2yy`X)|YVf1tS0|58Lj0E(8 zNiDKnHI1s}L*UphNy=WLRuByR3zB@d|-UG_iEzjg$g5>GvU3Z^BXJ8JO#NW%@wEStrMnx+CE5jwIGB zx@0j|#-CQy{%q&q>m3U}T8koGyid)$B$u?9kq5njy&m}pGwhrj<1>d5^ZIN%Y4j+! zOV?=6tRN+!hiobXhLk2Br7SiA@*|9bN3Q^oJ4@slt4mivV3`nqEphv;`tMN*%OuvG z>{1NJZdFlAt?_PH!+OrP0{Xq`0m=dJ*br>#3EHb@J>7liZ$Tu6dyl2v_%vR*i9LJN zz06QoJgPwuH|0|wH80zsg!XE}ETW!ygqIJ1}SV%0JQBn)R z`d*chz}czGYn^(~A7Ff-L>CKp!a%}#2c(-CAD`XpiuL1(O-LA33<6W8&NHh&!0ko# z#NdSrkm^|BHA?#X9hr9zq-zlcE4z@_Ve9Hx+2Jbq zi#22;<#YQ188lLT$hERXzV4&$SFS?z5zmP@~2XiA%Y$#*0G$({{oeD-i(B zC5x5UU`UK@XU8DLm0PQ!5_K!tes+{yYW(3oac7*689bUG7!)i}`x{bD^47#gE=VBS zdjReuOWlKdCF?*{_;40Zx(YCju&e6o)1aHFy`4P4j|@gw2DE-eceh(kw${-&@$QZKxWT)+ggBQq2H_*e8VG*B0icA z&1rn6qDV_Em8=NjlILM%^d>X2hZY*KeXa2?^e z&pwktr8lH#VDG0!K^x?5OwxWwgf(GxI0L+?mAhcTfCAdNxTthTMCa`Zl0mqr;C{E%@!?ch1|?r;c7OO5dPQzPVW=!2Qhc8PI?jJT{~Zq@~#tTL6@8xvJrM z1B#tiW7(PK(iU-5+m%w#DQap+7ZD_R+E%yWssyXkWBT6Y7x6UPz}mHi#m+) zEuKV!KukLj1g)QUJBw|dsRavMqhBxSplPh4tV=?4%qQ@qi61jCy~2LB?1@o3;QC9= zhOtLYjoOso3`c~&=O9Gi%tneHjXWVfrO#+!)k$cN*}lO-7ytLK0DqY%Xw!q8+!&JI z?Iz#=)DFsvkZa7{-=3qwJe_vIFsQO86L$x}*3Z3>V|;eq`q-f)gF>;Yg#FO#cPvIVGkxYa zLAKq~g@cP9_TDl%ydPmf;NeJj&Tq5IF`?y!n0gE=Q^Ec$X)!&nMok=7mNpgXtvOj^{9w48(r0 z9!RA64vENsZc#^T7W3!!% zi2Gq^rYnHliV+aFOos%XUJrRnXX?2($Lk8)LK2*3D?s&;0)4NYpNd#^<@jBoAg1qS zHa}fg?Ln=PUYD1F@$NxOSKUW4yEdC6)TGqXO^Ca(c1XOW1%I7cV(G!tavd8+;V=0L zOLsZQowG>lr+*z0%;1(GA74COp%{^B-KQ>zcV89Z6k2KA0p8!8#pS6Ig|lz^wc%HB zVlyG3VEmxgpmk*^h3AMz4V6_|f*h zOHW+h=Wrm`T$UetiS8J9%m+D_CGATN;OY!{@=u%gZFi)8=N#z>3%ZzNN8<_Sw02bF z#qiKaVT@e*7>GXhRFwve(wDs7&5T+>d;q&JD8%rm6dSAV@0B#34>Ia=PArmQ_g5p( z6Ie0$j+ms&c{(}J7M&o9(;ZAe*f8Ts=G8xA&eQ&uM!Jinr@sn+-J`^^#4sUl6z|AD zJMDl%_M)v&PAV$oP+wCUPVAAF%QVB0clh8%zd7D-r)?6BS0!v#$xOIr6h~~-dGp~G zR_Gmpa8NYs*{8P!lnrWn7?}=1vUyG|1h^9Ij!Z~-lrGK3^$)CGrd|^+KEl0%~ObswD(31j96Fwb=auLRyellsxlfL5LC$(8qm_^7C zZY*3mNHC$Huj8q(Y7P_(#lJ|1-$lAbW45JVhquc5b>^Rcd7WA?YH&;P!q$aiqT!Yd zHJwa8WK>%~z*AtV-eNH8UC@btsdGh&VuIs4G-6loF`&-f1x zB1UPJzYoA!2wl21y8G}~hmIQ`MU(V!u$~8{(Wb7jhVz@-QhjZ+3Pfbts6C<&5BB3j z4|RT}29F)R7t>CFF6j>GD2|-!lwLVcFg}*k5MRN~!F{-TELHy9i;_S3$!yvPiek3w zc}+{^ux@RLz|0IPS0QD&p%fgmH>VvNSC()YKU~zE?Vz>i4C|#w7%y$JPEJRZQ#2~^ zGV^PDr&uv)^gh;2M`lfoz-y4_*GRs;bxQh;na3;__`^b*8a07ANs$7}Ntwd7Onup7 zUthp9MzPKV#2#VBrCO1@CUEyi3xjC=onHu7=a>9634P5K90aPM{r(!M13D=Aj^{;^ zV}utfCL6{iT2%PLt#gid@-tTsEi6xUaqnANBQ)DJExXw^K$&6D#&|BQ*&C^&(P

    ;>=Tkz7?VaMix}#+$OZ?Q!|mZ^c=Ky8dH}+6TQVgykf` z=0@L;4~rLU1aeriI6Q~zhypa=DW`-SaQqzF=FaQ0Me5j<`_1PPt(6Z(1_sp$U9t5% z)N(AHql-NJ{Mxu+8AVIaJ%ws*G>oFi6=+hg*wy5ANaJlc+kNPNzLjNsEBR1W6YBy1mrTC4bgAVMK8v|HMr?$O*xh%lu447 zvQ)F_hpJ+6BUu6+pyU{Y_(Zqs10QlyyhPwWSAjDL7w*%F-1c8 z@Z`xN*+=MM5uf#7nYsGO?e%xj z04N)>U>7p})52%n9Zr0-uQKDG=iszTO4)rcoTJHK{?)#sjAz2*5XoSH4lrONX?lO{ z6N6~MSyETCx~2@YinaZjVP7RSdq_=ArL^FZ=;o`E#Q*6^D?i=HnHV5LGVN!@XG=%rc<`q=zj|!>V1#|SuR7lS3!GwW zN!0ef-^gIM_VT-$Cv67ni0H1Huc~l<#+ry`Y0wJ2ZHOJo3+e+w9Xl+A^FrF7Z87CG z+?+}oo*u#TcK)3}D8Hv?t5$z?d!?XX5Mi*I7_DPk*@}vctl;B zUDY*jFiyN?p49s(VTUP{iS<*w9Mkzl+G9sqwWJ+_7Pam~YeY3@*PuqNZpdNuTViC8 zq$aMQt#T)+z@`$9Z#E$eTh!HG~lI(ff)`!n0by{C-wY-$*GlIRhwF?6G&L-D5R=kfWB_YA{- zfQXbK@X2w7eVljS0H43+sIaBJyhU^tJYzaaReZc($b{oSN9_pxCWO8?H7)hfec7HM zV!8c`>y+LUItSy3CHwWdI*TaPA|aWH50gLTEv3I9`OEFY)>R7^L%c-Fu}4SbSEN(G zSy@ZIdVl}%v8?EcW$)qH@bt2$5uPHgk4*V=IBjzVto+d3g+OSTMyi<4mU66RG>p>| z2=$DDONH^7($pRmkpdf1ma;d{(7ul8V0v5bv5$@cB@71R8pChWU_ywK4+3ecfS$tYPo=YIN3^{XtQPk?M$}3Us;8x=j-rCPx zd(5K7%^mrxA%1coCrw#eGuCwl2g67Gi;>H|DDFpMbA0E+laiWwj`84ZB}6lo6VH)q z?0pfeR#fh@$EFWqOXzzV!Z#g)a7v$Gyf0iEZ!{>5owD85d`N0lp{(}wu&!DopXAk- z!uBNob>P<%Kk2@MZu!wTxVtS^+gT}#vUq}s7${?Dyz<8R(TqIu)2Bg0fy_c(9Q&mr z@>B&Rtj5j{t#!^Bh`LbkhlkFdOLWqZMn?;gvb~c(5+OhsQ+;@}6$!(@V|Dg78n zhX8@wF>?Ja%`JHlndz3a<0^Ke(|fC~O-aM{NR4(qWa%slJyu9Abga~I4d+AUQJNDQ zln+(`uH7OMPNu`I%}vjh7M`(2*R%Q@a+r%(^H}_yBN4ePiD`hOFk4uNaa?eLkTy=Y zjv5pnJGjwnc21jXjI)iGNarzOTS)L*JL)hjpN|KOyIgY=b8v9l z`FYFoRH+mR);o#1Hz2gnJi;v6!AA26dNohfrq_~TN>D-(S(4e;YSH}BPYI$2&a3YZ ztIkrC2e`g(R@vG?(ef*FSYq`~x#U>BSsq!Zgjb^=LyDe9r=wOvT1f4g2km)L!2|Ga zB_U+=0}nxhzDRGMs%CKQIx9cF1SR*Er=J)>3B{CPaoW9=Q0p^#>8B2R5dt>0fdWvL zLi{7V+2qq72cjNH8qX4P@eio5s|05&8ad4w^M*d^uRN|<(h;_0)0K$loFxL6&@l?d zz3)R;cWmPV+n6wF~pQ30q~9)&{>z zVu?f)y+z(|`JHG6xZtF+Gp<$U%@NTB4+tKLPsGQm`MBKziQ&S&%{7o89B(BAV;rln zhFiZ*$yyT$AO5AbL0*f0!p4`gFYCK12accEm=xk@$SUk_TWuH2XiWNMzcWZ6A03=X zUBn|!fREc#T;2Ie&f)liGK49;5^rv?YfEN$ab~kTr|7BD&0YP9mdl&oNAc*_qQ#B^ zTjAt-;cq4m&jQNM#EfMzqug0MklS4`eu{B5@L{X0 zH)$_`0Z63ShX=I~PW8%9SdY%~x$hHtv@f>omRm3)iT&X<6PmihsV7`Ih3tivUD(Yc zxMmgzFS{0HhDIE{9q}%|HR+p37QAHJ+L<2wS2=oWG<{*g76=Pl-LjzrV{_Ja$;Ed> zW<~eAW0*S*dX?C;EtTP+TE(llW>m{ze7u+G>ziu`Bt8~C0j2kMou^<{JWNxn9+p6I zJULSuGad;Zp92g!8`q@vUjJ=Y`b_*p$HG8m(` z6-gy~fE}hFH$C9-U=XGGmevM;J>x1lwu$3=a!MnuBu33&16#*a2zT1(| zeA}M+$ly)gxzNN4?qaawVC-`-*?4?9d2phoJ%(cL13m#zpKF%GN$jW;ye#gZI4MgUcQg&i_ zY`i(IIw`v-Eh**M#829cP}C_RxrhfynW?bp(NoxXr0>yOkbuNSV{#XE+IcSCO<^cd z&afq45q2wsg$=C>v)x`Ckt~xm?kk&mA8m}PF;@iNz@30vE33JU`-~ZdB&|3p-;0N0 zjtuG{&<9dU}k74ZD%iG*v^6HUg=OLVRtGy`S9{IGnKl0RL6}WC&2&ExwF*Nd*$N$$V%3#bid2W=*~YsACzfNc3EO>?7eO>qgh&dtDXCv1eyIGKBt%(1bqY!i3GAi#U9jqdseEL?dCy;#&Jc&$ z5}yG2GQ-jomMWbJMl=i~a19=X97e@ao$-E8@@6zX;u@2A^{~qgZBv$7qAQH@NUmc<+wF~QV{BueBGs(iT&2mTXca205)Z0g+M(UJFcG*W z4u$+Gj6gX#gCqt-UvD6+$2yJACcEA(a-L+F^0vbhZ44MxJ{Pzz%dDB+Iwmf4J?Ng^Kx zz@nT0e{S>ojNUWKuKC$D#w||dutTyWu&5? zZpg{;c@1X!xTr!&uiVV_P!cB{%Fgj_2$NXk{P-oWSN*5OlToVma&Z_TojtpwQpU-#AGP@RQ7#rW6$qQa{JxqdH#6(XI?Yk z<(%_9=X1X2{r;R!wJG_vD7&4|)twJU*GO=_un#F>S$aJOq0cWRawwiF5z&aXO62^( zFgy~?%HL^T?7p(%R9mFNTdLAPYbFkis6&S?v(98fezr@Df>FNlDMw`>uh1r^R=S_t zW!b}ktU@3^OHp`S`^3-*+7hN=pdGo|WpFb(B zZgdpj-lz;+Osu(-_-YpEbLl=SZ&(<-k9FqzWrgC8IBu&NRYNMF&#gPp{q7X*?i6lm z4CjBjw>Y#G(Go@~30@p-I~JRV}FmM>D&2Lu{m6a@?HXrP&AEGCTPjVX0qFG=0=` z_0$hHA$DKDh+QcZ*ZhoZRO{iN@D{Q|wZtrOmpJPWS(-kLf=lVz6EvP8s{w8IqI^Y+ z-&(Br@t;u|4CMv=v?S*=^wP|5fQaa|yv!)Lau|2TQ5W&sR8+XOtD(bEvaHYXU?=xQhiW)t!?Z8HlhQs2J4ihL zW$N=~)&e!+6e;HcvYyY6ZzIQKSYEWzdO>vnP@FyPGDc2<-WDIMCemAA(s*-Fap3-}Q{L5-nZk%iX9?<2 z3)EM>xJZFr?)DnH!5UMrcmV1$aSw`Fw@;_2T$X9JWPjQJ-f|NA^d2w9;ie?Hfz#0! zlN_1y-ni%-XtN4Y$>mf19b#(+0gsHqNsm)LsrJ-ebj#?KWfN}hI*%EO_FOZ&^W=;| zOklTb)5nC=S5F@CxmrknGZW7kqo@!Kn9svc6|eNSzdN^!OrTi$Jk&q+_Nhz^K7&FT zRUz{U++&a{LR`wT7(wxxsK?%s*0TnKh>;U0m-!sO!RXR-RrP~fE&01Ixb^dzmfhZe z7(uWRLp{8kAsiP$`k;3`c__Mce0hiLLUh;Rj5WmvE~uKw(1+8BMc<(Ljd1CWK{Wi} zLT(d8$k8Z4CXA8+eNf|c@=MD(FCseBe$W{ezgIc`ma1HX)my3t)M((N6seJ|UL@~$ zfao!Oc|vcF&(14}n5&C?(9eur>Uun?5{+%vC?9Fhg#&+&e!RVJVi-30M6c3CWSqJkC0{jw=Q1 z=;M;5D!7JAacEFN)yOTmx(qf4lvVRR1mobb7Hd_*otFzQa@cYv6UbK9AvL)o#hNN* zJ?8gc?4hVb(+baj$?hLia^8(;rl>Eri3Ui)&p-lA7~0S-ioLTCae&yVb+>#v>rNkA zL6O!SRld~aYhnDaIei!|7Wd&jIvS}q-l$v~_WGTKv_f{VIW^~Kf!%*CWJT;VX9z%< zqWPwWZ|r5Wb)4%!4TU=)uW}mt3@Uracc^wsRCLKVjuu9Ya#r#Vi!K)FECtSW># z{+Edm5`GX^9*6E+()>58`@=*??*b;!|M!$U2iW^*&1-!a<>Ljj7k>M7s_M*Bz)#6M zW!}2=VTjSe;Q1(Mf5J&E90=Yrzt%-$a|$6;Di+#IqpCA{?}3zTbgv7+8Ih5=AD#@j zvQa$HC6-Rs=l%qXuZLF`;5R1JuQfhWzcorNuKF`8>FBIgrCGnQ8LS3P&rZC0zrEd` z@lZnZxO?eXmfJm<_%kfz@Iy}S@RsipygTP4hv7J#dvV(+tIj;_IM^8D23UDBnwj>C zll!vQhY}pj0Gn63`F(xlBnc3MrIybr?aEc_bJLwH8r-e83uVe8m?vTGx@m27#%>vWYLpg1!D zjGFgi#~{(WNulTEET^$%8qJFwE~e3SO}OD(KDostub}d_Kw!G;cFE)@0g`F7=EP!(b{3$B0pRTH!eRklM`xW;HzZwY7xF@o zOQbqQtJ)n<%}9_+)!rCqxBYNXxwu~6soZht5@f6Sr_=W|CPa?V=kLQqz609X>h@5a zRS*%=JD?fSy#bPg&6&cAV6X9OJ@mHsa-?WgJ^Dku z_+*1VFX8c*(^Un0X7lkfO6n3XB#4^oo-zXv^Fv;#-o{eCmuiU+haV6$U}+x#!2RT3 z17>}b^U>(9`^Y-M!F2$_eUf&2T){u>GvHK(OsFH^^@rLAuy|+R;nS)!;!6Smd*z4V zAlQl!pq9IV@lkOgl$}dg`ob^}I2Ii1!R#-_D01Id-#vF>-R5J28r*F8Xh z9lLSE)nGgBsm|XUfKBmR>k8n(lMdk5z5E9UR#ruRKF3rMmai1{csyduSK0;RpcIpz zq_rYc37hp2o<~p=R-UI4=cw3qU>amX0DL~yZjL;^37#9QsMLr=L@IU|WF8~YEN$K6 z019hOk`m%gD`%1C+PPpBnxwsyzcamG!n6D%ft`}0T-3Z+=4)T}1Ib`n%HbKmZ6hNFe$Oavlj0U$RMM)Iq*e(O+sJ(GajKGX{tt zM#s+AA4xg7mWq} zn$EJ=edc&I`Ejtz+T;p)3}L!D|87!SfjLXF=MaK@h7D8l8}#vmbwye3ikl4>^p2n- zt3gzK-sdcyaJq4(Y^G9)_dpPzELAN2UBeCC#VMA}0Nj;x0N9TjsK%z;vgD)Mhl&hc znl@gk{HLeQIR-|O2^07m+pR9_Izdmd1kD3aS#m`@RDn*zUP#<1)W5<+O8POw$#+s* z23S_c#wrW;+q`B;5_gqkoPo6?byqT5!^!GZZ|@v#ng4UGmS6c6I_3GYBC+9VvXLqjvDU+r?a0<*;&}ctR%B5*7`DX<*$|#DekC_E_AD+FD`m*=#Frd52el5EPlih_OfUs zbuT9+*y>H$x+IQf6kkNP2=<1pe_i(>6&EU!-l;9YM^b1r!n+PaLL$At`C_?rKWua3Utwm^LkD zO+6QIjtibFaYSeJYre7WG_Rl{IeI;4t~b{P3-VqrK9SioX&pHArq2zXMKT|Ub8ikO zdonfI7^&F#J?jjERpfS2a_wX@p$aQas<$M{@L@79XM%`lMDf)dAe-yS(8ul8y~n)Tca- z-zJmMT{_t#B%n^{SboUISzzKB!iYv1$h18ZF&iZbe>K$WbJwoKQ*li$y@uTnk-V{N z;XBy+$$vZG@-@_hGJKyUb|2&V@vIXLgs|Qq)3b44&(vinUm*2h_B)x55#8%cP-o!< zw|fExqG+gO_fLoI70-{@O^lH6m5`kg!jwbautrIvhO!WD6j?I1mn;^Y8ObsPguh>@32uNiAVyd1s3dBsK8LaV{9_1^aRMW|^&LMMpw4 z&4XEIp7wdt%iv`8C!a*Ad+0(m9(aEX{-0^6;ypJvz?eM$O7ePkmw$xm<_>n&9qAER zm3;7O?y;YhjOQzSEjh9;yD*z*CwUc>PuyskS0Q{*^v4uvy%^GR38=zY)XvHbt<}8` zWQ}=bEFgPJgj-(u_Gg<(1mLxJD_F6XMf453XUgr-hPcHJt2L+F<@2e5G-6hTlK{m!6v9iu7*`i&c8HKq|3Il#2Z1kK^wr^Z5$)l4!s*2bND`&fTahl z9DsIE(#cOMf(*0^&S|fXKHJ!PUC{zA4U*EP8@158|p~(-5%k6xr5D zUH1q6-3?@~X^f0yO$4w2)BzLd$cFhd)`AVT_@#=X)|?4!PnjrjsO)SmqWj!NiUE=u zqN}h~iy#exGzTQZJPYWR#%19iS3NuAw%cp0cnE-=y4U@`r>+grFJv7j3HRVQ4uIgv zm?duZ&K8*BJIwy;+cLg)`>q*X>DW4S^%sL+@d3n8)%?t$!e&cz$JfBV9tYmY4x_D~ z?CS;714!?J)y#HW3!(x=(nxP`5NR`5(?sB(C@YN5RVYa8x6|jhrv)mByB!@6fg^!h zjh-q_Pb;@W@=*cc7tRPs_5GTvtnLlD*QZG@h-$0bKFuvTj&4qR>a7cT+cmvQkn6Q4 z?k3fyVk=+?IZFs;RwMUIE{bg5-oK_k8~4GK2p480B=$4DIk?qAw!Wm|#u63XZnm3m zAOF83M?J8UH|V?9_S)-Q`Pn1=(z^e3nE&w)j{wJ}&`TH71eME2kZj=Z>}j)8FAW?+ F{{zEKaF+l8 diff --git a/docs/_modules/ice/anomaly_detection/datasets.html b/docs/_modules/ice/anomaly_detection/datasets.html index da93e55..b6df286 100644 --- a/docs/_modules/ice/anomaly_detection/datasets.html +++ b/docs/_modules/ice/anomaly_detection/datasets.html @@ -414,6 +414,30 @@

    Source code for ice.anomaly_detection.datasets

    + + +
    [docs]class AnomalyDetectionRiethTEP(BaseDataset): + """ + Dataset of Tennessee Eastman Process dataset + Rieth, C. A., Amsel, B. D., Tran, R., & Cook, M. B. (2017). + Additional Tennessee Eastman Process Simulation Data for + Anomaly Detection Evaluation (Version V1) [Computer software]. + Harvard Dataverse. + https://doi.org/10.7910/DVN/6C3JR1. + """ + def __init__(self, num_chunks=None, force_download=False): + self.df = None + self.test_mask = None + self.name = None + self.public_link = None + self.set_name_public_link() + self._load(num_chunks, force_download) + self.target[self.target != 0] = 1 + self.train_mask[self.target != 0] = False + +
    diff --git a/docs/_modules/ice/anomaly_detection/metrics.html b/docs/_modules/ice/anomaly_detection/metrics.html index ac961de..1f94cf9 100644 --- a/docs/_modules/ice/anomaly_detection/metrics.html +++ b/docs/_modules/ice/anomaly_detection/metrics.html @@ -366,19 +366,56 @@

    Source code for ice.anomaly_detection.metrics

    -
    [docs]def accuracy(pred: list, target: list) -> float:
+import numpy as np
+from sklearn.metrics import confusion_matrix
+
+
+
    [docs]def accuracy(pred: np.ndarray, target: np.ndarray) -> float:
     """
     Accuracy of the classification is the number of true positives divided by 
     the number of examples.
 
     Args:
-        pred (list): predictions.
-        target (list): target values.
+        pred (np.ndarray): predictions.
+        target (np.ndarray): target values.
     
     Returns:
-        float: accuracy
+        float: accuracy.
     """
     return sum(pred == target) / len(pred)
    
+
+
+
    [docs]def true_positive_rate(pred: np.ndarray, target: np.ndarray) -> np.ndarray[float]:
+    """
+    True Positive Rate is the number of detected faults i divided by the 
+    number of faults i.
+
+    Args:
+        pred (np.ndarray): predictions.
+        target (np.ndarray): target values.
+    
+    Returns:
+        list: list of float values with true positive rate for each fault.
+    """
+    cm = confusion_matrix(target, pred, labels=np.arange(target.max() + 1))
+    correct = cm[1:, 1:].diagonal()
+    return list(correct / cm[1:].sum(axis=1))
    
+
+
+
    [docs]def false_positive_rate(pred: np.ndarray, target: np.ndarray) -> np.ndarray[float]:
+    """
+    False Positive Rate, aka False Alarm Rate is the number of false alarms i 
+    divided by the number of normal samples.
+
+    Args:
+        pred (np.ndarray): predictions.
+        target (np.ndarray): target values.
+    
+    Returns:
+        list: list of float values with true positive rate for each fault.
+    """
+    cm = confusion_matrix(target, pred, labels=np.arange(target.max() + 1))
+    return list(cm[0, 1:] / cm[0].sum())
    diff --git a/docs/_modules/ice/anomaly_detection/models/autoencoder.html b/docs/_modules/ice/anomaly_detection/models/autoencoder.html index 84bb12e..34a334d 100644 --- a/docs/_modules/ice/anomaly_detection/models/autoencoder.html +++ b/docs/_modules/ice/anomaly_detection/models/autoencoder.html @@ -366,8 +366,7 @@

    Source code for ice.anomaly_detection.models.autoencoder

    -from pandas import DataFrame
-from sklearn.preprocessing import StandardScaler
+import pandas as pd
 from torch import nn
 
 from ice.anomaly_detection.models.base import BaseAnomalyDetection
@@ -376,48 +375,41 @@ 
    Source code for ice.anomaly_detection.models.autoencoder
    [docs]class MLP(nn.Module):
     def __init__(
             self,
-            num_sensors: int,
+            input_dim: int,
             window_size: int,
-            num_layers: int,
             hidden_dims: list,
-            type: str
+            decoder: bool = False,
             ):
         super().__init__()
-        self.num_sensors = num_sensors
+        self.input_dim = input_dim
         self.window_size = window_size
-        self.hidden_dims = [num_sensors * window_size]
-        self.type = type
-        if hidden_dims is not None and self.type == 'encoder':
-            self.hidden_dims = self.hidden_dims + hidden_dims
-        elif hidden_dims is not None and self.type == 'decoder':
+        self.hidden_dims = [input_dim * window_size]
+        self.decoder = decoder
+        if self.decoder:
             self.hidden_dims = hidden_dims + self.hidden_dims
         else:
-            for i in range(num_layers):
-                dim = self.hidden_dims[i] // 2
-                self.hidden_dims.append(dim)
-            if self.type == 'decoder':
-                self.hidden_dims.reverse()
+            self.hidden_dims = self.hidden_dims + hidden_dims
         self.mlp = nn.Sequential(nn.Flatten())
-        for i in range(num_layers):
+        for i in range(len(hidden_dims)):
             self.mlp.append(nn.Linear(
                                 self.hidden_dims[i],
                                 self.hidden_dims[i + 1])
                                 )
-            if self.type == 'decoder' and i + 1 == num_layers:
+            if self.decoder and i + 1 == len(hidden_dims):
                 break
             self.mlp.append(nn.ReLU())
 
 
    [docs]    def forward(self, x):
         output = self.mlp(x)
-        if self.type == 'decoder':
-            return output.view(-1, self.window_size, self.num_sensors)
+        if self.decoder:
+            return output.view(-1, self.window_size, self.input_dim)
 
         return output
    
 
 
 
    [docs]class AutoEncoderMLP(BaseAnomalyDetection):
     """
-    Autoencoder (AE) consists of encoder and decoder parts. Each
+    MLP autoencoder consists of MLP encoder and MLP decoder parts. Each
     sample is reshaped to a vector (B, L, C) -> (B, L * C) for calculations
     and to a vector (B, L * C) -> (B, L, C) for the output. Where B is the
     batch size, L is the sequence length, C is the number of sensors.
@@ -425,18 +417,24 @@ 
    Source code for ice.anomaly_detection.models.autoencoder
    def __init__(
             self,
             window_size: int,
+            stride: int = 1,
             batch_size: int = 128,
             lr: float = 0.001,
             num_epochs: int = 10,
             device: str = 'cpu',
             verbose: bool = False,
             name: str = 'ae_anomaly_detection',
-            threshold: float = 0.95,
-            hidden_dims: list = [256, 128, 64],
+            random_seed: int = 42,
+            val_ratio: float = 0.15,
+            save_checkpoints: bool = False,
+            threshold_level: float = 0.95,
+            hidden_dims: list = [256, 128, 64]
             ):
         """
         Args:
             window_size (int): The window size to train the model.
+            stride (int): The time interval between first points of consecutive 
+                sliding windows in training.
             batch_size (int): The batch size to train the model.
             lr (float): The larning rate to train the model.
             num_epochs (float): The number of epochs to train the model.
@@ -444,20 +442,19 @@ 
    Source code for ice.anomaly_detection.models.autoencoder
                    `cuda` are possible.
             verbose (bool): If true, show the progress bar in training.
             name (str): The name of the model for artifact storing.
-            threshold (float): The boundary for anomaly detection.
+            random_seed (int): Seed for random number generation to ensure reproducible results.
+            val_ratio (float): Proportion of the dataset used for validation, between 0 and 1.
+            save_checkpoints (bool): If true, store checkpoints.
+            threshold_level (float): Takes a value from 0 to 1. It specifies
+                the quantile in the distribution of errors on the training
+                dataset at which the threshold value is set.
             hidden_dims (list): Dimensions of hidden layers in encoder/decoder.
         """
         super().__init__(
-            window_size, batch_size, lr, num_epochs, device, verbose, name, threshold
+            window_size, stride, batch_size, lr, num_epochs, device, verbose, name, random_seed, 
+            val_ratio, save_checkpoints, threshold_level
         )
-
-        self.window_size = window_size
-        self.num_layers = len(hidden_dims)
         self.hidden_dims = hidden_dims
-        self.threshold = threshold
-        self.loss_fn = nn.MSELoss(reduction='mean')
-        self.preprocessing = True
-        self.scaler = StandardScaler()
 
     _param_conf_map = dict(BaseAnomalyDetection._param_conf_map,
             **{
@@ -465,24 +462,20 @@ 
    Source code for ice.anomaly_detection.models.autoencoder
    }
         )
 
-    def _create_model(self, df: DataFrame):
-        num_sensors = df.shape[1]
+    def _create_model(self, input_dim: int, output_dim: int):
         self.model = nn.Sequential(
             MLP(
-                num_sensors,
+                input_dim,
                 self.window_size,
-                self.num_layers,
                 hidden_dims=self.hidden_dims,
-                type='encoder'
             ),
             MLP(
-                num_sensors,
+                input_dim,
                 self.window_size,
-                self.num_layers,
                 hidden_dims=self.hidden_dims[::-1],
-                type='decoder'
+                decoder=True
             )
-            )
    
+        )
    diff --git a/docs/_modules/ice/anomaly_detection/models/base.html b/docs/_modules/ice/anomaly_detection/models/base.html index 714b56e..71c0600 100644 --- a/docs/_modules/ice/anomaly_detection/models/base.html +++ b/docs/_modules/ice/anomaly_detection/models/base.html @@ -369,15 +369,16 @@

    Source code for ice.anomaly_detection.models.base

    from abc import ABC, abstractmethod import numpy as np import pandas as pd -from tqdm.auto import trange, tqdm - +from tqdm.auto import tqdm import torch -from torch.optim import Adam +import torch.nn as nn from torch.utils.data import DataLoader +from torch.optim import Adam +import optuna -from ice.anomaly_detection.utils import SlidingWindowDataset -from ice.base import BaseModel -from ice.anomaly_detection.metrics import accuracy +from ice.base import BaseModel, SlidingWindowDataset +from ice.anomaly_detection.metrics import ( + accuracy, true_positive_rate, false_positive_rate)
    [docs]class BaseAnomalyDetection(BaseModel, ABC): @@ -387,17 +388,23 @@

    Source code for ice.anomaly_detection.models.base

    def __init__( self, window_size: int, + stride: int, batch_size: int, lr: float, num_epochs: int, device: str, verbose: bool, name: str, - threshold: float = 0.95 - ): + random_seed: int, + val_ratio: float, + save_checkpoints: bool, + threshold_level: float = 0.95 + ): """ Args: window_size (int): The window size to train the model. + stride (int): The time interval between first points of consecutive + sliding windows in training. batch_size (int): The batch size to train the model. lr (float): The learning rate to train the model. num_epochs (float): The number of epochs to train the model. @@ -405,105 +412,98 @@

    Source code for ice.anomaly_detection.models.base

    `cuda` are possible. verbose (bool): If true, show the progress bar in training. name (str): The name of the model for artifact storing. + random_seed (int): Seed for random number generation to ensure reproducible results. + val_ratio (float): Proportion of the dataset used for validation, between 0 and 1. + save_checkpoints (bool): If true, store checkpoints. + threshold_level (float): Takes a value from 0 to 1. It specifies + the quantile in the distribution of errors on the training + dataset at which the threshold value is set. """ - super().__init__(batch_size, lr, num_epochs, device, verbose, name) - self._cfg.path_set(["TASK"], "anomaly_detection") + super().__init__(window_size, stride, batch_size, lr, num_epochs, device, verbose, name, random_seed, val_ratio, save_checkpoints) + self.val_metrics = False + + self.threshold_level = threshold_level + self.threshold_value = None - self.window_size = window_size - self.loss_fn = None - self.preprocessing = False - self.scaler = None - self.threshold = None +
    [docs] def fit(self, df: pd.DataFrame, target: pd.Series = None, + epochs: int = None, save_path: str = None, trial: optuna.Trial = None, + force_model_ctreation: bool = False): + """Fit (train) the model by a given dataset. + + Args: + df (pandas.DataFrame): A dataframe with sensor data. Index has + two columns: `run_id` and `sample`. All other columns a value of + sensors. + target (pandas.Series): A series with target values. Indes has two + columns: `run_id` and `sample`. It is omitted for anomaly + detection task. + epochs (int): The number of epochs for training step. If None, + self.num_epochs parameter is used. + save_path (str): Path to save checkpoints. If None, the path is + created automatically. + """ + if trial: + super().fit(df, target, epochs, save_path, trial=trial, force_model_ctreation=True) + else: + super().fit(df, target, epochs, save_path) + + error = [] + for sample, target in tqdm( + self.dataloader, desc='Steps ...', leave=False, disable=(not self.verbose) + ): + sample = sample.to(self.device) + with torch.no_grad(): + pred = self.model(sample) + error.append(self.loss_no_reduction(pred, sample).mean(dim=(1, 2))) + error = torch.concat(error) + self.threshold_value = torch.quantile(error, self.threshold_level).item() + if self.save_checkpoints: + self.save_checkpoint(save_path)
    _param_conf_map = dict(BaseModel._param_conf_map, **{ - "window_size": ["MODEL", "WINDOW_SIZE"], - "threshold": ["MODEL", "THRESHOLD"] + "threshold_level": ["MODEL", "THRESHOLD_LEVEL"] } ) - -
    [docs] def fit(self, df: pd.DataFrame): - """ Method fit for training anomaly detection models. + +
    [docs] def load_checkpoint(self, checkpoint_path: str): + """Load checkpoint. Args: - df (pd.DataFrame): data without anomaly states + checkpoint_path (str): Path to load checkpoint. """ - assert len(df) >= self.window_size, "window size is larger than the length of df." - if self.preprocessing: - self.scaler.fit(df) - df.loc[:] = self.scaler.transform(df) - self._create_model(df) - self._train_nn(df)
    - - def _fit(self, df: pd.DataFrame): - pass + super().load_checkpoint(checkpoint_path) + self.threshold_value = self._cfg['MODEL']['THRESHOLD_VALUE']
    + + def _prepare_for_training(self, input_dim: int, output_dim: int): + self.optimizer = Adam(self.model.parameters(), lr=self.lr) + self.loss_fn = nn.L1Loss() + self.loss_no_reduction = nn.L1Loss(reduction='none') def _predict(self, sample: torch.Tensor) -> torch.Tensor: input = sample.to(self.device) output = self.model(input) - return output.cpu() - - def _train_nn(self, df: pd.DataFrame): - self.model.train() - self.model.to(self.device) - self.optimizer = Adam(self.model.parameters(), lr=self.lr) - - dataset = SlidingWindowDataset(df, window_size=self.window_size) - self.dataloader = DataLoader(dataset, batch_size=self.batch_size, shuffle=True) - errors = [] - for e in trange(self.num_epochs, desc='Epochs ...', disable=(not self.verbose)): - for sample in tqdm(self.dataloader, desc='Steps ...', leave=False, disable=(not self.verbose)): - input = sample.to(self.device) - output = self.model(input) - loss = self.loss_fn(output, input) - self.optimizer.zero_grad() - loss.backward() - self.optimizer.step() - error = torch.sum(torch.abs(input - output), dim=(1, 2)) - errors = np.append(errors, error.detach().numpy()) - if self.verbose: - print(f'Epoch {e+1}, Loss: {loss.item():.4f}') - self.threshold_value = np.quantile(errors, self.threshold) - -
    [docs] def evaluate(self, df: pd.DataFrame, target: pd.Series) -> dict: - """Evaluate the metrics: accuracy. - - Args: - df (pandas.DataFrame): A dataframe with sensor data. Index has - two columns: `run_id` and `sample`. All other columns a value of - sensors. - target (pandas.Series): A series with target values. Indes has two - columns: `run_id` and `sample`. + error = self.loss_no_reduction(output, input).mean(dim=(1, 2)) + return (error > self.threshold_value).float().cpu() + + def _validate_inputs(self, df: pd.DataFrame, target: pd.Series): + if target is not None: + assert df.shape[0] == target.shape[0], f"target is incompatible with df by the length: {df.shape[0]} and {target.shape[0]}." + assert np.all(df.index == target.index), "target's index and df's index are not the same." + assert df.index.names == (['run_id', 'sample']), "An index should contain columns `run_id` and `sample`." + assert len(df) >= self.window_size, "window size is larger than the length of df." - Returns: - dict: A dictionary with metrics where keys are names of metrics and - values are values of metrics. - """ - assert df.shape[0] == target.shape[0], f"target is incompatible with df by the length: {df.shape[0]} and {target.shape[0]}." - assert np.all(df.index == target.index), "target's index and df's index are not the same." - assert df.index.names == (['run_id', 'sample']), "An index should contain columns `run_id` and `sample`." - - if self.preprocessing: - df.loc[:] = self.scaler.transform(df) - dataset = SlidingWindowDataset(df, window_size=self.window_size, target=target) - self.dataloader = DataLoader(dataset, batch_size=self.batch_size, shuffle=True) - target, pred = [], [] - for sample, _target in tqdm( - self.dataloader, desc='Steps ...', leave=False, disable=(not self.verbose) - ): - input = sample.to(self.device) - target.append(_target) - with torch.no_grad(): - output = self.predict(input) - error = torch.sum(torch.abs(input - output), dim=(1, 2)) - pred.append((error > self.threshold_value).float()) - target = torch.concat(target).numpy() - pred = torch.concat(pred).numpy() + def _calculate_metrics(self, pred: torch.tensor, target: torch.tensor) -> dict: metrics = { - 'accuracy': accuracy(pred, target) + 'accuracy': accuracy(pred, target), + 'true_positive_rate': true_positive_rate(pred, target), + 'false_positive_rate': false_positive_rate(pred, target), } - self._store_atrifacts_inference(metrics) - return metrics
    + return metrics + + def _set_dims(self, df: pd.DataFrame, target: pd.Series): + self.input_dim = df.shape[1] + self.output_dim = 1
    diff --git a/docs/_modules/ice/anomaly_detection/models/gnn.html b/docs/_modules/ice/anomaly_detection/models/gnn.html new file mode 100644 index 0000000..700432f --- /dev/null +++ b/docs/_modules/ice/anomaly_detection/models/gnn.html @@ -0,0 +1,607 @@ + + + + + + + + + + + ice.anomaly_detection.models.gnn — ICE documentation + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
    + + + + + + + + + + + +
    +
    +
    +
    + + + Ctrl+K +
    +
    + + + +
    +
    + +
    + + + +
    + + +
    + +
    + +
    + + + +
    + +
    +
    + +
    + +
    + + +
    +
    + + + + +
    + +
    + + +
    +
    + +
    +
    + +
    + +
    + + + + +
    + +
    + + +
    +
    + + + + + +
    + +

    Source code for ice.anomaly_detection.models.gnn

    +import torch
+from torch import nn
+from torch.nn import functional as F
+from pandas import DataFrame, Series
+
+from ice.anomaly_detection.models.base import BaseAnomalyDetection
+
+
+class GCLayer(nn.Module):
+    def __init__(
+            self,
+            in_dim: int,
+            out_dim: int
+            ):
+        super().__init__()
+        self.dense = nn.Linear(in_dim, out_dim)
+    
+    def forward(self, adj, x):
+        adj = adj + torch.eye(adj.size(0)).to(x.device)
+        x = self.dense(x)
+        norm = adj.sum(1)**(-1/2)
+        x = norm[None, :] * adj * norm[:, None] @ x
+
+        return x
+
+
+class Directed_A(nn.Module):
+    def __init__(
+            self,
+            num_sensors: int,
+            window_size: int,
+            alpha: float,
+            k: int
+            ):
+        super().__init__()
+        self.alpha = alpha
+        self.k = k
+
+        self.e1 = nn.Embedding(num_sensors, window_size)
+        self.e2 = nn.Embedding(num_sensors, window_size)
+        self.l1 = nn.Linear(window_size,window_size)
+        self.l2 = nn.Linear(window_size,window_size)
+    
+    def forward(self, idx):
+        m1 = torch.tanh(self.alpha*self.l1(self.e1(idx)))
+        m2 = torch.tanh(self.alpha*self.l2(self.e2(idx)))
+        adj = F.relu(torch.tanh(self.alpha*torch.mm(m1, m2.transpose(1,0))))
+        
+        if self.k:
+            mask = torch.zeros(idx.size(0), idx.size(0)).to(idx.device)
+            mask.fill_(float('0'))
+            s1,t1 = (adj + torch.rand_like(adj)*0.01).topk(self.k,1)
+            mask.scatter_(1,t1,s1.fill_(1))
+            adj = adj*mask
+        
+        return adj
+
+
+class GNNEncoder(nn.Module):
+    def __init__(
+            self,
+            num_sensors: int,
+            window_size: int,
+            alpha: float,
+            k: int
+            ):
+        super().__init__()
+        self.idx = torch.arange(num_sensors)
+        self.gcl1 = GCLayer(window_size, window_size // 2)
+        self.gcl2 = GCLayer(window_size // 2, window_size // 8)
+        self.A = Directed_A(num_sensors, window_size, alpha, k)
+    
+    def forward(self, x):
+        x = torch.transpose(x, 1, 2)
+        adj = self.A(self.idx.to(x.device))
+        x = self.gcl1(adj, x).relu()
+        x = self.gcl2(adj, x).relu()
+        return x
+    
+
+class Decoder(nn.Module):
+    def __init__(
+            self,
+            num_sensors: int,
+            window_size: int
+            ):
+        super().__init__()
+        self.num_sensors = num_sensors
+        self.window_size = window_size
+        self.decoder = nn.Sequential(
+            nn.Linear(window_size // 8 * num_sensors, window_size // 2 * num_sensors),
+            nn.ReLU(),
+            nn.Linear(window_size // 2 * num_sensors, num_sensors * window_size)
+        )
+    
+    def forward(self, x):
+        x = torch.flatten(x,1)
+        x = self.decoder(x)
+
+        return x.view(-1, self.window_size, self.num_sensors)
+
+
+
    [docs]class GSL_GNN(BaseAnomalyDetection):
+    """
+    GNN autoencoder consists of encoder with graph convolutional layers 
+    and MLP decoder parts. The graph describing the data is constructed 
+    during the training process using trainable parameters.
+    """
+    def __init__(
+            self,
+            window_size: int,
+            stride: int = 1,
+            batch_size: int = 128,
+            lr: float = 0.001,
+            num_epochs: int = 10,
+            device: str = 'cpu',
+            verbose: bool = False,
+            name: str = 'gnn_anomaly_detection',
+            random_seed: int = 42,
+            val_ratio: float = 0.15,
+            save_checkpoints: bool = False,
+            threshold_level: float = 0.95,
+            alpha: float = 0.2,
+            k: int = None
+        ):
+        """
+        Args:
+            window_size (int): The window size to train the model.
+            stride (int): The time interval between first points of consecutive 
+                sliding windows in training.
+            batch_size (int): The batch size to train the model.
+            lr (float): The larning rate to train the model.
+            num_epochs (float): The number of epochs to train the model.
+            device (str): The name of a device to train the model. `cpu` and 
+                `cuda` are possible.
+            verbose (bool): If true, show the progress bar in training.
+            name (str): The name of the model for artifact storing.
+            random_seed (int): Seed for random number generation to ensure reproducible results.
+            val_ratio (float): Proportion of the dataset used for validation, between 0 and 1.
+            save_checkpoints (bool): If true, store checkpoints.
+            threshold_level (float): Takes a value from 0 to 1. It specifies
+                the quantile in the distribution of errors on the training
+                dataset at which the threshold value is set.
+            alpha (float): Saturation rate for adjacency matrix.
+            k (int): Limit on the number of edges in the adjacency matrix.
+        """
+        super().__init__(
+            window_size, stride, batch_size, lr, num_epochs, device, verbose, name, random_seed, 
+            val_ratio, save_checkpoints, threshold_level
+        )
+        self.alpha = alpha
+        self.k = k
+    
+    _param_conf_map = dict(BaseAnomalyDetection._param_conf_map,
+            **{
+                "alpha": ["MODEL", "ALPHA"]
+            }
+        )
+
+    def _create_model(self, input_dim: int, output_dim: int):
+        self.model = nn.Sequential(
+            GNNEncoder(
+                input_dim,
+                self.window_size,
+                self.alpha,
+                self.k
+                ),
+            Decoder(input_dim, self.window_size)
+            )
    
+
    + +
    + + + + + +
    + +
    +
    +
    + +
    + + + + +
    +
    + +
    + +
    +
    +
    + + + + + +
    +
    + +
    + +
    +

    + + © Copyright 2024, ICE Team. +
    + +

    +
    + +
    +

    + Created using Sphinx 5.3.0. +
    +

    +
    + +
    + + + +
    + +

    + Built with the PyData Sphinx Theme 0.14.0. +

    + +
    + +
    + +
    + + \ No newline at end of file diff --git a/docs/_modules/ice/anomaly_detection/models/stgat.html b/docs/_modules/ice/anomaly_detection/models/stgat.html new file mode 100644 index 0000000..a6dacec --- /dev/null +++ b/docs/_modules/ice/anomaly_detection/models/stgat.html @@ -0,0 +1,733 @@ + + + + + + + + + + + ice.anomaly_detection.models.stgat — ICE documentation + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
    + + + + + + + + + + + +
    +
    +
    +
    + + + Ctrl+K +
    +
    + + + +
    +
    + +
    + + + +
    + + +
    + +
    + +
    + + + +
    + +
    +
    + +
    + +
    + + +
    +
    + + + + +
    + +
    + + +
    +
    + +
    +
    + +
    + +
    + + + + +
    + +
    + + +
    +
    + + + + + +
    + +

    Source code for ice.anomaly_detection.models.stgat

    +import pandas as pd
+import torch
+import torch.nn as nn
+from torch.nn import functional as F
+from torch_geometric.nn import GCNConv, GATConv
+
+from ice.anomaly_detection.models.base import BaseAnomalyDetection
+
+
+"""
+The code of Stgat-Mad is taken from:
+https://github.com/wagner-d/TimeSeAD
+"""
+
+
+class InputLayer(nn.Module):
+    """1-D Convolution layer to extract high-level features of each time-series input
+    :param n_features: Number of input features/nodes
+    :param window_size: length of the input sequence
+    :param kernel_size: size of kernel to use in the convolution operation
+    """
+    def __init__(self, n_features, kernel_size=7):
+        super(InputLayer, self).__init__()
+        self.padding = nn.ConstantPad1d((kernel_size - 1) // 2, 0.0)
+        self.conv = nn.Conv1d(in_channels=n_features, out_channels=n_features, kernel_size=kernel_size)
+        self.relu = nn.ReLU()
+
+    def forward(self, x):
+        x = x.permute(0, 2, 1)
+        x = self.padding(x)
+        x = self.relu(self.conv(x))
+        return x.permute(0, 2, 1)  # Permute back
+
+
+class StgatBlock(nn.Module):
+    def __init__(self, n_features, window_size, dropout, embed_dim=None):
+        super(StgatBlock, self).__init__()
+        self.n_features = n_features
+        self.window_size = window_size
+        self.dropout = dropout
+        self.embed_dim = embed_dim if embed_dim is not None else n_features
+
+        self.embed_dim *= 2
+
+        self.feature_gat_layers = GATConv(window_size, window_size)
+        self.temporal_gat_layers = GATConv(n_features, n_features)
+
+        self.temporal_gcn_layers = GCNConv(n_features, n_features)
+
+    def forward(self, data, fc_edge_index, tc_edge_index):
+        # x shape (b, n, k): b - batch size, n - window size, k - number of features
+        x = data.clone().detach()
+        x = x.permute(0, 2, 1)
+        batch_num, node_num, all_feature = x.shape
+
+        x = x.reshape(-1, all_feature).contiguous()
+        f_out = self.feature_gat_layers(x, fc_edge_index)
+        f_out = F.relu(f_out)
+        f_out = f_out.view(batch_num, node_num, -1)
+        f_out = f_out.permute(0, 2, 1)
+        z = f_out.reshape(-1, node_num).contiguous()
+
+        t_out = self.temporal_gcn_layers(z, tc_edge_index)
+        t_out = F.relu(t_out)
+        t_out = t_out.view(batch_num, node_num, -1)
+
+        return t_out.permute(0, 2, 1)
+
+
+class BiLSTMLayer(nn.Module):
+    def __init__(self, in_dim, hid_dim, n_layers, dropout):
+        super(BiLSTMLayer, self).__init__()
+        self.hid_dim = hid_dim
+        self.n_layers = n_layers
+        self.dropout = 0.0 if n_layers == 1 else dropout
+        self.bilstm = nn.LSTM(in_dim, hid_dim, num_layers=n_layers, batch_first=True, dropout=self.dropout, bidirectional=True)
+
+    def forward(self, x):
+        out, h = self.bilstm(x)
+        out = out.permute(1,0,2)[-1, :, :] # Extracting from last layer
+        return out
+
+
+class BiLSTMDecoder(nn.Module):
+    def __init__(self, in_dim, hid_dim, n_layers, dropout):
+        super(BiLSTMDecoder, self).__init__()
+        self.in_dim = in_dim
+        self.dropout = 0.0 if n_layers == 1 else dropout
+        self.bilstm = nn.LSTM(in_dim, hid_dim, num_layers=n_layers, batch_first=True, dropout=self.dropout, bidirectional=True)
+
+    def forward(self, x):
+        decoder_out, _ = self.bilstm(x)
+        return decoder_out
+
+
+class ReconstructionModel(nn.Module):
+    def __init__(self, window_size, in_dim, hid_dim, out_dim, n_layers, dropout):
+        super(ReconstructionModel, self).__init__()
+        self.window_size = window_size
+        self.decoder = BiLSTMDecoder(in_dim, hid_dim, n_layers, dropout)
+        self.fc = nn.Linear(2 * hid_dim, out_dim)
+
+    def forward(self, x):
+        # x will be last hidden state of the GRU layer
+        h_end = x
+        h_end_rep = h_end.repeat_interleave(self.window_size, dim=1).view(x.size(0), self.window_size, -1)
+        decoder_out = self.decoder(h_end_rep)
+        out = self.fc(decoder_out)
+        return out
+
+
+def get_batch_edge_index(org_edge_index, batch_num, node_num):
+    # org_edge_index:(2, edge_num)
+    edge_index = org_edge_index.clone().detach()
+    edge_num = org_edge_index.shape[1]
+    batch_edge_index = edge_index.repeat(1,batch_num).contiguous()
+
+    for i in range(batch_num):
+        batch_edge_index[:, i*edge_num:(i+1)*edge_num] += i*node_num
+
+    return batch_edge_index.long()
+
+# graph is 'fully-connect'
+def get_fc_graph_struc(n_features):
+    edge_indices = torch.tensor([[i, j] for j in range(n_features) for i in range(n_features) if i != j])
+    return edge_indices.T.contiguous()
+
+
+def get_tc_graph_struc(temporal_len):
+    edge_indices = torch.tensor([[i, j] for j in range(temporal_len) for i in range(j)])
+    return edge_indices.T.contiguous()
+
+
+
    [docs]class STGAT(nn.Module):
+    def __init__(
+        self,
+        n_features,
+        window_size,
+        embed_dim,
+        layer_numb,
+        lstm_n_layers,
+        lstm_hid_dim,
+        recon_n_layers,
+        recon_hid_dim,
+        dropout
+    ):
+        super(STGAT, self).__init__()
+
+        layers1 = []
+        layers2 = []
+        layers3 = []
+
+        self.layer_numb = layer_numb
+        self.h_temp = []
+
+        self.input_1 = InputLayer(n_features, 1)
+        self.input_2 = InputLayer(n_features, 5)
+        self.input_3 = InputLayer(n_features, 7)
+
+        for i in range(layer_numb):
+            layers1 += [StgatBlock(n_features, window_size, dropout, embed_dim)]
+        for i in range(layer_numb):
+            layers2 += [StgatBlock(n_features, window_size, dropout, embed_dim)]
+        for i in range(layer_numb):
+            layers3 += [StgatBlock(n_features, window_size, dropout, embed_dim)]
+
+        self.stgat_1 = nn.Sequential(*layers1)
+        self.stgat_2 = nn.Sequential(*layers2)
+        self.stgat_3 = nn.Sequential(*layers3)
+
+        self.bilstm = BiLSTMLayer(n_features * 3, lstm_hid_dim, lstm_n_layers, dropout)
+        self.recon_model = ReconstructionModel(window_size, 2 * lstm_hid_dim, recon_hid_dim, n_features, recon_n_layers, dropout)
+
+        # Register as buffers so that tensors are moved to the correct device along with the rest of the model
+        self.register_buffer('fc_edge_index', get_fc_graph_struc(n_features), persistent=False)
+        self.register_buffer('tc_edge_index', get_tc_graph_struc(window_size), persistent=False)
+
+
    [docs]    def forward(self, x):
+        # x shape (b, n, k): b - batch size, n - window size, k - number of features
+        fc_edge_index_sets = get_batch_edge_index(self.fc_edge_index, x.shape[0], x.shape[2])
+        tc_edge_index_sets = get_batch_edge_index(self.tc_edge_index, x.shape[0], x.shape[1])
+
+        x_1 = x
+        x_2 = self.input_2(x)
+        x_3 = self.input_3(x)
+
+        for layer in range(self.layer_numb):
+            if layer==0:
+                h_cat_1 = x_1 + self.stgat_1[layer](x_1, fc_edge_index_sets, tc_edge_index_sets)
+                h_cat_2 = x_2 + self.stgat_2[layer](x_2, fc_edge_index_sets, tc_edge_index_sets)
+                h_cat_3 = x_3 + self.stgat_3[layer](x_3, fc_edge_index_sets, tc_edge_index_sets)
+            else:
+                h_cat_1 = h_cat_1 + self.stgat_1[layer](h_cat_1, fc_edge_index_sets, tc_edge_index_sets)
+                h_cat_2 = h_cat_2 + self.stgat_2[layer](h_cat_2, fc_edge_index_sets, tc_edge_index_sets)
+                h_cat_3 = h_cat_3 + self.stgat_3[layer](h_cat_3, fc_edge_index_sets, tc_edge_index_sets)
+
+        h_cat = torch.cat([h_cat_1, h_cat_2, h_cat_3], dim=2)
+
+        out_end = self.bilstm(h_cat)
+        h_end = out_end.view(x.shape[0], -1)   # Hidden state for last timestamp
+
+        recons = self.recon_model(h_end)
+
+        return recons
    
+
+
+class STGAT_MAD(BaseAnomalyDetection):
+    """
+    Stgat-Mad was presented at ICASSP 2022: "Stgat-Mad : Spatial-Temporal Graph 
+    Attention Network For Multivariate Time Series Anomaly Detection".
+    https://ieeexplore.ieee.org/abstract/document/9747274/
+    """
+    def __init__(
+            self,
+            window_size: int,
+            stride: int = 1,
+            batch_size: int = 128,
+            lr: float = 0.001,
+            num_epochs: int = 10,
+            device: str = 'cpu',
+            verbose: bool = False,
+            name: str = 'stgat_anomaly_detection',
+            random_seed: int = 42,
+            val_ratio: float = 0.15,
+            save_checkpoints: bool = False,
+            threshold_level: float = 0.95,
+            embed_dim: int = None,
+            layer_numb: int = 2,
+            lstm_n_layers: int = 1,
+            lstm_hid_dim: int = 150,
+            recon_n_layers: int = 1,
+            recon_hid_dim: int = 150,
+            dropout: float = 0.2
+            ):
+        """
+        Args:
+            window_size (int): The window size to train the model.
+            stride (int): The time interval between first points of consecutive 
+                sliding windows in training.
+            batch_size (int): The batch size to train the model.
+            lr (float): The larning rate to train the model.
+            num_epochs (float): The number of epochs to train the model.
+            device (str): The name of a device to train the model. `cpu` and 
+                `cuda` are possible.
+            verbose (bool): If true, show the progress bar in training.
+            name (str): The name of the model for artifact storing.
+            random_seed (int): Seed for random number generation to ensure reproducible results.
+            val_ratio (float): Proportion of the dataset used for validation, between 0 and 1.
+            save_checkpoints (bool): If true, store checkpoints.
+            threshold_level (float): Takes a value from 0 to 1. It specifies
+                the quantile in the distribution of errors on the training
+                dataset at which the threshold value is set.
+            embed_dim (int) : Embedding dimension.
+            layer_numb (int) : Number of layers.
+            lstm_n_layers (int) : Number of LSTM layers.
+            lstm_hid_dim (int) : Hidden dimension of LSTM layers.
+            recon_n_layers (int) : Number of reconstruction layers.
+            recon_hid_dim (int) : Hidden dimension of reconstruction layers.
+            dropout (float) : The rate of dropout.
+        """
+        super().__init__(
+            window_size, stride, batch_size, lr, num_epochs, device, verbose, name, random_seed, 
+            val_ratio, save_checkpoints, threshold_level
+        )
+        self.embed_dim = embed_dim
+        self.layer_numb = layer_numb
+        self.lstm_n_layers = lstm_n_layers
+        self.lstm_hid_dim = lstm_hid_dim
+        self.recon_n_layers = recon_n_layers
+        self.recon_hid_dim = recon_hid_dim
+        self.dropout = dropout
+
+    _param_conf_map = dict(BaseAnomalyDetection._param_conf_map,
+            **{
+                "layer_numb": ["MODEL", "LAYER_NUMB"],
+                "lstm_n_layers": ["MODEL", "LSTM_N_LAYERS"],
+                "lstm_hid_dim": ["MODEL", "LSTM_HID_DIM"],
+                "recon_n_layers": ["MODEL", "RECON_N_LAYERS"],
+                "recon_hid_dim": ["MODEL", "RECON_HID_DIM"],
+                "dropout": ["MODEL", "DROPOUT"]
+            }
+        )
+
+    def _create_model(self, input_dim: int, output_dim: int):
+        self.model = STGAT(
+            n_features=input_dim,
+            window_size=self.window_size,
+            embed_dim=self.embed_dim,
+            layer_numb=self.layer_numb,
+            lstm_n_layers=self.lstm_n_layers,
+            lstm_hid_dim=self.lstm_hid_dim,
+            recon_n_layers=self.recon_n_layers,
+            recon_hid_dim=self.recon_hid_dim,
+            dropout=self.dropout
+            )
+
    + +
    + + + + + +
    + +
    +
    +
    + +
    + + + + +
    +
    + +
    + +
    +
    +
    + + + + + +
    +
    + +
    + +
    +

    + + © Copyright 2024, ICE Team. +
    + +

    +
    + +
    +

    + Created using Sphinx 5.3.0. +
    +

    +
    + +
    + + + +
    + +

    + Built with the PyData Sphinx Theme 0.14.0. +

    + +
    + +
    + +
    + + \ No newline at end of file diff --git a/docs/_modules/ice/anomaly_detection/models/transformer.html b/docs/_modules/ice/anomaly_detection/models/transformer.html new file mode 100644 index 0000000..65be837 --- /dev/null +++ b/docs/_modules/ice/anomaly_detection/models/transformer.html @@ -0,0 +1,778 @@ + + + + + + + + + + + ice.anomaly_detection.models.transformer — ICE documentation + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
    + + + + + + + + + + + +
    +
    +
    +
    + + + Ctrl+K +
    +
    + + + +
    +
    + +
    + + + +
    + + +
    + +
    + +
    + + + +
    + +
    +
    + +
    + +
    + + +
    +
    + + + + +
    + +
    + + +
    +
    + +
    +
    + +
    + +
    + + + + +
    + +
    + + +
    +
    + + + + + +
    + +

    Source code for ice.anomaly_detection.models.transformer

    +import torch
+from torch import nn
+import torch.nn.functional as F
+import numpy as np
+import math
+
+from ice.anomaly_detection.models.base import BaseAnomalyDetection
+
+
+"""
+The code of Anomaly Transformer is taken from official implementation:
+https://github.com/thuml/Anomaly-Transformer
+"""
+class TriangularCausalMask():
+    def __init__(self, B, L, device="cpu"):
+        mask_shape = [B, 1, L, L]
+        with torch.no_grad():
+            self._mask = torch.triu(torch.ones(mask_shape, dtype=torch.bool), diagonal=1).to(device)
+
+    @property
+    def mask(self):
+        return self._mask
+
+
+
    [docs]class AnomalyAttention(nn.Module):
+    def __init__(self, win_size, mask_flag=True, scale=None, attention_dropout=0.0, output_attention=True, device="cpu"):
+        super(AnomalyAttention, self).__init__()
+        self.scale = scale
+        self.mask_flag = mask_flag
+        self.output_attention = output_attention
+        self.dropout = nn.Dropout(attention_dropout)
+        window_size = win_size
+        self.distances = torch.zeros((window_size, window_size)).to(device)
+        for i in range(window_size):
+            for j in range(window_size):
+                self.distances[i][j] = abs(i - j)
+
+
    [docs]    def forward(self, queries, keys, values, sigma, attn_mask):
+        B, L, H, E = queries.shape
+        _, S, _, D = values.shape
+        scale = self.scale or 1. / math.sqrt(E)
+
+        scores = torch.einsum("blhe,bshe->bhls", queries, keys)
+        if self.mask_flag:
+            if attn_mask is None:
+                attn_mask = TriangularCausalMask(B, L, device=queries.device)
+            scores.masked_fill_(attn_mask.mask, -np.inf)
+        attn = scale * scores
+
+        sigma = sigma.transpose(1, 2)  # B L H ->  B H L
+        window_size = attn.shape[-1]
+        sigma = torch.sigmoid(sigma * 5) + 1e-5
+        sigma = torch.pow(3, sigma) - 1
+        sigma = sigma.unsqueeze(-1).repeat(1, 1, 1, window_size)  # B H L L
+        prior = self.distances.unsqueeze(0).unsqueeze(0).repeat(sigma.shape[0], sigma.shape[1], 1, 1).to(queries.device)
+        prior = 1.0 / (math.sqrt(2 * math.pi) * sigma) * torch.exp(-prior ** 2 / 2 / (sigma ** 2))
+
+        series = self.dropout(torch.softmax(attn, dim=-1))
+        V = torch.einsum("bhls,bshd->blhd", series, values)
+
+        if self.output_attention:
+            return (V.contiguous(), series, prior, sigma)
+        else:
+            return (V.contiguous(), None)
    
+
+
+
    [docs]class AttentionLayer(nn.Module):
+    def __init__(self, attention, d_model, n_heads, d_keys=None,
+                 d_values=None):
+        super(AttentionLayer, self).__init__()
+
+        d_keys = d_keys or (d_model // n_heads)
+        d_values = d_values or (d_model // n_heads)
+        self.norm = nn.LayerNorm(d_model)
+        self.inner_attention = attention
+        self.query_projection = nn.Linear(d_model,
+                                          d_keys * n_heads)
+        self.key_projection = nn.Linear(d_model,
+                                        d_keys * n_heads)
+        self.value_projection = nn.Linear(d_model,
+                                          d_values * n_heads)
+        self.sigma_projection = nn.Linear(d_model,
+                                          n_heads)
+        self.out_projection = nn.Linear(d_values * n_heads, d_model)
+
+        self.n_heads = n_heads
+
+
    [docs]    def forward(self, queries, keys, values, attn_mask):
+        B, L, _ = queries.shape
+        _, S, _ = keys.shape
+        H = self.n_heads
+        x = queries
+        queries = self.query_projection(queries).view(B, L, H, -1)
+        keys = self.key_projection(keys).view(B, S, H, -1)
+        values = self.value_projection(values).view(B, S, H, -1)
+        sigma = self.sigma_projection(x).view(B, L, H)
+
+        out, series, prior, sigma = self.inner_attention(
+            queries,
+            keys,
+            values,
+            sigma,
+            attn_mask
+        )
+        out = out.view(B, L, -1)
+
+        return self.out_projection(out), series, prior, sigma
    
+
+
+
    [docs]class PositionalEmbedding(nn.Module):
+    def __init__(self, d_model, max_len=5000):
+        super(PositionalEmbedding, self).__init__()
+        # Compute the positional encodings once in log space.
+        pe = torch.zeros(max_len, d_model).float()
+        pe.require_grad = False
+
+        position = torch.arange(0, max_len).float().unsqueeze(1)
+        div_term = (torch.arange(0, d_model, 2).float() * -(math.log(10000.0) / d_model)).exp()
+
+        pe[:, 0::2] = torch.sin(position * div_term)
+        pe[:, 1::2] = torch.cos(position * div_term)
+
+        pe = pe.unsqueeze(0)
+        self.register_buffer('pe', pe)
+
+
    [docs]    def forward(self, x):
+        return self.pe[:, :x.size(1)]
    
+
+
+
    [docs]class TokenEmbedding(nn.Module):
+    def __init__(self, c_in, d_model):
+        super(TokenEmbedding, self).__init__()
+        padding = 1 if torch.__version__ >= '1.5.0' else 2
+        self.tokenConv = nn.Conv1d(in_channels=c_in, out_channels=d_model,
+                                   kernel_size=3, padding=padding, padding_mode='circular', bias=False)
+        for m in self.modules():
+            if isinstance(m, nn.Conv1d):
+                nn.init.kaiming_normal_(m.weight, mode='fan_in', nonlinearity='leaky_relu')
+
+
    [docs]    def forward(self, x):
+        x = self.tokenConv(x.permute(0, 2, 1)).transpose(1, 2)
+        return x
    
+
+
+
    [docs]class DataEmbedding(nn.Module):
+    def __init__(self, c_in, d_model, dropout=0.0):
+        super(DataEmbedding, self).__init__()
+
+        self.value_embedding = TokenEmbedding(c_in=c_in, d_model=d_model)
+        self.position_embedding = PositionalEmbedding(d_model=d_model)
+
+        self.dropout = nn.Dropout(p=dropout)
+
+
    [docs]    def forward(self, x):
+        x = self.value_embedding(x) + self.position_embedding(x)
+        return self.dropout(x)
    
+
+
+
    [docs]class EncoderLayer(nn.Module):
+    def __init__(self, attention, d_model, d_ff=None, dropout=0.1, activation="relu"):
+        super(EncoderLayer, self).__init__()
+        d_ff = d_ff or 4 * d_model
+        self.attention = attention
+        self.conv1 = nn.Conv1d(in_channels=d_model, out_channels=d_ff, kernel_size=1)
+        self.conv2 = nn.Conv1d(in_channels=d_ff, out_channels=d_model, kernel_size=1)
+        self.norm1 = nn.LayerNorm(d_model)
+        self.norm2 = nn.LayerNorm(d_model)
+        self.dropout = nn.Dropout(dropout)
+        self.activation = F.relu if activation == "relu" else F.gelu
+
+
    [docs]    def forward(self, x, attn_mask=None):
+        new_x, attn, mask, sigma = self.attention(
+            x, x, x,
+            attn_mask=attn_mask
+        )
+        x = x + self.dropout(new_x)
+        y = x = self.norm1(x)
+        y = self.dropout(self.activation(self.conv1(y.transpose(-1, 1))))
+        y = self.dropout(self.conv2(y).transpose(-1, 1))
+
+        return self.norm2(x + y), attn, mask, sigma
    
+
+
+
    [docs]class Encoder(nn.Module):
+    def __init__(self, attn_layers, norm_layer=None):
+        super(Encoder, self).__init__()
+        self.attn_layers = nn.ModuleList(attn_layers)
+        self.norm = norm_layer
+
+
    [docs]    def forward(self, x, attn_mask=None):
+        # x [B, L, D]
+        series_list = []
+        prior_list = []
+        sigma_list = []
+        for attn_layer in self.attn_layers:
+            x, series, prior, sigma = attn_layer(x, attn_mask=attn_mask)
+            series_list.append(series)
+            prior_list.append(prior)
+            sigma_list.append(sigma)
+
+        if self.norm is not None:
+            x = self.norm(x)
+
+        return x, series_list, prior_list, sigma_list
    
+
+
+
    [docs]class AnomalyTransformerModel(nn.Module):
+    def __init__(
+            self,
+            win_size,
+            enc_in,
+            c_out,
+            d_model,
+            n_heads,
+            e_layers,
+            d_ff,
+            dropout,
+            activation,
+            device
+            ):
+        super(AnomalyTransformerModel, self).__init__()
+
+        # Encoding
+        self.embedding = DataEmbedding(enc_in, d_model, dropout)
+
+        # Encoder
+        self.encoder = Encoder(
+            [
+                EncoderLayer(
+                    AttentionLayer(
+                        AnomalyAttention(win_size, False, attention_dropout=dropout, device=device),
+                        d_model, n_heads),
+                    d_model,
+                    d_ff,
+                    dropout=dropout,
+                    activation=activation
+                ) for l in range(e_layers)
+            ],
+            norm_layer=torch.nn.LayerNorm(d_model)
+        )
+
+        self.projection = nn.Linear(d_model, c_out, bias=True)
+
+
    [docs]    def forward(self, x):
+        enc_out = self.embedding(x)
+        enc_out, series, prior, sigmas = self.encoder(enc_out)
+        enc_out = self.projection(enc_out)
+
+        return enc_out  # [B, L, D]
    
+
+
+
    [docs]class AnomalyTransformer(BaseAnomalyDetection):
+    """
+    Anomaly Transformer was presented at ICLR 2022: "Anomaly Transformer:
+    Time Series Anomaly Detection with Association Discrepancy".
+    https://openreview.net/forum?id=LzQQ89U1qm_
+    """
+    def __init__(
+            self,
+            window_size: int = 100,
+            stride: int = 1,
+            batch_size: int = 128,
+            lr: float = 0.0001,
+            num_epochs: int = 10,
+            device: str = 'cpu',
+            verbose: bool = False,
+            name: str = 'transformer_anomaly_detection',
+            random_seed: int = 42,
+            val_ratio: float = 0.15,
+            save_checkpoints: bool = False,
+            threshold_level: float = 0.95,
+            d_model: int = 256,
+            n_heads: int = 8,
+            e_layers: int = 3,
+            d_ff: int = 256,
+            dropout: float = 0.0,
+            activation: str = 'gelu'
+            ):
+        """
+        Args:
+            window_size (int): The window size to train the model.
+            stride (int): The time interval between first points of consecutive 
+                sliding windows in training.
+            batch_size (int): The batch size to train the model.
+            lr (float): The larning rate to train the model.
+            num_epochs (float): The number of epochs to train the model.
+            device (str): The name of a device to train the model. `cpu` and 
+                `cuda` are possible.
+            verbose (bool): If true, show the progress bar in training.
+            name (str): The name of the model for artifact storing.
+            random_seed (int): Seed for random number generation to ensure reproducible results.
+            val_ratio (float): Proportion of the dataset used for validation, between 0 and 1.
+            save_checkpoints (bool): If true, store checkpoints.
+            threshold_level (float): Takes a value from 0 to 1. It specifies
+                the quantile in the distribution of errors on the training
+                dataset at which the threshold value is set.
+            threshold_value (float): Threshold value is calculated after the model is trained.
+                It sets the error limit above which the data sample defines as anomaly.
+            d_model (int): Dimension of model.
+            n_heads (int): Number of heads.
+            e_layers (int): Number of encoder layers.
+            d_ff (int): Dimension of MLP.
+            dropout (float): The rate of dropout.
+            activation (str): Activation ('relu', 'gelu').
+        """
+        super().__init__(
+            window_size, stride, batch_size, lr, num_epochs, device, verbose, name, random_seed, 
+            val_ratio, save_checkpoints, threshold_level
+        )
+        self.d_model = d_model
+        self.n_heads = n_heads
+        self.e_layers = e_layers
+        self.d_ff = d_ff
+        self.dropout = dropout
+        self.activation = activation
+
+    _param_conf_map = dict(BaseAnomalyDetection._param_conf_map,
+            **{
+                "d_model": ["MODEL", "D_MODEL"],
+                "n_heads": ["MODEL", "N_HEADS"],
+                "e_layers": ["MODEL", "E_LAYERS"],
+                "d_ff": ["MODEL", "D_FF"],
+                "dropout": ["MODEL", "DROPOUT"],
+                "activation": ["MODEL", "ACTIVATION"]
+            }
+        )
+
+    def _create_model(self, input_dim: int, output_dim: int):
+        self.model = AnomalyTransformerModel(
+            win_size=self.window_size,
+            enc_in=input_dim,
+            c_out=input_dim,
+            d_model=self.d_model,
+            n_heads=self.n_heads,
+            e_layers=self.e_layers,
+            d_ff=self.d_ff,
+            dropout=self.dropout,
+            activation=self.activation,
+            device=self.device
+            )
    
+
    + +
    + + + + + +
    + +
    +
    +
    + +
    + + + + +
    +
    + +
    + +
    +
    +
    + + + + + +
    +
    + +
    + +
    +

    + + © Copyright 2024, ICE Team. +
    + +

    +
    + +
    +

    + Created using Sphinx 5.3.0. +
    +

    +
    + +
    + + + +
    + +

    + Built with the PyData Sphinx Theme 0.14.0. +

    + +
    + +
    + +
    + + \ No newline at end of file diff --git a/docs/_modules/ice/base.html b/docs/_modules/ice/base.html index 69f01bd..4870df6 100644 --- a/docs/_modules/ice/base.html +++ b/docs/_modules/ice/base.html @@ -371,17 +371,22 @@

    Source code for ice.base

     import pandas as pd
 import numpy as np
 import torch
-from tqdm.auto import tqdm
+from tqdm.auto import tqdm, trange
 import os
 import zipfile
 import requests
 import datetime
 import json
+import random
 from ice.configs import Config
+import time
+from torch.utils.data import DataLoader, Dataset, random_split
+import optuna
 
 
 
    [docs]class BaseDataset(ABC):
     """Base class for datasets."""
+
     def __init__(self, num_chunks=None, force_download=False):
         """
         Args:
@@ -397,40 +402,55 @@ 
    Source code for ice.base
             self.public_link = None
         self.set_name_public_link()
         self._load(num_chunks, force_download)
-    
+
 
-    
+
     def _load(self, num_chunks, force_download):
         """Load the dataset in self.df and self.target."""
-        ref_path = f'data/{self.name}/'
+        ref_path = f"data/{self.name}/"
         if not os.path.exists(ref_path):
             os.makedirs(ref_path)
-        zfile_path = f'data/{self.name}.zip'
+        zfile_path = f"data/{self.name}.zip"
 
         url = self._get_url(self.public_link)
         if not os.path.exists(zfile_path) or force_download:
             self._download_pgbar(url, zfile_path, self.name, num_chunks)
-            
+
         self._extracting_files(zfile_path, ref_path)
-        self.df = self._read_csv_pgbar(ref_path + 'df.csv', index_col=['run_id', 'sample'])
-        self.target = self._read_csv_pgbar(ref_path + 'target.csv', index_col=['run_id', 'sample'])['target']
-        self.train_mask = self._read_csv_pgbar(ref_path + 'train_mask.csv', index_col=['run_id', 'sample'])['train_mask']
+        self.df = self._read_csv_pgbar(
+            ref_path + "df.csv", index_col=["run_id", "sample"]
+        )
+        self.target = self._read_csv_pgbar(
+            ref_path + "target.csv", index_col=["run_id", "sample"]
+        )["target"]
+        self.train_mask = self._read_csv_pgbar(
+            ref_path + "train_mask.csv", index_col=["run_id", "sample"]
+        )["train_mask"]
         self.train_mask = self.train_mask.astype(bool)
         self.test_mask = ~self.train_mask
-    
+
     def _get_url(self, public_link):
-        r = requests.get(f'https://cloud-api.yandex.net/v1/disk/public/resources?public_key={public_link}')
-        return r.json()['file']
+        url = ""
+        r = requests.get(
+            f"https://cloud-api.yandex.net/v1/disk/public/resources?public_key={public_link}"
+        )
+        if r.status_code == 200:
+            url = r.json()["file"]
+        else:
+            raise Exception(r.json()["description"])
+        return url
 
-    def _read_csv_pgbar(self, csv_path, index_col, chunk_size=1024*100):
-        rows = sum(1 for _ in open(csv_path, 'r')) - 1
+    def _read_csv_pgbar(self, csv_path, index_col, chunk_size=1024 * 100):
+        rows = sum(1 for _ in open(csv_path, "r")) - 1
         chunk_list = []
-        with tqdm(total=rows, desc=f'Reading {csv_path}') as pbar:
-            for chunk in pd.read_csv(csv_path, index_col=index_col, chunksize=chunk_size):
+        with tqdm(total=rows, desc=f"Reading {csv_path}") as pbar:
+            for chunk in pd.read_csv(
+                csv_path, index_col=index_col, chunksize=chunk_size
+            ):
                 chunk_list.append(chunk)
                 pbar.update(len(chunk))
         df = pd.concat((f for f in chunk_list), axis=0)
@@ -439,14 +459,15 @@ 
    Source code for ice.base
         def _download_pgbar(self, url, zfile_path, fname, num_chunks):
         resp = requests.get(url, stream=True)
         total = int(resp.headers.get("Content-Length"))
-        with open(zfile_path, 'wb') as file: 
+        with open(zfile_path, "wb") as file:
             with tqdm(
                 total=total,
-                desc=f'Downloading {fname}',
-                unit='B',
+                desc=f"Downloading {fname}",
+                unit="B",
                 unit_scale=True,
-                unit_divisor=1024) as pbar:
-                i = 0    
+                unit_divisor=1024,
+            ) as pbar:
+                i = 0
                 for data in resp.iter_content(chunk_size=1024):
                     if num_chunks is not None and num_chunks == i:
                         break
@@ -454,20 +475,21 @@ 
    Source code for ice.base
                         pbar.update(len(data))
                     i += 1
 
-    def _extracting_files(self, zfile_path, ref_path, block_size=1024*10000):
-        with zipfile.ZipFile(zfile_path, 'r') as zfile:
+    def _extracting_files(self, zfile_path, ref_path, block_size=1024 * 10000):
+        with zipfile.ZipFile(zfile_path, "r") as zfile:
             for entry_info in zfile.infolist():
                 if os.path.exists(ref_path + entry_info.filename):
                     continue
                 input_file = zfile.open(entry_info.filename)
-                target_file = open(ref_path + entry_info.filename, 'wb')
+                target_file = open(ref_path + entry_info.filename, "wb")
                 block = input_file.read(block_size)
                 with tqdm(
-                    total=entry_info.file_size, 
-                    desc=f'Extracting {entry_info.filename}', 
-                    unit='B', 
-                    unit_scale=True, 
-                    unit_divisor=1024) as pbar:
+                    total=entry_info.file_size,
+                    desc=f"Extracting {entry_info.filename}",
+                    unit="B",
+                    unit_scale=True,
+                    unit_divisor=1024,
+                ) as pbar:
                     while block:
                         target_file.write(block)
                         block = input_file.read(block_size)
@@ -480,38 +502,57 @@ 
    Source code for ice.base
         """Base class for all models."""
 
     _param_conf_map = {
-            "batch_size" : ["DATASET", "BATCH_SIZE"],
-            "lr" : ["OPTIMIZATION", "LR"],
-            "num_epochs" : ["OPTIMIZATION", "NUM_EPOCHS"],
-            "verbose" : ["VERBOSE"],
-            "device" : ["DEVICE"],
-            "name" : ["EXPERIMENT_NAME"]
-        }
+        "batch_size": ["DATASET", "BATCH_SIZE"],
+        "lr": ["OPTIMIZATION", "LR"],
+        "num_epochs": ["OPTIMIZATION", "NUM_EPOCHS"],
+        "verbose": ["VERBOSE"],
+        "device": ["DEVICE"],
+        "name": ["EXPERIMENT_NAME"],
+        "window_size": ["MODEL", "WINDOW_SIZE"],
+        "stride": ["MODEL", "STRIDE"],
+        "val_ratio": ["DATASET", "VAL_RATIO"],
+        "random_seed": ["SEED"],
+    }
 
     @abstractmethod
     def __init__(
-            self, 
-            batch_size: int, 
-            lr: float, 
-            num_epochs: int, 
-            device: str, 
-            verbose: bool,
-            name: str
-        ):
+        self,
+        window_size: int,
+        stride: int,
+        batch_size: int,
+        lr: float,
+        num_epochs: int,
+        device: str,
+        verbose: bool,
+        name: str,
+        random_seed: int,
+        val_ratio: float,
+        save_checkpoints: bool,
+    ):
         """
         Args:
+            window_size (int): The window size to train the model.
+            stride (int): The time interval between first points of consecutive
+                sliding windows in training.
             batch_size (int): The batch size to train the model.
             lr (float): The larning rate to train the model.
             num_epochs (float): The number of epochs to train the model.
-            device (str): The name of a device to train the model. `cpu` and 
+            device (str): The name of a device to train the model. `cpu` and
                 `cuda` are possible.
             verbose (bool): If true, show the progress bar in training.
             name (str): The name of the model for artifact storing.
+            random_seed (int): Seed for random number generation to ensure reproducible results.
+            val_ratio (float): Proportion of the dataset used for validation, between 0 and 1.
+            save_checkpoints (bool): If true, store checkpoints.
         """
         self._cfg = Config()
         self._cfg.path_set(["MODEL", "NAME"], self.__class__.__name__)
         self._output_dir = "outputs"
 
+        self.window_size = window_size
+        self.val_ratio = val_ratio
+        self.val_metrics = None
+        self.stride = stride
         self.batch_size = batch_size
         self.lr = lr
         self.num_epochs = num_epochs
@@ -519,17 +560,29 @@ 
    Source code for ice.base
             self.verbose = verbose
         self.model = None
         self.name = name
+        self.random_seed = random_seed
+        self.save_checkpoints = save_checkpoints
+        self.input_dim = None
+        self.output_dim = None
+        self.train_time = "no date"
+        self.checkpoint_epoch = 0
+        self.direction = "minimize"
 
     def __setattr__(self, __name: str, __value: Any):
-        if __name not in ['cfg', 'param_conf_map'] and __name in self._param_conf_map.keys():
+        if (
+            __name not in ["cfg", "param_conf_map"]
+            and __name in self._param_conf_map.keys()
+        ):
             self._cfg.path_set(self._param_conf_map[__name], __value)
 
         super().__setattr__(__name, __value)
         if __name == "name":
-            self._training_path, self._inference_path = self._initialize_paths()
-    
+            self._training_path, self._inference_path, self._checkpoints_path = (
+                self._initialize_paths()
+            )
+
 
    [docs]    @classmethod
-    def from_config(cls, cfg : Config):
+    def from_config(cls, cfg: Config):
         """Create instance of the model class with parameters from config.
 
         Args:
@@ -543,63 +596,48 @@ 
    Source code for ice.base
     
         for key in cls._param_conf_map.keys():
             param_dict[key] = cfg.path_get(cls._param_conf_map[key])
-        
+
         return cls(**param_dict)
    
 
-
    [docs]    def fit(self, df: pd.DataFrame, target: pd.Series):
+
    [docs]    def fit(
+        self,
+        df: pd.DataFrame,
+        target: pd.Series = None,
+        epochs: int = None,
+        save_path: str = None,
+        trial: optuna.Trial = None,
+        force_model_ctreation: bool = False,
+    ):
         """Fit (train) the model by a given dataset.
 
         Args:
-            df (pandas.DataFrame): A dataframe with sensor data. Index has 
-                two columns: `run_id` and `sample`. All other columns a value of 
+            df (pandas.DataFrame): A dataframe with sensor data. Index has
+                two columns: `run_id` and `sample`. All other columns a value of
                 sensors.
-            target (pandas.Series): A series with target values. Indes has two
-                columns: `run_id` and `sample`.
+            target (pandas.Series): A series with target values. Index has two
+                columns: `run_id` and `sample`. It is omitted for anomaly
+                detection task.
+            epochs (int): The number of epochs for training step. If None,
+                self.num_epochs parameter is used.
+            save_path (str): Path to save checkpoints. If None, the path is
+                created automatically.
+            trial (optuna.Trial, None): optuna.Trial object created by optimize method.
+            force_model_ctreation (bool): force fit to create model for optimization study.
         """
-        assert df.shape[0] == target.shape[0], f"target is incompatible with df by the length: {df.shape[0]} and {target.shape[0]}."
-        assert np.all(df.index == target.index), "target's index and df's index are not the same."
-        assert df.index.names == (['run_id', 'sample']), "An index should contain columns `run_id` and `sample`."
-        self._create_model(df, target)
-        assert self.model is not None, "Model creation error."
-        self._fit(df, target)
+        self.train_time = datetime.datetime.now().strftime("%Y-%m-%d-%H-%M-%S")
+        if epochs is None:
+            epochs = self.num_epochs
+        self._validate_inputs(df, target)
+        if self.model is None or force_model_ctreation:
+            self._set_dims(df, target)
+            self._create_model(self.input_dim, self.output_dim)
+            assert self.model is not None, "Model creation error."
+            self._prepare_for_training(self.input_dim, self.output_dim)
+        self._train_nn(
+            df=df, target=target, epochs=epochs, save_path=save_path, trial=trial
+        )
         self._store_atrifacts_train()
    
 
-    def _initialize_paths(self):
-        artifacts_path = os.path.join(self._output_dir, self.name)
-        training_path = os.path.join(artifacts_path, 'training')
-        inference_path = os.path.join(artifacts_path, 'inference')
-
-        return training_path, inference_path
-    
-    def _store_atrifacts_train(self):
-        save_path = os.path.join(self._training_path, datetime.datetime.now().strftime("%Y-%m-%d-%H-%M-%S"))
-        os.makedirs(save_path, exist_ok= True)
-        self._cfg.to_yaml(os.path.join(save_path, 'config.yaml'))
-
-    def _store_atrifacts_inference(self, metrics):
-        save_path = os.path.join(self._inference_path, datetime.datetime.now().strftime("%Y-%m-%d-%H-%M-%S"))
-        os.makedirs(save_path, exist_ok= True)
-
-        self._cfg.to_yaml(os.path.join(save_path, 'config.yaml'))
-        with open(os.path.join(save_path, "metrics.json"), 'w') as json_file:
-            json.dump(metrics, json_file)
-
-    @abstractmethod
-    def _create_model(self, df: pd.DataFrame, target: pd.Series):
-        """
-        This method has to be implemented by all children. Create a torch 
-        model for traing and prediction. 
-        """
-        pass
-
-    @abstractmethod
-    def _fit(self, df: pd.DataFrame, target: pd.Series):
-        """
-        This method has to be implemented by all children. Fit (train) the model 
-        by a given dataset.
-        """
-        pass
-
 
    [docs]    @torch.no_grad()
     def predict(self, sample: torch.Tensor) -> torch.Tensor:
         """Make a prediction for a given batch of samples.
@@ -608,7 +646,7 @@ 
    Source code for ice.base
                 sample (torch.Tensor): A tensor of the shape (B, L, C) where
                 B is the batch size, L is the sequence length, C is the number
                 of sensors.
-        
+
         Returns:
             torch.Tensor: A tensor with predictions of the shape (B,).
         """
@@ -616,32 +654,492 @@ 
    Source code for ice.base
             self.model.to(self.device)
         return self._predict(sample)
    
 
-    @abstractmethod
-    def _predict(self, sample: torch.Tensor) -> torch.Tensor:
-        """
-        This method has to be implemented by all children. Make a prediction 
-        for a given batch of samples.
+
    [docs]    def optimize(
+        self,
+        df: pd.DataFrame,
+        target: pd.Series = None,
+        optimize_parameter: str = "batch_size",
+        optimize_range: tuple = (128, 256),
+        direction: str = "minimize",
+        n_trials: int = 5,
+        epochs: int = None,
+        optimize_metric: str = None,
+    ):
+        """Make the optuna study to return the best hyperparameter value on validation dataset
+
+        Args:
+            df (pd.DataFrame): DataFrame to use method fit
+            optimize_parameter (str, optional): Model parameter to optimize. Defaults to 'batch_size'.
+            optimize_range (tuple, optional): Model parameter range for optuna trials. Defaults to (128, 256).
+            n_trials (int, optional): number of trials. Defaults to 5.
+            target (pd.Series, optional): target pd.Series to use method fit. Defaults to None.
+            epochs (int, optional): Epoch number to use method fit. Defaults to None.
+            optimize_metric (str): Metric on validation dataset to use as a target for hyperparameter optimization.
+            direction (str): "minimize" or "maximize" the target for hyperparameter optimization
+
         """
-        pass
-    
-
    [docs]    @abstractmethod
+        param_type = type(self.__dict__[optimize_parameter])
+        self.direction = direction
+
+        defaults_torch_backends = (
+            torch.backends.cudnn.deterministic,
+            torch.backends.cudnn.benchmark,
+        )
+        self.dump = self._training_path
+
+        # make torch deterministic behavior
+        torch.backends.cudnn.deterministic = True
+        torch.backends.cudnn.benchmark = False
+        torch.use_deterministic_algorithms(True)
+        os.environ["CUBLAS_WORKSPACE_CONFIG"] = ":16:8"
+
+        if not epochs:
+            epochs = self.num_epochs
+
+        def objective(trial):
+
+            self._training_path = self.dump
+            """optuna objective
+
+            Args:
+                trial (optuna.Trial): optuna trial object
+
+            Raises:
+                AssertionError: Returns if optimize_parameter value is not numerical
+
+            Returns:
+                float: best validation loss to perform optimization step
+            """
+            if param_type == float:
+                suggested_param = trial.suggest_float(
+                    optimize_parameter, optimize_range[0], optimize_range[1]
+                )
+            elif param_type == int:
+                suggested_param = trial.suggest_int(
+                    optimize_parameter, optimize_range[0], optimize_range[1]
+                )
+            elif optimize_parameter == "lr":
+                suggested_param = trial.suggest_loguniform(
+                    optimize_parameter, optimize_range[0], optimize_range[1]
+                )
+            else:
+                raise AssertionError(f"{optimize_parameter} is not int or float value")
+
+            setattr(self, optimize_parameter, suggested_param)
+
+            if optimize_metric:
+                self.val_metrics = True
+
+            self._training_path = (
+                self._training_path + f"/parameter_{optimize_parameter} optimization"
+            )
+            os.makedirs(self._training_path, exist_ok=True)
+
+            self.checkpoint_epoch = 0
+            print(f"trial step with {optimize_parameter} = {suggested_param}")
+            self.fit(
+                df=df,
+                epochs=epochs,
+                target=target,
+                trial=trial,
+                force_model_ctreation=True,
+            )
+            # use the best key metric on validation dataset from training
+            # if there is no such metric, use the best validation loss
+
+            if optimize_metric:
+                return self.best_validation_metrics[
+                    optimize_metric
+                ]  # dict with all best metric achieved during training -> write it
+            else:
+                return self.best_val_loss
+
+        study = optuna.create_study(
+            direction=direction,
+            pruner=optuna.pruners.PercentilePruner(25.0),
+            study_name=f"/parameter_{optimize_parameter} study",
+        )
+        study.optimize(objective, n_trials=n_trials)
+        self._training_path = self.dump
+
+        print(f"Best hyperparameters: {study.best_params}")
+        print(f"Best trial: {study.best_trial}")
+
+        df_trials = study.trials_dataframe()
+        df_trials.to_csv(
+            self._training_path
+            + f"/parameter_{optimize_parameter} optimization"
+            + f"/parameter_{optimize_parameter}.csv",
+            index=False,
+        )
+
+        # restore standard torch deterministic values
+        torch.backends.cudnn.deterministic, torch.backends.cudnn.benchmark = (
+            defaults_torch_backends
+        )
+        del os.environ["CUBLAS_WORKSPACE_CONFIG"]
+        torch.use_deterministic_algorithms(False)
    
+
+
    [docs]    @torch.no_grad()
     def evaluate(self, df: pd.DataFrame, target: pd.Series) -> dict:
-        """This method has to be implemented by all children. Evaluate the 
-        metrics. Docstring has to be rewritten so that all metrics are clearly 
-        described. 
+        """Evaluate the metrics: accuracy.
 
         Args:
-            df (pandas.DataFrame): A dataframe with sensor data. Index has 
-                two columns: `run_id` and `sample`. All other columns a value of 
+            df (pandas.DataFrame): A dataframe with sensor data. Index has
+                two columns: `run_id` and `sample`. All other columns a value of
                 sensors.
             target (pandas.Series): A series with target values. Indes has two
                 columns: `run_id` and `sample`.
-        
+
         Returns:
             dict: A dictionary with metrics where keys are names of metrics and
                 values are values of metrics.
         """
-        pass
    
+        self._validate_inputs(df, target)
+        dataset = SlidingWindowDataset(df, target, window_size=self.window_size)
+        self.dataloader = DataLoader(dataset, batch_size=self.batch_size, shuffle=False)
+        target, pred = [], []
+        for sample, _target in tqdm(
+            self.dataloader, desc="Steps ...", leave=False, disable=(not self.verbose)
+        ):
+            sample = sample.to(self.device)
+            target.append(_target)
+            pred.append(self.predict(sample))
+        target = torch.concat(target).numpy()
+        pred = torch.concat(pred).numpy()
+        metrics = self._calculate_metrics(pred, target)
+        self._store_atrifacts_inference(metrics)
+        return metrics
    
+
+
    [docs]    @torch.no_grad()
+    def model_param_estimation(self):
+        """Calculate number of self.model parameters, mean and std for inference time
+
+        Returns:
+            tuple: A tuple containing the number of parameters in the
+                   model and the mean and standard deviation of model inference time.
+        """
+        assert (
+            self.model != None
+        ), "use model.fit() to create fitted model object before"
+        sample = iter(self.dataloader).__next__()
+        if len(sample) == 2:
+            x, y = sample
+        else:
+            x = sample
+
+        dummy_input = torch.randn(1, x.shape[1], x.shape[2]).to(self.device)
+        repetitions = 500
+        times = np.zeros((repetitions, 1))
+
+        if self.device == "cuda":
+            starter, ender = torch.cuda.Event(enable_timing=True), torch.cuda.Event(
+                enable_timing=True
+            )
+
+            for i in range(repetitions):
+                starter.record()
+                _ = self.model(dummy_input)
+                ender.record()
+                torch.cuda.synchronize()
+                curr_time = starter.elapsed_time(ender)
+                times[i] = curr_time
+
+        else:
+            for i in range(repetitions):
+                start_time = time.time()
+                _ = self.model(dummy_input)
+                end_time = time.time()
+
+                curr_time = (end_time - start_time) * 1000  # Convert to milliseconds
+                times[i] = curr_time
+
+        mean_inference_time = np.sum(times) / repetitions
+        std_inference_time = np.std(times)
+
+        num_params = sum(p.numel() for p in self.model.parameters())
+
+        return num_params, (mean_inference_time, std_inference_time)
    
+
+    @abstractmethod
+    def _create_model(self, input_dim: int, output_dim: int):
+        """
+        This method has to be implemented by all children. Create a torch
+        model for traing and prediction.
+        """
+        pass
+
+    @abstractmethod
+    def _prepare_for_training(self, input_dim: int, output_dim: int):
+        """
+        This method has to be implemented by all children. Prepare the model
+        for training by a given dataset.
+        """
+        pass
+
+    @abstractmethod
+    def _calculate_metrics(self, pred: torch.tensor, target: torch.tensor) -> dict:
+        """
+        This method has to be implemented by all children. Calculate metrics.
+        """
+        pass
+
+    @abstractmethod
+    def _predict(self, sample: torch.Tensor) -> torch.Tensor:
+        """
+        This method has to be implemented by all children. Make a prediction
+        for a given batch of samples.
+        """
+        pass
+
+    @abstractmethod
+    def _set_dims(self, df: pd.DataFrame, target: pd.Series):
+        """
+        This method has to be implemented by all children. Calculate input and
+        output dimensions of the model by the given dataset.
+        """
+        pass
+
+    def _validate_inputs(self, df: pd.DataFrame, target: pd.Series):
+        assert (
+            df.shape[0] == target.shape[0]
+        ), f"target is incompatible with df by the length: {df.shape[0]} and {target.shape[0]}."
+        assert np.all(
+            df.index == target.index
+        ), "target's index and df's index are not the same."
+        assert df.index.names == (
+            ["run_id", "sample"]
+        ), "An index should contain columns `run_id` and `sample`."
+        assert (
+            len(df) >= self.window_size
+        ), "window size is larger than the length of df."
+        assert len(df) >= self.stride, "stride is larger than the length of df."
+
+    def _train_nn(
+        self,
+        df: pd.DataFrame,
+        target: pd.Series,
+        epochs: int,
+        save_path: str,
+        trial: optuna.Trial,
+    ):
+        self.model.train()
+        self.model.to(self.device)
+        self.best_val_loss = (
+            float("inf") if self.direction == "minimize" else float("-inf")
+        )
+        self.best_validation_metrics = {}
+        self._set_seed()
+
+        dataset = SlidingWindowDataset(df, target, window_size=self.window_size, stride=self.stride)
+        val_size = max(int(len(dataset) * self.val_ratio), 1)
+        train_size = len(dataset) - val_size
+        train_dataset, val_dataset = random_split(
+            dataset,
+            [train_size, val_size],
+            generator=torch.Generator().manual_seed(self.random_seed),
+        )
+
+        self.dataloader = DataLoader(
+            train_dataset, batch_size=self.batch_size, shuffle=True
+        )
+        self.val_dataloader = DataLoader(
+            val_dataset, batch_size=self.batch_size, shuffle=False
+        )
+        for e in trange(
+            self.checkpoint_epoch,
+            self.checkpoint_epoch + epochs,
+            desc="Epochs ...",
+            disable=(not self.verbose),
+        ):
+            for sample, target in tqdm(
+                self.dataloader,
+                desc="Steps ...",
+                leave=False,
+                disable=(not self.verbose),
+            ):
+                sample = sample.to(self.device)
+                target = target.to(self.device)
+                logits = self.model(sample)
+                loss = self.loss_fn(logits, target)
+                self.optimizer.zero_grad()
+                loss.backward()
+                self.optimizer.step()
+            if self.verbose:
+                print(f"Epoch {e+1}, Loss: {loss.item():.4f}")
+            self.checkpoint_epoch = e + 1
+            if self.save_checkpoints:
+                self.save_checkpoint(save_path)
+
+            val_loss, val_metrics = self._validate_nn()
+
+            if self.direction == "minimize":
+                self.best_val_loss = (
+                    val_loss if val_loss < self.best_val_loss else self.best_val_loss
+                )
+            else:
+                self.best_val_loss = (
+                    val_loss if val_loss > self.best_val_loss else self.best_val_loss
+                )
+
+            if self.val_metrics:
+                for key, value in val_metrics.items():
+                    if key not in self.best_validation_metrics:
+                        self.best_validation_metrics[key] = value
+                    else:
+                        if self.direction == "minimize":
+                            self.best_validation_metrics[key] = (
+                                value
+                                if self.best_validation_metrics[key] > value
+                                else self.best_validation_metrics[key]
+                            )
+                        else:
+                            self.best_validation_metrics[key] = (
+                                value
+                                if self.best_validation_metrics[key] < value
+                                else self.best_validation_metrics[key]
+                            )
+
+            if self.verbose:
+                if self.val_metrics:
+                    print(
+                        f"Epoch {e+1}, Validation Loss: {val_loss:.4f}, Metrics: {val_metrics}"
+                    )
+                else:
+                    print(f"Epoch {e+1}, Validation Loss: {val_loss:.4f}")
+
+            if trial:
+                trial.report(val_loss, self.checkpoint_epoch)
+
+                if trial.should_prune():
+                    raise optuna.exceptions.TrialPruned()
+
+    def _validate_nn(self):
+        self.model.eval()
+
+        val_loss = 0
+        target_list, pred_list = [], []
+        with torch.no_grad():
+            for sample, target in self.val_dataloader:
+                sample = sample.to(self.device)
+                target = target.to(self.device)
+                logits = self.model(sample)
+                loss = self.loss_fn(logits, target)
+                val_loss += loss.item()
+
+                if self.val_metrics:
+                    pred = self.predict(sample)
+                    target_list.append(target.cpu())
+                    pred_list.append(pred.cpu())
+
+        val_loss /= len(self.val_dataloader)
+
+        if self.val_metrics:
+            target_tensor = torch.cat(target_list).numpy()
+            pred_tensor = torch.cat(pred_list).numpy()
+
+            val_metrics = self._calculate_metrics(pred_tensor, target_tensor)
+        else:
+            val_metrics = None
+
+        self.model.train()
+
+        return val_loss, val_metrics
+
+    def _set_seed(self):
+        torch.manual_seed(self.random_seed)
+        random.seed(self.random_seed)
+        np.random.seed(self.random_seed)
+
+    def _initialize_paths(self):
+        artifacts_path = os.path.join(self._output_dir, self.name)
+        training_path = os.path.join(artifacts_path, "training")
+        inference_path = os.path.join(artifacts_path, "inference")
+        checkpoints_path = os.path.join(artifacts_path, "checkpoints")
+        return training_path, inference_path, checkpoints_path
+
+    def _store_atrifacts_train(self):
+        save_path = os.path.join(self._training_path, self.train_time)
+        os.makedirs(save_path, exist_ok=True)
+        self._cfg.to_yaml(os.path.join(save_path, "config.yaml"))
+
+    def _store_atrifacts_inference(self, metrics):
+        save_path = os.path.join(self._inference_path, self.train_time)
+        os.makedirs(save_path, exist_ok=True)
+
+        self._cfg.to_yaml(os.path.join(save_path, "config.yaml"))
+        with open(os.path.join(save_path, "metrics.json"), "w") as json_file:
+            json.dump(metrics, json_file)
+
+
    [docs]    def save_checkpoint(self, save_path: str = None):
+        """Save checkpoint.
+
+        Args:
+            save_path (str): Path to save checkpoint.
+        """
+        if save_path is None:
+            checkpoints_path = os.path.join(self._checkpoints_path, self.train_time)
+            os.makedirs(checkpoints_path, exist_ok=True)
+            file_path = self.name + "_epoch_" + str(self.checkpoint_epoch)
+            save_path = os.path.join(checkpoints_path, file_path + ".tar")
+        torch.save(
+            {
+                "config": self._cfg,
+                "epoch": self.checkpoint_epoch,
+                "input_dim": self.input_dim,
+                "output_dim": self.output_dim,
+                "model_state_dict": self.model.state_dict(),
+                "optimizer_state_dict": self.optimizer.state_dict(),
+            },
+            save_path,
+        )
    
+
+
    [docs]    def load_checkpoint(self, checkpoint_path: str):
+        """Load checkpoint.
+
+        Args:
+            checkpoint_path (str): Path to load checkpoint.
+        """
+        checkpoint = torch.load(checkpoint_path, map_location=self.device)
+        self._cfg = checkpoint["config"]
+        self.input_dim = checkpoint["input_dim"]
+        self.output_dim = checkpoint["output_dim"]
+        self._create_model(self.input_dim, self.output_dim)
+        self.model.to(self.device)
+        assert self.model is not None, "Model creation error."
+        self._prepare_for_training(self.input_dim, self.output_dim)
+        self.model.load_state_dict(checkpoint["model_state_dict"])
+        self.optimizer.load_state_dict(checkpoint["optimizer_state_dict"])
+        self.checkpoint_epoch = checkpoint["epoch"]
    
+
+
+
    [docs]class SlidingWindowDataset(Dataset):
+    def __init__(
+        self, df: pd.DataFrame, target: pd.Series, window_size: int, stride: int = 1
+    ):
+        self.df = df
+        self.target = target
+        self.window_size = window_size
+
+        window_end_indices = []
+        run_ids = df.index.get_level_values(0).unique()
+        for run_id in tqdm(run_ids, desc="Creating sequence of samples"):
+            indices = np.array(df.index.get_locs([run_id]))
+            indices = indices[self.window_size :: stride]
+            window_end_indices.extend(indices)
+        self.window_end_indices = np.array(window_end_indices)
+
+    def __len__(self):
+        return len(self.window_end_indices)
+
+    def __getitem__(self, idx):
+        window_index = self.window_end_indices[idx]
+        sample = self.df.values[window_index - self.window_size : window_index]
+        if self.target is not None:
+            target = self.target.values[window_index]
+        else:
+            target = sample.astype("float32")
+        return sample.astype("float32"), target
    
 
    
 
                 
diff --git a/docs/_modules/ice/fault_diagnosis/datasets.html b/docs/_modules/ice/fault_diagnosis/datasets.html
index 42d66ab..82e834b 100644
--- a/docs/_modules/ice/fault_diagnosis/datasets.html
+++ b/docs/_modules/ice/fault_diagnosis/datasets.html
@@ -369,6 +369,20 @@ 
    Source code for ice.fault_diagnosis.datasets
    from ice.base import BaseDataset
 
 
+
    [docs]class FaultDiagnosisRiethTEP(BaseDataset):
+    """
+    Dataset of Tennessee Eastman Process dataset 
+    Rieth, C. A., Amsel, B. D., Tran, R., & Cook, M. B. (2017). 
+    Additional Tennessee Eastman Process Simulation Data for 
+    Anomaly Detection Evaluation (Version V1) [Computer software]. 
+    Harvard Dataverse. 
+    https://doi.org/10.7910/DVN/6C3JR1.
+    """
+
    
+
+
 
    [docs]class FaultDiagnosisSmallTEP(BaseDataset):
     """
     Cropped version of Tennessee Eastman Process dataset 
diff --git a/docs/_modules/ice/fault_diagnosis/models/base.html b/docs/_modules/ice/fault_diagnosis/models/base.html
index 6380b1c..79bafda 100644
--- a/docs/_modules/ice/fault_diagnosis/models/base.html
+++ b/docs/_modules/ice/fault_diagnosis/models/base.html
@@ -366,123 +366,43 @@
                 
    
                   
   
    Source code for ice.fault_diagnosis.models.base
    -from abc import ABC, abstractmethod
+from abc import ABC
 import pandas as pd
-from tqdm.auto import trange, tqdm
-
 import torch
 from torch import nn
 from torch.optim import Adam
-from torch.utils.data import DataLoader
 
-from ice.fault_diagnosis.utils import SlidingWindowDataset
 from ice.base import BaseModel
 from ice.fault_diagnosis.metrics import (
     accuracy, correct_daignosis_rate, true_positive_rate, false_positive_rate)
 
+
 
    [docs]class BaseFaultDiagnosis(BaseModel, ABC):
     """Base class for all fault diagnosis models."""
 
-    @abstractmethod
-    def __init__(
-            self,
-            window_size: int,
-            batch_size: int, 
-            lr: float, 
-            num_epochs: int, 
-            device: str, 
-            verbose: bool,
-            name: str
-        ):
-        """
-        Args:
-            window_size (int): The window size to train the model.
-            batch_size (int): The batch size to train the model.
-            lr (float): The learning rate to train the model.
-            num_epochs (float): The number of epochs to train the model.
-            device (str): The name of a device to train the model. `cpu` and 
-                `cuda` are possible.
-            verbose (bool): If true, show the progress bar in training.
-            name (str): The name of the model for artifact storing.
-        """
-        super().__init__(batch_size, lr, num_epochs, device, verbose, name)
-        self._cfg.path_set(["TASK"], "fault_diagnosis")
-
-        self.window_size = window_size
-        self.loss_fn = None
-
-    _param_conf_map = dict(BaseModel._param_conf_map,
-            **{
-                "window_size" : ["MODEL", "WINDOW_SIZE"]
-            }
-        )
-
-    def _fit(self, df: pd.DataFrame, target: pd.Series):
-        assert len(df) >= self.window_size, "window size is larger than the length of df."
-        num_classes = len(set(target))
-        weight = torch.ones(num_classes, device=self.device) * 0.5
-        weight[1:] /= num_classes - 1
+    def _prepare_for_training(self, input_dim: int, output_dim: int):
+        weight = torch.ones(output_dim, device=self.device) * 0.5
+        weight[1:] /= output_dim - 1
         self.loss_fn = nn.CrossEntropyLoss(weight=weight)
-        self._train_nn(df, target)
+        self.optimizer = Adam(self.model.parameters(), lr=self.lr)
 
     def _predict(self, sample: torch.Tensor) -> torch.Tensor:
         sample = sample.to(self.device)
         logits = self.model(sample)
         return logits.argmax(axis=1).cpu()
 
-    def _train_nn(self, df: pd.DataFrame, target: pd.Series):
-        self.model.train()
-        self.model.to(self.device)
-        self.optimizer = Adam(self.model.parameters(), lr=self.lr)
-
-        dataset = SlidingWindowDataset(df, target, window_size=self.window_size)
-        self.dataloader = DataLoader(dataset, batch_size=self.batch_size, shuffle=True)
-        for e in trange(self.num_epochs, desc='Epochs ...', disable=(not self.verbose)):
-            for sample, target in tqdm(self.dataloader, desc='Steps ...', leave=False, disable=(not self.verbose)):
-                sample = sample.to(self.device)
-                target = target.to(self.device)
-                logits = self.model(sample)
-                loss = self.loss_fn(logits, target)
-                self.optimizer.zero_grad()
-                loss.backward()
-                self.optimizer.step()
-            if self.verbose:
-                print(f'Epoch {e+1}, Loss: {loss.item():.4f}')
-
-
    [docs]    def evaluate(self, df: pd.DataFrame, target: pd.Series) -> dict:
-        """Evaluate the metrics: accuracy.
-        
-        Args:
-            df (pandas.DataFrame): A dataframe with sensor data. Index has 
-                two columns: `run_id` and `sample`. All other columns a value of 
-                sensors.
-            target (pandas.Series): A series with target values. Indes has two
-                columns: `run_id` and `sample`.
-        
-        Returns:
-            dict: A dictionary with metrics where keys are names of metrics and
-                values are values of metrics.
-        """
-        dataset = SlidingWindowDataset(df, target, window_size=self.window_size)
-        self.dataloader = DataLoader(dataset, batch_size=self.batch_size, shuffle=True)
-        target, pred = [], []
-        for sample, _target in tqdm(
-            self.dataloader, desc='Steps ...', leave=False, disable=(not self.verbose)
-        ):
-            sample = sample.to(self.device)
-            target.append(_target)
-            with torch.no_grad():
-                pred.append(self.predict(sample))
-        target = torch.concat(target).numpy()
-        pred = torch.concat(pred).numpy()
+    def _calculate_metrics(self, pred: torch.tensor, target: torch.tensor) -> dict:
         metrics = {
             'accuracy': accuracy(pred, target),
             'correct_daignosis_rate': correct_daignosis_rate(pred, target), 
             'true_positive_rate': true_positive_rate(pred, target),
             'false_positive_rate': false_positive_rate(pred, target),
         }
-        self._store_atrifacts_inference(metrics)
-        return metrics
    
+        return metrics
+    
+    def _set_dims(self, df: pd.DataFrame, target: pd.Series):
+        self.input_dim = df.shape[1]
+        self.output_dim = len(set(target))
    
 
    
 
                 
diff --git a/docs/_modules/ice/fault_diagnosis/models/mlp.html b/docs/_modules/ice/fault_diagnosis/models/mlp.html
index 736beff..0ade77f 100644
--- a/docs/_modules/ice/fault_diagnosis/models/mlp.html
+++ b/docs/_modules/ice/fault_diagnosis/models/mlp.html
@@ -380,13 +380,17 @@ 
    Source code for ice.fault_diagnosis.models.mlp
    def __init__(
             self, 
             window_size: int,
+            stride: int = 1,
             hidden_dim: int=256,
             batch_size: int=128,
             lr: float=0.001,
             num_epochs: int=10,
             device: str='cpu',
             verbose: bool=False,
-            name: str='mlp_fault_diagnosis'
+            name: str='mlp_fault_diagnosis',
+            random_seed: int = 42,
+            val_ratio: float = 0.15,
+            save_checkpoints: bool = False
         ):
         """
         Args:
@@ -399,10 +403,14 @@ 
    Source code for ice.fault_diagnosis.models.mlp
                    `cuda` are possible.
             verbose (bool): If true, show the progress bar in training.
             name (str): The name of the model for artifact storing.
+            random_seed (int): Seed for random number generation to ensure reproducible results.
+            val_ratio (float): Proportion of the dataset used for validation, between 0 and 1.
+            save_checkpoints (bool): If true, store checkpoints.
         """
         super().__init__(
-            window_size, batch_size, lr, num_epochs, device, verbose, name
+            window_size, stride, batch_size, lr, num_epochs, device, verbose, name, random_seed, val_ratio, save_checkpoints
         )
+        self.val_metrics = True
 
         self.hidden_dim = hidden_dim
 
@@ -412,14 +420,12 @@ 
    Source code for ice.fault_diagnosis.models.mlp
    }
         )   
 
-    def _create_model(self, df: DataFrame, target: Series):
-        num_sensors = df.shape[1]
-        num_classes = len(set(target))
+    def _create_model(self, input_dim: int, output_dim: int):
         self.model = nn.Sequential(
             nn.Flatten(),
-            nn.Linear(num_sensors * self.window_size, self.hidden_dim),
+            nn.Linear(input_dim * self.window_size, self.hidden_dim),
             nn.ReLU(),
-            nn.Linear(self.hidden_dim, num_classes),
+            nn.Linear(self.hidden_dim, output_dim),
         )
    
 
    
 
diff --git a/docs/_modules/ice/fault_diagnosis/models/tcn.html b/docs/_modules/ice/fault_diagnosis/models/tcn.html
index fa58700..f196d4f 100644
--- a/docs/_modules/ice/fault_diagnosis/models/tcn.html
+++ b/docs/_modules/ice/fault_diagnosis/models/tcn.html
@@ -493,6 +493,7 @@ 
    Source code for ice.fault_diagnosis.models.tcn
    def __init__(
             self, 
             window_size: int,
+            stride: int = 1,
             hidden_dim: int=256,
             kernel_size: int=5,
             num_layers: int=4,
@@ -503,7 +504,10 @@ 
    Source code for ice.fault_diagnosis.models.tcn
    num_epochs: int=10,
             device: str='cpu',
             verbose: bool=False,
-            name: str='tcn_fault_diagnosis'
+            name: str='tcn_fault_diagnosis',
+            random_seed: int = 42,
+            val_ratio: float = 0.15,
+            save_checkpoints: bool = False
         ):
         """
         Args:
@@ -520,10 +524,14 @@ 
    Source code for ice.fault_diagnosis.models.tcn
                    `cuda` are possible.
             verbose (bool): If true, show the progress bar in training.
             name (str): The name of the model for artifact storing.
+            random_seed (int): Seed for random number generation to ensure reproducible results.
+            val_ratio (float): Proportion of the dataset used for validation, between 0 and 1.
+            save_checkpoints (bool): If true, store checkpoints.
         """
         super().__init__(
-            window_size, batch_size, lr, num_epochs, device, verbose, name
+            window_size, stride, batch_size, lr, num_epochs, device, verbose, name, random_seed, val_ratio, save_checkpoints
         )
+        self.val_metrics = True
 
         self.hidden_dim = hidden_dim
         self.kernel_size = kernel_size
@@ -541,16 +549,14 @@ 
    Source code for ice.fault_diagnosis.models.tcn
    }
         )   
 
-    def _create_model(self, df: DataFrame, target: Series):
-        num_sensors = df.shape[1]
-        num_classes = len(set(target))
+    def _create_model(self, input_dim: int, output_dim: int):
         self.model = _TCNModule(
-            input_dim=num_sensors,
+            input_dim=input_dim,
             kernel_size=self.kernel_size,
             hidden_dim=self.hidden_dim,
             num_layers=self.num_layers,
             dilation_base=self.dilation_base,
-            output_dim=num_classes,
+            output_dim=output_dim,
             dropout=self.dropout,
             seq_len=self.window_size,
         )
    
diff --git a/docs/_modules/ice/health_index_estimation/datasets.html b/docs/_modules/ice/health_index_estimation/datasets.html
index 0ce534d..168321c 100644
--- a/docs/_modules/ice/health_index_estimation/datasets.html
+++ b/docs/_modules/ice/health_index_estimation/datasets.html
@@ -419,6 +419,7 @@ 
    Source code for ice.health_index_estimation.datasets
    inter_func = []
 
         for i in range(15):
+            data[i]["material"] = data[i]["material"].astype("float64")
             y = data[i].dropna().VB
             x = data[i].dropna().time
 
@@ -432,21 +433,16 @@ 
    Source code for ice.health_index_estimation.datasets
    else:
                 data[i] = data[i].fillna(0)
 
-        self.df = [data[i].drop(columns=["VB"]) for i in train_nums]
+        self.df = [
+            data[i].drop(columns=["VB", "Unnamed: 0", "case", "run"])
+            for i in train_nums
+        ]
         self.target = [data[i]["VB"] for i in train_nums]
 
-        self.test = [data[i].drop(columns=["VB"]) for i in test_nums]
-        self.test_target = [data[i]["VB"] for i in test_nums]
-
-    def _read_csv_pgbar(self, csv_path, index_col, chunksize=1024 * 100):
-        df = pd.read_csv(csv_path)
-        df.rename(columns={"cut_no": "run_id"}, inplace=True)
-        df = df.set_index(["run_id", "sample"]).drop(
-            columns=["Unnamed: 0", "case", "run"]
-        )
-        df["material"] = df["material"].astype("float64")
-
-        return df
    
+        self.test = [
+            data[i].drop(columns=["VB", "Unnamed: 0", "case", "run"]) for i in test_nums
+        ]
+        self.test_target = [data[i]["VB"] for i in test_nums]
    
 
    
 
                 
diff --git a/docs/_modules/ice/health_index_estimation/metrics.html b/docs/_modules/ice/health_index_estimation/metrics.html
index 9dfbdcf..6ded8b2 100644
--- a/docs/_modules/ice/health_index_estimation/metrics.html
+++ b/docs/_modules/ice/health_index_estimation/metrics.html
@@ -381,6 +381,19 @@ 
    Source code for ice.health_index_estimation.metrics
            float: rmse
     """
     return float(np.mean((pred - target) ** 2))
    
+
+
    [docs]def rmse(pred: list, target: list) -> float:
+    """
+    Mean squared error between real and predicted wear.
+
+    Args:
+        pred (list): numpy prediction values.
+        target (list): numpy target values.
+
+    Returns:
+        float: rmse
+    """
+    return float(np.sqrt(np.mean((pred - target) ** 2)))
    
 
    
 
                 
diff --git a/docs/_modules/ice/health_index_estimation/models/base.html b/docs/_modules/ice/health_index_estimation/models/base.html
index 2fc2116..8f1c574 100644
--- a/docs/_modules/ice/health_index_estimation/models/base.html
+++ b/docs/_modules/ice/health_index_estimation/models/base.html
@@ -367,134 +367,37 @@
                   
   
    Source code for ice.health_index_estimation.models.base
     from ice.base import BaseModel
-from abc import ABC, abstractmethod
+from abc import ABC
 import pandas as pd
-from tqdm.auto import trange, tqdm
-
-from torch.optim import Adam, AdamW
-from torch.utils.data import DataLoader
+from torch.optim import AdamW
 import torch
 from torch import nn
-from ice.health_index_estimation.utils import SlidingWindowDataset
-from ice.health_index_estimation.metrics import mse
+
+from ice.health_index_estimation.metrics import mse, rmse
 
 
 
    [docs]class BaseHealthIndexEstimation (BaseModel, ABC):
     """Base class for all HI diagnosis models."""
 
-    @abstractmethod
-    def __init__(
-        self,
-        window_size: int,
-        stride: int,
-        batch_size: int,
-        lr: float,
-        num_epochs: int,
-        device: str,
-        verbose: bool,
-        name: str,
-    ):
-        """
-        Args:
-            window_size (int): The window size to train the model.
-            stride (int): The time interval between first points of consecutive sliding windows.
-            batch_size (int): The batch size to train the model.
-            lr (float): The larning rate to train the model.
-            num_epochs (float): The number of epochs to train the model.
-            device (str): The name of a device to train the model. `cpu` and
-                `cuda` are possible.
-            verbose (bool): If true, show the progress bar in training.
-            name (str): The name of the model for artifact storing.
-        """
-        super().__init__(batch_size, lr, num_epochs, device, verbose, name)
-
-        self.window_size = window_size
-        self.stride = stride
-        self.loss_fn = None
-        self.newvalues = None
-
-        self.test_stride = 50
-
-    _param_conf_map = dict(
-        BaseModel._param_conf_map,
-        **{"window_size": ["MODEL", "WINDOW_SIZE"], "stride": ["MODEL", "STRIDE"]},
-    )
-
-    def _fit(self, df: pd.DataFrame, target: pd.Series):
-        assert (
-            len(df) >= self.window_size
-        ), "window size is larger than the length of df."
-        assert len(df) >= self.stride, "stride is larger than the length of df."
+    def _prepare_for_training(self, input_dim: int, output_dim: int):
         self.loss_fn = nn.L1Loss()
-        self._train_nn(df, target)
+        self.optimizer = AdamW(self.model.parameters(), lr=self.lr)
 
     def _predict(self, sample: torch.Tensor) -> torch.Tensor:
         sample = sample.to(self.device)
         predicted_rul = self.model(sample)
         return predicted_rul.cpu()
 
-    def _train_nn(self, df: pd.DataFrame, target: pd.Series):
-        self.model.train()
-        self.model.to(self.device)
-        self.optimizer = AdamW(self.model.parameters(), lr=self.lr)
-
-        dataset = SlidingWindowDataset(
-            df, target, window_size=self.window_size, stride=self.stride
-        )
-        self.dataloader = DataLoader(dataset, batch_size=self.batch_size, shuffle=True)
-        for e in trange(self.num_epochs, desc="Epochs ...", disable=(not self.verbose)):
-            for sample, target in tqdm(
-                self.dataloader,
-                desc="Steps ...",
-                leave=False,
-                disable=(not self.verbose),
-            ):
-                sample = sample.to(self.device)
-                target = target.to(self.device)
-
-                logits = self.model(sample)
-                loss = self.loss_fn(logits, target)
-
-                self.optimizer.zero_grad()
-                loss.backward()
-                self.optimizer.step()
-            self.newvalues.append(loss.item())
-            if self.verbose:
-                print(f"Epoch {e+1}, Loss: {loss.item():.4f}")
-
-
    [docs]    def evaluate(self, df: pd.DataFrame, target: pd.Series) -> dict:
-        """Evaluate the metrics: mse.
-
-        Args:
-            df (pandas.DataFrame): A dataframe with sensor data. Index has
-                two columns: `run_id` and `sample`. All other columns a value of
-                sensors.
-            target (pandas.Series): A series with target values. Indes has two
-                columns: `run_id` and `sample`.
-
-        Returns:
-            dict: A dictionary with metrics where keys are names of metrics and
-                values are values of metrics.
-        """
-        dataset = SlidingWindowDataset(
-            df, target, window_size=self.window_size, stride=self.test_stride
-        )
-        self.dataloader = DataLoader(dataset, batch_size=1, shuffle=True)
-        target, pred = [], []
-        for sample, _target in tqdm(
-            self.dataloader, desc="Steps ...", leave=False, disable=(not self.verbose)
-        ):
-            sample = sample.to(self.device)
-            target.append(_target)
-            with torch.no_grad():
-                pred.append(self.predict(sample))
-        target = torch.concat(target).numpy()
-        pred = torch.concat(pred).numpy()
+    def _calculate_metrics(self, pred: torch.tensor, target: torch.tensor) -> dict:
         metrics = {
             "mse": mse(pred, target),
+            "rmse": rmse(pred, target),
         }
-        self._store_atrifacts_inference(metrics)
-        return metrics
    
+        return metrics
+    
+    def _set_dims(self, df: pd.DataFrame, target: pd.Series):
+        self.input_dim = df.shape[1]
+        self.output_dim = 1
    
 
    
 
                 
diff --git a/docs/_modules/ice/health_index_estimation/models/mlp.html b/docs/_modules/ice/health_index_estimation/models/mlp.html
index 39c637a..188788b 100644
--- a/docs/_modules/ice/health_index_estimation/models/mlp.html
+++ b/docs/_modules/ice/health_index_estimation/models/mlp.html
@@ -390,6 +390,9 @@ 
    Source code for ice.health_index_estimation.models.mlp
    device: str = "cpu",
         verbose: bool = True,
         name: str = "mlp_fault_diagnosis",
+        random_seed: int = 42,
+        val_ratio: float = 0.15,
+        save_checkpoints: bool = False
     ):
         """
         Args:
@@ -403,10 +406,14 @@ 
    Source code for ice.health_index_estimation.models.mlp
                    `cuda` are possible.
             verbose (bool): If true, show the progress bar in training.
             name (str): The name of the model for artifact storing.
+            random_seed (int): Seed for random number generation to ensure reproducible results.
+            val_ratio (float): Proportion of the dataset used for validation, between 0 and 1.
+            save_checkpoints (bool): If true, store checkpoints.
         """
         super().__init__(
-            window_size, stride, batch_size, lr, num_epochs, device, verbose, name
+            window_size, stride, batch_size, lr, num_epochs, device, verbose, name, random_seed, val_ratio, save_checkpoints
         )
+        self.val_metrics = True
 
         self.hidden_dim = hidden_dim
         self.newvalues = []
@@ -416,12 +423,13 @@ 
    Source code for ice.health_index_estimation.models.mlp
    **{"hidden_dim": ["MODEL", "HIDDEN_DIM"]}
     )
 
-    def _create_model(self, df: DataFrame, target: Series):
-        num_sensors = df.shape[1]
+    def _create_model(self, input_dim: int, output_dim: int):
         self.model = nn.Sequential(
             nn.Flatten(),
-            nn.Linear(num_sensors * self.window_size, self.hidden_dim),
+            nn.Dropout(0.5),
+            nn.Linear(input_dim * self.window_size, self.hidden_dim),
             nn.ReLU(),
+            nn.Dropout(0.5),
             nn.Linear(self.hidden_dim, 1),
             nn.Flatten(start_dim=0),
         )
    
diff --git a/docs/_modules/ice/remaining_useful_life_estimation/datasets.html b/docs/_modules/ice/remaining_useful_life_estimation/datasets.html
index 9787a0f..c7a9020 100644
--- a/docs/_modules/ice/remaining_useful_life_estimation/datasets.html
+++ b/docs/_modules/ice/remaining_useful_life_estimation/datasets.html
@@ -398,7 +398,7 @@ 
    Source code for ice.remaining_useful_life_estimation.datasets
    if not os.path.exists(zfile_path) or force_download:
             self._download_pgbar(url, zfile_path, self.name, num_chunks)
 
-        self._extracting_files(zfile_path, ref_path)
+        self._extracting_files(zfile_path, "data/")
         self.df = [
             self._read_csv_pgbar(
                 ref_path + f"fd{i}_train.csv", index_col=["run_id", "sample"]
@@ -422,13 +422,50 @@ 
    Source code for ice.remaining_useful_life_estimation.datasets
    ref_path + f"fd{i}_test.csv", index_col=["run_id", "sample"]
             )["rul"]
             for i in range(1, 5)
-        ]
+        ]
    
+
+
+
    [docs]class RulCmapssPaper(BaseDataset):
+    """
+    Preprocessed to piece wise RUL data from the dataset:
+    Saxena A. et al. Damage propagation modeling for aircraft engine run-to-failure simulation
+    DOI: 10.1109/PHM.2008.4711414. Target is the minimum rul value for every test device. 
+
+    """
+
+
+
+    def _load(self, num_chunks, force_download):
+        """
+        Load the test dataset in list obects: self.df, self.target, self.test and self.test_target.
+        4 subdatasets fd001-fd004, list index corresponds to a subdataset number
+
+        """
+        ref_path = f"data/{self.name}/"
+        if not os.path.exists(ref_path):
+            os.makedirs(ref_path)
+        zfile_path = f"data/{self.name}.zip"
 
-    def _read_csv_pgbar(self, csv_path, index_col, chunksize=1024 * 100):
-        df = pd.read_csv(csv_path)
-        df.rename(columns={"unit_num": "run_id"}, inplace=True)
-        df = df.set_index(["run_id", "sample"])
-        return df
    
+        url = self._get_url(self.public_link)
+        if not os.path.exists(zfile_path) or force_download:
+            self._download_pgbar(url, zfile_path, self.name, num_chunks)
+
+        self._extracting_files(zfile_path, f"data/{self.name}/")
+        
+        self.test = [
+            self._read_csv_pgbar(
+                ref_path + f"fd{i}_test.csv", index_col=["run_id", "sample"]
+            ).drop(columns=["rul"])
+            for i in range(1, 5)
+        ]
+        self.test_target = [
+            self._read_csv_pgbar(
+                ref_path + f"/fd{i}_test.csv", index_col=["run_id", "sample"]
+            )["rul"]
+            for i in range(1, 5)
+        ]
    diff --git a/docs/_modules/ice/remaining_useful_life_estimation/metrics.html b/docs/_modules/ice/remaining_useful_life_estimation/metrics.html index 0bf0393..18f3641 100644 --- a/docs/_modules/ice/remaining_useful_life_estimation/metrics.html +++ b/docs/_modules/ice/remaining_useful_life_estimation/metrics.html @@ -397,7 +397,7 @@

    Source code for ice.remaining_useful_life_estimation.metrics

    return float(np.exp((-value / 13)) - 1 if value < 0 else np.exp((value / 10)) - 1)
    -
    [docs]def score(pred: list, target: list) -> float: +
    [docs]def cmapss_score(pred: list, target: list) -> float: """ Non-simmetric metric proposed in the original dataset paper. DOI: 10.1109/PHM.2008.4711414 diff --git a/docs/_modules/ice/remaining_useful_life_estimation/models/base.html b/docs/_modules/ice/remaining_useful_life_estimation/models/base.html index 52b7803..a1814c6 100644 --- a/docs/_modules/ice/remaining_useful_life_estimation/models/base.html +++ b/docs/_modules/ice/remaining_useful_life_estimation/models/base.html @@ -367,132 +367,37 @@

    Source code for ice.remaining_useful_life_estimation.models.base

     from ice.base import BaseModel
-from abc import ABC, abstractmethod
+from abc import ABC
 import pandas as pd
-from tqdm.auto import trange, tqdm
-
-from torch.optim import Adam, AdamW
-from torch.utils.data import DataLoader
+from torch.optim import AdamW
 import torch
 from torch import nn
-from ice.remaining_useful_life_estimation.utils import SlidingWindowDataset
-from ice.remaining_useful_life_estimation.metrics import rmse, score
+
+from ice.remaining_useful_life_estimation.metrics import rmse, cmapss_score
 
 
 
    [docs]class BaseRemainingUsefulLifeEstimation(BaseModel, ABC):
     """Base class for all RUL models."""
 
-    @abstractmethod
-    def __init__(
-        self,
-        window_size: int,
-        stride: int,
-        batch_size: int,
-        lr: float,
-        num_epochs: int,
-        device: str,
-        verbose: bool,
-        name: str,
-    ):
-        """
-        Args:
-            window_size (int): The window size to train the model.
-            stride (int): The time interval between first points of consecutive sliding windows.
-            batch_size (int): The batch size to train the model.
-            lr (float): The learning rate to train the model.
-            num_epochs (float): The number of epochs to train the model.
-            device (str): The name of a device to train the model. `cpu` and
-                `cuda` are possible.
-            verbose (bool): If true, show the progress bar in training.
-            name (str): The name of the model for artifact storing.
-        """
-        super().__init__(batch_size, lr, num_epochs, device, verbose, name)
-
-        self.window_size = window_size
-        self.stride = stride
-        self.loss_fn = None
-        self.loss_array = None
-
-    _param_conf_map = dict(
-        BaseModel._param_conf_map,
-        **{
-            "window_size": ["MODEL", "WINDOW_SIZE"],
-            "stride": ["MODEL", "STRIDE"],
-        },
-    )
-
-    def _fit(self, df: pd.DataFrame, target: pd.Series):
-        assert (
-            len(df) >= self.window_size
-        ), "window size is larger than the length of df."
-        assert len(df) >= self.stride, "stride is larger than the length of df."
+    def _prepare_for_training(self, input_dim: int, output_dim: int):
         self.loss_fn = nn.L1Loss()
-        self._train_nn(df, target)
+        self.optimizer = AdamW(self.model.parameters(), lr=self.lr)
 
     def _predict(self, sample: torch.Tensor) -> torch.Tensor:
         sample = sample.to(self.device)
         predicted_rul = self.model(sample)
         return predicted_rul.cpu()
 
-    def _train_nn(self, df: pd.DataFrame, target: pd.Series):
-        self.model.train()
-        self.model.to(self.device)
-        self.optimizer = AdamW(self.model.parameters(), lr=self.lr)
-
-        dataset = SlidingWindowDataset(
-            df, target, window_size=self.window_size, stride=self.stride
-        )
-        self.dataloader = DataLoader(dataset, batch_size=self.batch_size, shuffle=True)
-        for e in trange(self.num_epochs, desc="Epochs ...", disable=(not self.verbose)):
-            for sample, target in tqdm(
-                self.dataloader,
-                desc="Steps ...",
-                leave=False,
-                disable=(not self.verbose),
-            ):
-                sample = sample.to(self.device)
-                target = target.to(self.device)
-                logits = self.model(sample)
-                loss = self.loss_fn(logits, target)
-                self.optimizer.zero_grad()
-                loss.backward()
-                self.optimizer.step()
-            self.loss_array.append(loss.item())
-            if self.verbose:
-                print(f"Epoch {e+1}, Loss: {loss.item():.4f}")
-
-
    [docs]    def evaluate(self, df: pd.DataFrame, target: pd.Series) -> dict:
-        """Evaluate the metrics: rmse, c-mapss score.
-
-        Args:
-            df (pandas.DataFrame): A dataframe with sensor data. Index has
-                two columns: `run_id` and `sample`. All other columns a value of
-                sensors.
-            target (pandas.Series): A series with target values. Indes has two
-                columns: `run_id` and `sample`.
-
-        Returns:
-            dict: A dictionary with metrics where keys are names of metrics and
-                values are values of metrics.
-        """
-        dataset = SlidingWindowDataset(df, target, window_size=self.window_size)
-        self.dataloader = DataLoader(dataset, batch_size=1, shuffle=True)
-        target, pred = [], []
-        for sample, _target in tqdm(
-            self.dataloader, desc="Steps ...", leave=False, disable=(not self.verbose)
-        ):
-            sample = sample.to(self.device)
-            target.append(_target)
-            with torch.no_grad():
-                pred.append(self.predict(sample))
-        target = torch.concat(target).numpy()
-        pred = torch.concat(pred).numpy()
+    def _calculate_metrics(self, pred: torch.tensor, target: torch.tensor) -> dict:
         metrics = {
             "rmse": rmse(pred, target),
-            "score": score(pred, target),
+            "cmapss_score": cmapss_score(pred, target),
         }
-        self._store_atrifacts_inference(metrics)
-        return metrics
    
+        return metrics
+    
+    def _set_dims(self, df: pd.DataFrame, target: pd.Series):
+        self.input_dim = df.shape[1]
+        self.output_dim = 1
    diff --git a/docs/_modules/ice/remaining_useful_life_estimation/models/lstm.html b/docs/_modules/ice/remaining_useful_life_estimation/models/lstm.html new file mode 100644 index 0000000..2fa856b --- /dev/null +++ b/docs/_modules/ice/remaining_useful_life_estimation/models/lstm.html @@ -0,0 +1,556 @@ + + + + + + + + + + + ice.remaining_useful_life_estimation.models.lstm — ICE documentation + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
    + + + + + + + + + + + +
    +
    +
    +
    + + + Ctrl+K +
    +
    + + + +
    +
    + +
    + + + +
    + + +
    + +
    + +
    + + + +
    + +
    +
    + +
    + +
    + + +
    +
    + + + + +
    + +
    + + +
    +
    + +
    +
    + +
    + +
    + + + + +
    + +
    + + +
    +
    + + + + + +
    + +

    Source code for ice.remaining_useful_life_estimation.models.lstm

    +from torch import nn
+from ice.remaining_useful_life_estimation.models.base import BaseRemainingUsefulLifeEstimation
+from pandas import DataFrame, Series
+import torch
+
+
+
    [docs]class LSTM_model(nn.Module):
+    """
+    Long short-term memory (LSTM) is reccurent neural network type, 
+    pytorch realisation https://pytorch.org/docs/stable/generated/torch.nn.LSTM.html 
+    """
+    def __init__(
+            self, 
+            input_dim, 
+            hidden_size=512, 
+            device="cpu", 
+            num_layers=2,):
+        """
+        Args:
+            input_dim (int): The dimension size of input data, related to the sensor amount in industry probles.
+            hidden_size (int): The number of features in the hidden state of the model.
+            device (str): The name of a device to train the model. `cpu` and `cuda` are possible.
+            num_layers (int): The number of stacked reccurent layers of the classic LSTM architecture.
+        """
+        super(LSTM_model, self).__init__()
+        self.input_dim = input_dim
+        self.hidden_size = hidden_size
+        self.num_layers = num_layers
+        self.device = device
+
+        self.lstm = nn.LSTM(input_size=input_dim, hidden_size=hidden_size,
+                          num_layers=num_layers, batch_first=True)
+
+
    [docs]    def forward(self, x):
+        h_0 = torch.zeros(self.num_layers, x.size(0), self.hidden_size).to(self.device) #hidden state
+        c_0 = torch.zeros(self.num_layers, x.size(0), self.hidden_size).to(self.device) #internal state
+        
+        output, (hn, cn) = self.lstm(x, (h_0, c_0)) 
+        return output[:, -1, :]
    
+
+
    [docs]class LSTM(BaseRemainingUsefulLifeEstimation):
+    """
+    Long short-term memory (LSTM) model consists of the classical LSTM architecture stack and 
+    two-layer MLP with SiLU nonlinearity and dropout to make the final prediction.
+    
+    Each sample is moved to LSTM and reshaped to a vector (B, L, C) -> (B, hidden_size, C)
+    Then the sample is reshaped to a vector (B, hidden_size, C) -> (B, hidden_size * C)
+    """
+
+    def __init__(
+        self,
+        window_size: int = 32,
+        stride: int = 1,
+        hidden_dim: int = 512,
+        hidden_size: int = 256, 
+        num_layers: int =2,
+        dropout_value: float = 0.5,
+        batch_size: int = 64,
+        lr: float = 1e-4,
+        num_epochs: int = 35,
+        device: str = "cpu",
+        verbose: bool = True,
+        name: str = "mlp_cmapss_rul",
+        random_seed: int = 42,
+        val_ratio: float = 0.15,
+        save_checkpoints: bool = False
+    ):
+        """
+        Args:
+            window_size (int): The window size to train the model.
+            stride (int): The time interval between first points of consecutive sliding windows.
+            hidden_dim (int): The dimensionality of the hidden layer in MLP.
+            hidden_size (int): The number of features in the hidden state of the model.
+            num_layers (int): The number of stacked reccurent layers of the classic LSTM architecture.
+            batch_size (int): The batch size to train the model.
+            lr (float): The larning rate to train the model.
+            num_epochs (float): The number of epochs to train the model.
+            device (str): The name of a device to train the model. `cpu` and
+                `cuda` are possible.
+            verbose (bool): If true, show the progress bar in training.
+            name (str): The name of the model for artifact storing.
+            random_seed (int): Seed for random number generation to ensure reproducible results.
+            val_ratio (float): Proportion of the dataset used for validation, between 0 and 1.
+            save_checkpoints (bool): If true, store checkpoints.
+        """
+        super().__init__(
+            window_size, stride, batch_size, lr, num_epochs, device, verbose, name, random_seed, val_ratio, save_checkpoints
+        )
+        self.val_metrics = True
+
+        self.hidden_dim = hidden_dim
+        self.hidden_size = hidden_size
+        self.num_layers = num_layers
+        self.device = device
+        self.dropout_value = dropout_value
+
+        self.loss_array = []
+
+    _param_conf_map = dict(
+        BaseRemainingUsefulLifeEstimation._param_conf_map,
+        **{"hidden_dim": ["MODEL", "HIDDEN_DIM"],
+        "hidden_size": ["MODEL", "HIDDEN_SIZE"],
+        "num_layers": ["MODEL", "NUM_LAYERS"],
+        "dropout_value": ["MODEL", "DROPOUT"],
+        }
+    )
+
+    def _create_model(self, input_dim: int, output_dim: int):
+        self.model = nn.Sequential(
+            nn.Dropout(self.dropout_value),
+            LSTM_model(input_dim, self.hidden_size, self.device, self.num_layers),
+            nn.Flatten(),
+            nn.Linear(self.hidden_size, self.hidden_dim),
+            nn.SiLU(),
+            nn.Dropout(self.dropout_value),
+            nn.Linear(self.hidden_dim, 1),
+            nn.Flatten(start_dim=0),
+        )
    
+
    + +
    + + + + + +
    + +
    +
    +
    + +
    + + + + +
    +
    + +
    + +
    +
    +
    + + + + + +
    +
    + +
    + +
    +

    + + © Copyright 2024, ICE Team. +
    + +

    +
    + +
    +

    + Created using Sphinx 5.3.0. +
    +

    +
    + +
    + + + +
    + +

    + Built with the PyData Sphinx Theme 0.14.0. +

    + +
    + +
    + +
    + + \ No newline at end of file diff --git a/docs/_modules/ice/remaining_useful_life_estimation/models/mlp.html b/docs/_modules/ice/remaining_useful_life_estimation/models/mlp.html index f5e0c46..7f67280 100644 --- a/docs/_modules/ice/remaining_useful_life_estimation/models/mlp.html +++ b/docs/_modules/ice/remaining_useful_life_estimation/models/mlp.html @@ -383,13 +383,16 @@

    Source code for ice.remaining_useful_life_estimation.models.mlp

    self, window_size: int = 32, stride: int = 1, - hidden_dim: int = 256, - batch_size: int = 256, - lr: float = 5e-5, - num_epochs: int = 50, + hidden_dim: int = 512, + batch_size: int = 64, + lr: float = 1e-4, + num_epochs: int = 15, device: str = "cpu", verbose: bool = True, name: str = "mlp_cmapss_rul", + random_seed: int = 42, + val_ratio: float = 0.15, + save_checkpoints: bool = False ): """ Args: @@ -403,10 +406,14 @@

    Source code for ice.remaining_useful_life_estimation.models.mlp

    `cuda` are possible. verbose (bool): If true, show the progress bar in training. name (str): The name of the model for artifact storing. + random_seed (int): Seed for random number generation to ensure reproducible results. + val_ratio (float): Proportion of the dataset used for validation, between 0 and 1. + save_checkpoints (bool): If true, store checkpoints. """ super().__init__( - window_size, stride, batch_size, lr, num_epochs, device, verbose, name + window_size, stride, batch_size, lr, num_epochs, device, verbose, name, random_seed, val_ratio, save_checkpoints ) + self.val_metrics = True self.hidden_dim = hidden_dim self.loss_array = [] @@ -416,13 +423,13 @@

    Source code for ice.remaining_useful_life_estimation.models.mlp

    **{"hidden_dim": ["MODEL", "HIDDEN_DIM"]} ) - def _create_model(self, df: DataFrame, target: Series): - num_sensors = df.shape[1] - + def _create_model(self, input_dim: int, output_dim: int): self.model = nn.Sequential( + nn.Dropout(0.5), nn.Flatten(), - nn.Linear(num_sensors * self.window_size, self.hidden_dim), + nn.Linear(input_dim * self.window_size, self.hidden_dim), nn.ReLU(), + nn.Dropout(0.5), nn.Linear(self.hidden_dim, 1), nn.Flatten(start_dim=0), )
    diff --git a/docs/_modules/index.html b/docs/_modules/index.html index 962bcad..e9571bf 100644 --- a/docs/_modules/index.html +++ b/docs/_modules/index.html @@ -367,6 +367,9 @@

    All modules for which code is available

  • ice.anomaly_detection.metrics
  • ice.anomaly_detection.models.autoencoder
  • ice.anomaly_detection.models.base
    • +
  • ice.anomaly_detection.models.gnn
    • +
  • ice.anomaly_detection.models.stgat
    • +
  • ice.anomaly_detection.models.transformer
  • ice.anomaly_detection.utils
  • ice.base
  • ice.configs
    • @@ -384,6 +387,7 @@

    All modules for which code is available

  • ice.remaining_useful_life_estimation.datasets
  • ice.remaining_useful_life_estimation.metrics
  • ice.remaining_useful_life_estimation.models.base
    • +
  • ice.remaining_useful_life_estimation.models.lstm
  • ice.remaining_useful_life_estimation.models.mlp
  • ice.remaining_useful_life_estimation.utils
diff --git a/docs/_sources/advanced/index.rst.txt b/docs/_sources/advanced/index.rst.txt index 241fa75..16b47c1 100644 --- a/docs/_sources/advanced/index.rst.txt +++ b/docs/_sources/advanced/index.rst.txt @@ -4,4 +4,7 @@ Advanced materials ################## -2nd project stage problem \ No newline at end of file +.. toctree:: + :maxdepth: 2 + + optimization_tutorial \ No newline at end of file diff --git a/docs/_sources/advanced/optimization_tutorial.ipynb.txt b/docs/_sources/advanced/optimization_tutorial.ipynb.txt new file mode 100644 index 0000000..541fbc1 --- /dev/null +++ b/docs/_sources/advanced/optimization_tutorial.ipynb.txt @@ -0,0 +1,881 @@ +{ + "cells": [ + { + "cell_type": "markdown", + "id": "5908e829", + "metadata": {}, + "source": [ + "# Optimization tutorial" + ] + }, + { + "cell_type": "code", + "execution_count": 1, + "id": "7e7ea27e-aca8-4bdc-bcdb-dda20d6ac045", + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "C:\\Users\\user\\conda\\envs\\ice_testing\\Lib\\site-packages\\tqdm\\auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n", + " from .autonotebook import tqdm as notebook_tqdm\n" + ] + } + ], + "source": [ + "from ice.remaining_useful_life_estimation.datasets import RulCmapss\n", + "from ice.remaining_useful_life_estimation.models import MLP" + ] + }, + { + "cell_type": "markdown", + "id": "f9f612b9-f990-40f8-9191-af341827f7d9", + "metadata": {}, + "source": [ + "Create the MLP model and dataset class." + ] + }, + { + "cell_type": "code", + "execution_count": 2, + "id": "81f891c7-de45-448f-a65b-806f2446e6e5", + "metadata": { + "scrolled": true + }, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "Reading data/C-MAPSS/fd1_train.csv: 100%|██████████| 20631/20631 [00:00<00:00, 517496.66it/s]\n", + "Reading data/C-MAPSS/fd2_train.csv: 100%|██████████| 53759/53759 [00:00<00:00, 539376.73it/s]\n", + "Reading data/C-MAPSS/fd3_train.csv: 100%|██████████| 24720/24720 [00:00<00:00, 506176.62it/s]\n", + "Reading data/C-MAPSS/fd4_train.csv: 100%|██████████| 61249/61249 [00:00<00:00, 539069.75it/s]\n", + "Reading data/C-MAPSS/fd1_train.csv: 100%|██████████| 20631/20631 [00:00<00:00, 544734.35it/s]\n", + "Reading data/C-MAPSS/fd2_train.csv: 100%|██████████| 53759/53759 [00:00<00:00, 481589.37it/s]\n", + "Reading data/C-MAPSS/fd3_train.csv: 100%|██████████| 24720/24720 [00:00<00:00, 427933.68it/s]\n", + "Reading data/C-MAPSS/fd4_train.csv: 100%|██████████| 61249/61249 [00:00<00:00, 546792.80it/s]\n", + "Reading data/C-MAPSS/fd1_test.csv: 100%|██████████| 13097/13097 [00:00<00:00, 505412.69it/s]\n", + "Reading data/C-MAPSS/fd2_test.csv: 100%|██████████| 33991/33991 [00:00<00:00, 478933.98it/s]\n", + "Reading data/C-MAPSS/fd3_test.csv: 100%|██████████| 16598/16598 [00:00<00:00, 520419.66it/s]\n", + "Reading data/C-MAPSS/fd4_test.csv: 100%|██████████| 41214/41214 [00:00<00:00, 537036.66it/s]\n", + "Reading data/C-MAPSS/fd1_test.csv: 100%|██████████| 13097/13097 [00:00<00:00, 515225.24it/s]\n", + "Reading data/C-MAPSS/fd2_test.csv: 100%|██████████| 33991/33991 [00:00<00:00, 520644.44it/s]\n", + "Reading data/C-MAPSS/fd3_test.csv: 100%|██████████| 16598/16598 [00:00<00:00, 489809.10it/s]\n", + "Reading data/C-MAPSS/fd4_test.csv: 100%|██████████| 41214/41214 [00:00<00:00, 537023.31it/s]\n" + ] + } + ], + "source": [ + "dataset_class = RulCmapss()\n", + "data, target = dataset_class.df[0], dataset_class.target[0]" + ] + }, + { + "cell_type": "code", + "execution_count": 3, + "id": "6fc9029e-7ce9-4cd8-9353-087b66556b17", + "metadata": {}, + "outputs": [], + "source": [ + "model = MLP(device=\"cuda\")" + ] + }, + { + "cell_type": "markdown", + "id": "f8ff1a72-19f4-4389-9362-35a80b8fb851", + "metadata": {}, + "source": [ + "Optimization **without changing the complexity** of the training process. Tune the lr of the training procedure using validation loss as optimization target" + ] + }, + { + "cell_type": "code", + "execution_count": 4, + "id": "c9e23868-09cb-4693-ad67-f1104245b7df", + "metadata": { + "scrolled": true + }, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "[I 2024-08-13 09:53:33,784] A new study created in memory with name: /parameter_lr study\n" + ] + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "trial step with lr = 0.00018951382914416393\n" + ] + }, + { + "name": "stderr", + "output_type": "stream", + "text": [ + "Creating sequence of samples: 100%|██████████| 100/100 [00:00<00:00, 33442.07it/s]\n", + "Epochs ...: 0%| | 0/5 [00:00\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
FaultTPRFPR
000.96750.0000
110.97380.0000
230.96430.0000
340.95840.0000
450.97310.0000
560.96790.0000
670.96910.0000
790.96510.0000
8100.97880.0000
9110.95260.0000
10120.94180.0001
11130.97800.0000
12150.97520.0000
13160.96080.0000
14170.93570.0000
15180.97170.0000
16190.94820.0000
\n", + "" + ], + "text/plain": [ + " Fault TPR FPR\n", + "0 0 0.9675 0.0000\n", + "1 1 0.9738 0.0000\n", + "2 3 0.9643 0.0000\n", + "3 4 0.9584 0.0000\n", + "4 5 0.9731 0.0000\n", + "5 6 0.9679 0.0000\n", + "6 7 0.9691 0.0000\n", + "7 9 0.9651 0.0000\n", + "8 10 0.9788 0.0000\n", + "9 11 0.9526 0.0000\n", + "10 12 0.9418 0.0001\n", + "11 13 0.9780 0.0000\n", + "12 15 0.9752 0.0000\n", + "13 16 0.9608 0.0000\n", + "14 17 0.9357 0.0000\n", + "15 18 0.9717 0.0000\n", + "16 19 0.9482 0.0000" + ] + }, + "execution_count": 19, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "idx = np.array([1, 2, 4, 5, 6, 7, 8, 10, 11, 12, 13, 14, 16, 17, 18, 19, 20]) - 1\n", + "pd.DataFrame({\n", + " 'Fault': idx,\n", + " 'TPR': np.array(metrics['true_positive_rate'])[idx],\n", + " 'FPR': np.array(metrics['false_positive_rate'])[idx],\n", + "}).round(4)" + ] + }, + { + "cell_type": "code", + "execution_count": 25, + "id": "2ad2b5f0", + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Average TPR: 0.96\n" + ] + } + ], + "source": [ + "print(f'Average TPR: {np.array(metrics[\"true_positive_rate\"])[idx].mean():.2f}')" + ] + }, + { + "cell_type": "code", + "execution_count": 31, + "id": "c3a334ba", + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "96.75\n", + "97.38\n", + "96.43\n", + "95.84\n", + "97.31\n", + "96.79\n", + "96.91\n", + "96.51\n", + "97.88\n", + "95.26\n", + "94.18\n", + "97.80\n", + "97.52\n", + "96.08\n", + "93.57\n", + "97.17\n", + "94.82\n" + ] + } + ], + "source": [ + "for i in np.array(metrics[\"true_positive_rate\"])[idx]*100:\n", + " print(f'{i:.2f}')" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "878cf22c", + "metadata": {}, + "outputs": [], + "source": [] + } + ], + "metadata": { + "kernelspec": { + "display_name": "Python 3 (ipykernel)", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.11.7" + } + }, + "nbformat": 4, + "nbformat_minor": 5 +} diff --git a/docs/_sources/benchmark/hi_sota.ipynb.txt b/docs/_sources/benchmark/hi_sota.ipynb.txt new file mode 100644 index 0000000..c0aa68b --- /dev/null +++ b/docs/_sources/benchmark/hi_sota.ipynb.txt @@ -0,0 +1,338 @@ +{ + "cells": [ + { + "cell_type": "markdown", + "id": "f2da72a3-830f-4a34-b767-230ed0dbf154", + "metadata": {}, + "source": [ + "# Results of HI estimation using Stacked LSTM" + ] + }, + { + "cell_type": "markdown", + "id": "e3de787a-3db3-47c0-a1c6-a2b36cc6586f", + "metadata": {}, + "source": [ + "This notebook presents experimental results of hi estimation on the Milling dataset using the model MLP-256.\n", + "\n", + "Importing libraries." + ] + }, + { + "cell_type": "code", + "execution_count": 1, + "id": "0d2bf80e-d11f-4c3d-ac75-e4f0cfd418f6", + "metadata": { + "scrolled": true + }, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "C:\\Users\\user\\conda\\envs\\ice_testing\\Lib\\site-packages\\tqdm\\auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n", + " from .autonotebook import tqdm as notebook_tqdm\n" + ] + } + ], + "source": [ + "from ice.health_index_estimation.datasets import Milling\n", + "from ice.health_index_estimation.models import MLP, TCM, IE_SBiGRU, Stacked_LSTM\n", + "\n", + "import pandas as pd\n", + "import numpy as np\n", + "import torch\n", + "from tqdm.auto import trange\n", + " " + ] + }, + { + "cell_type": "markdown", + "id": "41bd9f60-51bf-4bb8-adc8-33a6abf923a1", + "metadata": {}, + "source": [ + "Initializing model class and train/test data split" + ] + }, + { + "cell_type": "code", + "execution_count": 2, + "id": "814ba92e-e394-49a4-ac06-7b68361cf078", + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "Reading data/milling/case_1.csv: 100%|██████████| 153000/153000 [00:00<00:00, 1268689.48it/s]\n", + "Reading data/milling/case_2.csv: 100%|██████████| 117000/117000 [00:00<00:00, 1248873.41it/s]\n", + "Reading data/milling/case_3.csv: 100%|██████████| 126000/126000 [00:00<00:00, 1276983.81it/s]\n", + "Reading data/milling/case_4.csv: 100%|██████████| 63000/63000 [00:00<00:00, 1374151.83it/s]\n", + "Reading data/milling/case_5.csv: 100%|██████████| 54000/54000 [00:00<00:00, 1290003.79it/s]\n", + "Reading data/milling/case_6.csv: 100%|██████████| 9000/9000 [00:00<00:00, 1003395.34it/s]\n", + "Reading data/milling/case_7.csv: 100%|██████████| 72000/72000 [00:00<00:00, 1245529.71it/s]\n", + "Reading data/milling/case_8.csv: 100%|██████████| 54000/54000 [00:00<00:00, 1321487.68it/s]\n", + "Reading data/milling/case_9.csv: 100%|██████████| 81000/81000 [00:00<00:00, 1377467.66it/s]\n", + "Reading data/milling/case_10.csv: 100%|██████████| 90000/90000 [00:00<00:00, 1347769.63it/s]\n", + "Reading data/milling/case_11.csv: 100%|██████████| 207000/207000 [00:00<00:00, 1180064.87it/s]\n", + "Reading data/milling/case_12.csv: 100%|██████████| 126000/126000 [00:00<00:00, 1276980.73it/s]\n", + "Reading data/milling/case_13.csv: 100%|██████████| 135000/135000 [00:00<00:00, 1242669.46it/s]\n", + "Reading data/milling/case_14.csv: 100%|██████████| 81000/81000 [00:00<00:00, 1310826.20it/s]\n", + "Reading data/milling/case_15.csv: 100%|██████████| 63000/63000 [00:00<00:00, 1316886.37it/s]\n", + "Reading data/milling/case_16.csv: 100%|██████████| 18000/18000 [00:00<00:00, 1128731.62it/s]\n", + "C:\\Users\\user\\conda\\envs\\ice_testing\\Lib\\site-packages\\scipy\\interpolate\\_interpolate.py:479: RuntimeWarning: invalid value encountered in divide\n", + " slope = (y_hi - y_lo) / (x_hi - x_lo)[:, None]\n" + ] + } + ], + "source": [ + "dataset_class = Milling()\n", + "\n", + "data, target = pd.concat(dataset_class.df), pd.concat(dataset_class.target) \n", + "test_data, test_target = dataset_class.test[0], dataset_class.test_target[0]" + ] + }, + { + "cell_type": "code", + "execution_count": 3, + "id": "75a22b4c-5b1c-4ce0-b74e-6886d1c61c90", + "metadata": {}, + "outputs": [], + "source": [ + "from sklearn.preprocessing import StandardScaler, MinMaxScaler\n", + "import pandas as pd \n", + "\n", + "scaler = MinMaxScaler()\n", + "trainer_data = scaler.fit_transform(data)\n", + "tester_data = scaler.transform(test_data)\n", + "\n", + "trainer_data = pd.DataFrame(trainer_data, index=data.index, columns=data.columns)\n", + "tester_data = pd.DataFrame(tester_data, index=test_data.index, columns=test_data.columns)" + ] + }, + { + "cell_type": "code", + "execution_count": 4, + "id": "049500f5-24a0-4076-a2eb-2e15970e3abb", + "metadata": {}, + "outputs": [], + "source": [ + "# path_to_tar = \"hi_sota/\"" + ] + }, + { + "cell_type": "code", + "execution_count": 5, + "id": "ac338ba4-ab10-41cc-916b-0413bed0d1c6", + "metadata": { + "scrolled": true + }, + "outputs": [], + "source": [ + "model_class = Stacked_LSTM(\n", + " window_size=64,\n", + " stride=1024, # 1024\n", + " batch_size=253, # 256\n", + " lr= 0.0031789041005068647, # 0.0004999805761074147,\n", + " num_epochs=55,\n", + " verbose=True,\n", + " device='cuda'\n", + " )\n", + "# model_class.fit(trainer_data, target)\n", + "model_class.load_checkpoint(path_to_tar + \"stack_sota.tar\")" + ] + }, + { + "cell_type": "code", + "execution_count": 6, + "id": "1512dcc2-1765-42f6-ab71-b8a44aa699df", + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "Creating sequence of samples: 100%|██████████| 14/14 [00:00<00:00, 2809.31it/s]\n", + " \r" + ] + }, + { + "data": { + "text/plain": [ + "{'mse': 0.0022332468596409335, 'rmse': 0.047257241346072384}" + ] + }, + "execution_count": 6, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "model_class.evaluate(tester_data, test_target)" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "defe310c-5d1e-412d-b68d-447a975ef937", + "metadata": {}, + "outputs": [], + "source": [] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "b58a2a2b-9b73-4ac8-959e-4f93db8cd8eb", + "metadata": {}, + "outputs": [], + "source": [] + }, + { + "cell_type": "code", + "execution_count": 7, + "id": "e7638bc7-d161-4266-b2d1-aa3545856853", + "metadata": { + "scrolled": true + }, + "outputs": [], + "source": [ + "model_class = TCM(\n", + " window_size=64,\n", + " stride=1024, # 1024\n", + " batch_size=253, # 256\n", + " lr= 0.0031789041005068647, # 0.0004999805761074147,\n", + " num_epochs=55,\n", + " verbose=True,\n", + " device='cuda'\n", + " )\n", + "# model_class.fit(trainer_data, target)\n", + "model_class.load_checkpoint(path_to_tar + \"TCM_sota.tar\")" + ] + }, + { + "cell_type": "code", + "execution_count": 8, + "id": "550c5609-a0ec-48d3-9149-a4de69ba8368", + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "Creating sequence of samples: 100%|██████████| 14/14 [00:00<00:00, 3511.98it/s]\n", + " \r" + ] + }, + { + "data": { + "text/plain": [ + "{'mse': 0.004014168163365719, 'rmse': 0.06335746335962102}" + ] + }, + "execution_count": 8, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "model_class.evaluate(tester_data, test_target)" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "ede9a8c3-27e5-42df-980e-bcdb1ddcec73", + "metadata": {}, + "outputs": [], + "source": [] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "e1ee17ee-09a8-4e7e-a81d-bea529af3f7f", + "metadata": {}, + "outputs": [], + "source": [] + }, + { + "cell_type": "markdown", + "id": "767c37e2-1750-46d5-9cd7-f628e7322305", + "metadata": {}, + "source": [ + "Training and testing with difference random seed for uncertainty estimation" + ] + }, + { + "cell_type": "code", + "execution_count": 9, + "id": "1c5e6783-49df-46b5-86a0-6f09ee501d0e", + "metadata": { + "scrolled": true + }, + "outputs": [], + "source": [ + "model_class = IE_SBiGRU(\n", + " window_size=64,\n", + " stride=1024, # 1024\n", + " batch_size=253, # 256\n", + " lr= 0.0011, # 0.0004999805761074147,\n", + " num_epochs=35,\n", + " verbose=True,\n", + " device='cuda'\n", + " )\n", + "# model_class.fit(trainer_data, target)\n", + "model_class.load_checkpoint(path_to_tar + \"IE_SBiGRU_sota.tar\")" + ] + }, + { + "cell_type": "code", + "execution_count": 10, + "id": "6425462a-14a5-4580-b082-92e9165d8984", + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "Creating sequence of samples: 100%|██████████| 14/14 [00:00<00:00, 2341.13it/s]\n", + " \r" + ] + }, + { + "data": { + "text/plain": [ + "{'mse': 0.004956771691496658, 'rmse': 0.07040434426579555}" + ] + }, + "execution_count": 10, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "model_class.evaluate(tester_data, test_target)" + ] + } + ], + "metadata": { + "kernelspec": { + "display_name": "Python 3 (ipykernel)", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.11.9" + } + }, + "nbformat": 4, + "nbformat_minor": 5 +} diff --git a/docs/_sources/benchmark/index.rst.txt b/docs/_sources/benchmark/index.rst.txt index e8a7f30..6cae700 100644 --- a/docs/_sources/benchmark/index.rst.txt +++ b/docs/_sources/benchmark/index.rst.txt @@ -9,6 +9,12 @@ Benchmarking :caption: Contents: fd_benchmark_mlp_256 + fd_benchmark_tcn ad_benchmark_autoencodermlp_256 + ad_benchmark_transformer + ad_benchmark_stgat + ad_benchmark_gnn rul_benchmark_lstm_256 - hi_benchmark_mlp_256 \ No newline at end of file + rul_sota + hi_benchmark_mlp_256 + hi_sota \ No newline at end of file diff --git a/docs/_sources/benchmark/rul_sota.ipynb.txt b/docs/_sources/benchmark/rul_sota.ipynb.txt new file mode 100644 index 0000000..4a7e0a9 --- /dev/null +++ b/docs/_sources/benchmark/rul_sota.ipynb.txt @@ -0,0 +1,187 @@ +{ + "cells": [ + { + "cell_type": "markdown", + "id": "f2da72a3-830f-4a34-b767-230ed0dbf154", + "metadata": {}, + "source": [ + "# Results of RUL estimation using IR" + ] + }, + { + "cell_type": "markdown", + "id": "e3de787a-3db3-47c0-a1c6-a2b36cc6586f", + "metadata": {}, + "source": [ + "This notebook presents experimental results of rul estimation on the CMAPSS dataset using the model lstm-256.\n", + "\n", + "Importing libraries." + ] + }, + { + "cell_type": "code", + "execution_count": 1, + "id": "0d2bf80e-d11f-4c3d-ac75-e4f0cfd418f6", + "metadata": { + "scrolled": true + }, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "C:\\Users\\user\\conda\\envs\\ice_testing\\Lib\\site-packages\\tqdm\\auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n", + " from .autonotebook import tqdm as notebook_tqdm\n" + ] + } + ], + "source": [ + "from ice.remaining_useful_life_estimation.datasets import RulCmapss\n", + "from ice.remaining_useful_life_estimation.models import IR \n", + "\n", + "import pandas as pd" + ] + }, + { + "cell_type": "markdown", + "id": "41bd9f60-51bf-4bb8-adc8-33a6abf923a1", + "metadata": {}, + "source": [ + "Initializing model class and train/test data split for fd001 subdataset" + ] + }, + { + "cell_type": "code", + "execution_count": 2, + "id": "814ba92e-e394-49a4-ac06-7b68361cf078", + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "Reading data/C-MAPSS/fd1_train.csv: 100%|██████████| 20631/20631 [00:00<00:00, 504872.87it/s]\n", + "Reading data/C-MAPSS/fd2_train.csv: 100%|██████████| 53759/53759 [00:00<00:00, 533656.45it/s]\n", + "Reading data/C-MAPSS/fd3_train.csv: 100%|██████████| 24720/24720 [00:00<00:00, 496051.49it/s]\n", + "Reading data/C-MAPSS/fd4_train.csv: 100%|██████████| 61249/61249 [00:00<00:00, 532245.75it/s]\n", + "Reading data/C-MAPSS/fd1_train.csv: 100%|██████████| 20631/20631 [00:00<00:00, 547602.44it/s]\n", + "Reading data/C-MAPSS/fd2_train.csv: 100%|██████████| 53759/53759 [00:00<00:00, 528810.42it/s]\n", + "Reading data/C-MAPSS/fd3_train.csv: 100%|██████████| 24720/24720 [00:00<00:00, 496134.57it/s]\n", + "Reading data/C-MAPSS/fd4_train.csv: 100%|██████████| 61249/61249 [00:00<00:00, 529770.68it/s]\n", + "Reading data/C-MAPSS/fd1_test.csv: 100%|██████████| 13097/13097 [00:00<00:00, 486699.50it/s]\n", + "Reading data/C-MAPSS/fd2_test.csv: 100%|██████████| 33991/33991 [00:00<00:00, 501537.62it/s]\n", + "Reading data/C-MAPSS/fd3_test.csv: 100%|██████████| 16598/16598 [00:00<00:00, 512334.66it/s]\n", + "Reading data/C-MAPSS/fd4_test.csv: 100%|██████████| 41214/41214 [00:00<00:00, 523435.48it/s]\n", + "Reading data/C-MAPSS/fd1_test.csv: 100%|██████████| 13097/13097 [00:00<00:00, 474171.77it/s]\n", + "Reading data/C-MAPSS/fd2_test.csv: 100%|██████████| 33991/33991 [00:00<00:00, 501537.62it/s]\n", + "Reading data/C-MAPSS/fd3_test.csv: 100%|██████████| 16598/16598 [00:00<00:00, 520419.66it/s]\n", + "Reading data/C-MAPSS/fd4_test.csv: 100%|██████████| 41214/41214 [00:00<00:00, 519167.37it/s]\n" + ] + } + ], + "source": [ + "dataset_class = RulCmapss()\n", + "\n", + "data, target = dataset_class.df[0], dataset_class.target[0]\n", + "test_data, test_target = dataset_class.test[0], dataset_class.test_target[0] " + ] + }, + { + "cell_type": "code", + "execution_count": 3, + "id": "717c0b98-a3cb-4cef-bf11-b9274617eddc", + "metadata": {}, + "outputs": [], + "source": [ + "from sklearn.preprocessing import StandardScaler, MinMaxScaler\n", + "import pandas as pd \n", + "\n", + "scaler = MinMaxScaler()\n", + "trainer_data = scaler.fit_transform(data)\n", + "tester_data = scaler.transform(test_data)\n", + "\n", + "trainer_data = pd.DataFrame(trainer_data, index=data.index, columns=data.columns)\n", + "tester_data = pd.DataFrame(tester_data, index=test_data.index, columns=test_data.columns)\n", + "\n" + ] + }, + { + "cell_type": "markdown", + "id": "767c37e2-1750-46d5-9cd7-f628e7322305", + "metadata": {}, + "source": [ + "Training and testing with difference random seed for uncertainty estimation" + ] + }, + { + "cell_type": "code", + "execution_count": 4, + "id": "f9bd5cb4-3cbd-48f2-bb70-8154f2a8e051", + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "C:\\Users\\user\\conda\\envs\\ice_testing\\Lib\\site-packages\\torch\\nn\\modules\\transformer.py:306: UserWarning: enable_nested_tensor is True, but self.use_nested_tensor is False because encoder_layer.activation_relu_or_gelu was not True\n", + " warnings.warn(f\"enable_nested_tensor is True, but self.use_nested_tensor is False because {why_not_sparsity_fast_path}\")\n" + ] + } + ], + "source": [ + "model_class = IR()\n", + "model_class.load_checkpoint(\"rul_sota.tar\")\n" + ] + }, + { + "cell_type": "code", + "execution_count": 5, + "id": "15689271-81f5-4197-934a-7ac71c32f057", + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "Creating sequence of samples: 100%|██████████| 100/100 [00:00<00:00, 11148.24it/s]\n", + " \r" + ] + }, + { + "data": { + "text/plain": [ + "{'rmse': 11.99217470692219, 'cmapss_score': 25394.12755711561}" + ] + }, + "execution_count": 5, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "model_class.evaluate(tester_data, test_target)" + ] + } + ], + "metadata": { + "kernelspec": { + "display_name": "Python 3 (ipykernel)", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.11.9" + } + }, + "nbformat": 4, + "nbformat_minor": 5 +} diff --git a/docs/_sources/reference/ice.anomaly_detection.models.rst.txt b/docs/_sources/reference/ice.anomaly_detection.models.rst.txt index a602716..a7bd8f1 100644 --- a/docs/_sources/reference/ice.anomaly_detection.models.rst.txt +++ b/docs/_sources/reference/ice.anomaly_detection.models.rst.txt @@ -6,15 +6,32 @@ BaseAnomalyDetection .. automodule:: ice.anomaly_detection.models.base :members: - :undoc-members: :show-inheritance: AutoEncoderMLP -------------- -.. automodule:: ice.anomaly_detection.models.autoencoder +.. autoclass:: ice.anomaly_detection.models.autoencoder.AutoEncoderMLP :members: - :undoc-members: :show-inheritance: +AnomalyTransformer +------------------ +.. autoclass:: ice.anomaly_detection.models.transformer.AnomalyTransformer + :members: + :show-inheritance: + +STGAT-MAD +------------------ + +.. autoclass:: ice.anomaly_detection.models.stgat.STGAT_MAD + :members: + :show-inheritance: + +GSL-GNN +------------------ + +.. autoclass:: ice.anomaly_detection.models.gnn.GSL_GNN + :members: + :show-inheritance: \ No newline at end of file diff --git a/docs/_sources/reference/ice.fault_diagnosis.models.rst.txt b/docs/_sources/reference/ice.fault_diagnosis.models.rst.txt index 47e43f4..f931a78 100644 --- a/docs/_sources/reference/ice.fault_diagnosis.models.rst.txt +++ b/docs/_sources/reference/ice.fault_diagnosis.models.rst.txt @@ -6,7 +6,6 @@ BaseFaultDiagnosis .. automodule:: ice.fault_diagnosis.models.base :members: - :undoc-members: :show-inheritance: MLP @@ -14,7 +13,6 @@ MLP .. automodule:: ice.fault_diagnosis.models.mlp :members: - :undoc-members: :show-inheritance: TCN @@ -22,6 +20,5 @@ TCN .. automodule:: ice.fault_diagnosis.models.tcn :members: - :undoc-members: :show-inheritance: diff --git a/docs/_sources/reference/ice.health_index_estimation.models.rst.txt b/docs/_sources/reference/ice.health_index_estimation.models.rst.txt index 86f5ce2..ac496e1 100644 --- a/docs/_sources/reference/ice.health_index_estimation.models.rst.txt +++ b/docs/_sources/reference/ice.health_index_estimation.models.rst.txt @@ -6,14 +6,11 @@ BaseRemainingUsefulLifeEstimation .. automodule:: ice.health_index_estimation.models.base :members: - :undoc-members: :show-inheritance: MLP --- -.. automodule:: ice.health_index_estimation.models.mlp +.. autoclass:: ice.health_index_estimation.models.mlp.MLP :members: - :undoc-members: :show-inheritance: - diff --git a/docs/_sources/reference/ice.remaining_useful_life_estimation.models.rst.txt b/docs/_sources/reference/ice.remaining_useful_life_estimation.models.rst.txt index 2ce815c..61d89a0 100644 --- a/docs/_sources/reference/ice.remaining_useful_life_estimation.models.rst.txt +++ b/docs/_sources/reference/ice.remaining_useful_life_estimation.models.rst.txt @@ -6,14 +6,18 @@ BaseHealthIndexEstimation .. automodule:: ice.remaining_useful_life_estimation.models.base :members: - :undoc-members: :show-inheritance: MLP --- -.. automodule:: ice.remaining_useful_life_estimation.models.mlp +.. autoclass:: ice.remaining_useful_life_estimation.models.mlp.MLP :members: - :undoc-members: :show-inheritance: +LSTM +---- + +.. autoclass:: ice.remaining_useful_life_estimation.models.lstm.LSTM + :members: + :show-inheritance: \ No newline at end of file diff --git a/docs/_sources/reference/ice.rst.txt b/docs/_sources/reference/ice.rst.txt deleted file mode 100644 index f83e0bb..0000000 --- a/docs/_sources/reference/ice.rst.txt +++ /dev/null @@ -1,27 +0,0 @@ -ice package -=========== - -Subpackages ------------ - -.. toctree:: - :maxdepth: 4 - :hidden: - - Anomaly detection - ice.configs - ice.fault_diagnosis - ice.health_index_estimation - ice.remaining_useful_life_estimation - -Submodules ----------- - -ice.base module ---------------- - -.. automodule:: ice.base - :members: - :undoc-members: - :show-inheritance: - diff --git a/docs/_sources/reference/modules.rst.txt b/docs/_sources/reference/modules.rst.txt deleted file mode 100644 index d4b70e8..0000000 --- a/docs/_sources/reference/modules.rst.txt +++ /dev/null @@ -1,7 +0,0 @@ -ice -=== - -.. toctree:: - :maxdepth: 4 - - ice diff --git a/docs/_sources/start/datasets.rst.txt b/docs/_sources/start/datasets.rst.txt deleted file mode 100644 index b178b4f..0000000 --- a/docs/_sources/start/datasets.rst.txt +++ /dev/null @@ -1,5 +0,0 @@ -.. _datasets: - -######## -Datasets -######## \ No newline at end of file diff --git a/docs/_sources/start/tasks/task_hi.rst.txt b/docs/_sources/start/tasks/task_hi.rst.txt index 54693e1..428e3b0 100644 --- a/docs/_sources/start/tasks/task_hi.rst.txt +++ b/docs/_sources/start/tasks/task_hi.rst.txt @@ -85,14 +85,14 @@ equipment :math:`i`, :math:`N` is number of test samples. References """""""""" -The reference results of state-of-the-art papers for the original [1] -dataset are presented in table 1. +.. The reference results of state-of-the-art papers for the original [1] + dataset are presented in table 1. -Since the test dataset involves processes with unknown characteristics, it limits most -related HI-based work approaches, especially similarity-based techniques, which require -information about similar process trajectories. + Since the test dataset involves processes with unknown characteristics, it limits most + related HI-based work approaches, especially similarity-based techniques, which require + information about similar process trajectories. -.. table:: Table 1: References for C-Milling dataset, cuts ‘1’, ‘3’, ‘7’ testing, with the original dataset wear target. + .. table:: Table 1: References for C-Milling dataset, cuts ‘1’, ‘3’, ‘7’ testing, with the original dataset wear target. +--------------------------------+-----------------------+ | Model | RMSE (cycles) | diff --git a/docs/_static/hi/HI_deviation.png b/docs/_static/hi/HI_deviation.png deleted file mode 100644 index aed36e9256ed17c68933447ea038ce208e6d043f..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 112570 zcmZs@1zeNs`#;V`gER=zAV^9aq-zq=AyU$aN_PwxsdPw(lF|(#5*slIQIJr&OIk)V zn%~3s|2gNV=da_n4ffRi-1l`|@3LY^6_JlD{`;09jfVPJ#EF)*(`0{q7S(PCi#-)jsERS?}j zuXRA2|N9IW10&oP1M7dEF#?{i|D^&yz}EkJ#>@r%yT@Gc-=D_Z$;JHpHJIpnvl`vH z0^kYXMftHi1_m|x^$*B8=R0r-5DXQC`?}tspIP{Rx+5o-Ij_I8ZnJ>EES7T#{?u3x zUF0jzbSm_X53lUYM~tS7HsmTtCr2kWDn|=Z_7Cr4PoSkC<52JfTKzBg?JH@HkT6oY zzh0|9zca8a)c-XLHsC*iCKed*7Y%xer5yC12XYhNFgdiy_W74bvegZ#IEfhe@qauZ z-E4z%lY(P?dP#r$1$Z0eOqLsjuH00#SNYG*B{)F@$Hcx1O%CMDpg$gStPs$)@mTce ze-G%P4W`^-IB0jDV1t1Fc(f`8foB`4s4Ln3+#&f|E*9s!2`ZuIUv2gp;UpSEAze@I z>*-P5)@Yi>$DUN7%Nsre&G zujB#!Dw9a_%e^$iwg72z3JS8DH^VYB=@k?e7Z7t7e2f9d3(wz*`5g;fp0CO-RW{6i zXO$Kge=J+xG~9@SZ|=NgI>&hZ`gPx%ILUY;>n))b9po57H=3INQ*ll}R5TwP zukB*XnT(FET$VxFx9;X%59SqceN~Uaml@8j-@0dE+bxYpllEFUilGBj4k8tX@4)5y zl@cu%e*Bj^;j+kekB!0e{jJ`CKW`{92}}2MC-?qK0sB#2t?c@atZ4A%FgrgdCo+tP z?m%)`xSC?yb1p)SzlBmoMa8hei9z?3pOKN#j8d$Oapi|Y&}R1Lwr+z)=&~qOYX7TH z-kkAGwYWi{b z7tzkD)?{iEmta2C;=QE-j(NQV3B6|2a+>)A3ya)Y^U$Q9b8Y6-pEoDob5ylmo`w~R zubw&0z!!GAnZj#d{%6uOM?xMv2pg51eVQaYDF+JNC`(OFX6EPDf;Cc*llQ|%d+e(^ zTgVqwf95z(`=e)E#OS>jLa9AUnos4*twuCn?CiLF&5~?T0aYK(G`VxszJKre+5h1~ zvgFv<2`?GrWAQ(ic^!LjR8&-bX103{urvb+6f`u5%F4^hNJ(?|XXpatb<0KdJT&n) zdwP1pm~swfBqV5;KE)1Mo}3+R$=leJ%qsbyMy%~#Ep^A;W@lfnuo}VX5U2~f`uTpq zV;1)K@#Fk6*V}&{G#=8;3><7jf`| zFBUq&H|0kw9>45ngxx^T1iBDH_bo4;2I{w-Dx!>Kt^a78I z!U*0OIIilB94j&U{HXIb=#HTmCnu3UARVnTjplDVCpWHl_%=eD@kYzPsDI97e9MdB zsJi)tTMkJb4&Tjnh>MOMI_FyS-{t=P{EzUbMVcEVtN2(i-5_*PcL1tqQ?;ONhGLNP zvf*#pz@8h{9{H))51*`lr@FuK;Qsb(i|fw(eSf|Yr;jPFdp}f6nfaxqO{bQ7=x1Hj zeSE~-+}u1ioPlJ=!6lQ!*mD|OQSiebodN%bp2Lie~iie7tlZ~5(L?gOib zIIYcmlbU~qQxqEnWFEJYE@(AaZotoL)=cBvdU8iM0$x^Ba(e4>MYQePEK1ycB{QMC z#cf&D?`*5VcMe{v+lttW+Malin-g$aT4hwd@{nXi{&Fc!b~GFP^{aXejbI+C=BDhK z-MC*gbr0sBR~Upfj@x<(wP+8)Kb{x}z!?FfvSFx?xJ-Vz?(YpM4BvLF7OTI`+aKFx z67t!5YI>j<+A($*W+1-TFc*M0s+B$a`TqEPe?ZV~nA6lo6F*>t)(IH|YTWOa{D9kX z>DPK}wEV8zCaA28+uvQDs^HN>Jo-Np6Yv&w8$#kIo!XP{zggMvWT8DYUCcG_sG{v~ z)YvFS=Hy3Gx@@4#ix)387(kt$K5d|hqO1MUh`CS+{TV#%ka(~ zm9HuO%qbAu>s-x2Ex9Re(YqPcqI`hJu$z3Xlfe(C6F3o!ytnpEtfakNE>p}Ed~t>z z>pGBtqGDW^J}tZ-kQw3v$1-(Q{&%$QvO_>~I#0~tZR;4BdO`AxDDs^!JnwK*sWa<_ zTv?5`1qBM!J8GJmqf@$&MaBfNtQ@U)Q#xjw?UcPft3hdT!m5fOu- z7khhq2c>$O(#t6ZDaYdfopRGkL3(R5-L>-l$?@^IJiE>_aJ}lb++2vkcH8k{v?Ln| z-lrj*^W+o1mxF( zqN0uPwu;Ej-^b9IOxA7to>B-zW+`15RF~na3|o3)JMF}Xifs}xfZXw%-=^1sfADTKsc(WKCjV%u8&=~3#~N)n4eiCn{=_Qo#Le0P+Z^<1GZUkRngKQD zs$wU@8&-=qMIorn*RY}S`l6M2>p!k_F#Bk+mN!|~A>2}oltLWq+X5yjZKA3aTQ0sf zazA>I8?4&QBU}wDyAx?^W zMH9lg3wMu)_yebMXuI=44NEgyDiWwNEcRFW4B;MJ*v)E`KBT{7E{M|Y`FlS(tEc%F z_KcT(azc7Jn7Nz?o+va=Ij}Gg;f_t+0QS`v^n(!bV->@v8&oV`AuW_Qp|d`#(~^4>e#X&oFYH=n(5iNtkt;(XuHDiSs!e{dYIi0@4&idXjZT<8q+i*uI@%-;o zja-Z)c;gM&7QcZ|u`ok>i}QhkO_383HRJcK%QHftU>k0s0t{W2sN_?a5Wg$jt`$o=*8R{iE}B{K41e6C%gLhYHL z_t2G_RN(oOn)|TJh3y zlv&~lGkIS3#w^OwUB7>08(0|^7X&a1e=h_Bhh`>3QBnF9ugoK9r%>R_a z{{6HlK#%O?CmtF#5jKUYJ-jWxyV)Ca1&L%zB+0Bv}U>Pqr?(>B?Djhewic*CK+IlM;P(leqRPw>g(&YqU? z$;!FSz8n}&b5_n31}@fEKCb(>*o1-aVF4!Ra~@WHb!LR0W`4&9<+Z*lLmBxliIj$P zB)G|C-4Vcdt(BkM%zis(W>6cuh3dewk%0?^j|L@GUzEI+B@&nT|AwnBuVn_Cw>5&>Pu4>zCXfF?lY_# z`Lx&Z)Vxk-Z@L_a8hDZ4C_2v=HL78~+HTwWzzT9^hcNz8Jk94|sShQW?LGd?W5()( z!p0-i9Kzm^d-=$Ms#+2k-}?;!VT&Wr7EW$-AXjq&Us-4S;80OfDXXY_1OPyI>7(M) z1-mA;UoJoR6(qNLk9jXDHMe;q7qy!G#zma9{Vy^L!j$Fr%BD)k#NJD^T=OlvWyk3n zBO@b5b+S81%ryX_h&YS~0RYkRb398#Tpaq{$e_t@`&CDpW$`egsW`c&_{@$hfSZW_ zjSRE!t_fSmvg96hhuWwnXq%ln@5A$oIa>?8)L>P~g4_r8*5%T(0>O=(gnV?)DQ%7Y z(ze>#q^IcZxce`0SeThHL^6n~+b$)Jwx*Oo+u3R#AOD(lC)0rlsv9*_dEa&Y`4@L6 z^PMO0Kj0*aF52LZ<*9V(VrLm(p|8B(+mP0}`DPqGmTZ<>bdJ5m_=x>4#!~#7>U7@@ zLiEKgU0in7ld`3J$gfM=+2bL)lNZ0`jpq0uZfyfyICf|uo%U9|6gnQY*hzcm1ZCPO zNVO7rbEx2zIzU2bRbB41>KAy{A{RW&AG{KqCnL8>kYiXqY8CBlfNq0=o zNKa1>W7?PgePI&?C+=k#H5%wMl54?O6PJo^v);qJuVb415~~@=+0CM6RMZYSEGUi8 zCWOGVXcBSX15Ti%6)z|(TsWJ%3W$zd1^t$(=JCN)df|$V&I-0Ce7m*btbEV!<+YzN ziwQj5o6gv7-V3Re!o?rv00L|djyXb|nS%qjx_OU4PF`LRiZZ$Qx%X5m{C>kk3T!67M(iMJG8Zu(ZPq-~Lyrc79MX@o8r;3;w0ak5%lBd4 z35EAUvEJMP;@Ul?fXN$RRiyRv=ZnClf7|+@%YR%`BfwQO=CO--uHI+*>+DcQJ_8r6 zC$kP9K6o1pumr0pN4zfSf>}V4d-wMs&|Z*@$;v`Sn(<-}1gb0Gvi#3?c;{9P+D264QbCj)8S zUmTII(7_{tSncdV7JF8Zv3GehnEpO`bu(1J1y^Y=3I1_*X7`b{x{(&L+zWeC9#z2j zRP?*Y#W61IV{dR1c22co86MKRI7r6}GNkX|#Y;@{3=TzjU8fo1K_As$VgQg0f=9^& zG;qtxw4!Y0-ooI<7K$Fn7M~eA%8cf$IFU?9mG?};Me*n+;|NvyDVxB?rI$!d zu`~4gW(45d{6Stp-0{4_)2#{CwVCWAFB# zpnxox)iYPJpQeLc0ko_gy27G!jQw`=PjA+s6fKt2K2WV#UtrRBzjUsS z9z~=9s}knpH9nzVZ5Eqzc~Fc;O#A={UwlnJ$feGAIv7%Mseg^=^L*etH}W6sO_SWN z_Ga%XZ1;1~X4Rp1&T_}C#EZFUQ+?~NIQ--+EH8hspColn8V2>WDcxU8a)E{j6LM!6 zp!gFt&$>DO9NNR=6^?(hzddT1@*>MJcu`MQYoVX(QIQ2ejhs$TOiZK;Srb3+dlL%6 z#AzI1)O_&EydHGLOD}q~+O!(o-BKV$2pSb&>r#JU^3{e*oBHQI^N-dLKFtPQxgXE~ zoPyX?7oRVlT5W@?X`is`$beWX0^VG=UETJa-}*^QX+XzXKagLaVVgN5@J;x8n9>8> zx3IboU-B#Tc``Lm2So%^#SRr8Q$OtAhsl_vMD1WojtQ9$SGuk9t|1mjd;9v(PsMsgK z*~eY0V1*z_JcYVd+-jsuFI^fY0>;GL9h4_|Ydw(NCmwV5+X_P3f;Br{wI?2Gn?b|h zV?A8B@HB25w6G0Ak;uzENBiOX?Gw(gN#t%nTl5ShP~O-kiC&_e1ahsPF~!l~`TM`X zxr|%lpJM64DL6DO`rZhqOG5eh+RoiPHb*O^H5QwGtsDG8gKYC^90-Ao3na*Byi(pd zaoc(+C_05IbF6<+9!b7i=N!N2QC4>T(;)bM1Gw)XEejfen!m!DG}t|nog;2)uQEkE z8O*(JSYrV~HShNme+i1d^GoSvBNGf@F0sdr%G70q_^S8EG)0x4ov24*rMPIz-YTm- zPwWm(>_A2^4TtqAK@XF4(tV zZrxqDsOgg)%-4qh)>eNsz+zXc#=V!g}dTL-;RA1c1KBW$D##t_?p|K}}8i0ZQS zS3kPIj&rClf%jEQJ)`ks8yili#s!=i5zQ=Jy@I%yn1k^XpWmw9IcAG=;m`a=eApYK z?Y6qAOlY94p=D25Uiu+ga23+KZHmF0$m*#ys+<5X4|*qytrHHaXP#@+tosfF!aQvI z!|ttK-tYB%tgLdfexGb;RexDz^Mc?a$`TD{JaeN|W^SV1`eCihh=1dClo7P|g+ z<*jS&=gv)H>0!DzHg%5_D->A%7f-u_=Koc ze%U<)yd|Go;DGR+HptW2n|b+?d{7u=7ZO&W(%5W$TcpiotCNoKMpLB4lUZqy6ZKA6 z@mbFtCN?72axN)+*iDlbQVl!v+dM~CQ0`p99Gm|7EsEQSf|payJo(X4W%A^VOi`My zj{qp}CVKM!v{nk|naO;Jqqci$PFmhzke_I5E zr)4L4AoY%k0w|`JqI(N}qmY7%@pUKT@^hcw?cqLJd(^EM3wi1-i=P%py?~mKLdT_w z?JkgGcDoUoh<^#Qz3uS%=aYmZkIG`5N1zICobf$c?T6Y^H>y`Iku>ge7tF8e0O;OG)K~m`mGQN!Qz4LFooW*@feaJIO9*jA=Mg^9N)km9fSF zsq0?qo~GVTQICc#iFz%%!!$aLs6(NR+%_ zEW^vJjtkqgIi4LGQ#$)~64P>MFLB;?ow(`q^zEmIF*Wfh9n*kZsbUpb^v)*|yZlPM zkvlTyqmldu8Q;xkJAZ!BHxc_9WoC=A)i7w(3OUOug> z6RfjaPo3tdsOMQlS)wNAHjVeieE=we*HYD&Q~gDs(j6lo=*vXmlTHm&fRYpq zFtF3E-3+l>^H#>ay}i3#)aLUu&^bkbdckgvGnIYwi@?{%Zb37`nR0%VxaX*iy#aPe z_)hM`%i7_)$-Uo~)opsVOr4S5)*cyZU#MsLI?|r6ZXqv0zo58L=`oWTsDxl2v5J{c zimWTPhati_5cb%a;aG390MN%+4$z!dv&}xoYx!~4y4rd;`3{|^Q>XiYbojNdw>DBv zU8I#`wq5ecoIByNTlSJH@TAvV;7KRmbl}y6vX)jFL8ktXl@+^$seGllPb6NJo3Gq( zYkydZPD3fQi{w0luqmE^$tVUr#uJ{@J?SDafo**_eI81vHqY{FL1(6pyqCpUEydaH zkH9-N-9w1r36i}eqx#15_G6xdv=m#79&G9g(T<$V!ohiFKh)Y++UQA&_SLP*#l zjh=o9sETEe!DgZJZNLuyzC_pZ4POQto4`KQ7vKcS8bz{(<7dE} zMA~h^>u>cBh@%~JkbvNz9F6|?5_i+%B_I!c`k=VO`D;+BX4AHny|b_<$!-!rr%ofG zE_3GzbG!U^MP9^uw>6=Dz`ZfbI7BdZIg##2dZSkql~DI&Jo%ijF})tLwc5U8nS>W7 z5f@|?$ZEMimynIhzMD4(W}W_!e=WLr7Dh)!<+iVmq~Eh8-{?FmjT3|&IY^zZFP5drYu;*7j45?6DDSvzO5pDo&MzgSb=g@+#|&SfsHAMKRXWf(SebXxajjW>38)|ly=2!0F`#O%L-ePF+{ zvQ=>icP6Vw3=x~`_*;ylgAALrKHdkNl6*WJM@`!%8vUb0MAcjmDxvm1i&oUak#Xh zV3y!W<){|z$#$Z|L7#!E72QOygHnH7I9D z|AX8qHj7S@pqPE@{%kr{S(adtYk82s&eio@zeZeCltk}YmDj_jO;fnXH@zPzgm0f9v5E_^4WNBFJI`3pWqVH2lYK4nuUGn zn+jXZBiQ5mc-g~KJ^wl3_9PvEh3_`qUvS*alm2&s4U*w=^1})*#G+J1>;?d^bc5XL z(`^yE^aD2%M{2f0syn0G_jZHp!l!(Yoe=^0CV>f5m_kO|o=|;Ej`u;a{Gi9%;J5F{ zW6yXj));=S&+<`ca7gn5Vf|4=GVWul>_-c8s@|gXJ(!HNv;^#VJlx#00e+v(YHRy=G& zCXkJ|!JPqHwI!*)~IBb`r!xl}J zkIVIaHi|-NiDaFZ*S*F5T}<&o8TcisE+fLBK(Pv7#*>TZK*VqTe$&YhQ5RBfY-ZaO zIWk4m+=N83knJXg8@le_;DHyPZ$rIiTJQSTr>XLz`;bv{D&yy@%x^_qPI^DI%os~h zqd7oYKs6()XhN0tXz|gZlfl9PQ8@V#cerV!4izz=(6TIhvbF(`!+LA7^btoEZrjr@ z#22f?pYAs<4@T`AR`ojClUs(xXq zucCP3#5&Ev%v&&eYlzXojy|YN;A#cT(K=(p%+>!Q+&Q%sb-VwSp<}8410wA=m=#xu zAMiFq*umNr=J4%%RfX?e1IcZefGk_w$idFOZnw*wX1tswO3&RXBPnSNTL)zNshuUE zS%8{n2C7)X@lOgBdxJc|)e2R<$4fR2Gnb8;RgwQLXurrHTUz)VzsFr>WGEhmPHN_R zB|_v+tXcm0>Cl~2JX#b$W}y4#_<2k4m<%*87pu{sc@I7nPdalt>f-fU>qFSt%IY;t zUoAw2x(62oXB)8$$b(TVgQql<=<#o;_a*wyFL8+EY;8-4otUf~r3cXE>GPnyVZhhYve50P^;?PTbjL#VjZJ7s&vxfq zimQ*`n&Ubrp31LwPAG3n6`OV;_TbynPv#A-Jmx1FG9zN^=Ap7TkXNSEgLb=BzVKC5 z`$jDk#%z)ds)sOeWD^OUE%t`)Q^ySXL;!-_Ata7b+KA9OCGBz{oM{n=8N(+#-EmA~ zp_9pD=SNe{vJ8yX#CoB?%`Shd(88@_AbO2<45`0hBHFB5N1di0^8B?-A{ymx2Gx z+{I+%mEW_7eZt12zj&&$9g+ktM{nJ^1uo&k991qkujRlv44Bp`tQwYwC@1&c*$3#a zYsTe}FPMCSMD|<&aP%;KeJzB$-FjBl4>j=bPtj(Q4Uh`BbdN-oiUL@-c&dm&%A2%j z-%RuxSZ%!vygXTbYgS(G$_)(wK=*i~YyQo1*^6~8GJ1LwC|=I5wXibM5lgqx@7W01 zkDGTDZM*oea6aWeYG;o~+St=qJL|#Fo3$Ns!W?MZLe5$tkMHYG3r&SNWf^fyQ%Myw zh&56kF9tAcHE&B7fvD9APdba8zcVOt@>Wcl}%kuwVwo53a@PYHF!qo3G#Ojp zin{|slg%46fz+t(R8PXWoc&CL#r5JfBF|2@7B+~1Vs?msv#fv2u^OhYHvzay_Jmf- zGOC@J{+J_@cqL{iAI08>CaoHn#L4f_IY&jW%F+^>3ZHOY370LYm3 ziX;+D+~<>RFc!AsE~q`93vy<{)SX3atjgRE@v8~xRkM~N25MJecEp;fveyi-f2Kn* z($I{%4Bgr`{%h@DSuou6=hQ*@d5b ziu=`Vjwh@*(NGxM3rq^SSjxi#ZW4OY-E&FeiUE;t@1wD`rk}}gJLzKyUp#_8evZ1r zG2Ee8-)hBP6H#ej4g{pli>^&nSX0I`3@9sz1Q1Km%npGvdVtNv7mcdi#L46_`j|5| z@Zn#O34xIUlyZW3vyRjx-NHR?QmR>Vxn(65w#Fnp=(4XYP>F-9~KW_>fULLoS<4eKCDbw8Np0N(J5H}2>;Ff_A%o<Pmq#HFQBY=#p6@a@m_vJxf@8q3+Mdj9gLIt!+!$Uif7U8O2>%u>t) z8{y^}wJ(b*H=frP+w=6rowEo$klG>3>fF3+E7Go%#0RDnKl@-E_r6ZP?dSxSD!ao) zOi`0?ZEIpC`s);k+;r;+`QRl*0InCctC z&6-j67S&&PW-rUb^^x)Z3(^uO*Si&Px zC38N~W1UB+Y!z`PY;EyR+uNysHMgCQ4b`MnBRNuI)4inS(oK6~;piE9a&C%r+ikQ~ zi7c#Pvv|{tf)bHte|0%|?&sMU&2%mPB}Lz2S5esbW+v}5Ga?`R`LlAxHJDE0D-3e3 zCT!pT2O=L6fkp_ba-UWnc*vHOe<|)97|cfJA{!t1ptjnE9nWqwUxrf9K{J%K-pyay zJJ2NCl2yYtop^PrFgMvv%%q1mgGE8q(|S`w(#&-iV;Y3Y$RPk~No!1#46J&TIrw}L zmsl<}*d}3=ig+z#{fC;jb0heBl(a;5OM3u~h?)TTlWssVqF}kN!a6;`#IlahL#V3m zcU&2KRTV$?GH;|!Xe4a?Ht*V-Y2WVeuSkME@M9>AjfVA-v?-N+)uJTvd%>aq$vr*@ zdcD5-Hy0o2m)5%y-sDBvSGR<@;4fE~mB|4JLJ-JYIYzj#zu0Y^nYGP)AOjCCAzt@uivPZBfYpq&K2nk`koYAVMZ0mhrcpKi& zb8WdFg#t5dU);Cd*-6vZA1sMe;U5EDVQ;1um)WQFd(D94^91nvCBCG}E&oQsAfX%W z<6N6m|1Yr{BuY_I!*neKChMV+fbuVbc)K_hW~fYU+jR4%a1qbP%P%?{H=~2%B5USH z)g03A4M!y(kK&_iDFBM1;O=x=a1S`Q`bQ&4^TX!t{KuxV_Ixr#w*S*5{N^x5(58n$ z1|-D4q=w*BvXZ)0fQjQ#V-xT0p0)=YJrp1jpfz1nt@k6M3C zBQzMLrexf;=|m-IT)%P(D-Tnen!^-nBcM0pe&K&Z;L;j0pcbs{#aR^>RjqIBG+ygDey06ndt()B=UP(WG?(diY$0)}ZDioeBw5nLk{tjpL>-b*^fP=wKU2+y-Q+GKjy8-_ z@H^lHh(QHyi;9ea(bM&MCV|~UBMuq!XUtrmUSt$axlDMA0;y55@n>8xAWg{q2jtd z=571u0eQ$ycoxp=$t2`HzJPBK_L!PE+RV1{Mw%?v&og+vu*s|@^U|VGH4Lr7*g`LL z3;!4!7{j*4nAvv`-@ofPOw?jOd-e<-Iz9jKC16%wXZZAu;NSVVLkcQ>57ryucv!*R ztCo%EB4?gG0!75aF@PAsSetiJTv{gbPQDqHqw0C$EK2iv?a9ZMv2mAOPAcwg-$p-A zE*Eb-N6!-moo6z(7y=iM=U+%>IMbm4--lQd2KM=DspEb{@bdEVwTsaVwj~Tu9c=9E z3*VHO3WJ~t93=m^rgx;ln)w7HBPXoW4Ys>gbh<-~praT*l0ghQ1x_>3AEc+luOLPp(lJVV2B%e6JR(L1{&^sjoGY-w@N(CT*fs2BcoX{|f%gxI$Q4 zHs@sh^|pScQ7DPbF-F_j%=Tclx$K3dqGH!;$z6tjSHdw1sMt~|Hs6pC-JVEPD~Cz; zoq&(e_eQ&%1Z+7fx>x<_mgp`V56!SARhNeC{(PdK7_n&vJ$!Gbmpq52s@XU(BFnkf zO(PDxn6^@AY$)-M=|OP>-t40!^kP8+GG{mJM=S0J1jx`6cRyw|i#!7is-*9&2ad#Z zx1Q}7Sqru%OSUevE>=1v?%=4i8NU?dXcW>}ZYp(M8)PY=6PiIlyIcLwNEsMnkB^Uw zh!=h_tIZExyH0>rt1N}WX18A!brf!YiC~d=o_XN5y^>;I?urracEgW#WLehCj#<3_ z#aYz{%Of`9@}D2n8k;xPTa7}__)7WaZ1G&W)R4#wBjLLwK!sofLFR)207;i-;)YKQ zB&LW6JS=Q%70tnk%f-Ar)=Vj%+Tc>ZDav4%@%Sf1a)r0%@0?Vj7%+Ex&SpOgmpMn1 zqc&U@JMpJI$4pO8UjYW{w$qVnQmSRs7WK`kzOjGFmU#k@JpCPFrKSVd0l9!()sn%J zn)xc_h%Dx8w1+{F!@gBAv^jUUlK?+0@X)<@ocN1t!{V92Bq_D$(e~~NZ9)6;J`}Wh zh2!A&xU9DuBZ&#~(>WF%fisGC8X&52o3Nzr%%`9)tYyz`B^MvnPdRlkK&OKNGaMFf z&enw9Sy=eKmY+L+-lTKE0kK}(5siL1RZ~+_Nm*HYG=D2z3ixf+A=A~dQWowUqsVUN zFK6xqE-K)aD1ewo0_VjWCTnV?yh`w$e_Ltj*_*8*;-ssI zyCNooO8C&RP6StWc9h@x>ZL@@9vaegJc~2Uys#G+g~>VOSmX|fSiv}0Qrz=typLbJ zv+7=~4!p3>Z*K{9^xM3R1$Z|SQd7gSrC}Rz?39$0YerM%bnQ0NV^H6}cCBD{3hn+v zqXj-jC>ct)pPUR;svMgY`ne{1Rc4S-M*NNa;*4|Vu}vh))}~!~Z=jM<$Arnx8yuV# z$~2d#fgtucOeMBu*(Euy01mlq$A#wk7eiQW9P@&pYUQ##Pe6qpVl-;wnEH6L1gn{R z&&S87W@^i@#aCFDPW+erx$_K&3gqn1rV@(MnSLn4v0Dm7 zb>O-}>w!m(j1E;PA*9%QObX>0-8X^fcSaL`QR+Pj0~xd#ZTd7|k8XVg2rZy5WBwsX z)&hyaAXhC{A$}f>JlH(4-fiC5CI$NucvWCh4Y zQtNZsQ9pjz7q%G5Qorn%4tJW-#N-cL>HK7RCrrijM9C@1mFRFrJRkR zEP5Y-4)|UjME0HbmE(dG)YRt=x%Tzk(t9VtzUI_&s1Nvh?^B5{?J$9R&H{SxHp~O9 z48zenKD#2un`N(0Nf@d^{^nVb2G{4)RVJ1d(lcf6rr21VtCmQP@~6Ao`cq?5BHMZ~ zNPhcK`0M z+ZgTAP?FkhE%C<4Urto3=83&y=YxjPauYQ z)1>h=CuV71HnEyw7f$P(%)|k~<=p*%((lX1Ma%DsRri@ z4L^;)=gqOHieb9fmyc?uXzw6MnImeFOG*Cr%8xyb9-osVJ@lGmHakigz&Jq!%NT&I7%>t%b8ga`I&-uF zQwHFN7%6t_3g5kZx7BJv`>T0-P!PXF`{YHUXGw)bRBo=AZK>Bg?`a{xHhQ~#CSNw% z6po=La?4wOvVBYH=@VVqlvT!v)*0SfoUx@zG+#2MwV&loPYXX&!VQWJump#(v)|Y)9{Cfhv$L3fb8XOivetDC4AtGqL?xt7eJFl zNb!^W8-PJ+_{vR%i+h!vyfX#TsNmp-$}w3$ivc5EYK>31qFyQ*^-#}ou|*N5)zEhS z>D%DmJL4|jTzGuFKEUbu%u_KdPK0W!i?tTb*|yzZhI2t4j!a+!GMMi}40fmaf+UAC6H zfmv-w{d4@CJoJ_~pdt9T2hjdwJD8WnQcveh@Ny}Gm;yu|8Jz4pg+Fh*{E%m;kNHkx zdHFk=mn=Rg&8J&Jg3KK+5LI}{`E31NBx2=%&-`_f9!3o_JZ7g_Mu4dMF#jr2Y4jr@F`>x@45P@(>$Gg{f2ap& zE_89yM~3-}e#0dXKY#i1@oc*-j)+d=9z4$2U|T6^IepnMj-n(4Q(siYTOMW6eVg)i zge}l4+3kI>_VBs|s&OqZdc*lwBGQHe9JUUz^#L#$1M4(NyB*O3hfdNjt^*2&$_CZd z+~&CK4TIMVBZgGHt5|E>FrGEl;V1-7s zoVT*54}a7M?gPb?$Y&Wueodc9w-J%8Gj|(Z-vDv}oB7-aY|5Vr^yr3QULV~bpw~Ky zHb^1}ny@ejfiAv8)UVy7e>R%jK~4%bG_0(~9&ap9Bi$(pGO*Wwdq9uki0A!?`jRSA zBK7?R45V31fq<&*Yv1DL0DQFnxI?EkkhbHBHQP!6#Fg4kTE81{V>`B_MkaA}Rz3tq z7|`4})p+x(HO-y+pMvioA!lX0@=g+lE-iTI(nX)`%(Y3{9<8Vb|5iY6AWNM^F1k&XM`saW8;+xTQU1*`cbkc0hlWc)q6J-xg{0FQIv zO%OXu@A!kSrlzLMrn}RQ6F8iN=@Y=VUt4CkMUqGVNx%SVD1I0k)gy1z2&lv(M@k{j zS-G3}+Y8eW2v$;Q(8?e_X0#~nv2P`p-kyd!jYAiEex*)D|A!KgUM^}7sPLS@X9d+T z1rCd^afR*Yv5EV$BNRM)D0iA=B}+ggD6X0?eO))E8+iW$~1B}-SUf|W)T+0x| zNb&C!Iq!~@+0U)3dgZn3%3nFo{)$l^?Gi_9u(0=9<^v=9=-8HkfH{yYBpmXfZ_Z^} zJHx)nJH{8u+-2mhIXw2|@CI+u}BwL*7g<( z@OM5r$afb*xg|*n#;Q3bd~aRkoY$12Q9&Efvv-Y9!-yH80HaKPULJU=-mwFKO@l2k z=r#TM6tA8n;4M%)ANHl0wfLUo9^QNX_hdn4;EBbm>Q1<`b(@ADel%scZ0NfPCcI=5 zoo)z$6*7!_ejZfZ(033dWv`sq%|0u%s*H&BxN9O~UV9TAh+W}N#wiQ(0K!UJaXq|Q z)Kn>MidED*X3y$F`$Z@1#U$;U7-ZdgjVCRUnoCT##v;B7GVA|f$z+K1y}!Q;;9J?V zOm3&qN@G!8Cd6Klj&aj>xRq559e#1(*`C-OslyY?68ursBL4Mzdi7Z&L*;RQ2a zN-?;8$Qj{ZXglV2Id!3Dl_9-z!|w=(Ux#5n@4_cT!4^J0#(t#Gyw`~9g}|y$7OcT~ zP}d6$;$R0sDX!P$Ck{5Yd34)VOVwY1PoPkK_2r8f@!8q0f-!O7F#pV79Ro;QykP>g z9~tU(NniCS1kZUu2h zpX|0<%k))Hp^?t8sSlE;;4mjNRTMppja@&zY4!+2srv5X?o}1lwrU*lYs{E4F4B>% z)t$xx3v1z+GQU#pHi}nEy_Whb5iZ+NL0|1H)uVQ0a z_-%d18xV+QMT=uICpn~~=s$k^7@Pi8<6pE10k>*G^iryUZn}r-VMYs8(L>KzWqj|o z%lTD%_{L&><`>+VTfQ@E{&P5Obq`-ZYmRcO`s+nkW+|3`iIsTB6q!Il9PC3r z-bt6eHY+1k7;s_TmU+%A_tL+bzu~w-Ti^Ccmp(>SMA2HNw&D7_s_$W%_XLu;sXoP_>)+_P*Uh_&h z#2%H4HK~qFOG{h2?!nj5AxE`c1>F5{byywvaUp&xY4w(#o*q1w9{bz}vTOBE%|O0` z57@TJERNpDQ6^W&C4}@ETgZDT$nG}A$t!E;X;tLCO(=t(qdIUdkKR#C9lcr0Cir@x ziSdcVjb*@;V0NMxAeV70)|Zfsr2;w3vHpC4_%wd@9YWA#G*}T8&&iLtZT*2bMadJt zLa>E&a1&>I5!Zu~)Pw^+^UmONu-S4$HSjeEK-pzyS4LcTwUXidhDxrw`Jk+XA>5f% zd!1ZZSom}$CaMY#4{va6Mzo?h<)5kHs00Z?=q5B6tln<35s&J9&qh8JmL9>P9Kjep z&ph|DtFnDri!$BYvz5ZUN|2O+cp4TF$_Y?qL@l;Uu!o$LF_jcVe|j$6H{c*#W1v*A3y{zI{uV@t0WXOA7Q9`zJ|D@PdXs z^2#J|eH3{LYMaSX_jS^p8i`X#2zL6+rge(&*{KqjT0bK*2(tv(MlS5gwi^0$McIw` zrbp89*O+TgeQXKQ)Xb31)3vK4efKyF%hp@TgFV{_F8B=iQQ{SYQrowMJ%KVy4Vo80 zRn#AEYT%mlfc9NvnSoJnWuencM;(rtcL#wm0roV5#u_`L26s!aBzikSolrIEen~E#A2SdX(76Q})Ux4rbe> zIbQLW{EnnQI`5K?id!&*FVeyeJVrpo%$n-1`)A*QfC&YBZu>rUv{*8Jx`m zupw71+x{dZQac!@M_VvWP@CtbzUes%h1f0%a{dUli1Wdcbj2GfaW&r3@FclMzCdHU z9&Y0!Gp5Z&;)P%f{-V8bu+dWHkNl@Q252ED$v9os!@XW*&!}K*$x>kacS&_)(gMsn zkj9m_PR=jFID0h^MCbB*(+vR4{+xO~m<}vZmV>{*@}T8?CX$wlSb?65h+~&;=2bo=EuhSe)ytkh zBEOb?$P0ywt2JLbJ_f6lX@5eKLDcx+Yu(g9da-8OHQvynwtdDyHSkTX)h5I$N2X}5 zWHpgdV|mnGxI$wZ0}i-@z3t-3xX!0`THczt zw+c^VxrGKTftq%*f-u(U5)(Dx6`FpKYx``pXDB%(WnrxCQW)sN@I3T6q;4rw0I;0D zukpnr%9}SA0f3y`#(8&b@`L?F-0`UtmP3uI(9MCl1sUKgZoIZ8vGv|VLi@_?MFD#( z&}|w4zLXk^VfvplCWxP0ps>u(bwUK-er~25S>X}HC>1wGlTpnYG;fpb8Ru#Qn$>gF zZqUxZr_yZR|G-j;ap#kA_F3$Z zt;w(IvtUqVQp!{;D&kEBd@)YW)%mIg$uW*a^MRwGA+GEwG#-!^P))NQdd*LuDH{G* zqgHfnsVl$%%9>m2`7j@(v}-5veXIS;Q9D4bK}bkA1|*=xCIZb_;(+;rJtYiaQyK%c zq9C3=33Kx*YD3GP|Gg0_XUJ3|uZfdvr^;3i-guz{x)ke9xsWh-vQ?TFoh|YK3UvZ) z)Mz-j7CAvA3@bz?Y*!t&l#IAlGFje{w_oh$ch$|V)kmjEf3ggHi6%902gHXkYId`z68HV5)itbs zC1|z*-Z4(|X347&EahYGDp~_IgqR6NSACGBqbUmD?c1B1yOi7m0|SCg%dk)HBLCOK z+e23q!mM@VDCLD%D$r=*sI(-c#mBpI2Tk6(7^?g3ujTV^!@ZNF+lU4`A zKgm%VNz@Z#ebynDTPgl73gtjddjNzb)j@wEgCeE+J(5ye2{M{ZJ7`{ay%Z_#+Q^u| z9C2MRirx)wSICnDc?7-S|Kxw{Bf2>$A-iq$q5T>1f?E48lBsa#i`$KOMFv4Z?MT>7 zmX~RP@W(U)B`Z9?n{%t`YXPqRUH)x!HT7+olv_ri7f!IMA5uBSb0Qr(jVSt(Q^6m6 zUH62AneSxaH4@e=1-}*m$s(1ehz>^F?|%A-W+|{OUm+s~OAcD!4@|%c9J4la3)JZ;-`y5CfYP8sdwS$VF4)omE>Q)FAS+>KA9*&96$ma zInf9dgbxoDG*n8moKrpjS)qKY9nz>Ahq66z;aEZgJ1_THE{ zeoh_nh>bJo05YihWpAus8WmqkR4QIgun<%%4E+rA)>Bgp0j`F@_7MDRT8WQ+kAe8s zD?^v^e;*r0_%fN)5ojPVEW#okIpTBeb{Q7?J{J3u{p38N&74dGWBXPOwl7eB$w$!S zv?wGoL6rgd7A0Z9f@Q5GHeu|iLgf`Woa=uVNW zPRMC3#Ga|h`zskSv1J@!MJ5=ly{TG1%kO{7|H#}vSZq**7y`8L9+1o;BvJC*5q zvHL3%Ux3-y|L4zC?GsdN@|8VT;h6tw;Xq_*{QMdSxL%-nEV+^S14e2f)5N{Vd%!qh=i(;~ zti(YO@Z5O$f1VqGk$@(jx#1A5{HaP-j8ZyLEne|`NX@wlqNw-{2;Rl-Jkgztqu`ZN zbhgN3VAxaa{|ULK@>0J;|3tVA>);l>AA)k8f+|ONk|`kJ-6Tui5P(C}v?1MdSm z9H2eMirw=kXbAD}5X=10_q&*Gw7v!`muke>Srcf!v9+-6q4oo8_jZQ>xXvR#90X_? zE}~J)@(3?R#U6zHfub#2qeA}?cGPyQd-S~<4Y=n9_>ytozU8>Ukr0Kyv`W=E{~u@k z;gK+|tn4N(3~f@@ucG~;RR(UC`f~Bzf&j|n5uNQ2t49;KJwUqDp&Rfc;#rU zfkDUp2c3eP#C8e#MN?R5u z`(K6+dA{QM>BHAsqZ7~(X)Ihc#c^-VRoSS8fFOxk0R{LbpVX1|Jt#DrJKA3p9A~^3 zy(@gm`-j|0N=0`_(tIAn5HibPE(N01*Vf6lKL~CQ+`p~(efzQDW0gbNV!oyYgG~mC zLKw=AJ8Mp}g}2x=3DF^OY>o-rLzl#wghdQ=Mw|fH*m@ln^eH|t>n|jVvIlfrUszA0 zV_{9M>Dg6K4oZ{>PXn}Rw>05XUBG7KVt=zP58Lf5TcLS2Lstspujr-0`30aWD}08& zCu*~I-GwI>;;rj8tq$5swK?85cxW6ify)oC9%hvRW$CNNT<8LHzWROjse6+Z?Z1!W z6$^s8RO^KQc|fP0%F}1>nx4JN1^jt#d|5z^`&$WF31C>HCB%4*#R|)!Dg~q|*)-@j zdSF}zF4Gz%xn`Kt==VJeFHS^s>-g3lC(xdq7L<=LoZ89ct|HbHE92VP%4#?y6aIS6 z`mJRMiXUo{$#L8>Z4jDY=kI@S zh48C$JJCX0J+FeY#53b`$Gb5!9n(c)*@U^T^7Mx zrI}gk(Dy5bG2zo%s9Umb*SdhCMso#8vLevM%1Xk3z54k(vs&4?^~I{&_AV7Q?9pp?82kW6`fM zz8UUTws00aUweD@+}zx_loZdIpHn!Y?0StmI|wf>NBPvoTLH6>RX|5`b$1j{;j*nB z)Ov<-(&#mP0-%>6g2~nuMfTRkcH0yUG0*rltEt}<>fKk_YZcEmZ1KDpV_&2QzwcNl zD33_wd979xq8*Q!p&{XfiHFw*NJE%cC|7&_gTGb)riIl9c)jtV69+q(u>l(9YUyk0 zY0xc+J#)CS=r?kieq@>h0Ed#GiUeVpALET%JBE_HL@BW3=Aq?nd0g&nL2NzJ&jnB{A+!G5HV!#_s#ZRxJ3u zjqX{~v+V5;Y>P`4C>B6p){zMz7Q@r*JW)eG98kr{_$->3n7GwmbF^N)3)6W3?$5-s z*R=Jc8{PP_y$AcbuM6`isIe+LVkk5u-LFVFTupsUA>49$=-zi7@^D{oFIDvFa5RN$ z-M1<0qCtim(4-~WHx!Agh2Eu7X-L#_nvZ~9 zRAP*rk}|eVy}cQ(uPRw&*1sKvKuYwKQwRUrJT)x9%xainC03TJ(Xm?+SITxu2^ozM zw92~7E#>rM!|I5SDj-$tmUU;y(!wP0^&Li-D6xwK7K4i@LXTSC5p={D$c4yvH-J_N5&hFrh^Xa zz{d~wkOzw+k@oeH^4bTFkz$y&PTd9|9HlnlB+_H#UknT@KKh31 zqB!R>JZ4??I3+@Fr=ymSNo+$&tn<#wM{Xn6%e8&4e=F6pSv@bi6JCEhoGxII1^Lm( zeWY*GvQO{~Q)u;;gX$eaAK`A2*PGITwn_xr{xfHqGJ$?st#-iGF7nact`!dDM)mwn z8xRn1c3*cOvBjWpv-C_B~4|MurV_)vBiwV{rmW|%)xKS6YWpx8!Q#)Yzd5) z2rsgc69NfAr^vQ~gR@?PT9C|`&WV=dpTl8o2FO03(E`t`!9F96`)oZ>U&O1GIc7=+ z?zPrTMmPGXYG3QBP@N$4Cz!cknPlF?;rxYAIOM2$P?^B;OAF+3yS9}^9LLwC3=_k6l`f=$j z#BD|OK+3t#i{_X&sRrJFgigQLi=PZV z_@jA8t5QFP7BNdh%%b5!IwH4s$gE}@FmRF<`CP@}?x{O99jcqLT>YlSvOd(0$=P7^ ziHYd&shCGv(MCL_x2J!i+=E!!q2abQyMiw=F&QYXpXWLa`m;CvR|Gzh7qVMleYUd! zsqWhoM<#uFm&WVL&&@C5M7@j!nSq^4eLnURv31~Sdcbd%o+i1sd&>nES~4=2`Lw?4 z^ttfk)h_8cfXC2yGDLRAChbee7I3v1Hhh>%Hu^x%wo4Om>*Ra4Z_o$0@Gea;E^P^X zJQDM{EY*7QP0d>?MYGCSs{Titg4%zlgoEf96ZV=o5&GU}facP%!hFzEthZxu+z7TBXQj7(1Cr&Bl!p7@je z8RM_QN!-1)qew!3Sgmhgav+V~eYfDjxugyJ$SnFXqX)N--I}?>7lr1|5H-^#WNe z`kxtW4Pg+%V6==DbIF~PJ?e%QqS)9E?s$cF$Ft}}|Z z4}2X&eUp$^;dU#+GUd z;<=8a3T8V?%sdwZe5B+Lo2GX7N`uIhImWLC;+YII9YGKG@U-;88cia0m7|9O4nK@p zJrBKjuvVnJA>~1%y@g`n=c#r42`X2AfXj#&mxK_O>5@gom3#G4c6uU*mm!WNY zL3~o(iizxMV)JBxk@F6`+t`+w>$eY=ODl@JU*T%ki(N=@bEAIut=AGRs0Ni=XX^ee z)d9@{5kvdwMGlDait>=X`GRnXj)87QZ?x`0Hj^jN{AQ8wvB@tpUeRh9!eROiIf4dujHu{JxU<78(6cp*`D1b-*!6}Wx!268hn!a zJko77yor@H4Jxf9#IQ*-)3-h8lL7Ui?m+KgV<0AsOJWdK&? zO_uA3#iyXOw&fyr)Fi*&j$U1fc(7yQ$+Y0UIYi%J>YplD6&pR=rd7o`sltLd?>WnX#IxBcI&>| z3KPaBGUwX<_GQlv(l6H1^nGbbO>oBMrbXhpb20^NyCBB)%mq*5>)p;;+VYMSMBtOjx(StbS7^AwPVp?}*4k z{UuaH;Yw{U@|$pn7gpkQUMuwV|Y8CT$ck#%&+u+*ZUk7(akj>dEv72`%TS93AMTB_gjV) zHPBK~9Z<-1Da8NYF4ieId05mZF?5vG{aF&*YNAG@$Fwl2PVSK70DVJDV+n`es#Q>}0$%MslaHN?$>j!TRR4S7i5 z=;-emQtmJCwe+}ql&<3Ce!OiaX*ZSz6NM=!krlNI{6=+?vf|-(U5UxUZR_r-d`ILt zaceg(A;xX@pth4FlMTbS>q|wlN0`|09p(`pXYRjt?k~$0)5qKhM^Wj#DRls+t@;a6~N%1T^6kl|CXF zn~{O7AKUF>L7aE5lAeUjcIegMxSyiVnHBvo$H(>{l;6D>G2(!|G=FN@-4abpy(EWm z{t_{RaplDgYx%i~273(j(Sn}KY>?P@Pc^F(V7Eko_}?t-FlVu9mG0th5{iV@XNLU~ zNUqos_T&=OskjJW&NKBoFFk_Lx{sR=S2QwM3H`IF?KvaI4J&K&f`Nf&cZc#A} zcLsw}U7dffWrWgBQRs=eg)^5i>HcMfhY|lM^Vq8|O@_~end+v;x0 z`!z9F?1Y5x(E~y6_#r{@P*gVxpOS0M{NU}t`TG9`mAeWzzS(f$Vu2)eV>(@<;}vvDI&nB|Y}_m$ajDSM811nhj~n$v%O4X(cPhaK48QzRn2p ztv7}7?*bk}+qw7SPKN?)alb$ERC|Dth&e#KUd`}W<}$tqd||o)+XOqR6*5*$!<+nA zO1|=0gKX#%y4SBKbVmUtB;bae%1lN=vdR%~jRm-;egW#YLE-zOhkFNMfDdPPNw$N# zUGAdx@0lNq6Svdjir&4f4uI2NI*&>iiVNTL>f23hBHiZ7*?_))VKseem)z?9WoTjC z)POJ;&u<}GU|sjX$YETs&CWt69p}J1rt^*@`<9CHfWjfe=V5ob6}rT{rr*aJa+hXV z2g7wt7{6Sg$Q<>=gpD?F8o$OTP>#$lxvQ*xYG+Y`z!yd$M1$aG(TTZ_mmGkcW%&$L zw>n)0oQa(+WI^@z*=kH2oX|zdep%B?pyAs)p7*{AFo zIT~W&R{G=>HF90U(qas;2G>P|%zI`urKO84hi_ zI?H;%c7i0g!eWI_(Sd**WPX{e$nsyt-ilmeJu)WmFE+?BNzWk4|8ksvFzk34B9{8| z>`y$j!vtmoo@zLfua@_|&6w3r5gT~-z#l1c0Crz15NOX7Qvp2?y%a^j*20v#5su@h zL^la(EMtb*32DE#-&|oOqwkjTKayUV;u^_%io~7V!0AN8X-J-NDiN`w@f*o)7p|=8 z|3&JSzVc=`SjZj`UJnbBRl)w};Z3C*dx6-xDr{_QOji%UH1vFYWG`O)`L|W@E){V z@H#m4jV!}A|nEA-K^3KBAnyz1*qMEGde z6Q^B_@~xl!8aA8wSWS^9H{L7NT;|1@K|S$kt+eX6iove!MweN^Qt@o8Kg^9b#K{C| z4n4j~F5)#mq4E!YvpNtZH;P<-%`x+b!_@DpABs5kmVWZ~#e}~hw@!Q2ccs_M?-da~ z`FDf{nP^sLDPcXNN*t%phgn~?y9mE=tZrS(@Fdv>hjFga9ND4F>M3Y#{{#YbGU1LZ z0*Bkt1fD-2f|zUa;J7@PH;EeF8p((R8a)vHNm)QVfap@-TJs~Y&ER?7W^25?nU<~> z9&Rh=1AqY&6|^0Pi2i@i;#8G|kJU9n!l(T46lj9h1}@%HsSZKxw?TK;B2+WOBgJ=< z!@_p!!PLCG2QK{`9W}}fi%Ti|e*ncz9R}>Xs%t_Km)=5KAa|4+lhT3y#WTZW!kZVh z>F2Ze7#?#%RQXO&IP=2fC8EHO0|`hN+i7{;{IVJuYBZ`b16EKj@f3~hTQetAF%JKz zn%K`m=GdWdL+QnJL0k7m6Cf+}W)D&zIwt()2bi$}iXmK=c#7SR@r+NCPoHv|8?JSS zGY)HmW8Et7(sVT%%U3&p3QtEr zHPQPTZJFmNhLPW@&;e&eIZ|+-1W7>H%;$P|yG#Uhp$ zF)tl|Y{$p+2r%suP5!(W|K2d!IVbGmz~8Ts^2A;W*DTf762`BUQ2@kk(T~}xD9tr- zMC*dPXh0Qh`#6b*VGo$H;7b^zP~wKjuv~DyI;~;wW)ok9^aKFp7IYy(N6kcSx0@VQ z=ED^E`S~k3VKgdeH#6V%NdL)iR*;aN%eUqVrim=ZR=0`-n;Bj6ooW-?Ree&JfK8xter)zAb7^u@Clwa8%(Ie+b*2GE&i^I;hqn zX6U#(VROY^r7pBF*-y4U!aje~I1o?YN zov*(wuJ84O(^H-(voDHUTyJtOk`M0q=zKwkVb>`Ut@;~__++S_D+D|thd+goK_`mz z7FwyrHslk<6kJ43P-)(AH(s;ok@qFJ-=Uofj55!8{U{y~!d7%@9O$iN`0_gp&JUXJ zTiS%2-Qh2sYK__bgYF5+xAUcYTU)=vr4@|=l%BaD5cwltWE^NyJ+fwiLhox#OfT?{ zyq;SZY}W+xEISjW)kEG@`k&>E@gf$Q$k>VE>?%;02!=ODj%!*T(_OxXrA}&n5(m4_ z4A;M6E;1ahta56rpV7nKEzdMsO`W%odO9V&>OVngMFy0H$MkZaQqEBf*S38PKxCVU z^S0qky{^pQAYF<>scyu0G7zVgD5p6EK7RkF7Lp>R@WN#(|-2%%$eWzCrC<37!tLKFZ#nE< zxf>Xbx64kubJTeo&R&{+)O(e~Agx1|e{R(^V0W;7uP9wEhWI&`dEqGE^FcTizEK#v zk2N}@6ubR9QSVb~eSpuNJ?hN-39l2s<6#E7oKEL8)1X}B3h`$4r74B*HQI_ASw@9} z0k+32v8It6*FW)W=lNrL-P3;%BLvOHh!PWpHH@w~?1ShwSIaXd?QX?ay-YQ=p zMS8U&v@4Mvru<)uNQv$#Oj7)l%{29vi#VC~9rfha-CeF+sxMGQM9iiv2X>^JnAHZa z3+M+odelcG<8YegWBnp+^n&oZv`$R6@9MR8Bx~;9i|%)H1)bC*sUE@XG_Kct>0-D2 zJ!63dDWF!P@WZMmgPv+|$@`OLY~1Y!E+Bf2WSG|6JrNEKTQ$}Sjl4MY|2Xt=k*RBw zY{EvA!#fu^Nmsa!&&Kz{%gL;J;z!oN^=7hD7Y^SF&>^M-`drjsF|_aSEpBRUW`PJn zbfLK0^1m(1MrQv3*Xktd?q8OeL8+)DUWQlbQT-<%mFx!NB`CgIuKe zKuMgtv-7aoH%4Y=3tCeZQh+MF=GS)z6Gha>$PsYzn}Wiiv0;ivv)UH-I#%Y2m1!RkvjgF(LMqh zPq%+Fyx7@%aac`*YW(Gm-(tCMUF!H{gbF*a>FdcAB*RhEOeCcJ2}I}@pJw}0u0l#f zw)6#=#{S6!=z;qrDz037JwsSWz|JdRI~CKXe~KkG5vI}TV7v^nf0wQLpz(dDF~65j z?gyCy!|`5NGkd^$Y3UiU0+^_RA>c>id3(}F51Tl;13GSR&CFit1B8g%HL|wHgtiLM zHOc*W-P1QJSyc27fV31Jh!jfzHIs{1l%{_Mb;~rqYIsIK+YH zDqhWt@hSPl63YBJI9K!8J-$yvKqBs8@8YmW z;DRvZRm3@=IqAWsZoMi(?c zd<)t&`tbqfU_Em~h^dhK@58~vFlL3QzYzWCQ{~WdLHe4To6CL?gr!==>adrTKm6D` zV2ogC>2a$Gev;h(LQknRxqO4uZjCh{b7aJei=M+7Im}Q5u8mJ zt$ce=W28Z%q3STb{Xek$H9}vNyoisR9XqD1yJ|D-+H7`ob|CV3xYp~Ni<9T5~;=9vq1<%L#7AfAy>tyipSGq<`Z z&$37MhuglcEYZZ$^-(#u$`dw22;-3pMycJy!LxK2COfa>kOTfJ-xA+6lv_88Im zgwG2qg8fXL4s%6vsm-z~SpSr3zUkscm{(ym-sq2`!|WS2NpJ6AZSSR|LJ8czd$Tdm zDT-VWi(aiD317~cq;LiId=y-e%VziB5Voj8bi18tF_r?PyZFR56aA$tTWwYAYio4# zBxtymwS(8Tm>-kLj11J&Ol26<+Uu`X_P+cZXI~*mFN&zU#*bz@LI&XLwn$_l$7ADP zqf>%b=xDL<>b*si+k)h|bfhe8v37BZIz9nZ-sst z8XzIJTHA*tM8y%txt{mKR+?4otH~gAXt6RIDzooqx*N^o4-N80EW7p4Z<|kSNVYrx zWHx@j+d_~IK%XU-@Ckdr`r+@aH*No`W7{|Ji;ec{*CvqU;vqvxj@P$r1U;5;b)Les&Lqdx z{bkXeYj^nvWit`(KDqs_m+aS2`zP)0Ur(gbGfbV5c<_BQN12LwId*$K0Q9Lg#75@`I zlOzTPQ=Zu}p}XxUMY?j6xx21IT>aa?&bxd*^?M(fmL}(bG3jUY`J}M?L8dlr+Z@MH zlNQf&CBseHz;F*Q>Mb3A4~y14y2zk2oQJ}1%i6oE3bXM0_S*W$cwtV*{69TDtugrn z)fPTWKcg$La9`Ukn`42Lk16k;ewbR?#69>EzF+nCf@-9TC;P4gbKfX|mGGIn6XqnI zTQ@W;*}F8YS^O0G4~hbkFKUF{S$%DPWRxBuCFKm1Nc&TQ920W8Ea@P@_FMO2GW#Xy z2D?coZz$XcOm&MF>LVSKKZpcrgG37uvDWq3=z(U(((NB0fp!l)xNqe|V$G1@h1)?+ z^}maRF3Nsgs9X=F_<B;dbV!&L^S++y~vD8A;V_1yK#jK{XwGXK7cepP{jCE?m+l%he$lZ`a zec~!yq*i#w;Cu`6^1%MKyFkP!hGvK|sKwAq-Ky;uJyT{L*6q@NhwY0On1 z)LZA6h>k(=Rn#wufya{*+>W9OGHCvAU~_g5N zq;@{>J7P-|>82T2y-1jN7{--qk%AGBC1!Ev{ZdRM3o-Wn#4)iyxwF{-%mBOvcz1n( zLUGpR18jwx;NdfwzGp`Y{-D*(_#W@8SA)9X>$B~&58=87PK6FsIY%9u=TrX#r)(T# z+q$=G?22!dcWD}nv$j0G&`!3$r+?+kfQWRXl8$<*y=t!tYO7B4Tb##*c0WavAPjWy z>xj0>W0h0}!Rt?1*dNfPP8=*RFY&SqLu%Ks<}sl|;uF4wS=A<~@Mb#uF0;Kjd25># zhPIe3uOm*Wdweo78RnV3C-nitX$-W(sQiGJng`_Aopk*|6fc&$Y{gTtP+i>0<9eFnK5$@yJ+QZ+XX_uA~Ybov<@tfYB(hWgl= z?stkY76fh?&Sc^=rJ&HlQf=D{oRo{hM0F-*4^eqhc58%ph;Z&)styQ_>#O~OygG$HySBGJ@fskQ z1@g@|otKM0UeVGH%hi4IuQe-0aDNDDS^N>f{&XYH9u-56Zy0_$WaNgP#>`X=#mF}VYi)>g7vHsbQ1%E+55D3-((O(JgL^3l{a z<&Dh@Q@77uPF`M~)9{l&4Nxu&e!D+(2|D_xda^`Ci*7z!6*Y595j(NCKc>NYLD_si8ffS}*} z6_D>Mn(_Et)e=MF$(>>kUsd8xFwel8xmL#%==l25aK+38$wg%X#TsLQ{G~xQH7%J*kH2yZjNzIZNHzw{+T*m}po!hi z-SO#)#0NZ?tk-(_75>W#fIn3;Y(GNAh zfxg9=7!!(jq5&@)LJmj{wQ2-QO0%_=(N`Lq`0J}m?P&Wgg=7);!c@@H4)%_dIf7%n?GzeS;EM*@|YVP<;Y(}_VEYliwNX#MHqXYh| z#>j56LkOR$OUlgY_Jcy{{R(`)7R+e&=fFW zd1rL7H_bS#zqi`6qY!pP6t|!H?OTwBz86Z7VixZULSTlk5HOWmZWIQz^}fp9#nDSR zj*8T@Pyb6268RE%IF*z83TyKKtgrtgJXNr8050`R&=>RFXEb{LY});_CULDB@>KxA z<~^*h+6}Dl72#SA1t&LfnUsbS5JTC19Bxyi#FXUg+AN`+e(SrNQ10{Sr`$|!Tlp@ z`hv5jQTVZJBj%@W{F!<@3c^J(-`PGvF$OBD?57%A>K%d}1dX?xN zXdPrh2;4fpHet=4g7J~)12J5AvWloOmLJH5&hLI6u!()Mug;0aPy5T*8 zGA`(+4vQWlR5FBh8P_aDP+9MRg-n0{8*1?wAM0dtTwr*K*!6&7lq00Zib%3jHviM& zPQl?~iTW?Ow1z|$4c;_kNN`2ZW`_3a$81h8`1nEw!?Z8w4Eq_=-PJGp${?^@-j6{y zKR~!E=hY2p6K(ZzfVmcr{y9NNKZN>Dz>xbH$z5P-bcdx}u&&@)$$j2GU_z@xyOVIK zm6*h8ru?cumHM;=NSw0$(jlBtUX;i4<k6w&Lvf}^hHF|$|da^_`x^-1cQz!i$8u*3|^u2)Vf9K?bpG?nMnDWcUSa< zgGWYizNC`K>qYUII*!KiG)6JU{79@b+EE$p){hB#QM=FemW|GrFn7BaF6Vg)D6Np2 z(Y^lVd^uV@LSjS2FW%@c4A_C|aCUchUotZXwY7->%09W9T1hK6UBsa3_kjr>L3_@q zBB>8fW91@eeH5Jy%`_jI=17r!f#ED0pEUXmXtv*ri?^ekuqzI@#Q+n{cYH+XS~MOr z7FQy~TDJn}QVliPl3S<30kSR#b@+$yx-jhbmiRb*x{aJH$!Gkb@MRv#{mVIgf~@|p zHIj>DinVm*r^C^EgqW~HV(7s-`|hb(QP91soxv6LrE`%|49&!%P{r*TNkM7VDE=xm zh39%bEO5b^VgqN*mp^%2m3i^06AbV6%9l#&I!>1-(Pg-8JX-6;;ycqV~Tn@b>btKo*S~ zPiRLCwH-cC62#)+15`xOz?gYjG5Gu_!0hFI$7DD=9)s$s7f<4GzrS1*8=`=jakGA)g>6X-JX;j7t!2D8 z-2IKJ7M%Mo(o1`PjAkNt1xyz57-&Av{uk4EjSyyTQtKKESY;Sufq3dZ3(2T9 zTG#nSO6zV=Nrpt59J!O2UJ(DXrZUY!xhw*EI#9wvlg0K*pRo);BhmX?nyhS=+8{9T z?M3I^m{>D#p{K%hQRMCskCJC-b4-q8pKgwdJ%gU;kNYQ}dc8>`-m59qx?vs(9q{d9 z?5CngN7EqBPg*MH2V$fx()LtBCkZi*N0yaipW++46+!o{Mvuo_GZ1h$dbM)l;y90- z-zox>53b~oHc|k8Bl_(2(Q&dF5VU#%Jv@4=b6}_hIoG>fJ{*5y-OF(|w11ZB$c!xP zP7>;a@#>di4u};c(KX(qDHxx4EQ`TanBiStO8;KLK@(GJQEo)-sx4amS!J|Gy|*J{ z)>iYK-qmRlxl@rN9;W3@Q(4lQru9NKF9MqW@ueeP%%wEAXjw77ae!L!R|(?F5sxU0 zr4GCHw=-ZerEb}X+bga1s22xIqRI2nnVxMrhNYT7lTmlYW&atF#khIjDEP*1<&9i? zR<|gWr|G|doFOfa;#-shZ9x0Za}6~4&jNPvt-qXW7(OZkXyz|LoA}CA(q)f)&BidL99hz$E!33d;G(Vf@a< ze4=XL_J+Wm^H{J)-CTds>SF(MGHCNs=c$%GyEBhSj&*8BBDV@h*~;bp(|*|kCf3Pu z3qO*ScdIl~s-PGuV_}UEfemg6x6zxkE)uK8EL)OUCfpl`0fyan2e% zhAcR05;uZ3zQAgPxIzX!a&Dk7K@tv4L|HTd<{rYS|t}%^${-bP+kt7Tsj)u zNH~=EDyiKmm*4tP+wA^oSW(hlaf^f0e!)VYf=p!EKiOY!Xigq?FWiGszU;f#tO+g* zRnQ}u+>lLS2*A&DY1(}iKFJ+`-MH!oU%^i9u{OwFbmWiFu^W0~J^sj6Q0p9T|wLTO?UKGza zHj9d5=}!$| z!=JmBbAL(5c#dYKlG{>WYm3fKtTI0L4nO0tfI5AuR(8LBe&`2OOd)1jye61w-^i}1 z=(`_SF&k(@%y}^{_VThp7Q5`t{e(33;8fVY#Jyzaw+JM(@N{#K{nxHaVmH1YSf@!U)8?0Iy%oSF&iVB!v zmw#APpuoBwDEe?k8qZoAr26nV$R>U9-RiAQ<6ztVVJfdViRvcpWnGBP#Y%D|0tc5Y zEx!djZ$9mL#HAml<#&z#sat+_epbY-itM~QVs{gVbWQqkpp~P39aVEAxzy02_F$nk zM1$JqPkeA^-W5UR{%xu)gn3Wr-^@A~-L*I{`Ft;xVpn##mHy}q25ajdZ}#&UD+6%3 zzo^4GBQP0@da|h{pY=zfYgFCrR#lzIpyWaxLm}tFavL9l)ntQA_fl8DC9T`>>3f<3 zQ=!^8x|P%)r1o0YKu$B+smB0)6ZIGNJ+WEgffu6ndBW>QZ)N(X>SeXtSb64jE&Gl0 z1GcSc;LlRWt42fvT{1Z-Q{IoW-z^{oQ%Wy{zgzy(-^fLvG0tO?1T%ftA(3zHVP zo_yjW12tKwQhw3(*+1{(jvylSs}*2?mm^>w$nf$jg!Pd8v7LE3z~nT_b|c8phyCw`P)*(t2qB%Ma4g5^5-)0dU@_neHY)_ zB0IvT?H6m>-?hz2BVGv;=Ss~BIHcY}A=RxJwNS3EK${0K2R|P=m{UQRr9(IlI?%Tou zvg$gKBnDSfhAyVx1zN%cJZwAONNGtURlf&2;6k>yCS_=SiU>&X&YI6UBgHgAUTp%U4Rv&=j+D@^#AMt+`-sO@B`lP z&YYO&s~F}}c_xhZLX3>6TW(G=vBZu&~7FsFQr?DSi3sX`EV`C7^ zo@M?VV_aB9kok=k$M(|cxEGbu!T$8pk(O^N^k~_}XaqU3^+`G7Mg0`Wq7*1XTxaUx zTtAHdSOosSL%4f-=dLU~bSM)Op@P;ldzKPaM{A7T-Bld$(9tdrt1iE{rps*EHtWb{ zUbWXC;ga&_*c-78R!z0TJT1sva_EnC?28E<;yAy37wj)xzVT+#@2?g~FkUKum7;f>;Q~T2*}C5 z=I1}ZJ&6sFw6(SEQ~VJ7H_-1GA-^h7%qcddl^Il(qlGD|z z)-`%DFF|ge@y)U+I(Fx!!C>RXk9A$o0|mZbyjr=Irj+S7GzW!WHj?Lr?(hx}MsF@T zrFXZ*26(Q0=7#*}oiSZG;M80=?CReA8qAz#D_gyk<>hz?Uc z&#=&MV-kDnCZN;ocIcMq7>XMRQHr<(ePq;IHv2?D(KP=?)aDLo%Rb_MjYvVbB*%#D zhcg^e(BrqDbOA=d=%CVwE#DbyQsN)&_Q?94_1rC@_SlX~4w^P2 zypHtaQSE330#tA0HWhO;AtS_I4IC_6?+|Quia1v7s`%7J1J6uZ^9NqG74ZkNpYr4l ziL~DzUdw%pH3!%4ZkdPTWYbeaKEwhl)(Nlw3a&lsWf@8d$Tpzh zH(587H^MDJV@V-7(|=gOv^@|382`ep?4h&VwYt5H_W!85%CM-`wN0neNJt|kEgb_$ ziBc*G(j_HbL#K3ybcdjV!r7M_NTs zPwzI%{ZCo`8fB1xwNw-k@I_^1mqdwH>z#uN|9ZihPrt}2Ki_es--rcco-pHs@9C`n^)MO`Y;x7MQzCmt8J|0E(eM6?ShxCb0- zPO(2RT*{90?hQr)+5$g(ZShpphE@g&Z>fbWL?vvm7pWiM&aTVfAE}sbHdv zj)@^aXbJ;lKXin7rqxUtG9be+Ssd1teUA7EcP8YtoTZ8|{?xIr6}R))(905e11Vay zd74N3U&5COBk0t2bj$C@J!m$iDVv+BWw{*UtnX73o_-(PiS2#@K z=~CyE?NaG%S9iR+FIeIh?-W{zu0yG|9{w9AB8XHn#nfrdF!0>)42C5?U#jp)YOQ1M zpKAASA_xu+Uc;2UY?C8&t zMe43bVjx%aGC|2{95bE(oB<(e{IeNL!YR_}DNu+xN%#RYgfCydW*0yE91COM`h-CT z;EdJQU%YMY1_lqNW9wo*ri)u46e2|SFoSVqOt((FAeQ>$1ST6Ls!int{7(#;zsa@B z5AMZcno@c{p-YWX89=A~kd^eKbM5eBs^a=7@y76v$+a7EZwIgs2Riw02wT&+9sGP( zK_ywZu@UERAzn#M441Iz@t8v-arAvnjn0?PNcFUuH9s@+><-mKTUn>-`L_PucARS= za=RP^blSs6)|N15Zb8gqF$u_zu*gSIVegPFyM{Q{;VrYgK2oDI0`c>o=e|RD#L3FK zl92*1klBOn_&4s$k}?@Nh?-{Q6k8$8)1vc@{KB|0T&VlI@h4CJoms9RC*-@)i0WY7 zJ3sehnsu>{(}tKuJIpOxA)g&mTa~lpZ*wP1>7pTXP=1d#0AmUxA0WBR$X5_P9WB#*eK zw2z>@v`-lJO?oUDCf*mjEDvYs7+cW7;LG<*`8|LE-9qg!H$eCU0uB9B_SvDJc@lt~ z(aVxGL3w%T@!fe7xCD@R-~1lMPe7my`(eeUN5+TB_T^$GPjgvfMs^Gfb^r5u*?G07d7TFs1yL0|F8%-^ww}lL`*vh^9mz`1V_(_DVe0jxNM!!d z27L0ykajO3%OPUQmOPM;SdaL??U9MHn5_tf@65`ONSDV`s}>^a{y18d>pCUlJY17)X8O?v#aKKOp zz+FEZ^Nz>`;l39rREEB>Gw_iORF$=p^5N2g<*;Q;an*ohzZQ7)BR`&hYefaxNzOdL zWzkNS=xR5|fz9-U2R3wk61R_)yvB>E3dK7jnesA!$aiF4&_8Va&hu&B%DH<3pNrLM&AS`&!DLo#q`MT{s_0JZ9sKJJmavm5~1w9QjNkg); zj490AY{0~`^kZqgqL`*v(%(A#lE;sW@K|C%rf^CHC2nk{#m8sCTTFXUfRdQ`I4S-Q zk5P)7?)4U%oSTWDoJT=u^|X5*!3tSu$VfRB-Yb7Bz_aC>p4958Dlcb$n|c_OI@s!C zmh*^ESK!_Zh7D!jp>gt4ZqUMGyq#(9eVI~M$*K&teNrigJ;dbVx5RYX(`7q&wM4Zw zCQFpva>ORBDu&9~aS1Dah=E{3^5{c~j@xJhj8+0$_SypIp6ibtDj$^0Tw{5N*oFv) zQ_b((53ukwGV^Xo#R%nDy?V_)R%te5DU8b@BarRQDr zE3%ZYU888q&m$l6Qzj43UA5)uDo_nY_?DSA-|vggn6Uh2%Yi9cFb8@(O0UQH(54hy z7~V+A`)dK%nI2?h+Wb?`yg~wi9k;i)53i>_E$=-X6Fn1UVqsZBIIJK5T56@7jtQbR z7CulwU!0QcVrnx!EiF6?(Az;cBS+Ygtw}_88;|vVJ-ol(EA768Ek|377gSZVd+GkI zJ3S+$-Vcfx%WAywifadjR~>8=n?*mDo%s-e@$i{YO+(768Le1SNoEwpPcg@C*?C5A zwaLHBi7}4cK5MUQ-+qc^cb@%}0(-fO;NpWCs_pE$nAMYzDUbDt^DN zO(VJ5W%yuqVJgSdN5e~!I8hy3{`%jz&+6At_oryusIi_0$J(b%$P|A4niBFWkyBgg zC!}^vi0Tv;LddA}3EZ(Ic5M#^R}(HF$|C!+`Nik+30cY|Zg4TGk;IWu`wnuXiJ z!l1@L|K7b1r7LMrcy;s9doVhEAVgw(9rf|oeIfCv+OEgH^-|?z>htWLI@b9UV6pQf z1C)*qPR^Bf%yFr^wKI!L@Bc7=vzrhFmg3`bK!k?u`E#bP7RAroETjf@$IAnE2bj^bAs;EP2gtl1)}yasUQkW`sQ&~lh>Mq^nTMDZ7wo~crFJan+_C0c z7Ztl<+T?CoV|IIqv>IQYx%(u-*V_G(N2!-O+<|Wu?KK2u5_uDsRyUjyv0BOE+0psA z%ftCcxM6_D6oU zxjE}xyu?wzseNZ>XIEKiW8=aXPy|IsddHc}uf{S`K1t2mZ!pPkKDtML?v$=_tKqFf zMq#>iQ;+N{@Ssw^HIR?J4D;kyTkoK-#y>8+t&I{S?g3P)?@12lffCNJB~n^`B6uYE z@NBxye$oloD_1hgVfu^Ow=5&s^fxCR`^#M4PYoROP#9PFTKak}3wO?G=Bs6rNGk;; zmf7dV$*W3_Cb?pv!im}3hWTDMRt6+$>l8w^;6fE^3aY8{`3*DHIeaw;pIoK|vXO+D zT(mF1o;;eBn2Z*i`d4EkJ03AcBrBiNYI0@CFe?fsBz#)6SFt_vMW&HM8la1&4EXkk zrk85h-~s+-ZC}{b=wk0SgRqtnc+7HcW>=&9ts)umhxzM<%gtV9 zWK2E)UalrMwJ)K+Jfw6wT5Nki_VCdoRcF4j+mpM#onyT|2sA=KYFCnN1-}9h6tsFR zZ?KxdA7xzcY57?{iC|Q(uX&%yLh_fETPS4fqif{$s^-I*8$pA14g+Dv1I{pTeW#cJ z$zicIEX15!$wb#fLT6ynC6KAjb_C}Gk#(U+#(yLjbf0GfyjqR)k*_prW6-aYX&j>b zC9_Fhuh4WJ9LRo9%)ffu1qe8N{Ith;=N|jzCMb5d3m#A8KUt*lcwEd)`By1b!|a=` z5qUOS!SDy#A*}rHJ7)crlbd8ba^hkC*3EI0&itd6ySy<0>EePSALs^jh5 z%zJva!Qqb^KZRHXotvXn>!R9PdmsZ!7p`LMJsxZ~3-nldgt{MRT;myz9K3X{C-eMt za^j1=t-?sOTMkrfPL>4IFr4v>7z?G{dNWc03q+jv>PHdl84irE*g|!w58ZdMIDYuF+H+5gpms+h3@YMOK+6XI zBrz6M|7xaDq}Z}1oOSL@%|IjbbGWHfNtrAHDGESAIIQ7Qr*fkje?xAz8rC;o$!Ja4 zss-66LK;56c1AX!y5@-#9=!z&;IBu7zV$(%on`Pr6RYTn#v!5lO|e<$QMPZ8Y;Oo$ z*W|)ngXk$;@gS}-Q+n%HxspEUpi`e&#bET1!3VsT@$-NKI=9wGVdvd1nq`;0s^0#V zC3OWUE_>sMIlo^N6P@z`!-6HqgtbWF#mg&0(Y~yld`Eg=R&4v&GXfgXxsc1KjF-c6 zV$&&ao(o;3>lE}o>j-I`^@JTkwPBV64@}o)FkAiZbBb;R zOW797Hd^z!%kXaJ60cmz7fW+5lj^}1SJghmb|JYS*b{J1HuX z8&!GiktW1Ahx_(-ucSfl;kyH?3qd6B&!sS^RUz;H-@j&ADfFAGBEX6};`rE(cqb8Z z)FmV9WoVc~I0_g#Me5s}*z$JAKXQ63WWz7-%CIsoma6@(wk&JJ?h!6C3gHduIFIB> zJ4Vv6lr)vswMX9Z&qm!xLX>!HFI@q7O?(B@g3O08faE^XhDS;=NLO!p6N=|^A(KeMZHm^*VWyl6BzHBPqVf%W*63)HfZM| zYrbKVd z2WzImFBYo1K&y7dnEXldBfnVZf-<64fsD>=kv)(omnor}sRB?&I2SsuIC$;x4YYWB z^H-fi>b5WK@N>+g)0B+so3mVNB>s(S2=6stNgt+P0(zeo011+Xe6dFIJ!+8qgzR$X z%jTalZ+j3!FXD{5(P-}gVA`9zq2R9Vvf$bf_`tB|XZQ=RL@9asAYPsupuD9kxn(#8 z783*%6yK$r!%Jgn^svu$PzJIc#DS&C+hVgfkiujVw<~R(^*nQVGEzamH!hFl_RPtU zzpnt6&?cmK)y(s*#-Ae@HSze!d^^=R5Heq}FSp(|SoQZb%Fd53V?sek2a7c17!Qn} zO}#6q#@8P`|GgnhWRbM(&1SjtHPH(uT1u!(VY$G#)_6N>tGz0aiyQvs7V>0H#NtY+ zOtOLH2uThOe4kVjTdyJbz?SLqdzW(a4h7(5{Rf)~B=c~fsX3cA9?5>5ENstk2Ve$L zonZHu(f766uR^gFn>=bPXoUdeBW~&B^w(+JXtNi(s!c%%TkbO;3lhz>IT>l_o(X2cK7jJ~EB3 z6}ERivinbgjmt7#+L{|_;BxPw6rI++?X&f?Vh2+7d@I`Fo|yywnDSkU7at>G7MzA`h=-fA?e06%0;lKg`m zh$kc0YX*%B1!ZuP^QMFP)ht@aUFHHvzIOgA%K)?s&3 zEr(1?4~VDfY+>O0Dd<*aG_x3cgT56n~xGc=i zPX@cR?|3Z3X=H!(6TH#qKK3#h62%8<)_QKgxj7>e4XRu_Re6)imv=t9!7IMV3L&~L zMD%(lplfcFm%YGuCS-xi79`{?aD_P9WRjW< zaVPT3>o~s@4Fjj`oPb|4M|X>ac{CCps2PM3QfH0-8>F&X2oOqqC@Q%Vh&7Wp&?TlN zS~X+5*6Q&BWPE=CUce`nS+@_kTW+4^H>Ka)0!)O}wB?%u-g--6TMH0Np5aH`M^ABZ zE{lR?7)~54KOyGMAnDfO6uxc_haQ1M`f%d0hGy>&k(F4+7X;)ppRm5(asJ&{)P_i9 z6#8FGTd+O_8SKt$9t>gGQC)X&7`H%qv5xcdC;}M{dbr%P?nc#7@Ss%S<1S|WYAU-A zQLT`RST1qb%D@z3PE&!HeK!bGVdKsT{H`4s!;o8>DLUE5WjS!etGRQJ*zSv3vXh3x zU3hcmyL)aVaG>kfOB1d;Z$(Y7-H7Y!NN!HXHag*Y(^oJ7A&-XrFQJfmprYtkNfoM- zCR!0X$Dka215tEuRBGSZkYJ!dm8YOy4_Cy03&}FTabdbyj137*H(@wBy?) zt6uaDS?y)Ia?Z|FaRL!;kP2x|`e)+K_2{M%3@aVu4%NZzDY3S+Ajyd7xkHORQo#e=Cv8U z^mmyT2bMjq$O<|184!P5Riv)PzC?SMvi@yXsJ&U0KfGOwU%Tn6>%(olXMNtM$!n^%+zYHme$$1z)Roh6h9JV~dd5qO!m=z>u%CGah z`~l>wl1}JW+|$$V0zs)oRL37sP9vw;WMtuh@S-(vZh>1sCFk(V+V1neaEaC$Y=r_f z%?*2C$-py~8Jb$p5C(l6fQJ8Ew#PgSie@T}t_D$S)a@CB9yp$M&b}{)dLGa`KH^)= zjQbtK#AD_kXlYi*^C`PC(2VTnDm^&CkYVntEqn*WpTLfZF@(AadX7AFeU268dVeN} znh>#4XtM#;{(#wn-zBG)_#Pt)t)u~$Tr58J`UwtL{uK2y8}za4cAC+8*+!PGYe24- zWTyvZ`cr7GBPJ!x73M2v_af4lJtoD|NagaR@3~sT*nfzg=L!f=^#-OEpjvJ3p5%TC zC_+&K#2q<+L#y26)IsU+U5xCk1^^vE6}H{q{!RZ<0uuFpHP{0DtXgv2|gdr#CM_H$QY(Z=*jH%5%S-;;9w`ysIO$HTl6S4*@9}m_^RIk0iy;TvwMt2 zE=e4`9i|zO6?AK^at8;qV}RQX#|sy^M7;GNj&00FJPR*1+NgO0ax+Xm>HM<)Zk`nW zuaqjlUr-@EHbT)*S2vw2X+ogo&IW~)51iK2KLL*Ae?*N@n#p{4E+&9hV><<{uc*RY z-sMY;)5Kc{AB>{y`7I$a@Av||j*)dihV@ztdlIT7uCZVIih^RfY`Co+L_^h=uO?R1M~-d0ayC*|1XVJxh#B&wRIT#CDXRV?7B@`i30QYBW!9-^ z%qy5H;qfG?C+Co*v`}i*{>zROJ=zUv@}j;cV@q>(pI$Y_LT^YE8iWAmBiB8F{2XS- zE1f>LAIZT2h0HBixQ)wyiVjv}ne1`2#*lLa_Q0PlU3Q_fs?l)0=`J;-5*|Tuvyt7g zh23*u*@!eg;V9w5Ao?6gcA^ndCX;Mmp7jCL%c`otvJBA>WL=#JPvD0}o=e32TyDSe zd0FVi5=rw9vRb=QlT~Qfa_!9Q z8XqPNl?YpotuoEoByF{R=gK$7=i50^CzA8nMzVGHL}Ic2G}n+?K%@U(k*Y)EFd+pJ z(g<*ZzbT6mBY!K|u$G?Oc--r(3>X?5dBem$nwlETR+zW~T`nR{ksfT#^!+TMQb>NF7cvMlAV_N9t6s@^y8MU<9`AiT#WNdoRz>iHb6XoIC$QY{!ZbsDZQ{ z>eJ>FO*UHcr^oz*mN6$oeAfr56hc$)_Q3@6og;nK)*R-==bIwcYd;5V`iL-pb)g6y zYv_o}a1BNPVhL&6*TBFy7$Apiw5wMbFcP{D9wWYV(1OJiJb}_TvV!6ga_*9=4!(>fjnb0RNdPik7WSa zGl&hKF8P;4Hchxqgv@O*=>gs6CDaa`n8I}qU3V9!2?r%Y-k zZX|FeCKSGO&8)XsURbn;R?@xcy#SwC=gOI_s4Y0LV&AsG6;AATRG z(KA{h@gTZ)3+L+yirQttPx5wLGEnN)xzvN2C^UebU5JnGvryBHvg@FwUgbK!HB^o?JBs8? zPKbRPBh0G>u->4C>?PkrZSR9ry-(v^_q3N^VIs5Rzb}Csbn~L?{XjuS(PkDq)SgE_ zfK;0~^5Kn`=$01ENt+)T9j(LC3v?|DW&PX|eLSFL3v1Yi9u@+HRiL5rSx!sz17sZt zb4b@mq&25{_w(@Ljyt-=6$|Suru1Pm(3G;-M%APezkXZ2y;YT`Ep1MdM+WV5Nq|$r zB$GL}aqC2|t7jw=pd|x#;2!P_m${m|3#xczBs7dCgd(`~q_jV`_*J9c{zm zD}L~8^q06vPUR1UXkeVkGEj3>G-HQwP{8{yFB&1q@DBsCzlx{2-YwsRIyr&i_ zhTTrbOY(v*gEw9lz4I!WA7U;X+lj+k|XdP$LJ z&TNN4QWfSpWrZKGCIP9jUAK4Ft#JkG^T92F|7C$Mn*q)jHNOhvMvmh!51+{*y+FVL zyVi0-w6nE;m&>;W5cxjNqrJt_zv|N{^MZpFz14?Q+xvy5dU~)B)|?B@)4HyJzuGb2 zAVHgvkzu1}bB2>q3+(0WfElDD=B|Dh{lQP*ckIGQUc1#;fm25$b{5tWo=c>qt+)tC zo7>9D9GHCj@!_C#aUoYVcDr*?X~0!j_}7>%CP=)o(ib)0@~LUmas^iiySa5sHg8D% z6;e=g-gz(ta%5r!zY7LEA#b7C&S$!%tfkzqig5a5DBkd_yYwq#Vk?M9m${5@Ha#iz zV^+4u&1+slYD)+og(02#3v*k!+5-o)-DFU6K~fj20wkEORc>wj=YasvrBx(XO7mVd z{nb~xes?xKtw-9LCW`%IPJk!zKqpkvew#q~;nWZgI@)ocRH=o&epbKi{iaY01w!>2 zk4ku@_|!7x!I6;9nGl)o?r<1xWjsIn&>QsAD>Qnq#HkB5)0iU{&2_$6Hg8ErKyD*C zC{+qDYy*|lSF-4z(zsEZE)`!91M256yfV*;cHFJM6zk-4WTZQ**2A~w_r~7 zfZD83n8x`)U(jq~9pu<#@A@zn3sp=!&S?r3k)I}L-BE=6*@{TTa zruo-~&8|=La)DAGuC&UuCN&S`JHtjnnTGqM5t^mmoIm%Ri#t^kAC;B1gd~Ec#ZVIx zT{2%uB{4yY7PO7N2+YAm6dU43dUI*$Ai>(z#B)0|`Yln1rvv_GgNRh{HzTO$qnMz& zcFNc;)Dkv;4;;8A->&^ADrq;9IHTAS@FgN||C1v<$ZLzO7c2@D4Yapc| z_nDVuSMV&>m0naDe*UHcKHnN$>UJhby~_mp$UXlttv*Vo_lQe$>26-&t$bJccwaM^ zuPpXrc3OYCp`UxcFVmR7OYryVH0~K{&+tAoWW17-@3U0?TwPbNdJ|=iyR_Ki+Rv1K z-(W%OYz`urIW*o~l)8lOtqbaFYwvw(V&`kpDJnJmc``csM*Lzzt@dg&&Ayk()2RHP z*I#CZEo-EZwsm84+@EgyBneuIb&t6jSwLb?xiq}bt8 zMKdDHhw1c*mL|g$4?M!k_x73C;pa4_F|0i)I5~z($`Nj=HNWg_LSVUzDtvB$w9E+cY0jcp+xe>3 z{{dR^bAY4;uXf*g0NJtQWo_s9ucMP@#;hEVo5n0pn_bQN?jvV)m+Q7OawVy0w}<|b zWIQEa4^imZVJ>~P%?Bg@9+R~aaXEF97FL%pN?br&M79kFf{kYxM^3#%pFVvW88fQ- zn8i4P{pg*g?$U!;tV%gP`jy$acY#tHTd$b5ho<(*&IHl>@{ktGun#J#x02a(%JmJlsZuS z$j6^z4Sip&;dQduz{LYdp=F~Kh{M2k!Ac#f<~?m4{GTwJqLE#U)N<~SaHBM3tc-4B zIdvS`HTRNj1n`h=$HTxNUsm88?s9cnyf^3H#c%;I%qqLr*2Zcpo^O1s)iKQjr87gf zn6dXgc4%6HtL8Kh!&e?*PmcCAhmJJ&GUci*f-MX6knePOsr*({!r_&{IJJ9b=nF7b z-SsVsN|o^Is1iD@xaSWfiCXiPk!e<;ym$K5sjpiK61dKlPt zU}PHI)ibqWI>{@J+e5S7TP0(S$v`{ULo5Ls!|y^`vp)yspLJFZ{->m<1TrgXqpz}a z8*G1eZ8w1Z@YdSa<1t#EJ7;`v?MJZxbtR!f{}peHgtfJGNzz#@m;<<@tidS1z|6Vq z1WR!{jLC}o;KP%rxAM6%a*&70RT?=#BYK2;Qal74-7hm-y6L3H-)CH{nBUMCT-6-i z4|FWzlZ~j7U9sNd&uLRg=|2rwUqg;qM%y_-5__}4C(<#zNEN8K^l3PZC zIs!J`Z}uixQW_VoLc(lek_B365g~#R50CcId}4rfQzn+H&HwuMY69_{6ul4cSXJ_t z-}`pjhZ-;~ZZ#6TR^)D)K=$g~spbtaj(QNva?7&1XYLug&zNTx6R(fQ2#E354kfofdck+$t zo67_AHv)uc!Ks9dj@`i@P9u%kn+7ObHj+?gu2r&|li*s1oN);tf#-YfgfnJPy9wVs zB&uZWcg~EByUb{5VdeX#CX(*)Rzv@uLU@>jmA#c#*bYf((TNyn#E<)1Zu*nc(|>6+ zpOTunB^?&2|7=Gr$bcQ8=w4$&oY=5LB)G1i)vZ{HSxotZ&!bA*h7^K>7{^F_jHg(8 zd&#~wO^VWFdJ$&&T%?&cKG$#l)dJ}DZ;nvRN@iQZgjD#zSZKiGVblb z!olYevYrF>3ZN^X*F$OhCoB3nzDKU>W)E7d>*B-KeJbN>93HSY^4gKD%?q?j3UZ3v z8ip8USoqa|>q(Azo~`~afs5a)oS@)Uf;|$|FOYsW8q$X}e?ZdJLz-no9=K13g}CHJ z_HGH@3dWB%k($rSnwzoJO7$%MAQOLCDplh;WkgoiKX%CZpqAqlOM~Z%iWm>vh(eFz zkzGIDvfOx7(fVK2IkH-IzLX~N9bze@2dkg{4Y)tbBC> zf4aRPCUgfk(@51r76@E(t+GzD!VQZw77XFn*(yF5RS2gY=9;$n3>|lZr`QJC-AypI z;wOh3j{2c=InWUKgQx6|!qCbSOqR9ykOw5FTi1t>dQQw+uxdAlljZKne7i0rAi-#d zEC&1S{#l$7s^&0L{EGF*$MzO}m~cH|>dHQ#qEa ziKjs7-gI^t({lT7O3c%tIh$iJrA=M%la5}PB(nG!>v2zI@p$G54=I!<31=aZFE!TP7xxJomG0OmL#TB#M`QP{r9A)V#&{vn1mUfR2@ZF|o zKg7e!ea$K*@cZw9?W);gxbw|TG_0Xs#f09zsBx8jY(x2u%HN9WLRy=HDAu(Jak{oc zWzDSWdbGtlaa=zbBvg5!K|#@|4Oz!_0YOK-=E(7c_>}=ra(YjA7ZO(qgX}#=NRAJ4 zEiH!W2fUI4>6ITYs9r+Xq1z0_%QPF;Ymq;Da`lo;bK5MW;r#s#UklAmR4)Z)Ej00+ zbv_&1ilVVoQr}{}TOE!Y$ykM3H~gmw9HK?ghV91d3TAfMOk zUly14pzB%|udL0SOFK05PA$93$3K8)p)e7>^l1&*S- zk~hRDG2$v9?QBUHQZuHOIWc?*-lFmO#61K3ugZSwJEE-Efj7gN@bv+%m;JnT&F>+x zqhH6zW$|9#-ZX$e*f0J23w=a2kKw&nc+P%?7%HX3X@urg6eT=N6g#KEt=z15|cS#@Jy~M~uSlZ+L z*;&paQ)Ze->$)Qx`v9K8<~@Vl@OB)NeV)*~KIRcr01YbW-k~9Oe=QZjVlVD^3;ZzH z|4PA8@tZ@tEc{ykdXmGmrYcKih-%jca zkX-}f+vXb9IK>~(-|iSDy6mw)U=0x`b>g3Y%o@d-PQ2GwV7C!MJI0&CJ_It0b4`U* z670?k&Ds4MEnOMytfE4e0fkT^I9&aiRL*RNYk07XBhAM5jD`4qJci2{_H zS4h1nNR&IBOWm?gMJ2{oaImz8P;0BINp`9?IK|9W0lR@zL1ijoJ^nAr$bh5TJH+vw~23KhB z7}`Op2)Hx{>8@BZIqP;f*lK&iF6-Yvc%iW=I{ge% zf=@65pT)m=1nw-~N^#}V3eV=)vlb&LHnOPs(4m>wx0vsdCo~~crKk#=m z!!Ns&?y8GV!5=0mi!V9FMLQPfJQ@^0ez$L;;&+deEl4e^oVu~REx7E?Ibx4KA}i!( z5?u-a@Rx~>H-lpG@VyP2Y4sdP^gfMYf;)=wHAYxdRKvPQv=Q z)nfpqcm{g0H*E~yiDqi;836;FfpwL{NVh;Q=6xq6KSy;xj>T2JS>7Zzi44Gr06LnR zkU=lu6j&MsBb?FRVOZYpYqrR!O3FjxYl|Mwm5?&v2BL2__C8;Pq1h*rcRb5!zFzYi zPL4+#qCxShe4AcCSaBuZI$ z^4u$(>jVt|t(sL3N0R%IWZf>Tt-vULm{qI0B z(GU%3vB8?^>P(9F``>_q!A||@YXi3$C2$z_RzVB{#0=};`v8_7sB_OSm(Mh`wD`k@ zZxh2@S29$6i_1q{Zy!f^$3km#<65H4^}^mC)fcgsICDM4`H7L2jRMUwsst5(11_8KEvz0@r~zRY!hYS=CD)AV%~F4Rv?=%48W}W zbDjI2q`I>J6DXVG0js3PkL7ofkR^s6xL6ILH&;iuE*-AUe3moUM|PNH`_EhqIKy!T zDbO|`$P6wl=}OEe=i{nk$r`sXFE})(rG+*N7#r(|)RmNQd9E|kwsF=hoK|p;fDtiv zs{|#p`@wph1(X8w&>>W{;6=<@g@GQ$T7pxFdKrx+OIgk6`%KNz;tU81c2HdZfiBmH znv;b_Ze7l2G_W*$f&VI{{NRL=c7{7fOx$qMS!>pEJEugjW8Lu zca-ZP{&Rnl@3~X_rC6P=SvvgprS*6L+8lSLG(2z|=-b=W@_Ll<=o^CTa4IY)M@o^%Gw44x69RQ9(jY(7tWIyQzl3G zUzEyl_er;F1hl%Rn600IArb})?;I*yuy=4c6BM4{s(hh zsdC*>nh@JGDgG{U==bOAo7-=Aha&x7iB9`U4N@lRA2C2&t~M6mO%49d82^;ETYe}e zn3pFf_7M>g!XK6^&G2d5x9QCyB^}nWET6ux-$>BN&CBaZND;$7uTkU&J`7^k6*sqb zcYnKPUn@QAZDyU68vgh%*y}qy2NfONc^M%ZPQ-H!!c1?T@zqsru+eG2&^Nu*$RNW? zu0s_iqWfz=n!R8SkNpmL_-XTFQ@(IHM^%yIRAh~5gl9-1G}pPQVh$6U$h9J!XiEIe zWxRfAwlI$>@M4VDjEo#6(M|4(ha|3u;q!&}V0ffY!d}={l-u|HBt>Ho@2)lffzVdm zcAY7P*Yv|&=MKUpv99vraV803*-zPFttv^E*VG{s#mr@}5vC=FoG(t*-~SXn|GZ}) zU~AhpO|o&B`v8oihMD=fxhM}74%AzFlkx;$4Up%Hzi%<6rz&EuI$DvbKuceSKR&0j zY>h1Tvusp>@l(3Cr_d^-B3SxCTs2r4{Zu3Jnrd6Vk9LlCO z%}53RmW$~Q61j@gps6auo$KJ31~(?INws>mcW}VGKI*80SAMhXJZ-`4VLCOBEG%#9 zCz!!L7=LNCPuZ5522Y8wK`MV5B*rn|m|cKUH~VURRY`EVmKx}pYrz8XpG^-zkqH3r&Ln%GeN{C4KJ9j=i1{xJORM}3?CtkdII=3;{tzP&(>0lfmh}% zOFQ3d=x*%(>VH-G(FM_Puk63imNI}PKU`v8C$G%RHSvn!-G8C8DpJ>3gpP^^{lHGS zSMo8mhGse;VrcG!l%SZM8B{oQrCHZTJcfV7TuhtooZbA3G`0{hWycGavagi~iFxC@m^ z&ytibD)H%33=1yL$>XET_ciGD`|1QHbvHg$XuKb>7VRlEoyyGfE#7eWtiO-8Yf)bW z8%Mw6X;___G{f2}E(qdbviitBmdc6dtTjblOgxj|de~{83sfQld{Rjs(jLrvhWD{l z#moE8<8-hN}hx6=} z{!f?!=J!=NUr1MyGTk$0zi`#ApiJAg+=3SX;p)qoZi&r`F?@XdhWca0ni@BtEc>8| z;~Jrwq*`4-v$#4Qp{e==|B82i=lRDEo18uMOMX%VCc&ka3%lAgJ4CctndMoT2~0*O zRybcOq@=i&9nrQ=o@Zxq9hcxvi43prXXig)L(*!{Y(98f-!`7x{KH(c^oNM|z%{O} zQ)ag_Dw_B)mGLy~VnTE8>FVfx*CZMJYF3!Aini7ueECTW)l1JGlD3D8t$Cp(@u zP(0)9tL`Pt4*{Lu$`jI$R|zG9!i0?SeYZq3T)F{{llQ=eHLxJe z5^G;cv|DRG|1w^Ds;;hX-I3@~Aqq55Q)?1^8}g%pZ7UdWglNdsaZ(HaM5EqIpt4Q> zfF(KJMEC2$)2FI>YWJ6SZIyK>P%5q!Kg{?6+2!ceu#TA#SF|H7M>J4UM4homV_u3U zqqK7q|50hKNXt47uzmiY8=H(`qLFQdREl|z?|u9kZwA5Dz})x0`oIy?&(-h%VS#B= z`x`?;zN9)4Js9z5T5(>t27^rkR0}Y%_MCzXwLOD27I7l13uj43?NE1rd>D%yp)q^V z@T@CKT&u)a5%jQd#UKMdc`3(yU})9q;*<2akz=po2YX$5YW16+&Yea2_v(`6>6`DH zIf* zVFarB<&)`$4ue-vi?F|prn3k`X1b;on=+gKwN2CvOV0a8pHv3iB(`lIEH+D9Q8!uw zM@0EBJJbh~Z`4T#MtdkFEUEqoO9v+L9rO8Q4XUgJjB>Ft z)KM4!=ddDr3M{U(fA7$fkH+YQYI^wr#j4PSOj0l{0Mml#U~Pe&so0&MtoM=elwz& zYLL$qrSxPDSo%zd*DP#cfU&ZpkcZ@x2(ZV)A+yiFw_lN`=5z^?TMR$gxN3U z>W^Co`>4O_jTK<%cu@W#Kbz{AKri`6$Jxt0X8dJ8))#fICIOJ`ZQ3?%L@PI3?6ba8 z;_IQeSQj!%Jp`su>2c#0)PA2&96->IpQ`y1*EWHA^?JOeG9p{~Sy56>Loo^CIeG`y z-r1C=)+MCHU>41Vh+@T4a(qG!>gy90;G*2!U&no&rJ;xPrOo(B2#x1kM?{%cKsrNb};|Bt^66Tm_u^J zIX3wJO$75r3MP&VtM~gi?sM}wir794rOlz{_iFN|Vdg0pVkz4Tb(r7_)79jfpGTr! z>SFNwrxP3{TRTdOJ9+!DfSzBBCR@y`mrbBs0}Qbuh2=@JNC!?Es5F^1Pk;aoM*-pK zUGOm9*meK!_UL@&AGZP&Lmphttxgpc`p~{;6F%ExeS#&SSV z{md*2#96gl5f*Ey`05B5))BTf%KD~PYGckIpkw;Q>TB`r_|zPZN|MPZ!E=fMp?Qpd z;(#>l2rJpNCsrToX>ZjQYq{&F0_+#-+>$UQg++Jb)I54_o-d!?JU5kQ#hZH484mxM zp!Qks|GQ)1DHu4%@Y(yu3dv`bx(Bs`09;fv5BM<{RWJ1F8{6Mc*m*7K$X-4y*S06% zQ}w*mZVQdeE4m9Oz#Q<^R6sD%?)*2F+Nt`STwB`aM&Kus=<;;@PTJ%XIaBE3Qeay3 zaf~?$R^h{W(Up7dlLfwa=F~0s-!0>qsY0)2IuvA~GvcHRHh)MD zh0@=c&X9aVswR%cfkBAtb}Lzp%N?%PPH3T2=dx{df@+2t!-duJC{p18T>O2{XnCV| za&YS`dYRFa1bpp5oO?cZ%REu@yz1(?GdgusHOE`ATiVrk_qYP8`_RLnF`OigR&XTBh1N=Xi*cPnK7zF+x!wWLUKq)NUi?PmY0!9`PPkmjP;&JUUVM<4P_ zF4}94iFaRK+0H!0Q{Pk-t>&fWIJk#RaLaMjbD$?|1IwWwLXm2wLxr9xJXf&SBg ztXc2T2GRiy=lYZqtX(zh4uuLSMAJlNE*5t=MeP~hStxqWk3pVeU<6{ks{wm#->EyS zMP1W9)UKcagMyGMErYohUbs{CWc7W{<*yNRui1c7+xttLl%Oez1Pz+U$#t7*Xqk?vJP8{Qg(Kt@mwG^$YQ(3` z0@CAgC+6M~?Ii@!oa;1eXb>8$J7up=fSkDZVq^*i%Y7M7y1A>ZWc@&xjlk6~96u3? z*~|~tmTW%6cgFvB1@>PGduewJj-Ty&xf@dFO9-VzAAu%03~Of(!$%AIID&0NW@Et_ z-z}T1E;AT?uJLILUoVJ|7coNhM{I(jlO^`!g@ADS7;|clZEt;CteTOKq{*!FI>N!g zZ8h)-b4LvR&v{= zM&7~F%jOe93|ERl6^Ejy?S$(a8J^F2Zdi_THV1!0_^!Te{?Hr^ld?9SB2pvve7)ms zZ(tM^Tnr>>h{oDWZiilTXaXwx&-FypBy_)oeN&Hnx=y1;ACWGA(-J%3Xc1Bxs`_=S za(;8IzKPMW+0S9iN|~ajs3{#fP$sIkYWBveKCz#P!ujh{q0;wwV^r2(wN3u#{}rZJ z(CvtMrO(yBXC%)On$@1*^c9)$kETryt(^HVLMcH(pqVH{!ieGmj}?Q)ELQP&5fMc7 zIm|AG-{=ttM+45%6~8B*OYwhxL75f}<29YBJV%oMVC!Ka-(X0amz$``q~S>-{C@}{ z3{DtV${ADM$y7Cu%67&GBG{k})z_hX8rSr)e`YlDvNAuL@eWfl@@wfMoMXS%aHPY-!qsXg(qZZwzT3}{3Nwc=R_(IK zwmuWte2qPCYd|vF$Yk8~mGd>)(6rUIJC&V}5zutV?^ZqTXSslCO{lfd1F5gD%&v7t zgk*jzhAGs(?&iJWErK8~6CZAhzv~jO*hn_#%>87}@^44uWu*)@M|!MH8oo$0J_7R- zrD{772Uwdl*JfwT2Sh^KT@!aZu>OPSsBR@E*+YDv=ikHVw@Z*E7Mh!p(OYx(w;?#xOGHHrpc%(cI! z;zvwN=mDa>Cf>6$sxRKaFnZnvS!CzQ>6p;h0xgEt)!nOI$RkanKDYh$P-iUhbQVR% zsT?tZEfgd;^PoS_vdzr26sWWGIC|R;|DuurK$?4SrmRFhCV2uxqz^CQq>1tp#Qi)} z7d>?@A)%gUk1-BP4h~qFj;DO-oM>5UJq!}vzA&^2u8kSmR-(!LDQccL>P)}wiaSV{ zHoW}@i9KDtjWqA|1TrXkk&Q1pMeI0Ot+&_>)B`0)jeVR^6kw%74yN@6)|m&!4UKhB zWRT|BzG?1uYL2C^{%PCH-_3MHU$TSsR;&Thb)C-*)t?i_5ABro=YfA%wziupbL=oi z@ZLL+T;|KYG~)R00mHrtqvXjd>o+`9fM^r}hojW|uWnCMl%t39R%~&Z(oYfFSdi#! z_%0_2BCYYO-|=+~BhAakkkI4c-_c0_x3mCt!pN$R?|6j}j#_R<%C6qe2r8s^$aNf! zN>#MO{$97AA&LF12lD+&|BNRwepLQh5H)Ly5s;>b(Vk~6c)(qAjq+zYw~==j%~9DO z!scE?C@e+$jD@vE_=ZVAxrwt=D<93ix3!^}F)hD*1#OHvHtl2k`1g*_uZAnUm3%lG z%mDCW$P?a%Z7|`(ez+dR$*T=Sd+R*)QEWxC5ID*Ri9!#@!rSWEBSw9C=AB8cd}|M9 zm_$_|Hx5spUZ~L{0duI*(U_@&l-T5t@c96%7qS z4WQ-cSpLqdpJO{HGcwQRJ?WEH%yDmTFR#^l2w&d~5}^aKyojPU>JF0#49+~Cihsz*3oSyU@_uNJbvv>g^F`yG7dkUJXbl)v|<|khN@2Yi<9&X zJ$yfgzaQ?^UrIPSFN@yxi`nYReD<7)cbES8mvQQ@R*IhIvh+KrjCr<8VXaoOx)Dm% zIA4ZH^=Dkd607CLPO^sH%zm-ge2}o28Qj5;7vT{1bDCadhVJKQRNs3;1*_1lJ5c+g z10-tQY1`4(-~0<*3G0A9`pLdIgNpThWRUcWy>U<5g;E{u21| z*MV^+-a}u^kh!Ue@k|yWumnXZNwX0Z8M_2Jr3$quebAWI zR`0ZF!5-^uog~_yN%)571t z)aOl2ZmN=9StR5g?#)fbFyiS2D1ii)dp~$K;d!F58!PJ8tiG<8+6XgR<%t-25v(c^ zwIQ`;Hs!CXQi`=igB}!KlMqKbSF&SGx}|_qGlt0H?Dru_W*ElP(RC}}81edaRA3O~ zt$`+6h3@6yG>a5Ic~WcBX@*@-|2&iZYJ_~+9%~;PJpiX~?`~8J*a2I`-uK4p#p#h! z4wAy#JJ4*L*Vaw_)Ui#^V{q-+B)fb|<;ExA8>(#}T67fNRgJe3jwH{rZ)eLlcOM-_ zZzs8dd%PAzScm&Z-Vjc5r@{2y8G)hIWlZgn?fl&*jOBM2`-x5{DAMs#$^~o}{mjLC0;tSl>u=ITD@BjQyz1u+xK_dSW zoTZ}PX+Q6a#P4=I!3#W!KUXIJ#jo?|)%ESi_wQRS+v@Ecz;F98n^MzhTgfxY0&k{PqroGBkr?nRIP!8`nQ{S7k{mdXcrWp{BhM8oN5ep&wZz^s(C{Bw+xDNHdKviGB#T4gczeRAk*2S) znqr(2|JX*Eaa)KUwz^0WAJ}NI#ay4vh@;iWGa0@-qAU(29nTCm*3zY}ycE!HtY6>6 z;Yq2;b+ISN;={)J3cR!6t##%iycJpV_>m{~Qa4q)ok>nHp;oCISZks4oVwDNjIu55 zRfBo9H^|>vw08%{<&7vgxuq7;Gx?^es_qqU%2=v3U?_$`p-Mv8fD-U!3qm-9MLE@40txyS3VjlbMAzvb`x zI7#bZ?1I(=1D@Oqkf z``Q!N#C25Jxrp3Z`WQI4mmyo3e{w#ZM3k|9i0$o0s$$x zUq1CPUnqW%C|!>yVwMq%EZ-HJ9v)#*MZEmJKJHwxHUnjDE=qRqBsOX7922mRg(}D1 zaP2MWj?tWRwRTL^TST7Ynd8kw1d@VY#E-QTC)tkpWOuhqSehDT{hc(&@Y2HGVmI>X z&w#`)`8r;fn3zmsZ@^s1R;L=f5}6A>n0cFwEQ#!4wveoLdYe7jEt zsHsK5o~RWR8G5*ly%!9hmpl6P2ckf=WbCh_3i{NdjK4`$A=i|UxICs!*tRza>CIW) z<&5Yg4D5_fZHlKtb1mSFwTY-iizZg`$86|^Jn5|A4P7R_@NrW4jN*r~cnO~WO7AMz zm#ZY4gU=^_a7|oMvHg_0c}LVW=x7SG@nW!EC2_(>yu*uG(p;URDp%lC?1(_~?-aj( zEMsDV%lBucDoE>SErGOcRmQ+r9vDEAu^@%ZwRQ&=xELd@s?xO7%PQ>c4AoNqCLZ|R z;{;2o^SsHJqavJJ%*?Mdgy7)jUs}yL2Ll3Xv|O&X2;^ojxvl8B;fC6Jq{eo%t`GYj zEgzVTbRnzUTA=@U&pNM&b=IDqHJ;1e9rr~x$W(Vd+^G2RGbquXJf8BSQsm?uNR>QJ z_@^65OqOR{JFmA@{GjzbH~(ZBF?T~xR8>_|sGLYi`>PUDK4{hUUNNz-jbP}-RS3kx zPwrkdhtczvw?Clmik4ny&3ef7MFW&t$%oigRlm_aQwsh_({u6$%aJ8sOJ@>0m>_uy z#mG6R)O;=|6VhH~H6WWltDsA40Wk=7i(7}WFM2ZYEDof3H~Nx0;qQfenc5dxu)FLv zmIn-<=4`ikSgLPtNR9FLZT*-PJkjatQ0?CMp4n|C83GIRD7S6SsoAOzE^GW@HWO=p zvQ^}-)wBCiNSmN*+$;H=gF$dqOn~~@1u^rb}nAnf2TlWMU6<&}Jh$~Y*-f3ySk|KU!`;i8#hXRSP8%hz z$UeOA8Oc?SPu*zChftY+TfciUO?Q9TVicpt2bqz5SD!+SzGMq5=`8x~5u=ehV{UlH zR44N)e(6UD_Sy_RjgQCnXpHu&5-Qm}o}FZRk=dQRU}4FLBds#&mZCN(HTR|Isz&*L zSFk)f=(OtR^mYgBZ$lr|Ug%IL1ffs@pnLFy3iaC3QLuP*+o6<~udlMz<3AxkmJeH& z$2E^PHGHmS)q5Qs9b1x;j4FIbDE9VcK)|iX(n7M9WgQ~M-sr8A;uWE&c`Q>1sxxK3U?ev_QY;R`hRt#OvY*YqRR!qSYhP`LTc6tIS|SHwKa7l9e4 z%;rV$)NxKJPF5Y!94nYAE!_s24}xBVW+r$(YiViY(#Z0a=97i+X2&!my9dwEs)lJM zgxWG@5m_*+NIu=Vtzf9k2SNMFmM-lxQ#h&zF#F6Eu<&JJf{<`r3N=_t(*#s&9-`0I z_@0@KvFyq6Bo16n?i6i_;qucorzMZ5_2=*R6qfoVe#X>|j$b`SmC-KK)o! zhm%MwC(BF!$X@=p>jAi)j<`0=cZiB{>F?G{VcN*}nSBf?p^x%zyifM%Dmcat6K z>?x@AIcA`6SHijIdhFflk4}f!m42Iq2J+N|lEIT%UfIhcdA`<<4-eAK-tx3rm79F; zq#w8G({2&id$1(8#H1n{fx|3$84nQWb3xd9&J?v$Z{w%lRd6BmFLT#&PqKs{XwlT( z?mnVg8cS8R|2eT)QrI6NcRR8eOGI*Le&QkNhY{)+eN4kH1VOji+uqzoK1s+UtR z3TBzzwNLO8+;qHrnZKh8yI)INcgDdRUCXHUYSe{2o$a4o!rUu+7$#Lq{`dfM=) z4goYb635t@q44#}I}NPyzNgVouJWjKKQLLY(J@*AZwmEJpw|#n&mSYoH_v#}_e7iT z$>W#mEyleaO+Ut7=7X!=SqL53U8^BPkBYO_ngp4W$9(sTy!%s^e2VV*wOf|nZ{ULo zO{He~t_2~1RwR4*OR`|<5?2#y-Mx&h&%D`UyAERyN>yHg^9|`wmPM-z19sK}9U0sn z3kHKc@*;5*L(o(r07cF2hW2D}<8BcE62Yj!gcwj(iAD`*yKnU-{vk>VDPTFJ_E)Qg z{ykjfbj%~;YvHJk&1ZAV8ur2MB$Rj%$CbAtv@qm|w3YalnHnL6%Wy5p8>EjUBRRXR zUipO=vSli{QuK}I@?c_9A~)o|7;KpFl-lU`rUbmr&*Bi8t&uUwHHtWXyiNgDoxvq; zw)wPbqPt()3t zUo!lc<#lgG$wbyAUzbafjjrRue-^I)1wxbL?$4+EZc)}qrBsjpo__;B#}jipA<)f% zT?y|x;#7cyO!B{$81X2Ao${!A%^Dt?7v54dn;=Gkw z`hOy&O*%6pK0Pr!LpR-~a;EVTj1G)s-NIJJ3Y5+Dkn6rCJ+l5hNTdSJfjkHwCq*k> zqE$x!UhgwGQJl1#mFgHcyxSACVs=6<2~a~3Z}d`Vpx2Bp{LO|fIc+a?4SymkMFU4< z^*z_MRXHn4hTij!0LwM^b9mP4y|`0gEt(#ZzY)G?^fVLtC)=Z3qi@D8Mu3H19W_?GkG5r_GN2(K3h%Rtz$X~ z>KgfFv-j>%D?gdE?Wo zW1SQ^*Ok3+$Dtz4BOKlWn`2y1rhd@!^`3DgOEEBKhyBpkUZe_{&{d&vS!2osaaZkW zxzc%@ATXwdp%(jdav+5G+4bXjfRP!(f;!FlSm_YKlRZEt-=7LGKn*eVX{kbB=V!c- z`z657WP%&Ga=_=V2kR(;KP-%P32*srT@&h4i`luQT;TTFa?sC@4jQVW{t>FlKz10baj#rI!bZ6aDwIDOkF#-Q7@#< zrd}9u@%k3r?>2cNJPOWs4~-Lo9xpB5+p7_Z-S^3_PuN!+2Z)EDQBB~SlddWsG!9?> z)J-)};8ez+1*H8RXc6?{RC|+4qtgFdg5@6gIq|M{TIAIVJWz`g+jO_Xbf5|%W%*34 zwm6*j5B`?OSX*=xp5r$loAtwBfFd68-u+?lZcd&2(X7WUiwPZzBU=yB}hL<&kE+;)*T5 z3W+_fQEoQ-qBDsm7Y`Yw4Wh|1i5aL4#Iy1hX*c}qxMX8)P-ym9d9Iue;)18m!j~E6 z#+0<3R%`9uWWe=fvK*hK)+@K|UvZ1MYt9?5XH5b2rI`_UUOuCeo?1Lxf_NBlocoFlg(_f4j7t1M_)CE{;) zr`H~5;MC1qP05n}>&cqR%u{h=c@_*HyLW8ec6(B1f9%k*I{wQ0INnQ*5|dML$N#t# zQK-`$>Lj3bxh_eirU#{Lb$_+rN;&_h?KO-2?&a%s?Lpqj>*qbN-@mc8w<+6hVH}}3 z??*fobIcxerhnw=V3M_Jn;;WhV(e3?EF)+aElY7PUkwbef!Hd`l)g%0CLQp;Y7~FpS%}}jEBY$&Y7$+RXwnSP{Z67P zT0X3Ua>hoeYBmc_XTRi`1B1PUfL8gAgZx{}$=hvP-pDS|^kjy->A9ZN_mAL)fx3R^ zjH9>Fzw?%1fVySyW%F0_twci05#E@T7{`;kL#@+#2rOw zM!k6Do5%Nx^9RB77Tu9eDNadvHBTnwPFvyVX(fe%be1<*Cf6ST_AV-tG*4cE(cOz3 zudLHOSb@v8uspVESOzs&hb9?Eoml6{^5|i zri}SPcU66T=%_c*8wcz8>?pQkJmllb7FnRYCOMB6Q|l$&Z$QOM8=)9c&Cp+R(lOP3 z@^qNEy5jZJClNaNQKG*Wt`o~D;bAx#UTw4WHHw{uxVxj;KTR_diBCPglT4@^viS}- z@+ki0ierNtxDWB4Y&ch|d8Vg3Vi+oN^ag$n${@+BN5ZcrC^wz6-E6>(p}t1IpUA)y&YmScur6Fek&sa-!JT3Ji&(;UcJodab>7f0UM<|~i`Dl>U5ywy8x{vRSDN(<4 zV{3w)0QOL1$-@wp1V-`XM^@s$_mdy_5Yi#?|jY>$~8RNiw~V!rgG^Zg|bIpiIhj)KFUqv~S$KVAx#3 zgTNKkSgN>k!&B{&clGy)RRgBGQ4K5%Y}Gy;MA*`s`832{UAq2%5##M_*nbI*yUwJ$ zHk?XBwzw_B{`TWfVxQFe$)f@_`0_I=SAAg~xO^Swlc`>c{75Z~T~v?vn*76tyjhCB z?U3cdhjgP0|F8#ClvhgUdHXfW2-VhuW?Q(@6(aA7Dh!5yEM)UT#mGcLdMytTfumpA zhp}bbE$H6G2P~hs27bfe3*N1HpnbfM4|&#w%#a_r?Z9;Lx~{E?A&CgR&(7-e{9W+*u{||*$VoX zW_4Q`#g0STb7q7_C}7P>x76~V`JSq1msN2=9q}{nP*E`H3!QB~MhB4@dux@wTW6*ddGM-~~_W^z@V+pg=UJ}d_u z^~+wu8t|6u(@@zA0H0-y<|Us+;QAG+{0QETJ6mb8@wX#c`zqcokd+aI90#ruTDnJ& zEjGLP@(=aa44-m6;jA?hYHMie|1t)RG~DR2cqG1avEdCy_C3ah+$XH`a;gP0gaj{z z^jU6FyIae|cLnM32G?+e&qqct-81?ye2}N_e&Fn*0jIj2vRL1Asy*{`FdMtN7S#-g zep5PaWgD;*hWvG8ydYyN+?5iEH;;&7tzns-TIdiCon5kVpqKxAvm+-mbExv}jMX8d zlTacr2tyBOPLnQ6Is*tAF{G>*b=G&*PA06gjCNSwqtr&oh zjXUerAA&iNLZgwN@8Wp$StYdGqIdx16$Pc0goSeS zDvd&sewb@nq~_u0c&TTN;Ux~|8Jt*}Km{w$6pt2qN{QiJO6H*ORr!|8@rSeZA@^Rr z?`EqmUSBp^-$l%Ik~y&dsDRve#Cq>JSES+9{^kl)_Tbr=rlIu@L@oBq^c9+>)bMs# z`@vop`?dzpdKUd)`JUG*d@L+(8V^J|TmzHPu7>&T}r8+Hy$SayP z|J&D#B2~n*(yKm0i!q^$Yb}sZOFb=_PT=oGw=5CF52Br`$&3L?zdrtFf$$GHY^&RS zpPT-ACTF{j?|I(|j9Hy_Wh+TePVO+&Saj;yvuEb}SDqAY*?VUO?KdUz_m1t4_eZ`= zA*cx0!m=`aR0$!;@dn^=ZYG)<@(#OBlWgLhE>{x>{(YjJZ3xpY<=4Q5piI;xbro z$ru+S{!BAQP7$5k@nQCNmtc&f5mxs#Hj_=w$EOW{?A0u%FgeT^y9}&+X5iZHfXuTzGLr_!f;DFlwzHrT8!RWDqGe?MDWMg_M2^T~R7<{Rxo^1I)aj z+#w64so~VTe8^K3@{C!&w7K^GXe5_Vee1eMYyqm|xN{~AHMs`12-K%Pe@mqJisQ~v zib9ctK{19ccdIdRBSH>vQHYfkBY=3RR!|DJQqc3J8KF0Gbp9;hNdH}J*VGHo$N!ur!nd@~9gkS(|O{sqXpU1Bld@bH>8@e!azMW(b zi6{!W_nou+Z}@R_`UMl_flY#J6e@9kqr|o3SBVYP-1V^hx6LK0 zC}KM77tS*f2xNBmfGc+#2A}y1wTh_!3*ilRbO|v3u7rtooO=s{*8<_%;fCnynUT?8m;9_FevCc_;C4r>vLHAmID!3@6y~ zYYN}pvhTq1u*zXL@xCMyP7Bn+yudSWiJprtxy~D<+oL$!!`2jNTM*%0`lQ;r)@(~^ zoj`C->!wreLwkrB7~&@sIcG)_|DCpymSLdvOP{f55FW!Bu5PBx`fctRvm`}X&&MUY zVKJA1K!_brql?n7I>Ha;ttmH4COQ7~Qu0saN-K8nG|pOa49Q+*ASZ?PO-Aj3SxF-e zFoC`&jdJ~)?n5FQA5HAs!Xj{952SE5YUb#8W>ahA|MF*?(Zbhh{lzug5zaAVZ0a$&Fy{VG753D}PUmm%cBG1E>+!(94@JjP1 z)E`&Z>v`!Y*=yr5l)Sgmp6O&#_wXS0fG~Y(TCsKWk80HP2##X?XezG5?LDa|f_wFOQXNFgHar#G|7lwP7zo%vNoms2wcu>h9F--o zTQI9X0o~|^ZRz&5S_LL)W9sPaP515I3e(=q*8avtyC^1e^~ zY(E;k2Uk?L{!LEgh7{$lGNo(U2}ZQ9nt3m7J=9#?*(|QE<^l7MK9DzUB&X^P_=mM$ zloba#$|onsELa3fKSQ%HV*UI`Ol$?;4{l!@ymdbqe84|DPguGAb-8+<`6zv;Jy!T_ zXk1Jva&;rK+{2%B_ev!rG>!O4ta}&Ub-Or-$TAN*1MrZq<~*R6Yuq&b#is3lAbN`m9_W$fO^k_{tl#b;#Nm@K)IcjC7W$<`WK zo?2bYdfQcOr>SN^1WiviDjAK2z+U~CMuwo$Lc6AYq^r8Pk2X?{zZ);86@yEo#XE7;<8&GlV5P$h$%qGQwPx>SL$JbuG-N2UY{ckrVSfT!V zC5~UM*bzVPXr}|mpRYv~f0ern!yKgV_l{lMlQ~aNW7{E&{Hi1mhoLEXpmjgNa2UP$(z`T*^dn`813p3eC4EIgn0m(W54#S>_?HH}0ME|-jZ+@We zCg*dCa~k2hFMrAG<&531@U!s~sup)iD9k^b1}tMd|~-syI+<;7+G<;~F) z-2?aG;7ZA($%MFJF0)pXPH+;q5_;#2;gp?wyTf}J{|@*lwLZam{3;e+CCvgh)H4&W zOV0&dCgfskA^o|u5{TJ$8;t0zCZ~@~nRN|IUN)$ni}cSEtCdrNAKX%j)nF>c6e<&vpy zaRv@$iZcT>IJdM+W{EI)SbE}ibQWrwR(TM=Q&=PsS<@jERktkPJCb`%$M=x>7{-6# zhbR4*AV27qFtZWX;@`ei=D3eHcwB-gukAp7LcMQ6`~zn4YBZ0`8P@a0MPN6%TZm8& z_w(TPq}N+l<6{;~O`v_p-a{I5bE?~Nnz=R5tW$A+@MDyeOty=%LUW=(Q=_RYaOfwJU!0cRy#>y(sD?TSGHu@0j@@B2Itj%Cc++3R)e9#x`Mx0Kx-UG;GLpOJnr_{{nTdnc z*w?HtsM~EAC6Fh1Q7U$<2V#G?fp>S|+4lzX5-!ujcD*rjNPOi|T9AR}M@ro_uTmw_z7$=KVN*&GfgUW?f6qGVhKCl34lF|4SBlK0qCMBND{B&Shb&)Ecw9g#cMVh6bdl zhfM4K-l_C6GozWINX%rxadEAazm>4^UtYX`h;=A4mrb&G(qh9;yVf5%R;z$MyJd@ z$tv(+`9Kq`S) z(4i%A{byT&fq~F2)nko($#o$@0~lC^hg?WlgcfZ1zBo_)*bFO>a9?#h0$7HcN-a;= zIBA`%to!rIaz0_BY_Uocz|Bk4AVg$=8JJ=*l4BmL!#!wG6+$;@gBkyGY{`#L98}Qkv^oeiCrEm3b z7<;Ojrv)q$c*271DqfIzIIkGkydki?;_+g+ZjHj>W`D&Bx$=lF6%q{KyE*ELJu>j! z;P)9@4`HY9^YVSVm>a*2WO&o=VNRo!{8d}KOa>#I07C&=aaA%tdO=}WcVLgG3QlV} zHsxim1I%vSQyjA=f5m1mn3mPoInDF-Ob;)CQFUpYxiOBYAYkQw2XuF<`I|w!FpL z6$kW+9?eRY%G_jhJ?vPHBn!ZDQs>Q2I8rg+qOw~36ZRS{c zCiroX{u=G^ACL>YUk}M5ysyJw`DupYzH33#8Z`#NHSu%$j6Y2 zWz6>K3hm3_1g{v7TMD~Iwdc|6{ogMl6BAW2KKwxw7HC}CfCvBqhr{k`nUJmJ7LqX{ zBG@qzW-1-IP-J8x)?_wgiIjHMy2U!nyA&Ri5ht~Sz@s+ke3BWfbW$MOeQHCakM!;S z{UPht-E5clIL8aHn{4=UYloFoFxw13;qT2s+Jy4->(^>JhKw16C3(e-7|%2y9`5Rr zUpJ{IfuXan2CnQWK6pAzk1+4xg$g+q4R0bpxc@-D4gDds$|wVJlBlTQkkCGbGe6LL z5cYj@?CCH6i@Uldb_n&7Ok54F^{tPz;BMyfZB9CpM0CVT*rBTP_~K_lE8#~iyTJ1P zmgBZ>`{Mgzc&!l{%dumt^nJk%z`(p>)*ta55gp&h3M0Bln6TWYH>chxfl0PaXT{K4 zI4aW^!VB|Y=&cWi(btkwY(|lVXtJMjJ23cc$;KP5{LiK20bCWuHdtGrc=5fK)61C% zuO6vXQ4hQS?R`cqW$wA%liYTW9)FfdMUX>-9?o(vVQ?MqVSjjbwDcjkPs7!>9F=c& z^@b_g@^p;k$9TNd=UdD+kFiPtx#RQ%tC2hQZidIO#@u^u!_I4#;-I@&@W=kQckRWP z2?oAaI`;fNGoNa!IU;F!Tzstkli(JXJzKz;AFeu~1#k7LCz2TKXcfF4N<=?(=BTF$AZkR~=9zC1nDWX%RfH^p#qNkgi z+|PrIBQC}k=^geZ!?_^A{)x9@Sv{~jEtWmbR;!bM*{S}a$Rjfpf>|RGi9dj4%sN8} zuVgXIalRYUR{!#$g?d=J`?6r80~O)obD~XhjD1afZrkrxaBwLQ0vZ~*0^weBl^0={ zsyA{P-ET1GdQ7x0inw0N5Zw&|>k;bXE^opcIU=!8loFD!8Zx;=7VU}NiVH`D%rQc> z8Jboij=;7sOi!1MIs*Ff^G|ZZAd7IvqZW_hf*jh{)8BKc!oYIMrrPl#hB)0Y%S+Wq zWQ-IbT8VS#tQjKo->KKAH;fgSAnrvsw`7)3eQxz=n^Ro{Y+*>KmE^K`AN>2R0$;rb zhKj|nv%xxQ65Hm2hln$GgKOW!Ul(MnJuCSRgLW{#ttV-h`SOEQkdjq-lrho9g>q!c zYvTaV`gU;$?E4Pzl#;`GWSXtg&gFXj{2S)vH{H02<4D%~SH7~oufqFA#X@JVr)O1v zHcc0T^9K~~*D$-KkMdmlxE-wbu8a3&8aJOZNfjk^U>vAyIq-6+;8vY8G&=d73(=bA zsyliMGiQCASI6RT1_PZhmUBP=H}o0dakkLKYElgQkyMox+#ufD>Jhp9YeW5>3^!|r z*2;SnD#+WeKh&d7bkN*!>tA>w1WX8}m*n1JTq-W>$? z>sDOo3S72}(Nw-mGipYYv|y&W@vQ$#|-tGaPz)!f)ZaSeDwFALyFL$YN&j%8n- zp@X3<`&ygF#p3fmDuBHJstm{ zJyZlnIK`p%5P~saUvSJEv^QPGx!?R`<6T&}0%KL6b!99=cWjuG$jFQSQI+3NyiPTGAqg6nl6;m~GKPt0|*oI7$c~{?MB*zDKVt4M@hE z?0IR6p>~9Xl@w+NiGOlO)K3!b%R*#(%rALrkJd|Vwal6F7`L1Q@Qcn(R|6Wz0qBMl zlw1VcX!5;+3T}yVu80GjZSIJjwK*-CuczjP^i9K`adcSukqo&8U0ihT=bRd#vIds> zrj^W>U*5GfRTk?MwsE@6EIp&{_=`c;QH@U}9#$X}Y%={;p~~5}JBY=BB-|H9J}Ueo z7^VdVzosJ@xIPT77x%@!0RA7YzB8<;t?QN+y7VGSv(P(;NDZK30clEyPy;H`J0YPe z(h;P06%gsYh6L#yq}R|pp%VfDZanv%?|a|-E5GtQd+)W@Tyu^&<`_-v<8C{FG6j@V8&{=@>{krjPQ6uM;Yno|EV_1F#`uuw( zsl0lHvIkO9!`Y)uFzJH>25%8$<}wjvSn7u0B_=_nk~s5kC(ebJjlW`OtM%Q^zgPs_ zVe(X`i2$6voA0fXMtV_~J;^SU5BqLVC;S4?Gf7tC*mEKyoBO3asC#ZCmM$lvJ)q8Kon^lo+qg=0{hfObNG6R{3>vt5G z(lNo6FxnfTbVz%%(u1T+>+1gLHoX8YAUpRDu1B|8VX7A}RmjOOjQvBzB3*50_wMyx zMHwNk?UH;l;~&D*KMI?<$e%=hCMLI=@q+xzCaCj7vVE@0jb}O3$H4R9nEj;)v2VYA zQR2-GIgxv_^^Ag2e7!gnt{;_MQE^!0*aiA}MoAt}Lj8t23Ei*Ne!8sE6g*u)X?m8F z^$Z;ZM&3h^WkT6~oE(t(C3VqRVTI0GQyQ}J(;3DNd5=r)U1uPy zRo5MZCP3^^uEm#?iL>dMiuSUTiSGd`#2q>+hA#Ad9e~pt7MbHd)VpId-8oh9`E)jb z?T!!u60tBZns^7{@4 z6~}q=iOkYvemU&f#pI}EWWxfoL9d*FZ70#3Yk~`8a`%{&ee?;AQjl)Go(wE2X(v!u znAJ|nzoMyuuZ6^a5tJS}s@@rN*IPND5B8CZ6KRkyw{~)?mDtp8 zN{PizBrUQ{Rk;ej^;VD`PYU&2CGr}Bx-;WtMd7Bq+o$DjUvpt2R0PY*^j5Mp>36e# z1SxC% z+_U<2!sc)LCMj=Ys!1OSaO^Sw`Ur>#5EkBfGuc@6g#HQvZs!XMlVU~Yc_daliLt)@pe-z&SBynjb!`J7^^ zJ?&qA3jF<~1Mo&1PYnI=NJ&XK!;c|tjRhL$pC_`|Zqj_z*4O{EIg%gaB)z*es6nW@ zK%M6#mM34u%F0z}$$Su1Cl0-pXa>C}IdHEH=*e|)iPUP2pfV_F?GbWpjv90z?z2X# zW?-cl7lflVZAEc$FDEXOgc3nUXj*u0bEnZ(trNk#?*JiGzqxX2+XF!2O9*>thOVqP zTTk1G%6e=xsaKvB@2dX~EwuMWJ2wU(Njj>;Ov&ACdCOZ&YKAL(6a?9 zU+d{wnyAOCm0d~~w32*x_he87k}4A@PL8GTn3WbL9^MDAJE#epu$`A9_ib z1cs*|jQhJ9m*%Bj~`VPNLH{rK_@A=3(4SE&|AnhJK9Kf&=TIdTmmrBdwF#;VBob z)Sca7BnlSh`stBVIC)7S1N{>M+ZBHLZ~p}sseuUh#>Fv+ z)j-j+^?MZiuJx=dc=TvF(ZsQ(dIBWx$|^N5SoMBPPys&&1RB3`feE!X;AE^kIJ;!% z99eP*Sjl=MK5i`aKZ2hUIqz{G(@KfHh P9n5~~J!mz&1Ny=2|GtdNU%mD`!tHl* zitpil?Vi)7=1<&O$tc%0Fbr+;;XS8_U3iLl#~=5K>;#+p!{@>v=3mA3)@ik$VV)|FVQszc1wji`&QeKa#AEzN?Ze-c-T|oXmZpIDKq%|( z2?ueuSgjx8Yo^sH03BYCh?UKfWq~BVH8m z1K!uDcHXzJTf&!8_~w|?5{w#$=i+|?2O1PL5j0mw z4WZZWq&Xn#{mGH&to4LAnm!|RgZ~)WrmI*osu!zpa>G7w-H@I;VWC9VV^vR?XC3>M z$2`#|<8j6mx)`nnP@rlS-b(2L19NT=`nAA`FKD72>l3p5Lsd* z;8uG(tNG4Y7}se8?^$4cKN0ZQ#W7lX!DWZ1qt;sVko=ID|Aq~n=h885b{Uxg1bBY1 zA`q0SL7lOlT}7%N^kxdPljkk3#gSk6zNB2s1AHl1n%$eS@}6<4^`ba<$hdL)T^J*^ z%?N|}Q>xh@Z)$j=)J^$Pu`7XenvvuO>9yX2_Rj-|-$a185upvu7c^3yE@mv}+!W8i z+j+yP=JiAMb3&g$B^Q>wfv;s1?5zt4EcM2*IV=#BOP#BMw`oP^qSc)01is!}-=Xtb zv+E4oV1^Sgd@%146n1I@XI&`VMh&@qa6%vNjX!?0$6NeNwn1}S&U(C%?fX@UGQ4}t zpmgzECCFDyWJ};2n(L;i)C&Hpp$U?yl-0?1PEEmOXx<)2g^ide?MnX(OYo)<*#>gO z^EF#&5d7ouadowK4bwos{np%VL+Y(Ka%xEXbpZ`vzyFft=56!v2yy1kB81|?{p-H( zNVBRgCrFy(g3(69)$6mhK72u2Zu-PX?DASby)!@j_s?t-=t?BN{HKgpmy=zoM^0s? z84SxDKgKlCwQ~I$y!olG2$BbzV=YRBF`IX6IoCf5Xg>UX5IHZZ18mzB!QWW}5FSy_I%!A0Ek%JLx_odoPqC->L<=arh53_n)ir3*9R*K{MXQNWjhm5r*TaQU z(yc}@BPV9)y&h6A>v}=$F05L;h|N>h?7{s869WO6|EL$mB`I$rVxu;U{6f0f=Y;RTe9bh9YqzU)F7(g-SUB}8@(JD~-4@tXfp-VYl^40->-zcrrOipl zqEeGe?M3aa^Mk_{C53?>-ey%qJr#V0$wJw7?3Y%mmk*4wUzxN%!MWclrD)&5T)2Fs zbxkw(&e>*AnL-!V4$BQvF!Hp`U`fjk-sLUvnt{cmLDk#&m4k+D6<#A8LqaeR3+_kY zJJyWu>9;~i+Ej+QRr=O3t+%c+b}&Dq9l&u(;}j2YBJX{kPxOT9LCR;(+<)u z&D?5%(_(uj#k|}?PdlTH<(s?b%rZ2hl|;SoB{kL$DJOZ=H&TsaDxa80^9|C_{Q>JZ z*|9hB_B=9QGa6f6gXswP8(xIrje$TnE!T{5C zV9ZBvdHtBsvfGFbTVds9vcso=4+d|29!qN3E=BAF;^PydEn`t$8`^zxcF``g#AVXE zUpVWIo9Ll7&EQgbSaAO2$0InYk2@Ux?e2!h_49=IY4;z4ZV3N~#PfU%W|1(l#ClRh zpqiPh-T`Tq}l-Ia7_;$`{L|xj_8OUNhQ@?lN%rsnkT}pToTJ{yhx9&wt-%&`h zc`=FUiW|@=6V;x8z4G3~JTBSIt;!p`lWi_-=Cb)Bld7qw%1E+)IV<%)zvh!S_#;2B zWh+AusDj*r09JlAKF|)4#cN@A%bqyAZF^c?aK@cCT&%QGe`MAE-Wz(PCN^b;mGW}-F_q+VXW2Ty(lJ9*l$ZI-3k(zr!!XMtEr>g*NK=p{0 znKy=pb<}C$e&c*OBl~#zget4|mV}M~S%12-;vV5PQMJX!LvZcwZmA#lJlIZ+4T6p&zwW)x@ zX~UU{J`Sq>fo{vhisLQIirdVUA!7qE+^^`otq`x1!l?EPB6uq%FS|s5S){n&t%Lg0 zqv1F1UU}c{OF!TCkyJ8F*-<%n*Mt>3nEC21x4E5!ABg>bdYpfM;@{vasI$@0H-31n z?vtLbSGv&MnrAwJ4}#96-3x)hhk%$r#aBi!*mK86? zJ^J0F(aiPL&--R`de0PZqSYMr$to%vV$-Gq1+ZrPU1((0Z_hq{r-h~)@CTFbj;~B| zhXLk^zU{NUu3}U$VCVN8w(>WhnkUzPGxov$q0RpD?T|lSBQLIi#zq^tws=JhMEbDp z79&~h*!cL;{$lHV^X#mtM%u%EdD}k){-iPrQ}L7Vq&MNcb6oP_z~N#TZ3cvDvjM_k z>npuNrVf-G%yCFe8-!NwkVf%7OsQ7ojX&ipAS)rZ*_zwuvL9Z!5l4F704Vn-yGx835Ft1B9Pw4-DF}0l$Nfd571l`Y!t9SiO0? zj<1u32I;84-3GJph!!r5627EMnuc|u;>_!UL`ZXQV%DNcGXXFSFLi!N{{51ShEu9| zC+c516yA3xZZfLd&ceNNh!zeQ^Ss8J<;FDbHQ$z&mga1_xTvhG{41C5o3h0!YQti0 z&-Yh$FX%qs2f8)S&W)0R5EEL)Nt5z zm1|)(_I|d0&GtuW^UZ5hIRsy{>R>8F5 zDMoIQjE>G50M{DAczHD6byuydITc8B<3$B zY5%1)BAIT!08lSH)9rtqX@5)z>W;zJrJ=W-n*P$YpBbo=@7nwCiYJm0a|epVi^WKeD`XTlN;~-@x2Fgd z3A2>Nx_YC}IJ#^VV0q&ya>_H6ldnC#(K|HNb81!`AKv|*f&6ldNKtB5sQ2<}w`jA| zj^&P#AjQ19?|#=D=*5AcVGJz&!K(`P-RPVIfF>gwt) z2korOGpMVFF#KG^((p5`_HbsQ*c2ajo2Z@snJL~nqMoN&x1YDH_Guu}PnH3P+f%iY z)|X3$grbjSd;DVaWyp0$lG&i7W|ivn%{%WyygsO{F%k2u8S?r1hSSjR;o`z#w~4 z)nfEMQmJG0LRny8bVhcAi$p<&FgtZNsFUE&*_4pbgx8K+A$)_X$)RQy86K% zUr?;yl7Cd=(inX}kiM>D8%ZoHX02TNVDtGfVD*EH)DT} zG2KX_Uoq!tLxi=0if}kXrSWX)1VUq{OI~}&1E%aU@-J9*?X%6xw+|d#d7JC=KVWkAVij&U$7Xkw;h%C2b@>|0%|v)|<0kU}f2`wcsd6BcbY3GX zCTjC&Uu{}plxJNMYfRnkK1C}&e(SK!BA$iF8#QLS{na9y*37HZg)3$g-&vC0GnS^H zU;Oy&`4Ih8?v10=Ttsc3!_#yz!S&!5>&+U*J<@}q>`T#Qk3-Yd_%bjE{qAf z@MY7|;zgD&h&r6=>q!@-<6rpM1Xp8(F}mN-rWV z-JggxQ!WaXJhs?+aGkL1%4FWP1U(c!<;$m4DW*M{_nvLA828#9uizyjt#L78B&<*_ zq}l>6q8BGe_P9~S^6GU(vjaqE-kZnsAa-}*u%U$RRbkKHQeh2+PD&3?YAk6X`c?Gp zuNt+d%Wl+HL&~d8Jo@gro3Zw&LB}7AG7mg6J0M~gO%FYF?-;A(VO)7ka;_b4Pnteb zJ*fp;0-5==v+R58)$}2MEMFkm?AKR`j#n}S3&AthA~Z|$E=rM&4ymJ$>bK^3reo-J zkvvE$L9fel5mA^J&}{2GU&j%1iy0zjfNSxFtmD&HUMtww!RV}{M@#m28Rhzw#*z?U zwt2~;mgJKw!(E5}fW|co{&pH(cI~1NTwE8Eoti#`oqIoj{W=u3x^d%1=U#=!Yl|mL z&(|lmQb1~Zliiq9nK+*+H#H3qNWSgNy^gTt7oyr~VRI9A8h%-|iwc#` zgb)Hsl|Is5Z-tU8;N;H>q`pjV^=?jIs-$Ygs3)#xy$y;YdX1X9oHUVlJ7$`MXloMe z#Q-Ly&l8<%W2m`mBjttaHCAy>NUIVaX7=qP(>ym>S1Q`i?qbi3y@dXJ?|;7i!Rl^+ z7aI_99K)iUu5&h5f&>!#IzPyQTry$g%bhZr;-g`w8Td2rA z0lzQdG*H|slNsIQC(25c87`>_;GDP>VA}*IPwmEJpn!4H7~ea9c%(uDMmgA~qU0TjM1e`W3I?poj6?h_7QS3CE%j?n`-4_UFIO zOUMa^g2(Q%*B=V{?M1!%T;$I=QG^PREiJY-YA`2=7$h4&kQn;VZ+<~nj?cd)L`f<> z7qWXSlJ9%NdgT7rp)vZ}x-f86;^Y9!jL_-2h7@ zHKTXHl|JEja#+wBr)~^F+`T@N+nrN;%Uq8ao1D!oN{6>YASz9i3-va>K^LOu#L_+i zrPg40l6rGP&l0%JrkJa0GT%)$!`Z0>8-LbnSL^tH$H&CqG2he+F!)eVF~pTsg?4{W z7OX`oOBC&kM~m_Lnx$eNWgGa?eW#cI#ZB0c7iHu_ZL7%FWfoLVE=U&8K8 za*fqC2OdlDXHBfVdCY!)qGfZ)I?JXp&Ik*@4lkY8wzi|$6S1+rG7ykW8M7-xZ25p;RAH$-U2B1;QdBi=?dh1#pHJSRuIH!>oT)v^??*8K7p1D?t_?sX7*&sxfVZ{kA%>>iX$$bB1A48JaGBE1E=Vv9cZ?} zZ-w@W!kvuZ3sEN{wxhJ0fu)b2G8~*oOK8ABflD20*M+i}m!F9J=*7AJOtg=aiNwEX zINk!Y1o>09!s)4j(W4WocP|6%@aahsWtZy1;BK^Sna0%Jj#j7dCUE5|S+!q+5IBK9 z%Y}Q8UuyGyMeafGNoXa}nX(%7;E(mCn5RMqL$e8n6Q(ihTQ$pvk4>Ck^t1eMnj>$* z3FpL!&bqgBD*X0iECl2v3UbDy4Px=^bqI_v4 z&mI8D(`NN&(=LTT)?0qW;+^71Y%0M^KIrc=*6qyWAG<9^wN@;8)K13B z^UE z)eV}kwlqi2_8bP#eH9Brob$KJ-ZL-mErx!{%5rkOA3&rtPL7G6+f@-lFkvii5yCi^ zSELh3zb?b}BZ{`NfWwnf5ZY%4v%^Xxt(K++>7VT5rpx^HOuD8T0?{X>l~q%X)Ue%e zR1Ymdcy8qTLF1$3muJ1zc3A{5u(KrYY}P)Z#>*VVfy+#zJ-Dyi6!rJmLg=QjU5bGu zoZV3>o2$9mbEQjAxy~Tbk8300iPPJ=iY;)ic+|5Qxa8^u&nd~?A-VRFg3-b1H?9K$ z^OWOO-1=`q)i%h#{R~Efz!@h5$88zMw&L1(ITxqHS=3Ze2K24Di3!-L}= z^nG-WweSy-IC1C~y#o?u+z_A#u|4Le#QIfI1qf?SCih;R`F1d#yvKTJ4vDJ6w|iny zY^5%!7M|QGw-=ZGRi;nrhlxt1M_P+V%bbRnBZ%j zRWjW(@I%h(-^*O`4DB2w*d+2p70p~|)BPJMqxW-oW(YF~2)*vwJZNt22`(Q`JR>UPL&~D@Huf zTXGzW?90s9}T~8)Pv*{kX>Ue$TA(! zD)>ir5{zCL;SoqqVNtd0;`8C0oLs<%ykg>q!!OOBO?(&=F-XT*jVGJegir0f%V6!Y z8?v$=_t8SRU>=KNAP)%}QLfr3-w<2PPj75iB_e|;X!50iK@CgmfWra}@0$`@kuI2Y zz9Ybq(+?z+ZbQE5zHboZMAXSC3nQggmt@k+Wha(|Z*a2IzBQ7-1o`p|yM5<27dRtz z`=Vi6usi=xvcLLF+xsA)r zP_~ycE2-RkuT_XU922dC7abA#oQU@-8{PJ&{An}ih?jwMxIcur!S3+mU1cyM-dd2~ zZU-0nl*IQ`fe4!Ha`V?K{GE#zQ1E^e;EV^BXx6~hPM8Id(P8zg?$h(Pl@k&bvlq@{ z$zINbhyoWRW^sP-r&t#zsq^~-&;EY^!27@OUOLu?HY|@uP{~b6@Gt@Kdgh~(2C)-jFZ*N*%3(_p|jAkdBKS=7J#9t4F$q7N`1A}cfr%E_^m8+}Pxw^&@Dj62F z{fXDFC6c(aZm&#eq4-PX4!8V0sD_>h?QXkQ`>)E5-#6F#=^yy*D6Onw8ZwThN~VkH z6#NcXR#HV$yQajm z60VJ4OYXLHj-|K#OgCxqyE;vH)2FJukPG#dt|~zjOS1v4ce3VWP=VdPu6XEUSlR~p zeeqoQ`A;v$GuQ6HwU7c~%#vN1M#6W593{&OtZ&}Vw&$Lg+t<&4Z6QmsG zdc!tQ>noo`#l$dsgrt5mjnkj@D*^|N^N;CISs`O-<~O_Y6O=xiWn7-Qc(C443eTHY zGBu%VL4D1&q_AF*iZ}W`bs?hE?y+m5apK!BT{w=j{XIX-;5i?~)x=Cdp>AE^M=rO~pGL@|JV` zGMh6JK*qq8L5aSE&Aydn7Y?{7S&M4cd7)n*76_~TU!=NJ2Yud6cooI-$6m69VKZZF zq>L3$u0!6}MGE8BzO4V_f1G4S9FuhU!w=aJmuZ=o{98QlbRgr!(71hNsVTxjWLy$-HE@n_e{e;!itzI=Cq5LuvA0 zwubjCcKhy`9?sm{8GV_5!DJf5Lk{SZe;4P+wDS-jcou}2#RGuqivv4g(tW$UiF)~I zf~(ng{-UV*q{PvT5Fevez!66kO)lLUI3;Vw{PRh5!C5eSiUk;$CG!%qVr03N?h|Iaxg{5A?LasR%zfXQ`m!f6vS)f=$o&+ z$Z-gk!@bc!x5-h_zNVm(eQ3}~g-S6{>E4X;ynr?E-;5Ujln?NCUY1(=ml^y=b~TPh zr{HgR1Jo0x##vcC)$UNAgxnd8WTztwrOzql?C!O;%lKNV$?tx^Z)XOj_Cq&wtKW@m z1$N|zk}3tgVO7B_=NuZ*r@Zlt6k9S5?rdmqf`|?Ki9*A)*pD%N`1`1vL` zfa`Q=!6qvd%et#rHmb4P!5SylV@n)_6+C70TiZ11%FZ zS7M4~tc>IPNL#WW@(J*|lmVDOm{0@?m_yC5Wxcfm?fjsW^Omj+u?D-dnYz5l{h< z;~=c8-?*Sz{;lKxdI@nr$CE%ySRq%ueYp(2va>j_m+Lu$`J_m)!V_-V`) zV0jf1*g@LyMoK01{jk&b>Z{FXfxy|t0PV1Sn-tceZD}`_KdB~*17IMd7NOBh!n!is zmY%Dl=wlux*3u^O@TAiDwb@;-10$dtZP}?z2u8!A$DLC>IVqtzZqP8E@c(-ewWonn zxQ9T&2@5_vV=AyxTUoqx#j~PsP)+oW^;K~BySLA`kOM2WA1FCnO1}CsF>|RE@r9UI z*)5zx6?l2L{B>eK4zc-8K$12Hq26Vg-lTH{>b-#xqFyZLB{bP3pOmi?iO0u!3lf34^(FitOFLUU>NDFz$&wn$ayOPr{@Fi|xfQ6E3atz(t zNZ841!?Cj4la z=zl2H|J?oQ)%cTPCz1glVtTFtXo|IMQGn=k2i?7U`TK|JEFD_KtcZB~+q1Ife72mg z@2fuAAF#*tZTGiTtSIz7p&8CvD2h*ac>ST;YH93U?1?d;xT05|apN6c6=d%%8I7LQ zg=}$MBgN_1MF8+mc;U{5ubH4o^G6;Mr63$N5p-1}JBa(|BZz%i4*t_caWZeoepG(J z!*~m{k$EtzrZloE+6Y%0`$k*&sV1N~DE>hZb~A!ax5e)Le*m_B2LS)(-n1ry@(Kqzja)2TYb`>`3ed8q}PNk zBEBy2BT@vCNs%A9){50oq`FhquSZ@^|vTBnSWa7c)LYA>_n2{G==0M=jIghuOWTQ^^}*!|SNn8?1Uv=jkSRSm2& zyII+m1Y=)*db2W2)%yP(j{yAkBG#P@!l#r>9t@u2*`gdlAW~0Hfvb&F%-gXo|S2d2*$K5PGbFaz@r|y`RTFP|6+?!reDg`Mg-+n6* zq!4)dHex&IJnx>U<$MmY@~+AR;I!iGeVod;Zy;2(xQQnOTM&Y^GX4}n{!y;B<4bDf z%{on!N4Z2jUfG#iP82Qm6|=R6Y&Ao2DCBzn4{x$bbAKYkYyP=iaQC{K&}(u)@i7s^ zXy?BOvf^R*^~q|ZMLEtf?KAUh-N;su4q8pIEYj>ns*5Ok=hZU~?(Fg30pgblhe-rw=rLWxpUL0gIi$O83i{t5TW z`IlCb^IJqpEoFh@^HBHRz4u-zha-P{IE^75B|8ds_@P`@24QgfV1Bz{K&hdh{rCKU zt-mZrzRyO4PgR%dZ8`VZfzuS75J=6Zh zMQFzn*Y$i2Y>o0KNrvy#2h6m-Qsuq$kS`w1j$A@)JFfcY98Pt=$N-0V7Ih)YAo!0z z?PFP3`5B>qaV9*Ghv}xw-?}{Q8JP7wWn>1CN=JTaAEobQE8tF%{Z_~8C`Io@X96eb zka9u!5WRYZPuy8)e}aA8mk6Y0xrpqCMpVVJX*&Xyvh>f1j5qOUJQjp^?YP=x6Zwe& z_jLXEn0R^CcXD~@QpP%4DO1?F*c|38)&~^kwat++Ll%h2*Iv0Jly-W?ja2x z**9+uQLTZel6sr}gB33h;ElyE9pi@ObTNLCHb+^{m~Lx=`E(J>+c3_!X8v~b_q3Y+ zr(WvCo+fwGxcG?MBg7k&?u!I>nm>t~50(d7AZFa^fvV|R-@ZdXb7ysA1}UgQVcZ{I z98R*CaB1r{mcM-w83Rlh1MP`<;cVyAl|fMc6^P?`Kh0T!0C4(6*}4aQqUd|p8!a@J zZRhG|7)c=XJa$-Xp)(fXHo?^qJP7s?vHVwN zb@Lg1vv+bO5?o-{vpQ-0&mK4J|E9u~aD*r;w2Ao}n0YWapO2MVO;vzfeXE=hmk(j* zBU!%d_D*=rK=D@8A;eF{wm)#j(@2x7FbM6m&Dm?K&=!u1EK7*WIs!Hun{*2=1*bdO=$JL{yrit)=cSdl}YvHTp7=5g&@s_l0<&GI?AZYwTBigkenD0QaFQf%ygDZwWj5qP@u{|Q_VBDZi zZ0U;t8dca0v)}S~CEhqOFQ}_hkR@3Lk(N>9W7EOx1P}|fe(fUBk7qeP0+y^S){E#_ zeW#FfbYhtiZ?=9k(CzZAnnAPfrx+sr4lQ*&U02~=z8-$ixL=IeN#Rxh&)lMeU$4bv z@XsoG&9W};v-nnzXXu2#Gf3(7Ey$ySWsj1J-qej@SFceIxGia^%XDPe>5m=Q|Av?a zC_V+`V%x1P_k(y?mwic_B;VOg~8er2UDz3Y5%OA+|Jv7{0U_Q?>`# z{=;|%&SkBmT1#NkDnCvMIe$O|-&R3WMEgF!EXxWZFiP1O)0!ap@Um=J5F5QCc-t``n}<3JHof&rb;)*ZXMEObH&D*PJ8_5*xrlgLKL0R6Zuot z3MrY6;|R;vef*F~wE<-3>>ex=~! zwdI=;4J19p3Rfu&Enl2m5jjSsej0Ypi(*s(RjBZ=kN)_i4B`J^-{|ECn)$jKa->^= zb~W0qby~3LHnwBa%JEQgqLjz1i_e4N*%gD19iuA}sRowkLLH6z6${T}p;s=c1p5z0 ze_rH@635>&woL*Z{))9zmgA#YG0M@V^zrRW@y~lXdGPP*1PF+?FfYhZ+E*ka&Xywh z#l$-ClW~wpZdqB`vqr>)yuv{nT?qPU>!RU&zj0c0)*SWm`oK}1k^(&;0lPBp=+WZqN>Ypa;zw!tS_I2R;_HF% zX!ljDg%HAL_Y&@Fji0}-lBj4->cHKsNT$}3I1Vplon&?G`&O-_8q3;8>-lscYpMTI zMx#ncJ0GNQ(pRF92qT1BAmUlf2hFnd&Hd}uXKW~Jlb%kRpPsszC0eWe8V8hevX9)F z<9i-$l6yH^`KEKxh`_d$H~*Imrx!^kO+_c+5x|+eWbgXH{2$^+o{s1;Eh$^JhEV8J&ujr_Ux1))+`&nY zyPtz`s@)ev#KA%MmutOOWTO2P%WF$a!eEw*xaSO#T2Buj&)?;td%|;I3eEmfO>&4Q zKk)WZ?TmSeznGVgER6YMEXL{w6v&+2*BjS#7Tbmw9p@0tU2IO5mYkitaaseN-R@N_ z7`0RU>}dkV^+${)f>Y*cSLS9 zDa{)Q7LsdZP`nWhF__mJ_f24-W*GSI%!Cww!@jn`n+0ueyDWZW{r2`4qT z9|Aw0ae&uV)-d}2+Uu}3F=hj1MW%)8I^QTh1e08`H!RTxuabeeYh~3Tw^+_hy4}Mw zFxFcfvlotMyR#F|WNZ%9OyH~1M=>8iX#OAe-ZCJn?t25JLn&#bK?J2i>24Gl z3`(S1y1PrdOJYDoL^`A!=>Z1moIyHg=!Sdn{k`$O-|x5kk(ptieb!!koqg7N*7HOr zYwb_7s0j7u$PW61OuFhU$(rkdLKLm0{^`(Ad4~BIqy5rsE?L>Ho>QebiUa<@-0oya zd+zPx?%CsQcM%g3%nMsvmm;x2w_aLzIIHA&gvILZ{JKOhRPmj++NcZY6z1PX;m(~d z>9RqDuimTsRc+sR&gR~uZ)i~&(7mQ=zWv=z-m9F+GlQ}2o6Ad4r2k3X^7kto34RvL zsEif~QA(-Ebv0jdL78u9&s(C45aZYn{e8V~_g=)aPF?01mcKjB#61~EEH}Z}6?W;> zdhZKB&QrZ%UggK?Gq0~#eAg$~?qmn;(v~`_ z*jJj0^AO`!XAtK&imEqfeSRf{U1}fkf6iGR;Cm`K0ck1k_7IA3bhNcJ&TEkU7cF`K z6dEtd#h?MR4{bf^x$Tsra%7TNlpCVu^tUMNJl5T>d)&bHc%o1|>iuaOGL+jx4|_ng zy={1`73%g0ZCNVt*{2yv-A`M`$0bj0w`H?E(p^^(!t1N$fbh9g!Nn4i#M?4<@)s_dhbZ=bGwx< z3{%zhdI5;fJD8_V7jE(8ZO>-yY_r&N`CPUr7T|Vr&)40X_bzIuw9EFMP|)S>oK&as zQ7zodO zF9x>cE_SpSY^hydh}UauPG<3CIw1XP`PAG?D(RcJ z<5yRWU!(f>g3j3Atx9%soV>A0p?Gu&72SHhAM*=$ev3@*YFU@tI$Cf{0VY>rZ$8_c zm(OwTG@`XB`dqKBu>@&VX(-m9KbkAv;;&4M!T5wRbvOGT zX2MK^R`sto6eT;sx5B0TcP<4NC#(Ocvo%3jMY?buat7nkyeSrq{XNH_=+2_Ief#|x zW?JQ2Jts6NwfKNik)thE>w80+hNiwGORX7Fu+GX9C71}0+cJ^u6<13R@@D!g$3@k2 zA<4!9<$(fcA->Ihr_8x>o9nf&{v~1SE~?7AYsAnA`({j`S)Areg20@zleroC3rq`F zzaHYinjWCJWDCgESx>MF>!bO!jug3_to?gjaR>sQpUGbUf}1owH!3>Ml?Q)FAoqKy zvEQ#2>kyt9+#VYwBqRim_}?7%;;_}17ZvR%j;dzxg=$1&ahgd2vaIjqyiUc=&x?P zt>Su>nT9bIyi4?zxH>!k8a^Rn8kaYMeocqwCm36GnF~{(xOOYE2Eg{xdlOYpVWM-F z>h>Whi6$y2ea+`q(PtKo;`S=5RongZYpG-PBLF4i+cg@=;=cb-WEw@E)`P2JN?8LJ^F;$@k8Zc8e^*c2 zNtGCY%K0*(tl5Hq-TcR2aDn!!e$r35PRG0EW(V9(}lk;f*d*@d(*5eC3V-5 zGPee zZ+pFW*Aj5iL6Cg~cH;J{DOcfo&~GmX!%UlNKf-+*zlN3$+ov6Z9x6OHtp`NvQ<8N% zWtI+?5!=06x$G_<>MJn(AIi;580RR;IDPI(A*}BkTaYLZnb|GJ(0t#)hbzvZqa~P5 z1d=MJ^y(g2Cl`$e%cgltx4f9v`A^JIE(ExduM)I0HN8^ZSn8VY^oF?s#n_zV_cOb9 z9Xm@=23Nmv=?PN6Z%R*?0sfz)7Qgl3>>M+FZYV~;#lGtPQ5BTBoxVI8Xj^5~3i#F~ zmx#*Ee_i-8aeuUv$IbO5o==;YpYnVK}2U$Py)UADxO_g>oqDQH8e-6oL6I!_CL=i!_Vw+ z-7p;ZPgoRITZkEF775rLt^uqGfX!(UniLz=>0V%A3>8VltS~Pnv@*k7UKP0Iq+@WqPX|a>2C1+4Bol>L2rH{XsJ27Y77z23O=H zJp9>f<=d8pyZ61xYhTJ&aL*#^igRubDS2A6+%LLRw}IQN?_N59-@E;^tf;51sxZN)sC{Xl5&hu3VPyGa|C(>*=>i3 zTSyaM%B0tSUZ8lmfb=ANTHFxV9mZJ=$eHs5%p3BVe|c=AYKMa@a=Fg|IEL33YSyr8&^YrXuRAD-DZ)S~-aLZqNVbiR#E%uVj`0M4P297)Ubun5;g>!Y9 ztKR^gg`>f~bncG_mL)9Eu~pyzaBc?w(HmGI_HH9%sWs8c0ppPDuwJeGZ@}B#n*nYS z5Q}6a$eY;iGVNBFd{pd`+~OO@&xMVyh2zyU2k&PvO<`81`MPGFN$%tGv~aKPA>^3N1xta zUuU1G+9qFO92%A6X|w`-Dgan2L|tliZCs;_poA4Rx$|L!UU=ojWWS8RC1KLA23bJ zonFK!dheqxuY&EiI!vN7(pYYMI-B{X!HV#aT95Mwql;>Z4EIb|>PwDSHTHr?1rZ%r zsJV9Z4h|F*nacYDEq#2P82&ozg(~jK`)?c7>@L0={H+w&4$PoO4p?vAgqg*kz#qhV z0DOKL43g5iWm{e{Fv0Sk^0rDq98^tno6xjy98#EuS7r_NlXHVN;wH)vDrD+cO&1WB zy%eWv$@ja4t!~`gPz2VZ5BW4QBnr%3mirtmW$+?CO~U}-M|uMLPcbh#5;*bXGZkl# z-_4;K&8f02b!KzI-Ti2Ztn-T9r4r^*(_1}E5uUHDI_0Ia0>J6l`D^deGK^&E=H9N| z!>7k0f|s)1o94G%>Mymt8|D!68zo3Ie*npHz=GjzRyR_3|G(7@N?yzz-Sv;kR;&Qm zYhc8;c9q7dV*kcRX6-FE@|{=j%&TtS_)@G9EpkWdu2AsdF4mA%OQ`)E30xo{Q|}ki znVU%SMcIfD2}sLqX$YVn)qX5-x`%Zw3r$Fl#|nue7Ipk7rJ zRH6SV1v}jP*4oTBUR-x{vaAzywXEV%h~W`*VI0)-VnQ#OR*sDGD5Yj#Jq&Mc_mIDX z?Bf^IYJe}N>?_{sWnJaom1gWtBp7zUm;1}@ySQWw4|i^wnM5H}XZ_8u1?%0ye_IQ) z-U4@a5)vJ8e>U^e{J{4@ zCy|Stl5(+smkr2au;=2PMrLbiS${Y2Y~L8 zi#Hkm*PHO}P!kNHoPPp4@`jX+xx~Nh*~IpoPU0S~OtbN-zwT710sYny&~;p<#=~9Q zoVZ%Re+lri8@^_Xf61ezb=Cao{m!YKli8MOf&YG-OBcc=)uVeDWU#U|Y%wMR9tO^d z6T1B(h1u87{ueQ7?jpvlqP^f=mLl7q*yB} z@JYOn|J@A%6=+DLX`a6gJ|FdY3;0cpA*h;+p@7e8f0@{gxtMXCBt|nk|_&i%h0I-GHUS-4z z>O}6fUa;jbKyDG@yCp1yCB_F$*7rK=lgmzC_@wp6JW+`RIwsQJ-b)_J>(0LZmH8g< z@Kue^cD-1J+g_771}_Z>8~9wuCEce^Kw;q`9aEYCl*@^Iufkn8xtyY&oA|F4PboF=pvn_NKZ zSk@>2m-XN58HqCBZxKh&b^(#iH-_(@IQY5YY~70Rhq-dRbVn?)(M2@P4x-(5G!vB# zfjU+j7)|-3090cb?#Wr|f!$lT8_!QPJFlpRMY!3oN!n29ZhttJQb0!Z2_QyV z*Ph)&)-nv=%EeCZKRW*#=@M;lhu}q2SF;0;jzHYcQURKsFD*T_Zaa4H#AMUkz*IXx zQc_ZB+b84_^Kgnn&cV!&QW)Y>{@N`S;uqwDvLi4IU?G0!dSz^TxjKti>^{dnnzPE& z3cI|~PkRQElC)q^{}F_H0G!-;xV0|7tnm9Ac!&PI&m8`gGJ6Tz7Z(Aus|4oV!>}^D zRo9>1Zn@?agU!cWQURox{^{Z>=${9EEvQo(5sURe!V|^0jiRn-zklWgiD14}^dn1J zDk`4XbLQdxrsaRF+%s|7aOSDAUW{{8z3BwbPy}Lj&cghz_~uUdUsskgW;a9uPmlDL z4FlJGT!|KRKB(iihrT13PTLg~opH--q_4-a)u}SeS$qw&qic=vNruk+!8@Rku@yO7 zC&rBtgB;2i32ZF*#9UdryqEke0>L6lJpOb%X?hX+Zp$fdIIVes?LCo>m^`xz z@#~%iyR)E5IuqlV)!jWKfLX+E+}-*GN!9kBTZcco$7Um_!eErgQ8G+mf?0|%C|BxQ zJJA;#cz75`yms5>K8M^DxBbS{ZlD|AN!R0)S^?aUiWeqY7J&($Quz}-i{l{>;|DZ zRTK7#YnbMZfYCTwL-tN+r8uy@MQ8s5H()>|1+Ym0;f85{5zi>Qkv+KWz5t9YrFLE=(`-k3%sR1vt!i_*A9k`RRKR(OV=oHSS3?`_&MFp7Uj1C$Q}a)5!B0 z-CmV2lpa2q%p%NJ(%JuexBu-K$#%E)od$msNePleCpE9x(jBrvhLE_&mEIY+(4Bzk>ZwU!Ng_OewD% z2u^xb+p97LvVxbpv@P0-THa>0{;BF=4K)Bln@Nw%`BFngS=N_jjkb zwE0EopP&E$XZlp!Bp?Ruzg&XOw*Y3My`Ao`$;I2YmcxL${BJjBIJiV!xOxl>&yN

!YW_n}=dI))t0#yrZ@@nM*~dquR;Z#FgyG;MgjS}N(kFI;o2@dg zqUSPFI5?<`3>YB+M3#@xoZ!^TnyxSIKH}NOf$$$EM@VL!tD1XDil0n+uR+?<_I?zc z)Xni-MuELW^-M3<`m4u}fZpAm@ko*?$aKARa96eb)UrG%pL8S49`*l-&tr}sp z>R|-N)$WbC;JlmWpRtb!fw^8{Yl&;z%@rzyF@4!uKU8&C*S*uPuX>sze1N!sT#2ti zoGBMwYwi7dp1@u(4+R7x?<&mdgY$*sVwwOeU7=s4RiR3his+-E zl9_Xb7))1%f5uH0!|4kIiwPBz-u83;L{n%)q>mEg5DZL(s1}n zzSjOmaHKWGyiRMJ(_(NxEoV=u@`+(X{&d-*=A_s2mI3 z@WFN4N#EXv*IrB6r?zcZqMHPwqdC5Zbu9;KjSw0yZ+5+R#*KNk7K^E?b0t?e&MWJ6 zZ}u!|4FeMYvs@uktbk!fc}L0q@`>~i9Edsgd?h4DlGuCIUNHP7HLbS*8|EXKeRR+^ zrY?%G^K*tfu4z@Yg@)lPJA=gKKh2r%gm1MRCdEJWO8#ePbqkD@_Nff74^4L;<%Z!n zRM7f2jfiu!m{uP)f7*MTnbJ07zQOJ5_x_@L%)USv;%G2=P$IIdo>_1k$1mB0iH3tU zZM~iHj|~79WqS~hF3Hwkrgk3{m~JC_pbmr)=22U89Y-#1=)G?5bDr%|Iy8K^cwVee zlO}|)Gs;Xb=r>>U6ukcIMnt#j(Xw##W*`$m@Zvk!-wqqnz9J<>aJ3?(9EvVH3vs^C+?C|)qsr_UKY*l^8pFY;wEPP0lUKcq89T6N#w!%6*u&CM2 z@+$w|>fDVT_?3wPyR|4;qwUWfJXbXI5{1~I&k2suZ$9xxT)9Gq=bF~+f3TcF#^cx{ zYXXFk#So!KrQ(Ng1zt)%8!{e5glEc@w;VpFNBPkDk6odGfFlqnxk&%zZ;cNE8+Nk~ zMIrsLXNzJ4T}?e%o-`Ni29#-B9&YcS?!X4teu5j%@T?d`XY>#E7nGELDF5rJ+Q1rt zIGmoc{P~&ezPc}R|H22H^HbW6%cWLN!IHpz?sZ?O{``T^|7}Q|KtI52 zK9T|774O|G24F_{fYpQTx!2w|$^Th1=v$1hmLGb@x&QL7e*?vz->Av}VUf ze~rziif1!y;vG2DRW1s64;PfjtSYgRwqBS9-~z4*|$ z_R0DeM>QmI5m0E8NZE~pV2~=wzdGt&K8Em5`CU9 zE=?c!leLSv+d2t##qXPp%U|Cv=g-uPsb%gpZkHXM%-NqhzdZ5~Mn0!PkK$^y6uCYc z8%utpCJg7K<1SOFyr6`nqG0!Yz4=P)Tn{BTp`ou`RR0)vm zzW1^)xD-bHWvVflhGZPb9|6U+^i*x* z2z^!(1JBiz8ld(Bg7T+W{eFP7ddpALJI$=1c4hL`se%!iPk-H)vg&uaD>&V#zkMTd z(8p{A_B~O)tFL@mi0ha<)6mpmxjrB9r?XiobM>#A{ZQ=k^MT?QK*_9X@|sd5Msh9P z>9&H18Tk%fjN|U>@l5Z2ZrEgmK z*q+wFV1g)%K8Hg`!=i}s@}=PBZYW;MSF)_*hd92Y)=rC;X|?ZUU~OjtC!23npKEAw ztlM+6Uyt+F8dfMO7wO}u2rjrpnodQTN+iwU>*jpGnhOx1Xz2mJ_9co}D<&)*;FK?a zGI1YuN6~h5o8Z%V*w(9jg+zZ`*MeO74c4P4KP z=qA@KQ7F}XQzR5{!AVH?_l(A!*-K0fT(`y01iDT}qhqS$RH|Oi8;#KrF$nY_MS1hxv{&yrDy#vhm zv;e?2u*g#_^0wUGTo$r|eX4r-xSNqOzf=4NIqTkk0xmL^v3Rso*-N(xFkEd^s$|r*>dT=rXh{n{alRF|XT6d+T&-unYkgMfo5#+Qz*nYdk{&Vq_hTh4HjV%} zl+@~4zguAF7vVHcAKTn%ycn_FW~~Zcfa*-$!CT++Zf?JXNdtf1w2o>BU)11=!R^IV zjPNmCMkS*4@+)Ap*-N3R_s-{+pA*cQPn)u!s#^MWNG8p_e}tDqZ|bC zeldQSQtrdT2`sM!7x_C&#`4n+Z2FxdC9mz$x?fKUvl2Cd|C9uop3GRBy3;Z9dlzrd zT9>N=Tm`|EeC5m^0B^a6>=uEPjwc*ylNnrw!m+XfJbSMeOG5k1k=i ztuX?9&v|z6%P$daZNUNtW@zOoC@2Eq%5u_gueY?u?i9;&zJLER)GKkqTi#aJc*iMT zv*0%O!?LSB*xfsSXb^zq5{MzCBtckSPIOQ}Gkx=!IV4^Ga>8W!a2h-mfBWG$PE)r1!tt&r!;b zX9R8+v_L0Q>%7*jKyABX@}*ZMcr^^SmiMIa5r&ssJLNZ?$6drvqxbKTWOWmW?x;^2 zNruwOy}p9oUK_-jxh}Wiq<3hicRJ9hE|H2Nnw^b?-y4Nl7_oYjn@3Z?uGR!^b>YGG zQ+n=tMu1nOo9LyZaH!ytxdniXP*XaC&}hw6GDNa-GlaRiDca7uJMqdplckIl>>RK5 zd))zRQEDmENHP2qGG=LE4W6G7s1wuMlh&2sUmN4{vYC%;`ynA-9Sn=~^*=v^VI_a2|3q$ywHq~BT z@tD1v_WmYL`0NfWw(jz*dvz)|u0^CMKZ{B_lu9L($}*Iyv^%Z!sKB_JOzWsob%xBm z&o#Lc1!NSKBssOHwP(iB{wCdMZ-f_sD;B_Bm8nt0il0oQ?VJUlLX6nmfY{X;^e(ie zxP?cyVj{Z`>^I9BkQ43MBPwRtlXUnK*d6l&nf|=tvj`9nA%{&22}ky6Jcgrxi#7ST z0a8?1YAd|q9lco1_@K-{y>!R%S7G;~@ja%aL22|y-9cFTQ_|S9Z?)TVtKREz-2mb* zPCN18NIdO#@L$IS+B!T1JnLl>v>tmaNgy{J^cYT*-(vhQpF=V?c;ao@!>+X&GICSL zc?m!`N#v&Q_0|u#>VCN7Ri1{TPoP13_)4I{hMZ|V)o1-FT-Gf0>EaD0@9i-(`5drM zrfqG;;O}qVQJl4G@e{5}8@@fPE{2f2OR9C?NaAUBxcj3Z6w^=KABvSD!zx8xQEh0Z zsAGu^QU^hM(ClJGz19F7c@ssd`#*)np~BMCJu*d@ z&%GK(!wpq6^=RXW-{eDEEWkVI&5x6qeHGR^D&ax#siSE_WJB$m9EFo3VDyz;E|)e# zUoVUAXWt_VjwPy0oQL++fO1@=@p6oyn~DdPkJj@Wu*dI+&LIBSDAjxC+6opO;Ejxz z+$}Xu9J{D)%E z%sn45=jQ3KC}w$;#+15-;*-!Nl{O`p7HU71+!slHqK2yy`X)|YVf1tS0|58Lj0E(8 zNiDKnHI1s}L*UphNy=WLRuByR3zB@d|-UG_iEzjg$g5>GvU3Z^BXJ8JO#NW%@wEStrMnx+CE5jwIGB zx@0j|#-CQy{%q&q>m3U}T8koGyid)$B$u?9kq5njy&m}pGwhrj<1>d5^ZIN%Y4j+! zOV?=6tRN+!hiobXhLk2Br7SiA@*|9bN3Q^oJ4@slt4mivV3`nqEphv;`tMN*%OuvG z>{1NJZdFlAt?_PH!+OrP0{Xq`0m=dJ*br>#3EHb@J>7liZ$Tu6dyl2v_%vR*i9LJN zz06QoJgPwuH|0|wH80zsg!XE}ETW!ygqIJ1}SV%0JQBn)R z`d*chz}czGYn^(~A7Ff-L>CKp!a%}#2c(-CAD`XpiuL1(O-LA33<6W8&NHh&!0ko# z#NdSrkm^|BHA?#X9hr9zq-zlcE4z@_Ve9Hx+2Jbq zi#22;<#YQ188lLT$hERXzV4&$SFS?z5zmP@~2XiA%Y$#*0G$({{oeD-i(B zC5x5UU`UK@XU8DLm0PQ!5_K!tes+{yYW(3oac7*689bUG7!)i}`x{bD^47#gE=VBS zdjReuOWlKdCF?*{_;40Zx(YCju&e6o)1aHFy`4P4j|@gw2DE-eceh(kw${-&@$QZKxWT)+ggBQq2H_*e8VG*B0icA z&1rn6qDV_Em8=NjlILM%^d>X2hZY*KeXa2?^e z&pwktr8lH#VDG0!K^x?5OwxWwgf(GxI0L+?mAhcTfCAdNxTthTMCa`Zl0mqr;C{E%@!?ch1|?r;c7OO5dPQzPVW=!2Qhc8PI?jJT{~Zq@~#tTL6@8xvJrM z1B#tiW7(PK(iU-5+m%w#DQap+7ZD_R+E%yWssyXkWBT6Y7x6UPz}mHi#m+) zEuKV!KukLj1g)QUJBw|dsRavMqhBxSplPh4tV=?4%qQ@qi61jCy~2LB?1@o3;QC9= zhOtLYjoOso3`c~&=O9Gi%tneHjXWVfrO#+!)k$cN*}lO-7ytLK0DqY%Xw!q8+!&JI z?Iz#=)DFsvkZa7{-=3qwJe_vIFsQO86L$x}*3Z3>V|;eq`q-f)gF>;Yg#FO#cPvIVGkxYa zLAKq~g@cP9_TDl%ydPmf;NeJj&Tq5IF`?y!n0gE=Q^Ec$X)!&nMok=7mNpgXtvOj^{9w48(r0 z9!RA64vENsZc#^T7W3!!% zi2Gq^rYnHliV+aFOos%XUJrRnXX?2($Lk8)LK2*3D?s&;0)4NYpNd#^<@jBoAg1qS zHa}fg?Ln=PUYD1F@$NxOSKUW4yEdC6)TGqXO^Ca(c1XOW1%I7cV(G!tavd8+;V=0L zOLsZQowG>lr+*z0%;1(GA74COp%{^B-KQ>zcV89Z6k2KA0p8!8#pS6Ig|lz^wc%HB zVlyG3VEmxgpmk*^h3AMz4V6_|f*h zOHW+h=Wrm`T$UetiS8J9%m+D_CGATN;OY!{@=u%gZFi)8=N#z>3%ZzNN8<_Sw02bF z#qiKaVT@e*7>GXhRFwve(wDs7&5T+>d;q&JD8%rm6dSAV@0B#34>Ia=PArmQ_g5p( z6Ie0$j+ms&c{(}J7M&o9(;ZAe*f8Ts=G8xA&eQ&uM!Jinr@sn+-J`^^#4sUl6z|AD zJMDl%_M)v&PAV$oP+wCUPVAAF%QVB0clh8%zd7D-r)?6BS0!v#$xOIr6h~~-dGp~G zR_Gmpa8NYs*{8P!lnrWn7?}=1vUyG|1h^9Ij!Z~-lrGK3^$)CGrd|^+KEl0%~ObswD(31j96Fwb=auLRyellsxlfL5LC$(8qm_^7C zZY*3mNHC$Huj8q(Y7P_(#lJ|1-$lAbW45JVhquc5b>^Rcd7WA?YH&;P!q$aiqT!Yd zHJwa8WK>%~z*AtV-eNH8UC@btsdGh&VuIs4G-6loF`&-f1x zB1UPJzYoA!2wl21y8G}~hmIQ`MU(V!u$~8{(Wb7jhVz@-QhjZ+3Pfbts6C<&5BB3j z4|RT}29F)R7t>CFF6j>GD2|-!lwLVcFg}*k5MRN~!F{-TELHy9i;_S3$!yvPiek3w zc}+{^ux@RLz|0IPS0QD&p%fgmH>VvNSC()YKU~zE?Vz>i4C|#w7%y$JPEJRZQ#2~^ zGV^PDr&uv)^gh;2M`lfoz-y4_*GRs;bxQh;na3;__`^b*8a07ANs$7}Ntwd7Onup7 zUthp9MzPKV#2#VBrCO1@CUEyi3xjC=onHu7=a>9634P5K90aPM{r(!M13D=Aj^{;^ zV}utfCL6{iT2%PLt#gid@-tTsEi6xUaqnANBQ)DJExXw^K$&6D#&|BQ*&C^&(P

    ;>=Tkz7?VaMix}#+$OZ?Q!|mZ^c=Ky8dH}+6TQVgykf` z=0@L;4~rLU1aeriI6Q~zhypa=DW`-SaQqzF=FaQ0Me5j<`_1PPt(6Z(1_sp$U9t5% z)N(AHql-NJ{Mxu+8AVIaJ%ws*G>oFi6=+hg*wy5ANaJlc+kNPNzLjNsEBR1W6YBy1mrTC4bgAVMK8v|HMr?$O*xh%lu447 zvQ)F_hpJ+6BUu6+pyU{Y_(Zqs10QlyyhPwWSAjDL7w*%F-1c8 z@Z`xN*+=MM5uf#7nYsGO?e%xj z04N)>U>7p})52%n9Zr0-uQKDG=iszTO4)rcoTJHK{?)#sjAz2*5XoSH4lrONX?lO{ z6N6~MSyETCx~2@YinaZjVP7RSdq_=ArL^FZ=;o`E#Q*6^D?i=HnHV5LGVN!@XG=%rc<`q=zj|!>V1#|SuR7lS3!GwW zN!0ef-^gIM_VT-$Cv67ni0H1Huc~l<#+ry`Y0wJ2ZHOJo3+e+w9Xl+A^FrF7Z87CG z+?+}oo*u#TcK)3}D8Hv?t5$z?d!?XX5Mi*I7_DPk*@}vctl;B zUDY*jFiyN?p49s(VTUP{iS<*w9Mkzl+G9sqwWJ+_7Pam~YeY3@*PuqNZpdNuTViC8 zq$aMQt#T)+z@`$9Z#E$eTh!HG~lI(ff)`!n0by{C-wY-$*GlIRhwF?6G&L-D5R=kfWB_YA{- zfQXbK@X2w7eVljS0H43+sIaBJyhU^tJYzaaReZc($b{oSN9_pxCWO8?H7)hfec7HM zV!8c`>y+LUItSy3CHwWdI*TaPA|aWH50gLTEv3I9`OEFY)>R7^L%c-Fu}4SbSEN(G zSy@ZIdVl}%v8?EcW$)qH@bt2$5uPHgk4*V=IBjzVto+d3g+OSTMyi<4mU66RG>p>| z2=$DDONH^7($pRmkpdf1ma;d{(7ul8V0v5bv5$@cB@71R8pChWU_ywK4+3ecfS$tYPo=YIN3^{XtQPk?M$}3Us;8x=j-rCPx zd(5K7%^mrxA%1coCrw#eGuCwl2g67Gi;>H|DDFpMbA0E+laiWwj`84ZB}6lo6VH)q z?0pfeR#fh@$EFWqOXzzV!Z#g)a7v$Gyf0iEZ!{>5owD85d`N0lp{(}wu&!DopXAk- z!uBNob>P<%Kk2@MZu!wTxVtS^+gT}#vUq}s7${?Dyz<8R(TqIu)2Bg0fy_c(9Q&mr z@>B&Rtj5j{t#!^Bh`LbkhlkFdOLWqZMn?;gvb~c(5+OhsQ+;@}6$!(@V|Dg78n zhX8@wF>?Ja%`JHlndz3a<0^Ke(|fC~O-aM{NR4(qWa%slJyu9Abga~I4d+AUQJNDQ zln+(`uH7OMPNu`I%}vjh7M`(2*R%Q@a+r%(^H}_yBN4ePiD`hOFk4uNaa?eLkTy=Y zjv5pnJGjwnc21jXjI)iGNarzOTS)L*JL)hjpN|KOyIgY=b8v9l z`FYFoRH+mR);o#1Hz2gnJi;v6!AA26dNohfrq_~TN>D-(S(4e;YSH}BPYI$2&a3YZ ztIkrC2e`g(R@vG?(ef*FSYq`~x#U>BSsq!Zgjb^=LyDe9r=wOvT1f4g2km)L!2|Ga zB_U+=0}nxhzDRGMs%CKQIx9cF1SR*Er=J)>3B{CPaoW9=Q0p^#>8B2R5dt>0fdWvL zLi{7V+2qq72cjNH8qX4P@eio5s|05&8ad4w^M*d^uRN|<(h;_0)0K$loFxL6&@l?d zz3)R;cWmPV+n6wF~pQ30q~9)&{>z zVu?f)y+z(|`JHG6xZtF+Gp<$U%@NTB4+tKLPsGQm`MBKziQ&S&%{7o89B(BAV;rln zhFiZ*$yyT$AO5AbL0*f0!p4`gFYCK12accEm=xk@$SUk_TWuH2XiWNMzcWZ6A03=X zUBn|!fREc#T;2Ie&f)liGK49;5^rv?YfEN$ab~kTr|7BD&0YP9mdl&oNAc*_qQ#B^ zTjAt-;cq4m&jQNM#EfMzqug0MklS4`eu{B5@L{X0 zH)$_`0Z63ShX=I~PW8%9SdY%~x$hHtv@f>omRm3)iT&X<6PmihsV7`Ih3tivUD(Yc zxMmgzFS{0HhDIE{9q}%|HR+p37QAHJ+L<2wS2=oWG<{*g76=Pl-LjzrV{_Ja$;Ed> zW<~eAW0*S*dX?C;EtTP+TE(llW>m{ze7u+G>ziu`Bt8~C0j2kMou^<{JWNxn9+p6I zJULSuGad;Zp92g!8`q@vUjJ=Y`b_*p$HG8m(` z6-gy~fE}hFH$C9-U=XGGmevM;J>x1lwu$3=a!MnuBu33&16#*a2zT1(| zeA}M+$ly)gxzNN4?qaawVC-`-*?4?9d2phoJ%(cL13m#zpKF%GN$jW;ye#gZI4MgUcQg&i_ zY`i(IIw`v-Eh**M#829cP}C_RxrhfynW?bp(NoxXr0>yOkbuNSV{#XE+IcSCO<^cd z&afq45q2wsg$=C>v)x`Ckt~xm?kk&mA8m}PF;@iNz@30vE33JU`-~ZdB&|3p-;0N0 zjtuG{&<9dU}k74ZD%iG*v^6HUg=OLVRtGy`S9{IGnKl0RL6}WC&2&ExwF*Nd*$N$$V%3#bid2W=*~YsACzfNc3EO>?7eO>qgh&dtDXCv1eyIGKBt%(1bqY!i3GAi#U9jqdseEL?dCy;#&Jc&$ z5}yG2GQ-jomMWbJMl=i~a19=X97e@ao$-E8@@6zX;u@2A^{~qgZBv$7qAQH@NUmc<+wF~QV{BueBGs(iT&2mTXca205)Z0g+M(UJFcG*W z4u$+Gj6gX#gCqt-UvD6+$2yJACcEA(a-L+F^0vbhZ44MxJ{Pzz%dDB+Iwmf4J?Ng^Kx zz@nT0e{S>ojNUWKuKC$D#w||dutTyWu&5? zZpg{;c@1X!xTr!&uiVV_P!cB{%Fgj_2$NXk{P-oWSN*5OlToVma&Z_TojtpwQpU-#AGP@RQ7#rW6$qQa{JxqdH#6(XI?Yk z<(%_9=X1X2{r;R!wJG_vD7&4|)twJU*GO=_un#F>S$aJOq0cWRawwiF5z&aXO62^( zFgy~?%HL^T?7p(%R9mFNTdLAPYbFkis6&S?v(98fezr@Df>FNlDMw`>uh1r^R=S_t zW!b}ktU@3^OHp`S`^3-*+7hN=pdGo|WpFb(B zZgdpj-lz;+Osu(-_-YpEbLl=SZ&(<-k9FqzWrgC8IBu&NRYNMF&#gPp{q7X*?i6lm z4CjBjw>Y#G(Go@~30@p-I~JRV}FmM>D&2Lu{m6a@?HXrP&AEGCTPjVX0qFG=0=` z_0$hHA$DKDh+QcZ*ZhoZRO{iN@D{Q|wZtrOmpJPWS(-kLf=lVz6EvP8s{w8IqI^Y+ z-&(Br@t;u|4CMv=v?S*=^wP|5fQaa|yv!)Lau|2TQ5W&sR8+XOtD(bEvaHYXU?=xQhiW)t!?Z8HlhQs2J4ihL zW$N=~)&e!+6e;HcvYyY6ZzIQKSYEWzdO>vnP@FyPGDc2<-WDIMCemAA(s*-Fap3-}Q{L5-nZk%iX9?<2 z3)EM>xJZFr?)DnH!5UMrcmV1$aSw`Fw@;_2T$X9JWPjQJ-f|NA^d2w9;ie?Hfz#0! zlN_1y-ni%-XtN4Y$>mf19b#(+0gsHqNsm)LsrJ-ebj#?KWfN}hI*%EO_FOZ&^W=;| zOklTb)5nC=S5F@CxmrknGZW7kqo@!Kn9svc6|eNSzdN^!OrTi$Jk&q+_Nhz^K7&FT zRUz{U++&a{LR`wT7(wxxsK?%s*0TnKh>;U0m-!sO!RXR-RrP~fE&01Ixb^dzmfhZe z7(uWRLp{8kAsiP$`k;3`c__Mce0hiLLUh;Rj5WmvE~uKw(1+8BMc<(Ljd1CWK{Wi} zLT(d8$k8Z4CXA8+eNf|c@=MD(FCseBe$W{ezgIc`ma1HX)my3t)M((N6seJ|UL@~$ zfao!Oc|vcF&(14}n5&C?(9eur>Uun?5{+%vC?9Fhg#&+&e!RVJVi-30M6c3CWSqJkC0{jw=Q1 z=;M;5D!7JAacEFN)yOTmx(qf4lvVRR1mobb7Hd_*otFzQa@cYv6UbK9AvL)o#hNN* zJ?8gc?4hVb(+baj$?hLia^8(;rl>Eri3Ui)&p-lA7~0S-ioLTCae&yVb+>#v>rNkA zL6O!SRld~aYhnDaIei!|7Wd&jIvS}q-l$v~_WGTKv_f{VIW^~Kf!%*CWJT;VX9z%< zqWPwWZ|r5Wb)4%!4TU=)uW}mt3@Uracc^wsRCLKVjuu9Ya#r#Vi!K)FECtSW># z{+Edm5`GX^9*6E+()>58`@=*??*b;!|M!$U2iW^*&1-!a<>Ljj7k>M7s_M*Bz)#6M zW!}2=VTjSe;Q1(Mf5J&E90=Yrzt%-$a|$6;Di+#IqpCA{?}3zTbgv7+8Ih5=AD#@j zvQa$HC6-Rs=l%qXuZLF`;5R1JuQfhWzcorNuKF`8>FBIgrCGnQ8LS3P&rZC0zrEd` z@lZnZxO?eXmfJm<_%kfz@Iy}S@RsipygTP4hv7J#dvV(+tIj;_IM^8D23UDBnwj>C zll!vQhY}pj0Gn63`F(xlBnc3MrIybr?aEc_bJLwH8r-e83uVe8m?vTGx@m27#%>vWYLpg1!D zjGFgi#~{(WNulTEET^$%8qJFwE~e3SO}OD(KDostub}d_Kw!G;cFE)@0g`F7=EP!(b{3$B0pRTH!eRklM`xW;HzZwY7xF@o zOQbqQtJ)n<%}9_+)!rCqxBYNXxwu~6soZht5@f6Sr_=W|CPa?V=kLQqz609X>h@5a zRS*%=JD?fSy#bPg&6&cAV6X9OJ@mHsa-?WgJ^Dku z_+*1VFX8c*(^Un0X7lkfO6n3XB#4^oo-zXv^Fv;#-o{eCmuiU+haV6$U}+x#!2RT3 z17>}b^U>(9`^Y-M!F2$_eUf&2T){u>GvHK(OsFH^^@rLAuy|+R;nS)!;!6Smd*z4V zAlQl!pq9IV@lkOgl$}dg`ob^}I2Ii1!R#-_D01Id-#vF>-R5J28r*F8Xh z9lLSE)nGgBsm|XUfKBmR>k8n(lMdk5z5E9UR#ruRKF3rMmai1{csyduSK0;RpcIpz zq_rYc37hp2o<~p=R-UI4=cw3qU>amX0DL~yZjL;^37#9QsMLr=L@IU|WF8~YEN$K6 z019hOk`m%gD`%1C+PPpBnxwsyzcamG!n6D%ft`}0T-3Z+=4)T}1Ib`n%HbKmZ6hNFe$Oavlj0U$RMM)Iq*e(O+sJ(GajKGX{tt zM#s+AA4xg7mWq} zn$EJ=edc&I`Ejtz+T;p)3}L!D|87!SfjLXF=MaK@h7D8l8}#vmbwye3ikl4>^p2n- zt3gzK-sdcyaJq4(Y^G9)_dpPzELAN2UBeCC#VMA}0Nj;x0N9TjsK%z;vgD)Mhl&hc znl@gk{HLeQIR-|O2^07m+pR9_Izdmd1kD3aS#m`@RDn*zUP#<1)W5<+O8POw$#+s* z23S_c#wrW;+q`B;5_gqkoPo6?byqT5!^!GZZ|@v#ng4UGmS6c6I_3GYBC+9VvXLqjvDU+r?a0<*;&}ctR%B5*7`DX<*$|#DekC_E_AD+FD`m*=#Frd52el5EPlih_OfUs zbuT9+*y>H$x+IQf6kkNP2=<1pe_i(>6&EU!-l;9YM^b1r!n+PaLL$At`C_?rKWua3Utwm^LkD zO+6QIjtibFaYSeJYre7WG_Rl{IeI;4t~b{P3-VqrK9SioX&pHArq2zXMKT|Ub8ikO zdonfI7^&F#J?jjERpfS2a_wX@p$aQas<$M{@L@79XM%`lMDf)dAe-yS(8ul8y~n)Tca- z-zJmMT{_t#B%n^{SboUISzzKB!iYv1$h18ZF&iZbe>K$WbJwoKQ*li$y@uTnk-V{N z;XBy+$$vZG@-@_hGJKyUb|2&V@vIXLgs|Qq)3b44&(vinUm*2h_B)x55#8%cP-o!< zw|fExqG+gO_fLoI70-{@O^lH6m5`kg!jwbautrIvhO!WD6j?I1mn;^Y8ObsPguh>@32uNiAVyd1s3dBsK8LaV{9_1^aRMW|^&LMMpw4 z&4XEIp7wdt%iv`8C!a*Ad+0(m9(aEX{-0^6;ypJvz?eM$O7ePkmw$xm<_>n&9qAER zm3;7O?y;YhjOQzSEjh9;yD*z*CwUc>PuyskS0Q{*^v4uvy%^GR38=zY)XvHbt<}8` zWQ}=bEFgPJgj-(u_Gg<(1mLxJD_F6XMf453XUgr-hPcHJt2L+F<@2e5G-6hTlK{m!6v9iu7*`i&c8HKq|3Il#2Z1kK^wr^Z5$)l4!s*2bND`&fTahl z9DsIE(#cOMf(*0^&S|fXKHJ!PUC{zA4U*EP8@158|p~(-5%k6xr5D zUH1q6-3?@~X^f0yO$4w2)BzLd$cFhd)`AVT_@#=X)|?4!PnjrjsO)SmqWj!NiUE=u zqN}h~iy#exGzTQZJPYWR#%19iS3NuAw%cp0cnE-=y4U@`r>+grFJv7j3HRVQ4uIgv zm?duZ&K8*BJIwy;+cLg)`>q*X>DW4S^%sL+@d3n8)%?t$!e&cz$JfBV9tYmY4x_D~ z?CS;714!?J)y#HW3!(x=(nxP`5NR`5(?sB(C@YN5RVYa8x6|jhrv)mByB!@6fg^!h zjh-q_Pb;@W@=*cc7tRPs_5GTvtnLlD*QZG@h-$0bKFuvTj&4qR>a7cT+cmvQkn6Q4 z?k3fyVk=+?IZFs;RwMUIE{bg5-oK_k8~4GK2p480B=$4DIk?qAw!Wm|#u63XZnm3m zAOF83M?J8UH|V?9_S)-Q`Pn1=(z^e3nE&w)j{wJ}&`TH71eME2kZj=Z>}j)8FAW?+ F{{zEKaF+l8 diff --git a/docs/advanced/index.html b/docs/advanced/index.html index 4247bc8..3fe31ec 100644 --- a/docs/advanced/index.html +++ b/docs/advanced/index.html @@ -49,7 +49,7 @@ - + @@ -244,7 +244,7 @@
    -
    +
    @@ -320,6 +320,17 @@
    +
    +
    +
    +
    @@ -371,7 +382,11 @@

    Advanced materials#

    -

    2nd project stage problem

    +
    + +
    @@ -394,11 +409,11 @@

    next

    -

    API reference

    +

    Optimization tutorial

     diff --git a/docs/advanced/optimization_tutorial.html b/docs/advanced/optimization_tutorial.html new file mode 100644 index 0000000..530435a --- /dev/null +++ b/docs/advanced/optimization_tutorial.html @@ -0,0 +1,995 @@ + + + + + + + + + + + + Optimization tutorial — ICE documentation + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
    + + + + + + + + + + + +
    +
    +
    +
    + + + Ctrl+K +
    +
    + + + +
    +
    + +
    + + + +
    + + +
    + +
    + +
    + + + +
    + +
    +
    + +
    + +
    + +
    +
    +
    + + +
    +
    + + + + +
    + +
    + + +
    +
    + +
    +
    + +
    + +
    + + + + +
    + +
    + + +
    +
    + + + + + +
    + +
    +

    Optimization tutorial#

    +
    +
    +
    from ice.remaining_useful_life_estimation.datasets import RulCmapss
+from ice.remaining_useful_life_estimation.models import MLP
+
    +
    +
    +
    +
    C:\Users\user\conda\envs\ice_testing\Lib\site-packages\tqdm\auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html
+  from .autonotebook import tqdm as notebook_tqdm
+
    +
    +
    +
    +

    Create the MLP model and dataset class.

    +
    +
    +
    dataset_class = RulCmapss()
+data, target = dataset_class.df[0], dataset_class.target[0]
+
    +
    +
    +
    +
    Reading data/C-MAPSS/fd1_train.csv: 100%|██████████| 20631/20631 [00:00<00:00, 517496.66it/s]
+Reading data/C-MAPSS/fd2_train.csv: 100%|██████████| 53759/53759 [00:00<00:00, 539376.73it/s]
+Reading data/C-MAPSS/fd3_train.csv: 100%|██████████| 24720/24720 [00:00<00:00, 506176.62it/s]
+Reading data/C-MAPSS/fd4_train.csv: 100%|██████████| 61249/61249 [00:00<00:00, 539069.75it/s]
+Reading data/C-MAPSS/fd1_train.csv: 100%|██████████| 20631/20631 [00:00<00:00, 544734.35it/s]
+Reading data/C-MAPSS/fd2_train.csv: 100%|██████████| 53759/53759 [00:00<00:00, 481589.37it/s]
+Reading data/C-MAPSS/fd3_train.csv: 100%|██████████| 24720/24720 [00:00<00:00, 427933.68it/s]
+Reading data/C-MAPSS/fd4_train.csv: 100%|██████████| 61249/61249 [00:00<00:00, 546792.80it/s]
+Reading data/C-MAPSS/fd1_test.csv: 100%|██████████| 13097/13097 [00:00<00:00, 505412.69it/s]
+Reading data/C-MAPSS/fd2_test.csv: 100%|██████████| 33991/33991 [00:00<00:00, 478933.98it/s]
+Reading data/C-MAPSS/fd3_test.csv: 100%|██████████| 16598/16598 [00:00<00:00, 520419.66it/s]
+Reading data/C-MAPSS/fd4_test.csv: 100%|██████████| 41214/41214 [00:00<00:00, 537036.66it/s]
+Reading data/C-MAPSS/fd1_test.csv: 100%|██████████| 13097/13097 [00:00<00:00, 515225.24it/s]
+Reading data/C-MAPSS/fd2_test.csv: 100%|██████████| 33991/33991 [00:00<00:00, 520644.44it/s]
+Reading data/C-MAPSS/fd3_test.csv: 100%|██████████| 16598/16598 [00:00<00:00, 489809.10it/s]
+Reading data/C-MAPSS/fd4_test.csv: 100%|██████████| 41214/41214 [00:00<00:00, 537023.31it/s]
+
    +
    +
    +
    +
    +
    +
    model = MLP(device="cuda")
+
    +
    +
    +
    +

    Optimization without changing the complexity of the training process. Tune the lr of the training procedure using validation loss as optimization target

    +
    +
    +
    # model_class.optimize(data, target, optimize_parameter, optimize_range, direction, n_trials, epochs, optimize_metric)
+model.optimize(data, target, optimize_parameter="lr", optimize_range=(5e-5, 1e-3), direction="minimize", n_trials=3, epochs=5) # if optimize_metric is None, than validation loss is using as optimization target
+
    +
    +
    +
    +
    [I 2024-08-13 09:53:33,784] A new study created in memory with name: /parameter_lr study
+
    +
    +
    trial step with lr = 0.00018951382914416393
+
    +
    +
    Creating sequence of samples: 100%|██████████| 100/100 [00:00<00:00, 33442.07it/s]
+Epochs ...:   0%|          | 0/5 [00:00<?, ?it/s]
+Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  10%|█         | 24/232 [00:00<00:00, 237.23it/s]
+Steps ...:  43%|████▎     | 99/232 [00:00<00:00, 535.34it/s]
+Steps ...:  75%|███████▌  | 174/232 [00:00<00:00, 631.00it/s]
+Epochs ...:  20%|██        | 1/5 [00:00<00:01,  2.31it/s]    
+
    +
    +
    Epoch 1, Loss: 39.3309
+Epoch 1, Validation Loss: 39.3238, Metrics: {'rmse': 50.063830627114406, 'cmapss_score': 1009001.2641988464}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  30%|██▉       | 69/232 [00:00<00:00, 686.55it/s]
+Steps ...:  62%|██████▏   | 143/232 [00:00<00:00, 715.14it/s]
+Steps ...:  94%|█████████▍| 218/232 [00:00<00:00, 730.32it/s]
+Epochs ...:  40%|████      | 2/5 [00:00<00:01,  2.54it/s]    
+
    +
    +
    Epoch 2, Loss: 36.8534
+Epoch 2, Validation Loss: 32.2449, Metrics: {'rmse': 40.68462185124231, 'cmapss_score': 326161.11634528585}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  33%|███▎      | 76/232 [00:00<00:00, 751.21it/s]
+Steps ...:  66%|██████▌   | 152/232 [00:00<00:00, 755.41it/s]
+Steps ...:  98%|█████████▊| 228/232 [00:00<00:00, 756.70it/s]
+Epochs ...:  60%|██████    | 3/5 [00:01<00:00,  2.68it/s]    
+
    +
    +
    Epoch 3, Loss: 22.9493
+Epoch 3, Validation Loss: 29.7487, Metrics: {'rmse': 38.11920046818511, 'cmapss_score': 251687.82706875}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  32%|███▏      | 75/232 [00:00<00:00, 745.06it/s]
+Steps ...:  65%|██████▌   | 151/232 [00:00<00:00, 750.90it/s]
+Steps ...:  98%|█████████▊| 227/232 [00:00<00:00, 753.53it/s]
+Epochs ...:  80%|████████  | 4/5 [00:01<00:00,  2.74it/s]    
+
    +
    +
    Epoch 4, Loss: 27.0963
+Epoch 4, Validation Loss: 28.3393, Metrics: {'rmse': 36.542765927986274, 'cmapss_score': 207206.28175345066}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  32%|███▏      | 74/232 [00:00<00:00, 735.12it/s]
+Steps ...:  65%|██████▌   | 151/232 [00:00<00:00, 752.66it/s]
+Steps ...:  98%|█████████▊| 227/232 [00:00<00:00, 755.28it/s]
+Epochs ...: 100%|██████████| 5/5 [00:01<00:00,  2.70it/s]    
+[I 2024-08-13 09:53:36,351] Trial 0 finished with value: 26.89373407131288 and parameters: {'lr': 0.00018951382914416393}. Best is trial 0 with value: 26.89373407131288.
+
    +
    +
    Epoch 5, Loss: 30.0107
+Epoch 5, Validation Loss: 26.8937, Metrics: {'rmse': 34.96109612817986, 'cmapss_score': 171362.21049284632}
+trial step with lr = 0.0007874022874638446
+
    +
    +
    Creating sequence of samples: 100%|██████████| 100/100 [00:00<00:00, 12542.02it/s]
+Epochs ...:   0%|          | 0/5 [00:00<?, ?it/s]
+Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  17%|█▋        | 39/232 [00:00<00:00, 389.34it/s]
+Steps ...:  48%|████▊     | 111/232 [00:00<00:00, 581.84it/s]
+Steps ...:  81%|████████  | 187/232 [00:00<00:00, 661.02it/s]
+Epochs ...:  20%|██        | 1/5 [00:00<00:01,  2.48it/s]    
+
    +
    +
    Epoch 1, Loss: 32.6104
+Epoch 1, Validation Loss: 28.4569, Metrics: {'rmse': 36.641393207296055, 'cmapss_score': 206445.81209835963}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  33%|███▎      | 76/232 [00:00<00:00, 755.00it/s]
+Steps ...:  66%|██████▌   | 153/232 [00:00<00:00, 760.84it/s]
+Steps ...:  99%|█████████▉| 230/232 [00:00<00:00, 752.50it/s]
+Epochs ...:  40%|████      | 2/5 [00:00<00:01,  2.69it/s]    
+
    +
    +
    Epoch 2, Loss: 29.5834
+Epoch 2, Validation Loss: 22.3682, Metrics: {'rmse': 29.68363304759782, 'cmapss_score': 98649.45005642212}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  33%|███▎      | 77/232 [00:00<00:00, 764.93it/s]
+Steps ...:  66%|██████▋   | 154/232 [00:00<00:00, 498.98it/s]
+Steps ...:  91%|█████████ | 210/232 [00:00<00:00, 435.48it/s]
+Epochs ...:  60%|██████    | 3/5 [00:01<00:00,  2.10it/s]    
+
    +
    +
    Epoch 3, Loss: 17.5193
+Epoch 3, Validation Loss: 17.9444, Metrics: {'rmse': 24.369740561776197, 'cmapss_score': 82673.16229614464}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  16%|█▌        | 37/232 [00:00<00:00, 363.96it/s]
+Steps ...:  32%|███▏      | 74/232 [00:00<00:00, 359.81it/s]
+Steps ...:  47%|████▋     | 110/232 [00:00<00:00, 355.65it/s]
+Steps ...:  63%|██████▎   | 146/232 [00:00<00:00, 352.98it/s]
+Steps ...:  79%|███████▉  | 183/232 [00:00<00:00, 357.91it/s]
+Steps ...:  94%|█████████▍| 219/232 [00:00<00:00, 355.44it/s]
+Epochs ...:  80%|████████  | 4/5 [00:02<00:00,  1.72it/s]    
+
    +
    +
    Epoch 4, Loss: 16.3577
+Epoch 4, Validation Loss: 16.1891, Metrics: {'rmse': 22.407311130072234, 'cmapss_score': 90375.15431472883}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  16%|█▌        | 36/232 [00:00<00:00, 357.63it/s]
+Steps ...:  31%|███       | 72/232 [00:00<00:00, 343.62it/s]
+Steps ...:  47%|████▋     | 108/232 [00:00<00:00, 346.81it/s]
+Steps ...:  63%|██████▎   | 146/232 [00:00<00:00, 356.10it/s]
+Steps ...:  79%|███████▉  | 183/232 [00:00<00:00, 360.17it/s]
+Steps ...:  95%|█████████▍| 220/232 [00:00<00:00, 358.45it/s]
+Epochs ...: 100%|██████████| 5/5 [00:02<00:00,  1.76it/s]    
+[I 2024-08-13 09:53:39,216] Trial 1 finished with value: 16.189056094099836 and parameters: {'lr': 0.0007874022874638446}. Best is trial 1 with value: 16.189056094099836.
+
    +
    +
    Epoch 5, Loss: 18.7968
+Epoch 5, Validation Loss: 16.4563, Metrics: {'rmse': 22.47353109963666, 'cmapss_score': 73324.81707642713}
+trial step with lr = 0.00012086132027556038
+
    +
    +
    Creating sequence of samples: 100%|██████████| 100/100 [00:00<00:00, 12541.65it/s]
+Epochs ...:   0%|          | 0/5 [00:00<?, ?it/s]
+Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  15%|█▍        | 34/232 [00:00<00:00, 337.76it/s]
+Steps ...:  31%|███       | 71/232 [00:00<00:00, 351.34it/s]
+Steps ...:  46%|████▌     | 107/232 [00:00<00:00, 354.90it/s]
+Steps ...:  62%|██████▏   | 143/232 [00:00<00:00, 356.58it/s]
+Steps ...:  78%|███████▊  | 180/232 [00:00<00:00, 357.98it/s]
+Steps ...:  94%|█████████▎| 217/232 [00:00<00:00, 358.82it/s]
+Epochs ...:  20%|██        | 1/5 [00:00<00:02,  1.35it/s]    
+
    +
    +
    Epoch 1, Loss: 42.6350
+Epoch 1, Validation Loss: 41.5396, Metrics: {'rmse': 53.309621676839306, 'cmapss_score': 1652572.5547515503}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  16%|█▌        | 36/232 [00:00<00:00, 355.25it/s]
+Steps ...:  31%|███       | 72/232 [00:00<00:00, 356.65it/s]
+Steps ...:  47%|████▋     | 108/232 [00:00<00:00, 357.09it/s]
+Steps ...:  62%|██████▏   | 144/232 [00:00<00:00, 355.91it/s]
+Steps ...:  78%|███████▊  | 181/232 [00:00<00:00, 360.10it/s]
+Steps ...:  94%|█████████▍| 218/232 [00:00<00:00, 359.01it/s]
+Epochs ...:  40%|████      | 2/5 [00:01<00:02,  1.35it/s]    
+
    +
    +
    Epoch 2, Loss: 36.4397
+Epoch 2, Validation Loss: 37.1115, Metrics: {'rmse': 46.95681782704423, 'cmapss_score': 653010.5664245718}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  16%|█▌        | 36/232 [00:00<00:00, 351.65it/s]
+Steps ...:  31%|███       | 72/232 [00:00<00:00, 353.10it/s]
+Steps ...:  47%|████▋     | 109/232 [00:00<00:00, 356.46it/s]
+Steps ...:  62%|██████▎   | 145/232 [00:00<00:00, 356.14it/s]
+Steps ...:  78%|███████▊  | 181/232 [00:00<00:00, 355.91it/s]
+Steps ...:  94%|█████████▎| 217/232 [00:00<00:00, 354.12it/s]
+Epochs ...:  60%|██████    | 3/5 [00:02<00:01,  1.34it/s]    
+
    +
    +
    Epoch 3, Loss: 25.7715
+Epoch 3, Validation Loss: 32.9065, Metrics: {'rmse': 41.430416515203575, 'cmapss_score': 349818.50039921864}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  16%|█▌        | 37/232 [00:00<00:00, 360.43it/s]
+Steps ...:  32%|███▏      | 74/232 [00:00<00:00, 359.15it/s]
+Steps ...:  48%|████▊     | 111/232 [00:00<00:00, 359.73it/s]
+Steps ...:  64%|██████▍   | 148/232 [00:00<00:00, 360.01it/s]
+Steps ...:  80%|███████▉  | 185/232 [00:00<00:00, 359.48it/s]
+Steps ...:  95%|█████████▌| 221/232 [00:00<00:00, 356.30it/s]
+Epochs ...:  80%|████████  | 4/5 [00:02<00:00,  1.35it/s]    
+
    +
    +
    Epoch 4, Loss: 29.5199
+Epoch 4, Validation Loss: 30.6848, Metrics: {'rmse': 39.035223216989905, 'cmapss_score': 274871.78833234054}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  16%|█▌        | 36/232 [00:00<00:00, 350.68it/s]
+Steps ...:  31%|███       | 72/232 [00:00<00:00, 352.70it/s]
+Steps ...:  47%|████▋     | 108/232 [00:00<00:00, 348.67it/s]
+Steps ...:  62%|██████▏   | 143/232 [00:00<00:00, 332.92it/s]
+Steps ...:  77%|███████▋  | 178/232 [00:00<00:00, 338.04it/s]
+Steps ...:  92%|█████████▏| 214/232 [00:00<00:00, 343.43it/s]
+Epochs ...: 100%|██████████| 5/5 [00:03<00:00,  1.34it/s]    
+[I 2024-08-13 09:53:42,980] Trial 2 finished with value: 29.60851450664241 and parameters: {'lr': 0.00012086132027556038}. Best is trial 1 with value: 16.189056094099836.
+
    +
    +
    Epoch 5, Loss: 33.5629
+Epoch 5, Validation Loss: 29.6085, Metrics: {'rmse': 37.93360429205696, 'cmapss_score': 242353.786303807}
+Best hyperparameters: {'lr': 0.0007874022874638446}
+Best trial: FrozenTrial(number=1, state=1, values=[16.189056094099836], datetime_start=datetime.datetime(2024, 8, 13, 9, 53, 36, 351082), datetime_complete=datetime.datetime(2024, 8, 13, 9, 53, 39, 216093), params={'lr': 0.0007874022874638446}, user_attrs={}, system_attrs={}, intermediate_values={1: 28.456903271558808, 2: 22.368176297443668, 3: 17.944359942180355, 4: 16.189056094099836, 5: 16.456260332247105}, distributions={'lr': FloatDistribution(high=0.001, log=False, low=5e-05, step=None)}, trial_id=1, value=None)
+
    +
    +
    +
    +

    Optimization with changing the complexity of the training process. Tune the MLP hidden dimension size using MSE metric as optimization target

    +
    +
    +
    # model_class.optimize(data, target, optimize_parameter, optimize_range, direction, n_trials, epochs, optimize_metric)
+model.optimize(data, target, optimize_parameter="hidden_dim", optimize_range=(256, 1024), direction="minimize", optimize_metric="rmse", n_trials=3, epochs=5)
+
    +
    +
    +
    +
    [I 2024-08-13 09:53:42,992] A new study created in memory with name: /parameter_hidden_dim study
+
    +
    +
    trial step with hidden_dim = 702
+
    +
    +
    Creating sequence of samples: 100%|██████████| 100/100 [00:00<00:00, 11148.24it/s]
+Epochs ...:   0%|          | 0/5 [00:00<?, ?it/s]
+Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  14%|█▍        | 32/232 [00:00<00:00, 315.23it/s]
+Steps ...:  29%|██▉       | 67/232 [00:00<00:00, 333.10it/s]
+Steps ...:  44%|████▍     | 102/232 [00:00<00:00, 338.76it/s]
+Steps ...:  59%|█████▉    | 137/232 [00:00<00:00, 339.63it/s]
+Steps ...:  74%|███████▍  | 172/232 [00:00<00:00, 340.10it/s]
+Steps ...:  89%|████████▉ | 207/232 [00:00<00:00, 341.73it/s]
+Epochs ...:  20%|██        | 1/5 [00:00<00:03,  1.28it/s]    
+
    +
    +
    Epoch 1, Loss: 41.7009
+Epoch 1, Validation Loss: 41.0446, Metrics: {'rmse': 52.46324288978529, 'cmapss_score': 1459642.2102222845}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  16%|█▌        | 36/232 [00:00<00:00, 350.68it/s]
+Steps ...:  31%|███       | 72/232 [00:00<00:00, 354.74it/s]
+Steps ...:  47%|████▋     | 108/232 [00:00<00:00, 354.46it/s]
+Steps ...:  62%|██████▏   | 144/232 [00:00<00:00, 354.32it/s]
+Steps ...:  78%|███████▊  | 181/232 [00:00<00:00, 357.14it/s]
+Steps ...:  94%|█████████▎| 217/232 [00:00<00:00, 354.29it/s]
+Epochs ...:  40%|████      | 2/5 [00:01<00:02,  1.31it/s]    
+
    +
    +
    Epoch 2, Loss: 34.5571
+Epoch 2, Validation Loss: 35.5294, Metrics: {'rmse': 44.69688917194367, 'cmapss_score': 487095.93532787054}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  14%|█▍        | 33/232 [00:00<00:00, 328.99it/s]
+Steps ...:  29%|██▉       | 67/232 [00:00<00:00, 335.00it/s]
+Steps ...:  44%|████▍     | 103/232 [00:00<00:00, 342.81it/s]
+Steps ...:  59%|█████▉    | 138/232 [00:00<00:00, 342.07it/s]
+Steps ...:  75%|███████▍  | 173/232 [00:00<00:00, 340.46it/s]
+Steps ...:  90%|████████▉ | 208/232 [00:00<00:00, 342.89it/s]
+Epochs ...:  60%|██████    | 3/5 [00:02<00:01,  1.30it/s]    
+
    +
    +
    Epoch 3, Loss: 24.0595
+Epoch 3, Validation Loss: 31.7725, Metrics: {'rmse': 40.124605880566875, 'cmapss_score': 305081.3116466362}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  15%|█▌        | 35/232 [00:00<00:00, 344.29it/s]
+Steps ...:  30%|███       | 70/232 [00:00<00:00, 344.29it/s]
+Steps ...:  46%|████▌     | 106/232 [00:00<00:00, 347.96it/s]
+Steps ...:  61%|██████    | 142/232 [00:00<00:00, 350.98it/s]
+Steps ...:  77%|███████▋  | 178/232 [00:00<00:00, 350.24it/s]
+Steps ...:  92%|█████████▏| 214/232 [00:00<00:00, 349.78it/s]
+Epochs ...:  80%|████████  | 4/5 [00:03<00:00,  1.30it/s]    
+
    +
    +
    Epoch 4, Loss: 27.7565
+Epoch 4, Validation Loss: 29.9569, Metrics: {'rmse': 38.28838337636238, 'cmapss_score': 252440.2810893617}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  14%|█▍        | 32/232 [00:00<00:00, 314.78it/s]
+Steps ...:  28%|██▊       | 66/232 [00:00<00:00, 324.48it/s]
+Steps ...:  44%|████▍     | 102/232 [00:00<00:00, 337.97it/s]
+Steps ...:  59%|█████▊    | 136/232 [00:00<00:00, 338.51it/s]
+Steps ...:  73%|███████▎  | 170/232 [00:00<00:00, 330.41it/s]
+Steps ...:  88%|████████▊ | 204/232 [00:00<00:00, 324.28it/s]
+Epochs ...: 100%|██████████| 5/5 [00:03<00:00,  1.29it/s]    
+[I 2024-08-13 09:53:46,884] Trial 0 finished with value: 37.12774637917532 and parameters: {'hidden_dim': 702}. Best is trial 0 with value: 37.12774637917532.
+
    +
    +
    Epoch 5, Loss: 31.4343
+Epoch 5, Validation Loss: 28.8321, Metrics: {'rmse': 37.12774637917532, 'cmapss_score': 219477.3266793877}
+trial step with hidden_dim = 662
+
    +
    +
    Creating sequence of samples: 100%|██████████| 100/100 [00:00<00:00, 12542.40it/s]
+Epochs ...:   0%|          | 0/5 [00:00<?, ?it/s]
+Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  15%|█▍        | 34/232 [00:00<00:00, 334.45it/s]
+Steps ...:  30%|██▉       | 69/232 [00:00<00:00, 342.21it/s]
+Steps ...:  45%|████▍     | 104/232 [00:00<00:00, 337.83it/s]
+Steps ...:  60%|█████▉    | 139/232 [00:00<00:00, 340.97it/s]
+Steps ...:  75%|███████▌  | 174/232 [00:00<00:00, 342.74it/s]
+Steps ...:  90%|█████████ | 209/232 [00:00<00:00, 344.41it/s]
+Epochs ...:  20%|██        | 1/5 [00:00<00:03,  1.26it/s]    
+
    +
    +
    Epoch 1, Loss: 41.5648
+Epoch 1, Validation Loss: 41.1247, Metrics: {'rmse': 52.64622625632633, 'cmapss_score': 1497278.352392366}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  14%|█▍        | 32/232 [00:00<00:00, 319.07it/s]
+Steps ...:  28%|██▊       | 66/232 [00:00<00:00, 330.08it/s]
+Steps ...:  44%|████▎     | 101/232 [00:00<00:00, 338.13it/s]
+Steps ...:  59%|█████▊    | 136/232 [00:00<00:00, 341.91it/s]
+Steps ...:  74%|███████▍  | 172/232 [00:00<00:00, 346.34it/s]
+Steps ...:  89%|████████▉ | 207/232 [00:00<00:00, 346.80it/s]
+Epochs ...:  40%|████      | 2/5 [00:01<00:02,  1.27it/s]    
+
    +
    +
    Epoch 2, Loss: 37.5466
+Epoch 2, Validation Loss: 35.6801, Metrics: {'rmse': 45.00773698916855, 'cmapss_score': 511059.8643841665}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  14%|█▍        | 33/232 [00:00<00:00, 321.46it/s]
+Steps ...:  28%|██▊       | 66/232 [00:00<00:00, 320.47it/s]
+Steps ...:  43%|████▎     | 99/232 [00:00<00:00, 315.95it/s]
+Steps ...:  57%|█████▋    | 132/232 [00:00<00:00, 318.65it/s]
+Steps ...:  71%|███████   | 165/232 [00:00<00:00, 321.90it/s]
+Steps ...:  86%|████████▌ | 200/232 [00:00<00:00, 330.28it/s]
+Epochs ...:  60%|██████    | 3/5 [00:02<00:01,  1.25it/s]    
+
    +
    +
    Epoch 3, Loss: 26.5062
+Epoch 3, Validation Loss: 31.7830, Metrics: {'rmse': 40.07700663323968, 'cmapss_score': 305080.57834935177}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  14%|█▍        | 32/232 [00:00<00:00, 318.04it/s]
+Steps ...:  28%|██▊       | 65/232 [00:00<00:00, 323.80it/s]
+Steps ...:  44%|████▎     | 101/232 [00:00<00:00, 337.28it/s]
+Steps ...:  59%|█████▊    | 136/232 [00:00<00:00, 341.38it/s]
+Steps ...:  74%|███████▎  | 171/232 [00:00<00:00, 342.43it/s]
+Steps ...:  89%|████████▉ | 207/232 [00:00<00:00, 345.26it/s]
+Epochs ...:  80%|████████  | 4/5 [00:03<00:00,  1.26it/s]    
+
    +
    +
    Epoch 4, Loss: 28.9049
+Epoch 4, Validation Loss: 29.9922, Metrics: {'rmse': 38.29525364434387, 'cmapss_score': 252989.93131717102}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  15%|█▌        | 35/232 [00:00<00:00, 345.28it/s]
+Steps ...:  31%|███       | 71/232 [00:00<00:00, 350.49it/s]
+Steps ...:  46%|████▌     | 107/232 [00:00<00:00, 352.15it/s]
+Steps ...:  62%|██████▏   | 143/232 [00:00<00:00, 349.53it/s]
+Steps ...:  77%|███████▋  | 179/232 [00:00<00:00, 351.74it/s]
+Steps ...:  93%|█████████▎| 215/232 [00:00<00:00, 348.30it/s]
+Epochs ...: 100%|██████████| 5/5 [00:03<00:00,  1.27it/s]    
+[I 2024-08-13 09:53:50,841] Trial 1 finished with value: 37.207252080735245 and parameters: {'hidden_dim': 662}. Best is trial 0 with value: 37.12774637917532.
+
    +
    +
    Epoch 5, Loss: 31.7666
+Epoch 5, Validation Loss: 28.9132, Metrics: {'rmse': 37.207252080735245, 'cmapss_score': 221679.17640028123}
+trial step with hidden_dim = 879
+
    +
    +
    Creating sequence of samples: 100%|██████████| 100/100 [00:00<00:00, 11148.54it/s]
+Epochs ...:   0%|          | 0/5 [00:00<?, ?it/s]
+Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  15%|█▌        | 35/232 [00:00<00:00, 340.94it/s]
+Steps ...:  31%|███       | 71/232 [00:00<00:00, 348.66it/s]
+Steps ...:  46%|████▌     | 107/232 [00:00<00:00, 351.77it/s]
+Steps ...:  62%|██████▏   | 143/232 [00:00<00:00, 354.07it/s]
+Steps ...:  77%|███████▋  | 179/232 [00:00<00:00, 351.60it/s]
+Steps ...:  93%|█████████▎| 215/232 [00:00<00:00, 353.62it/s]
+Epochs ...:  20%|██        | 1/5 [00:00<00:03,  1.32it/s]    
+
    +
    +
    Epoch 1, Loss: 40.2073
+Epoch 1, Validation Loss: 40.6210, Metrics: {'rmse': 51.9431875404865, 'cmapss_score': 1331357.9658768093}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  16%|█▌        | 36/232 [00:00<00:00, 350.68it/s]
+Steps ...:  31%|███       | 72/232 [00:00<00:00, 348.69it/s]
+Steps ...:  47%|████▋     | 108/232 [00:00<00:00, 349.60it/s]
+Steps ...:  62%|██████▎   | 145/232 [00:00<00:00, 354.45it/s]
+Steps ...:  78%|███████▊  | 181/232 [00:00<00:00, 353.68it/s]
+Steps ...:  94%|█████████▎| 217/232 [00:00<00:00, 352.66it/s]
+Epochs ...:  40%|████      | 2/5 [00:01<00:02,  1.32it/s]    
+
    +
    +
    Epoch 2, Loss: 37.4625
+Epoch 2, Validation Loss: 34.5596, Metrics: {'rmse': 43.41147259207038, 'cmapss_score': 421283.66798380984}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  16%|█▌        | 36/232 [00:00<00:00, 358.52it/s]
+Steps ...:  31%|███       | 72/232 [00:00<00:00, 356.73it/s]
+Steps ...:  47%|████▋     | 109/232 [00:00<00:00, 358.43it/s]
+Steps ...:  62%|██████▎   | 145/232 [00:00<00:00, 355.35it/s]
+Steps ...:  78%|███████▊  | 182/232 [00:00<00:00, 358.45it/s]
+Steps ...:  94%|█████████▍| 218/232 [00:00<00:00, 356.98it/s]
+Epochs ...:  60%|██████    | 3/5 [00:02<00:01,  1.33it/s]    
+
    +
    +
    Epoch 3, Loss: 23.8398
+Epoch 3, Validation Loss: 30.9023, Metrics: {'rmse': 39.31126980464057, 'cmapss_score': 286766.446915035}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  16%|█▌        | 36/232 [00:00<00:00, 357.63it/s]
+Steps ...:  31%|███       | 72/232 [00:00<00:00, 357.63it/s]
+Steps ...:  47%|████▋     | 109/232 [00:00<00:00, 358.92it/s]
+Steps ...:  62%|██████▎   | 145/232 [00:00<00:00, 358.41it/s]
+Steps ...:  78%|███████▊  | 182/232 [00:00<00:00, 359.14it/s]
+Steps ...:  94%|█████████▍| 218/232 [00:00<00:00, 359.17it/s]
+Epochs ...:  80%|████████  | 4/5 [00:02<00:00,  1.34it/s]    
+
    +
    +
    Epoch 4, Loss: 28.2913
+Epoch 4, Validation Loss: 29.3355, Metrics: {'rmse': 37.68203820452062, 'cmapss_score': 236872.37425031906}
+
    +
    +
    Steps ...:   0%|          | 0/232 [00:00<?, ?it/s]
+Steps ...:  15%|█▌        | 35/232 [00:00<00:00, 347.69it/s]
+Steps ...:  30%|███       | 70/232 [00:00<00:00, 347.69it/s]
+Steps ...:  45%|████▌     | 105/232 [00:00<00:00, 345.31it/s]
+Steps ...:  60%|██████    | 140/232 [00:00<00:00, 345.55it/s]
+Steps ...:  76%|███████▌  | 176/232 [00:00<00:00, 348.00it/s]
+Steps ...:  91%|█████████▏| 212/232 [00:00<00:00, 351.78it/s]
+Epochs ...: 100%|██████████| 5/5 [00:03<00:00,  1.33it/s]    
+[I 2024-08-13 09:53:54,617] Trial 2 finished with value: 36.46475764089671 and parameters: {'hidden_dim': 879}. Best is trial 2 with value: 36.46475764089671.
+
    +
    +
    Epoch 5, Loss: 31.4287
+Epoch 5, Validation Loss: 28.2309, Metrics: {'rmse': 36.46475764089671, 'cmapss_score': 203393.37691007284}
+Best hyperparameters: {'hidden_dim': 879}
+Best trial: FrozenTrial(number=2, state=1, values=[36.46475764089671], datetime_start=datetime.datetime(2024, 8, 13, 9, 53, 50, 842965), datetime_complete=datetime.datetime(2024, 8, 13, 9, 53, 54, 617905), params={'hidden_dim': 879}, user_attrs={}, system_attrs={}, intermediate_values={1: 40.62102061946218, 2: 34.55964553646925, 3: 30.902342726544635, 4: 29.335512719503264, 5: 28.230895298283276}, distributions={'hidden_dim': IntDistribution(high=1024, log=False, low=256, step=1)}, trial_id=2, value=None)
+
    +
    +
    +
    +

    The best results are printed at the end of optimization processand saved in the outputs/task_name/traininig/param_name_optimization folder

    +
    + + +
    + + + + + + + +
    + + + +
    + +
    +
    + + Show Source + +
    +
    + +
    + + +
    +
    + +
    + +
    +
    +
    + + + + + +
    +
    + +
    + +
    +

    + + © Copyright 2024, ICE Team. +
    + +

    +
    + +
    +

    + Created using Sphinx 5.3.0. +
    +

    +
    + +
    + + + +
    + +

    + Built with the PyData Sphinx Theme 0.14.0. +

    + +
    + +
    + +
    + + \ No newline at end of file diff --git a/docs/benchmark/ad_benchmark_autoencodermlp_256.html b/docs/benchmark/ad_benchmark_autoencodermlp_256.html index fe0527a..3ec2291 100644 --- a/docs/benchmark/ad_benchmark_autoencodermlp_256.html +++ b/docs/benchmark/ad_benchmark_autoencodermlp_256.html @@ -50,7 +50,7 @@ - + @@ -477,12 +477,12 @@

    Results of anomaly detection using AutoEncoderMLP-256

    previous

    -

    Results of fault diagnosis using MLP-256

    +

    <no title>

     - Datasets — ICE documentation + Results of anomaly detection using GSL-GNN — ICE documentation @@ -46,11 +46,11 @@ - + - - + + @@ -147,8 +147,8 @@

    @@ -396,8 +393,79 @@
    -
    -

    Datasets#

    +
    +

    Results of anomaly detection using GSL-GNN#

    +
    +
    +
    from ice.anomaly_detection.datasets import AnomalyDetectionRiethTEP
+from ice.anomaly_detection.models import GSL_GNN
+from sklearn.preprocessing import StandardScaler
+import numpy as np
+import pandas as pd
+
    +
    +
    +
    +

    Download the dataset.

    +
    +
    +
    dataset = AnomalyDetectionRiethTEP()
+
    +
    +
    +
    +
    +
    +

    Normalize the data.

    +
    +
    +
    scaler = StandardScaler()
+dataset.df[dataset.train_mask] = scaler.fit_transform(dataset.df[dataset.train_mask])
+dataset.df[dataset.test_mask] = scaler.transform(dataset.df[dataset.test_mask])
+
    +
    +
    +
    +

    Create the GNN model.

    +
    +
    +
    model = GSL_GNN(
+    window_size=32, 
+    num_epochs=30, 
+    device='cuda',
+    verbose=True,
+    val_ratio=0.1,
+    save_checkpoints=True,
+    threshold_level=0.98
+    )
+
    +
    +
    +
    +

    Load the checkpoint.

    +
    +
    +
    model.load_checkpoint('gnn_anomaly_detection_epoch_30.tar')
+
    +
    +
    +
    +

    Evaluate the model on the test data.

    +
    +
    +
    metrics = model.evaluate(dataset.df[dataset.test_mask], dataset.target[dataset.test_mask])
+metrics
+
    +
    +
    +
    +
    {'accuracy': 0.8517518472906404,
+ 'true_positive_rate': [0.82365725],
+ 'false_positive_rate': [0.019373853211009175]}
+
    +
    +
    +
    @@ -411,20 +479,20 @@

    previous

    -

    Installation

    +

    Results of anomaly detection using STGAT-MAD

    next

    -

    Tasks

    +

    Results of RUL estimation using lstm-256

     @@ -439,7 +507,7 @@
    - + Show Source
    diff --git a/docs/reference/modules.html b/docs/benchmark/ad_benchmark_stgat.html similarity index 57% rename from docs/reference/modules.html rename to docs/benchmark/ad_benchmark_stgat.html index 8c9e7c0..3f91f99 100644 --- a/docs/reference/modules.html +++ b/docs/benchmark/ad_benchmark_stgat.html @@ -9,7 +9,7 @@ - ice — ICE documentation + Results of anomaly detection using STGAT-MAD — ICE documentation @@ -46,9 +46,11 @@ - + + + @@ -160,7 +162,7 @@
  • - + API reference
    • @@ -173,8 +175,8 @@ -
  • - +
  • + Benchmarking
    • @@ -242,7 +244,7 @@
    -
    +
    @@ -275,7 +277,7 @@
  • - + API reference
    • @@ -288,8 +290,8 @@ -
  • - +
  • + Benchmarking
    • @@ -318,6 +320,27 @@
    +
    +
    +
    +
    @@ -350,7 +373,10 @@ -
  • ice
    • + +
  • Benchmarking
    • + +
  • Results of anomaly detection using STGAT-MAD

@@ -367,32 +393,78 @@
-
-

ice#

-
- +
+

Results of anomaly detection using STGAT-MAD#

+
+
+
from ice.anomaly_detection.datasets import AnomalyDetectionRiethTEP
+from ice.anomaly_detection.models import STGAT_MAD
+from sklearn.preprocessing import StandardScaler
+import numpy as np
+import pandas as pd
+
+
+
+
+

Download the dataset.

+
+
+
dataset = AnomalyDetectionRiethTEP()
+
+
+
+
+
+
+

Normalize the data.

+
+
+
scaler = StandardScaler()
+dataset.df[dataset.train_mask] = scaler.fit_transform(dataset.df[dataset.train_mask])
+dataset.df[dataset.test_mask] = scaler.transform(dataset.df[dataset.test_mask])
+
+
+
+
+

Create the GNN model.

+
+
+
model = STGAT_MAD(
+    window_size=32, 
+    num_epochs=30, 
+    device='cuda',
+    verbose=True,
+    val_ratio=0.1,
+    save_checkpoints=True,
+    threshold_level=0.98
+    )
+
+
+
+
+

Load the checkpoint.

+
+
+
model.load_checkpoint('stgat_anomaly_detection_epoch_30.tar')
+
+
+
+
+

Evaluate the model on the test data.

+
+
+
metrics = model.evaluate(dataset.df[dataset.test_mask], dataset.target[dataset.test_mask])
+metrics
+
+
+
+
+
{'accuracy': 0.861279659277504,
+ 'true_positive_rate': [0.8352755],
+ 'false_positive_rate': [0.019435206422018347]}
+
+
+
@@ -406,6 +478,24 @@

ice# @@ -417,7 +507,7 @@

ice#
- + Show Source
diff --git a/docs/benchmark/ad_benchmark_autorncodermlp_256.html b/docs/benchmark/ad_benchmark_transformer.html similarity index 60% rename from docs/benchmark/ad_benchmark_autorncodermlp_256.html rename to docs/benchmark/ad_benchmark_transformer.html index 010e475..44c3f47 100644 --- a/docs/benchmark/ad_benchmark_autorncodermlp_256.html +++ b/docs/benchmark/ad_benchmark_transformer.html @@ -9,7 +9,7 @@ - Results of anomaly detection using AutoEncoderMLP-256 — ICE documentation + Results of anomaly detection using AnomalyTransformer — ICE documentation @@ -46,10 +46,11 @@ - + - + + @@ -161,7 +162,7 @@
  • - + API reference
    • @@ -276,7 +277,7 @@
  • - + API reference
    • @@ -326,7 +327,15 @@

    Contents:

    @@ -367,7 +376,7 @@
  • Benchmarking
    • -
  • Results of anomaly detection using AutoEncoderMLP-256
    • +
  • Results of anomaly detection using AnomalyTransformer

    • @@ -384,79 +393,80 @@
    -
    -

    Results of anomaly detection using AutoEncoderMLP-256#

    -

    This notebook presents experimental results of anomaly detection on the Tennessee Eastman Process dataset using the model AutoEncoderMLP-256.

    -

    Importing libraries.

    +
    +

    Results of anomaly detection using AnomalyTransformer#

    -
    import numpy as np
-import torch
+
    from ice.anomaly_detection.datasets import AnomalyDetectionRiethTEP
+from ice.anomaly_detection.models import AnomalyTransformer
 from sklearn.preprocessing import StandardScaler
-from tqdm.auto import trange
-from ice.anomaly_detection.datasets import AnomalyDetectionReinartzTEP, AnomalyDetectionSmallTEP
-from ice.anomaly_detection.models import AutoEncoderMLP
+import numpy as np
+import pandas as pd
    -

    Downloading the TEP dataset.

    +

    Download the dataset.

    -
    dataset = AnomalyDetectionReinartzTEP()
+
    dataset = AnomalyDetectionRiethTEP()
    -
    +
    -

    Training the model and calculation of metrics.

    +

    Normalize the data.

    -
    metrics = []
-for i in trange(5):
-    torch.random.manual_seed(i)
-    model = AutoEncoderMLP(
-        window_size=32,
-        batch_size=512,
-        lr=0.001,
-        num_epochs=20,
-        verbose=False,
-        device='cuda'
+
    scaler = StandardScaler()
+dataset.df[dataset.train_mask] = scaler.fit_transform(dataset.df[dataset.train_mask])
+dataset.df[dataset.test_mask] = scaler.transform(dataset.df[dataset.test_mask])
+
    
+
    
+
    +
    +

    Create the AnomalyTransformer model.

    +
    +
    +
    model = AnomalyTransformer(
+    window_size=32, 
+    lr=0.001, 
+    num_epochs=30, 
+    device='cuda', 
+    verbose=True, 
+    val_ratio=0.1,
+    save_checkpoints=True,
+    threshold_level=0.98,
+    d_model=32, 
+    e_layers=1,
+    d_ff=32, 
+    dropout=0.0
     )
-    model.fit(
-        dataset.df[dataset.train_mask])
-    metrics.append(
-        model.evaluate(
-            dataset.df[dataset.test_mask], dataset.target[dataset.test_mask]))
    -
    -
    -

    Printing metrics.

    +

    Load the checkpoint.

    +
    +
    +
    model.load_checkpoint('transformer_anomaly_detection_epoch_30.tar')
+
    +
    +
    +
    +

    Evaluate the model on the test data.

    -
    acc = []
-tpr = []
-fpr = []
-for metrics_i in metrics:
-    acc.append(metrics_i["accuracy"])
-    tpr.append(metrics_i["true_positive_rate"])
-    fpr.append(metrics_i["false_positive_rate"])
-tpr_mean, tpr_std = np.mean(tpr, axis=0), np.std(tpr, axis=0)
-fpr_mean, fpr_std = np.mean(fpr, axis=0), np.std(fpr, axis=0)
-
-print(f'Accuracy: {np.mean(acc):.4f} ± {2*np.std(acc):.4f}')
-for i in range(len(tpr_mean)):
-    print(f'TPR/FPR: {tpr_mean[i]:.4f} ± {2*tpr_std[i]:.4f} / {fpr_mean[i]:.4f} ± {2*fpr_std[i]:.4f}')
+
    metrics = model.evaluate(dataset.df[dataset.test_mask], dataset.target[dataset.test_mask])
+metrics
    -
    Accuracy: 0.8028 ± 0.0007
-TPR/FPR: 0.7378 ± 0.0022 / 0.0366 ± 0.0034
+
    {'accuracy': 0.8588669950738916,
+ 'true_positive_rate': [0.830817625],
+ 'false_positive_rate': [0.012466169724770642]}
    @@ -474,13 +484,22 @@

    Results of anomaly detection using AutoEncoderMLP-256

    previous

    -

    Results of fault diagnosis using MLP-256

    +

    Results of anomaly detection using AutoEncoderMLP-256

    +
    +     + +
    +

    next

    +

    Results of anomaly detection using STGAT-MAD

    +
    @@ -493,7 +512,7 @@

    Results of anomaly detection using AutoEncoderMLP-256
    - + Show Source
    diff --git a/docs/benchmark/fd_benchmark_mlp_256.html b/docs/benchmark/fd_benchmark_mlp_256.html index 0fe9d50..5df2dd1 100644 --- a/docs/benchmark/fd_benchmark_mlp_256.html +++ b/docs/benchmark/fd_benchmark_mlp_256.html @@ -49,7 +49,7 @@ - + @@ -528,11 +528,11 @@

    Results of fault diagnosis using MLP-256

    next

    -

    Results of anomaly detection using AutoEncoderMLP-256

    +

    <no title>

     diff --git a/docs/benchmark/fd_benchmark_tcn.html b/docs/benchmark/fd_benchmark_tcn.html new file mode 100644 index 0000000..0c2cc60 --- /dev/null +++ b/docs/benchmark/fd_benchmark_tcn.html @@ -0,0 +1,753 @@ + + + + + + + + + + + + Results of fault diagnosis using TCN — ICE documentation + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
    + + + + + + + + + + + +
    +
    +
    +
    + + + Ctrl+K +
    +
    + + + +
    +
    + +
    + + + +
    + + +
    + +
    + +
    + + + +
    + +
    +
    + +
    + +
    + +
    +
    +
    + + +
    +
    + + + + +
    + +
    + + +
    +
    + +
    +
    + +
    + +
    + + + + +
    + +
    + + +
    +
    + + + + + +
    + +
    +

    Results of fault diagnosis using TCN#

    +
    +
    +
    from ice.fault_diagnosis.datasets import FaultDiagnosisRiethTEP
+from ice.fault_diagnosis.models import TCN
+from sklearn.preprocessing import StandardScaler
+import numpy as np
+import pandas as pd
+
    +
    +
    +
    +

    Download the dataset.

    +
    +
    +
    dataset = FaultDiagnosisRiethTEP()
+
    +
    +
    +
    +
    +
    +

    Normalize the data.

    +
    +
    +
    scaler = StandardScaler()
+dataset.df[dataset.train_mask] = scaler.fit_transform(dataset.df[dataset.train_mask])
+dataset.df[dataset.test_mask] = scaler.transform(dataset.df[dataset.test_mask])
+
    +
    +
    +
    +

    Create the TCN model.

    +
    +
    +
    model = TCN(
+    window_size=60,
+    batch_size=128,
+    num_layers=1,
+    kernel_size=3,
+    hidden_dim=32,
+    lr=1e-4,
+    num_epochs=30,
+    verbose=True,
+    device='cpu',
+    save_checkpoints=True,
+    val_ratio=0.1,
+)
+
    +
    +
    +
    +

    Load the checkpoint.

    +
    +
    +
    model.load_checkpoint('tcn_fault_diagnosis_epoch_30.tar')
+
    +
    +
    +
    +

    Evaluate the model on the test data.

    +
    +
    +
    metrics = model.evaluate(
+    dataset.df[dataset.test_mask],
+    dataset.target[dataset.test_mask]
+)
+
    +
    +
    +
    +
    +
    +
    +
    +
    idx = np.array([1, 2, 4, 5, 6, 7, 8, 10, 11, 12, 13, 14, 16, 17, 18, 19, 20]) - 1
+pd.DataFrame({
+    'Fault': idx,
+    'TPR': np.array(metrics['true_positive_rate'])[idx],
+    'FPR': np.array(metrics['false_positive_rate'])[idx],
+}).round(4)
+
    +
    +
    +
    +
    + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
    FaultTPRFPR
    000.96750.0000
    110.97380.0000
    230.96430.0000
    340.95840.0000
    450.97310.0000
    560.96790.0000
    670.96910.0000
    790.96510.0000
    8100.97880.0000
    9110.95260.0000
    10120.94180.0001
    11130.97800.0000
    12150.97520.0000
    13160.96080.0000
    14170.93570.0000
    15180.97170.0000
    16190.94820.0000
    +
    +
    +
    +
    +
    print(f'Average TPR: {np.array(metrics["true_positive_rate"])[idx].mean():.2f}')
+
    +
    +
    +
    +
    Average TPR: 0.96
+
    +
    +
    +
    +
    +
    +
    for i in np.array(metrics["true_positive_rate"])[idx]*100:
+    print(f'{i:.2f}')
+
    +
    +
    +
    +
    96.75
+97.38
+96.43
+95.84
+97.31
+96.79
+96.91
+96.51
+97.88
+95.26
+94.18
+97.80
+97.52
+96.08
+93.57
+97.17
+94.82
+
    +
    +
    +
    +
    + + +
    + + + + + + + +
    + + + +
    + +
    +
    + + Show Source + +
    +
    + +
    + + +
    +
    + +
    + +
    +
    +
    + + + + + +
    +
    + +
    + +
    +

    + + © Copyright 2024, ICE Team. +
    + +

    +
    + +
    +

    + Created using Sphinx 5.3.0. +
    +

    +
    + +
    + + + +
    + +

    + Built with the PyData Sphinx Theme 0.14.0. +

    + +
    + +
    + +
    + + \ No newline at end of file diff --git a/docs/benchmark/hi_sota.html b/docs/benchmark/hi_sota.html new file mode 100644 index 0000000..1c9ad6f --- /dev/null +++ b/docs/benchmark/hi_sota.html @@ -0,0 +1,665 @@ + + + + + + + + + + + + Results of HI estimation using Stacked LSTM — ICE documentation + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
    + + + + + + + + + + + +
    +
    +
    +
    + + + Ctrl+K +
    +
    + + + +
    +
    + +
    + + + +
    + + +
    + +
    + +
    + + + +
    + +
    +
    + +
    + +
    + +
    +
    +
    + + +
    +
    + + + + +
    + +
    + + +
    +
    + +
    +
    + +
    + +
    + + + + +
    + +
    + + +
    +
    + + + + + +
    + +
    +

    Results of HI estimation using Stacked LSTM#

    +

    This notebook presents experimental results of hi estimation on the Milling dataset using the model MLP-256.

    +

    Importing libraries.

    +
    +
    +
    from ice.health_index_estimation.datasets import Milling
+from ice.health_index_estimation.models import MLP, TCM, IE_SBiGRU, Stacked_LSTM
+
+import pandas as pd
+import numpy as np
+import torch
+from tqdm.auto import trange
+    
+
    +
    +
    +
    +
    C:\Users\user\conda\envs\ice_testing\Lib\site-packages\tqdm\auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html
+  from .autonotebook import tqdm as notebook_tqdm
+
    +
    +
    +
    +

    Initializing model class and train/test data split

    +
    +
    +
    dataset_class = Milling()
+
+data, target = pd.concat(dataset_class.df), pd.concat(dataset_class.target) 
+test_data, test_target = dataset_class.test[0], dataset_class.test_target[0]
+
    +
    +
    +
    +
    Reading data/milling/case_1.csv: 100%|██████████| 153000/153000 [00:00<00:00, 1268689.48it/s]
+Reading data/milling/case_2.csv: 100%|██████████| 117000/117000 [00:00<00:00, 1248873.41it/s]
+Reading data/milling/case_3.csv: 100%|██████████| 126000/126000 [00:00<00:00, 1276983.81it/s]
+Reading data/milling/case_4.csv: 100%|██████████| 63000/63000 [00:00<00:00, 1374151.83it/s]
+Reading data/milling/case_5.csv: 100%|██████████| 54000/54000 [00:00<00:00, 1290003.79it/s]
+Reading data/milling/case_6.csv: 100%|██████████| 9000/9000 [00:00<00:00, 1003395.34it/s]
+Reading data/milling/case_7.csv: 100%|██████████| 72000/72000 [00:00<00:00, 1245529.71it/s]
+Reading data/milling/case_8.csv: 100%|██████████| 54000/54000 [00:00<00:00, 1321487.68it/s]
+Reading data/milling/case_9.csv: 100%|██████████| 81000/81000 [00:00<00:00, 1377467.66it/s]
+Reading data/milling/case_10.csv: 100%|██████████| 90000/90000 [00:00<00:00, 1347769.63it/s]
+Reading data/milling/case_11.csv: 100%|██████████| 207000/207000 [00:00<00:00, 1180064.87it/s]
+Reading data/milling/case_12.csv: 100%|██████████| 126000/126000 [00:00<00:00, 1276980.73it/s]
+Reading data/milling/case_13.csv: 100%|██████████| 135000/135000 [00:00<00:00, 1242669.46it/s]
+Reading data/milling/case_14.csv: 100%|██████████| 81000/81000 [00:00<00:00, 1310826.20it/s]
+Reading data/milling/case_15.csv: 100%|██████████| 63000/63000 [00:00<00:00, 1316886.37it/s]
+Reading data/milling/case_16.csv: 100%|██████████| 18000/18000 [00:00<00:00, 1128731.62it/s]
+C:\Users\user\conda\envs\ice_testing\Lib\site-packages\scipy\interpolate\_interpolate.py:479: RuntimeWarning: invalid value encountered in divide
+  slope = (y_hi - y_lo) / (x_hi - x_lo)[:, None]
+
    +
    +
    +
    +
    +
    +
    from sklearn.preprocessing import StandardScaler, MinMaxScaler
+import pandas as pd 
+
+scaler = MinMaxScaler()
+trainer_data = scaler.fit_transform(data)
+tester_data = scaler.transform(test_data)
+
+trainer_data = pd.DataFrame(trainer_data, index=data.index, columns=data.columns)
+tester_data = pd.DataFrame(tester_data, index=test_data.index, columns=test_data.columns)
+
    +
    +
    +
    +
    +
    +
    # path_to_tar = "hi_sota/"
+
    +
    +
    +
    +
    +
    +
    model_class = Stacked_LSTM(
+        window_size=64,
+        stride=1024, # 1024
+        batch_size=253, # 256
+        lr= 0.0031789041005068647, # 0.0004999805761074147,
+        num_epochs=55,
+        verbose=True,
+        device='cuda'
+    )
+# model_class.fit(trainer_data, target)
+model_class.load_checkpoint(path_to_tar + "stack_sota.tar")
+
    +
    +
    +
    +
    +
    +
    model_class.evaluate(tester_data, test_target)
+
    +
    +
    +
    +
    Creating sequence of samples: 100%|██████████| 14/14 [00:00<00:00, 2809.31it/s]
+                                                             
+
    +
    +
    {'mse': 0.0022332468596409335, 'rmse': 0.047257241346072384}
+
    +
    +
    +
    +
    +
    +
    model_class = TCM(
+        window_size=64,
+        stride=1024, # 1024
+        batch_size=253, # 256
+        lr= 0.0031789041005068647, # 0.0004999805761074147,
+        num_epochs=55,
+        verbose=True,
+        device='cuda'
+    )
+# model_class.fit(trainer_data, target)
+model_class.load_checkpoint(path_to_tar + "TCM_sota.tar")
+
    +
    +
    +
    +
    +
    +
    model_class.evaluate(tester_data, test_target)
+
    +
    +
    +
    +
    Creating sequence of samples: 100%|██████████| 14/14 [00:00<00:00, 3511.98it/s]
+                                                             
+
    +
    +
    {'mse': 0.004014168163365719, 'rmse': 0.06335746335962102}
+
    +
    +
    +
    +

    Training and testing with difference random seed for uncertainty estimation

    +
    +
    +
    model_class = IE_SBiGRU(
+        window_size=64,
+        stride=1024, # 1024
+        batch_size=253, # 256
+        lr= 0.0011, # 0.0004999805761074147,
+        num_epochs=35,
+        verbose=True,
+        device='cuda'
+    )
+# model_class.fit(trainer_data, target)
+model_class.load_checkpoint(path_to_tar + "IE_SBiGRU_sota.tar")
+
    +
    +
    +
    +
    +
    +
    model_class.evaluate(tester_data, test_target)
+
    +
    +
    +
    +
    Creating sequence of samples: 100%|██████████| 14/14 [00:00<00:00, 2341.13it/s]
+                                                            
+
    +
    +
    {'mse': 0.004956771691496658, 'rmse': 0.07040434426579555}
+
    +
    +
    +
    +
    + + +
    + + + + + + + +
    + + + +
    + +
    +
    + + Show Source + +
    +
    + +
    + + +
    +
    + +
    + +
    +
    +
    + + + + + +
    +
    + +
    + +
    +

    + + © Copyright 2024, ICE Team. +
    + +

    +
    + +
    +

    + Created using Sphinx 5.3.0. +
    +

    +
    + +
    + + + +
    + +

    + Built with the PyData Sphinx Theme 0.14.0. +

    + +
    + +
    + +
    + + \ No newline at end of file diff --git a/docs/benchmark/index.html b/docs/benchmark/index.html index c25bc45..f47e1c3 100644 --- a/docs/benchmark/index.html +++ b/docs/benchmark/index.html @@ -327,9 +327,15 @@

    Contents:

    @@ -390,9 +396,15 @@

    Contents:

    diff --git a/docs/benchmark/rul_sota.html b/docs/benchmark/rul_sota.html new file mode 100644 index 0000000..a26a94c --- /dev/null +++ b/docs/benchmark/rul_sota.html @@ -0,0 +1,593 @@ + + + + + + + + + + + + Results of RUL estimation using IR — ICE documentation + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
    + + + + + + + + + + + +
    +
    +
    +
    + + + Ctrl+K +
    +
    + + + +
    +
    + +
    + + + +
    + + +
    + +
    + +
    + + + +
    + +
    +
    + +
    + +
    + +
    +
    +
    + + +
    +
    + + + + +
    + +
    + + +
    +
    + +
    +
    + +
    + +
    + + + + +
    + +
    + + +
    +
    + + + + + +
    + +
    +

    Results of RUL estimation using IR#

    +

    This notebook presents experimental results of rul estimation on the CMAPSS dataset using the model lstm-256.

    +

    Importing libraries.

    +
    +
    +
    from ice.remaining_useful_life_estimation.datasets import RulCmapss
+from ice.remaining_useful_life_estimation.models import IR 
+
+import pandas as pd
+
    +
    +
    +
    +
    C:\Users\user\conda\envs\ice_testing\Lib\site-packages\tqdm\auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html
+  from .autonotebook import tqdm as notebook_tqdm
+
    +
    +
    +
    +

    Initializing model class and train/test data split for fd001 subdataset

    +
    +
    +
    dataset_class = RulCmapss()
+
+data, target = dataset_class.df[0], dataset_class.target[0]
+test_data, test_target = dataset_class.test[0], dataset_class.test_target[0]  
+
    +
    +
    +
    +
    Reading data/C-MAPSS/fd1_train.csv: 100%|██████████| 20631/20631 [00:00<00:00, 504872.87it/s]
+Reading data/C-MAPSS/fd2_train.csv: 100%|██████████| 53759/53759 [00:00<00:00, 533656.45it/s]
+Reading data/C-MAPSS/fd3_train.csv: 100%|██████████| 24720/24720 [00:00<00:00, 496051.49it/s]
+Reading data/C-MAPSS/fd4_train.csv: 100%|██████████| 61249/61249 [00:00<00:00, 532245.75it/s]
+Reading data/C-MAPSS/fd1_train.csv: 100%|██████████| 20631/20631 [00:00<00:00, 547602.44it/s]
+Reading data/C-MAPSS/fd2_train.csv: 100%|██████████| 53759/53759 [00:00<00:00, 528810.42it/s]
+Reading data/C-MAPSS/fd3_train.csv: 100%|██████████| 24720/24720 [00:00<00:00, 496134.57it/s]
+Reading data/C-MAPSS/fd4_train.csv: 100%|██████████| 61249/61249 [00:00<00:00, 529770.68it/s]
+Reading data/C-MAPSS/fd1_test.csv: 100%|██████████| 13097/13097 [00:00<00:00, 486699.50it/s]
+Reading data/C-MAPSS/fd2_test.csv: 100%|██████████| 33991/33991 [00:00<00:00, 501537.62it/s]
+Reading data/C-MAPSS/fd3_test.csv: 100%|██████████| 16598/16598 [00:00<00:00, 512334.66it/s]
+Reading data/C-MAPSS/fd4_test.csv: 100%|██████████| 41214/41214 [00:00<00:00, 523435.48it/s]
+Reading data/C-MAPSS/fd1_test.csv: 100%|██████████| 13097/13097 [00:00<00:00, 474171.77it/s]
+Reading data/C-MAPSS/fd2_test.csv: 100%|██████████| 33991/33991 [00:00<00:00, 501537.62it/s]
+Reading data/C-MAPSS/fd3_test.csv: 100%|██████████| 16598/16598 [00:00<00:00, 520419.66it/s]
+Reading data/C-MAPSS/fd4_test.csv: 100%|██████████| 41214/41214 [00:00<00:00, 519167.37it/s]
+
    +
    +
    +
    +
    +
    +
    from sklearn.preprocessing import StandardScaler, MinMaxScaler
+import pandas as pd 
+
+scaler = MinMaxScaler()
+trainer_data = scaler.fit_transform(data)
+tester_data = scaler.transform(test_data)
+
+trainer_data = pd.DataFrame(trainer_data, index=data.index, columns=data.columns)
+tester_data = pd.DataFrame(tester_data, index=test_data.index, columns=test_data.columns)
+
    +
    +
    +
    +

    Training and testing with difference random seed for uncertainty estimation

    +
    +
    +
    model_class = IR()
+model_class.load_checkpoint("rul_sota.tar")
+
    +
    +
    +
    +
    C:\Users\user\conda\envs\ice_testing\Lib\site-packages\torch\nn\modules\transformer.py:306: UserWarning: enable_nested_tensor is True, but self.use_nested_tensor is False because encoder_layer.activation_relu_or_gelu was not True
+  warnings.warn(f"enable_nested_tensor is True, but self.use_nested_tensor is False because {why_not_sparsity_fast_path}")
+
    +
    +
    +
    +
    +
    +
    model_class.evaluate(tester_data, test_target)
+
    +
    +
    +
    +
    Creating sequence of samples: 100%|██████████| 100/100 [00:00<00:00, 11148.24it/s]
+                                                          
+
    +
    +
    {'rmse': 11.99217470692219, 'cmapss_score': 25394.12755711561}
+
    +
    +
    +
    +
    + + +
    + + + + + + + +
    + + + +
    + +
    +
    + + Show Source + +
    +
    + +
    + + +
    +
    + +
    + +
    +
    +
    + + + + + +
    +
    + +
    + +
    +

    + + © Copyright 2024, ICE Team. +
    + +

    +
    + +
    +

    + Created using Sphinx 5.3.0. +
    +

    +
    + +
    + + + +
    + +

    + Built with the PyData Sphinx Theme 0.14.0. +

    + +
    + +
    + +
    + + \ No newline at end of file diff --git a/docs/genindex.html b/docs/genindex.html index 3002e7b..4964942 100644 --- a/docs/genindex.html +++ b/docs/genindex.html @@ -345,9 +345,12 @@

    Index

    | C | E | F + | G | I + | L | M | N + | O | P | R | S @@ -363,11 +366,15 @@

    A

  • (in module ice.fault_diagnosis.metrics)
    • +
  • AnomalyDetectionReinartzTEP (class in ice.anomaly_detection.datasets) +
    • +
  • AnomalyDetectionRiethTEP +
  • AnomalyDetectionSmallTEP diff --git a/docs/reference/ice.anomaly_detection.html b/docs/reference/ice.anomaly_detection.html index 608e9bf..fea602c 100644 --- a/docs/reference/ice.anomaly_detection.html +++ b/docs/reference/ice.anomaly_detection.html @@ -418,16 +418,22 @@

    Anomaly detectionModels

  • Datasets
  • Metrics
  • Utils
      diff --git a/docs/reference/ice.anomaly_detection.metrics.html b/docs/reference/ice.anomaly_detection.metrics.html index b7d3359..cc3036b 100644 --- a/docs/reference/ice.anomaly_detection.metrics.html +++ b/docs/reference/ice.anomaly_detection.metrics.html @@ -419,18 +419,18 @@

      Anomaly detection metrics#

      -ice.anomaly_detection.metrics.accuracy(pred: list, target: list) → float[source]#
      +ice.anomaly_detection.metrics.accuracy(pred: ndarray, target: ndarray) → float[source]#

      Accuracy of the classification is the number of true positives divided by the number of examples.

      Parameters:
        -

      • pred (list) – predictions.

        • -

      • target (list) – target values.

        • +

      • pred (np.ndarray) – predictions.

        • +

      • target (np.ndarray) – target values.

      Returns:
      -

      accuracy

      +

      accuracy.

      Return type:

      float

      @@ -438,6 +438,48 @@
      +
      +
      +ice.anomaly_detection.metrics.false_positive_rate(pred: ndarray, target: ndarray) → ndarray[float][source]#
      +

      False Positive Rate, aka False Alarm Rate is the number of false alarms i +divided by the number of normal samples.

      +
      +
      Parameters:
      +
        +

      • pred (np.ndarray) – predictions.

        • +

      • target (np.ndarray) – target values.

        • +
      +
      +
      Returns:
      +

      list of float values with true positive rate for each fault.

      +
      +
      Return type:
      +

      list

      +
      +
      +
      + +
      +
      +ice.anomaly_detection.metrics.true_positive_rate(pred: ndarray, target: ndarray) → ndarray[float][source]#
      +

      True Positive Rate is the number of detected faults i divided by the +number of faults i.

      +
      +
      Parameters:
      +
        +

      • pred (np.ndarray) – predictions.

        • +

      • target (np.ndarray) – target values.

        • +
      +
      +
      Returns:
      +

      list of float values with true positive rate for each fault.

      +
      +
      Return type:
      +

      list

      +
      +
      +
      +
    • @@ -484,6 +526,8 @@ diff --git a/docs/reference/ice.anomaly_detection.models.html b/docs/reference/ice.anomaly_detection.models.html index 2142208..f35abb4 100644 --- a/docs/reference/ice.anomaly_detection.models.html +++ b/docs/reference/ice.anomaly_detection.models.html @@ -421,13 +421,15 @@

    Anomaly detection models

    BaseAnomalyDetection#

    -class ice.anomaly_detection.models.base.BaseAnomalyDetection(window_size: int, batch_size: int, lr: float, num_epochs: int, device: str, verbose: bool, name: str, threshold: float = 0.95)[source]#
    +class ice.anomaly_detection.models.base.BaseAnomalyDetection(window_size: int, stride: int, batch_size: int, lr: float, num_epochs: int, device: str, verbose: bool, name: str, random_seed: int, val_ratio: float, save_checkpoints: bool, threshold_level: float = 0.95)[source]#

    Bases: BaseModel, ABC

    Base class for all anomaly detection models.

    Parameters:
    -
    -fit(df: DataFrame)[source]#
    -

    Method fit for training anomaly detection models.

    +
    +load_checkpoint(checkpoint_path: str)[source]#
    +

    Load checkpoint.

    Parameters:
    -

    df (pd.DataFrame) – data without anomaly states

    +

    checkpoint_path (str) – Path to load checkpoint.

    @@ -479,13 +482,13 @@

    Anomaly detection models -

    AutoEncoderMLP#

    +
    +

    AutoEncoderMLP#

    -class ice.anomaly_detection.models.autoencoder.AutoEncoderMLP(window_size: int, batch_size: int = 128, lr: float = 0.001, num_epochs: int = 10, device: str = 'cpu', verbose: bool = False, name: str = 'ae_anomaly_detection', threshold: float = 0.95, hidden_dims: list = [256, 128, 64])[source]#
    +class ice.anomaly_detection.models.autoencoder.AutoEncoderMLP(window_size: int, stride: int = 1, batch_size: int = 128, lr: float = 0.001, num_epochs: int = 10, device: str = 'cpu', verbose: bool = False, name: str = 'ae_anomaly_detection', random_seed: int = 42, val_ratio: float = 0.15, save_checkpoints: bool = False, threshold_level: float = 0.95, hidden_dims: list = [256, 128, 64])[source]#

    Bases: BaseAnomalyDetection

    -

    Autoencoder (AE) consists of encoder and decoder parts. Each +

    MLP autoencoder consists of MLP encoder and MLP decoder parts. Each sample is reshaped to a vector (B, L, C) -> (B, L * C) for calculations and to a vector (B, L * C) -> (B, L, C) for the output. Where B is the batch size, L is the sequence length, C is the number of sensors.

    @@ -493,6 +496,8 @@

    Anomaly detection modelsParameters:

    • window_size (int) – The window size to train the model.

      • +

    • stride (int) – The time interval between first points of consecutive +sliding windows in training.

    • batch_size (int) – The batch size to train the model.

    • lr (float) – The larning rate to train the model.

    • num_epochs (float) – The number of epochs to train the model.

      • @@ -500,37 +505,135 @@

      Anomaly detection models +

      AnomalyTransformer#

      -
      -class ice.anomaly_detection.models.autoencoder.MLP(num_sensors: int, window_size: int, num_layers: int, hidden_dims: list, type: str)[source]#
      -

      Bases: Module

      -

      Initializes internal Module state, shared by both nn.Module and ScriptModule.

      -
      -
      -forward(x)[source]#
      -

      Defines the computation performed at every call.

      -

      Should be overridden by all subclasses.

      -
      -

      Note

      -

      Although the recipe for forward pass needs to be defined within -this function, one should call the Module instance afterwards -instead of this since the former takes care of running the -registered hooks while the latter silently ignores them.

      -
      +
      +class ice.anomaly_detection.models.transformer.AnomalyTransformer(window_size: int = 100, stride: int = 1, batch_size: int = 128, lr: float = 0.0001, num_epochs: int = 10, device: str = 'cpu', verbose: bool = False, name: str = 'transformer_anomaly_detection', random_seed: int = 42, val_ratio: float = 0.15, save_checkpoints: bool = False, threshold_level: float = 0.95, d_model: int = 256, n_heads: int = 8, e_layers: int = 3, d_ff: int = 256, dropout: float = 0.0, activation: str = 'gelu')[source]#
      +

      Bases: BaseAnomalyDetection

      +

      Anomaly Transformer was presented at ICLR 2022: “Anomaly Transformer: +Time Series Anomaly Detection with Association Discrepancy”. +https://openreview.net/forum?id=LzQQ89U1qm_

      +
      +
      Parameters:
      +
        +

      • window_size (int) – The window size to train the model.

        • +

      • stride (int) – The time interval between first points of consecutive +sliding windows in training.

        • +

      • batch_size (int) – The batch size to train the model.

        • +

      • lr (float) – The larning rate to train the model.

        • +

      • num_epochs (float) – The number of epochs to train the model.

        • +

      • device (str) – The name of a device to train the model. cpu and +cuda are possible.

        • +

      • verbose (bool) – If true, show the progress bar in training.

        • +

      • name (str) – The name of the model for artifact storing.

        • +

      • random_seed (int) – Seed for random number generation to ensure reproducible results.

        • +

      • val_ratio (float) – Proportion of the dataset used for validation, between 0 and 1.

        • +

      • save_checkpoints (bool) – If true, store checkpoints.

        • +

      • threshold_level (float) – Takes a value from 0 to 1. It specifies +the quantile in the distribution of errors on the training +dataset at which the threshold value is set.

        • +

      • threshold_value (float) – Threshold value is calculated after the model is trained. +It sets the error limit above which the data sample defines as anomaly.

        • +

      • d_model (int) – Dimension of model.

        • +

      • n_heads (int) – Number of heads.

        • +

      • e_layers (int) – Number of encoder layers.

        • +

      • d_ff (int) – Dimension of MLP.

        • +

      • dropout (float) – The rate of dropout.

        • +

      • activation (str) – Activation (‘relu’, ‘gelu’).

        • +
      +
      +
      -
      -
      -training: bool#
      -
      +

    +
    +

    STGAT-MAD#

    +
    +
    +class ice.anomaly_detection.models.stgat.STGAT_MAD(window_size: int, stride: int = 1, batch_size: int = 128, lr: float = 0.001, num_epochs: int = 10, device: str = 'cpu', verbose: bool = False, name: str = 'stgat_anomaly_detection', random_seed: int = 42, val_ratio: float = 0.15, save_checkpoints: bool = False, threshold_level: float = 0.95, embed_dim: Optional[int] = None, layer_numb: int = 2, lstm_n_layers: int = 1, lstm_hid_dim: int = 150, recon_n_layers: int = 1, recon_hid_dim: int = 150, dropout: float = 0.2)[source]#
    +

    Bases: BaseAnomalyDetection

    +

    Stgat-Mad was presented at ICASSP 2022: “Stgat-Mad : Spatial-Temporal Graph +Attention Network For Multivariate Time Series Anomaly Detection”. +https://ieeexplore.ieee.org/abstract/document/9747274/

    +
    +
    Parameters:
    +
      +

    • window_size (int) – The window size to train the model.

      • +

    • stride (int) – The time interval between first points of consecutive +sliding windows in training.

      • +

    • batch_size (int) – The batch size to train the model.

      • +

    • lr (float) – The larning rate to train the model.

      • +

    • num_epochs (float) – The number of epochs to train the model.

      • +

    • device (str) – The name of a device to train the model. cpu and +cuda are possible.

      • +

    • verbose (bool) – If true, show the progress bar in training.

      • +

    • name (str) – The name of the model for artifact storing.

      • +

    • random_seed (int) – Seed for random number generation to ensure reproducible results.

      • +

    • val_ratio (float) – Proportion of the dataset used for validation, between 0 and 1.

      • +

    • save_checkpoints (bool) – If true, store checkpoints.

      • +

    • threshold_level (float) – Takes a value from 0 to 1. It specifies +the quantile in the distribution of errors on the training +dataset at which the threshold value is set.

      • +

    • embed_dim (int) – Embedding dimension.

      • +

    • layer_numb (int) – Number of layers.

      • +

    • lstm_n_layers (int) – Number of LSTM layers.

      • +

    • lstm_hid_dim (int) – Hidden dimension of LSTM layers.

      • +

    • recon_n_layers (int) – Number of reconstruction layers.

      • +

    • recon_hid_dim (int) – Hidden dimension of reconstruction layers.

      • +

    • dropout (float) – The rate of dropout.

      • +
    +
    +
    +
    +
    +
    +

    GSL-GNN#

    +
    +
    +class ice.anomaly_detection.models.gnn.GSL_GNN(window_size: int, stride: int = 1, batch_size: int = 128, lr: float = 0.001, num_epochs: int = 10, device: str = 'cpu', verbose: bool = False, name: str = 'gnn_anomaly_detection', random_seed: int = 42, val_ratio: float = 0.15, save_checkpoints: bool = False, threshold_level: float = 0.95, alpha: float = 0.2, k: Optional[int] = None)[source]#
    +

    Bases: BaseAnomalyDetection

    +

    GNN autoencoder consists of encoder with graph convolutional layers +and MLP decoder parts. The graph describing the data is constructed +during the training process using trainable parameters.

    +
    +
    Parameters:
    +
      +

    • window_size (int) – The window size to train the model.

      • +

    • stride (int) – The time interval between first points of consecutive +sliding windows in training.

      • +

    • batch_size (int) – The batch size to train the model.

      • +

    • lr (float) – The larning rate to train the model.

      • +

    • num_epochs (float) – The number of epochs to train the model.

      • +

    • device (str) – The name of a device to train the model. cpu and +cuda are possible.

      • +

    • verbose (bool) – If true, show the progress bar in training.

      • +

    • name (str) – The name of the model for artifact storing.

      • +

    • random_seed (int) – Seed for random number generation to ensure reproducible results.

      • +

    • val_ratio (float) – Proportion of the dataset used for validation, between 0 and 1.

      • +

    • save_checkpoints (bool) – If true, store checkpoints.

      • +

    • threshold_level (float) – Takes a value from 0 to 1. It specifies +the quantile in the distribution of errors on the training +dataset at which the threshold value is set.

      • +

    • alpha (float) – Saturation rate for adjacency matrix.

      • +

    • k (int) – Limit on the number of edges in the adjacency matrix.

      • +
    +
    +
    @@ -581,19 +684,26 @@

    Anomaly detection models
  • BaseAnomalyDetection
    • -
  • AutoEncoderMLP
      +
    • AutoEncoderMLP diff --git a/docs/reference/ice.base.html b/docs/reference/ice.base.html index 9b307e3..c7d3446 100644 --- a/docs/reference/ice.base.html +++ b/docs/reference/ice.base.html @@ -438,12 +438,15 @@
      -class ice.base.BaseModel(batch_size: int, lr: float, num_epochs: int, device: str, verbose: bool, name: str)[source]#
      +class ice.base.BaseModel(window_size: int, stride: int, batch_size: int, lr: float, num_epochs: int, device: str, verbose: bool, name: str, random_seed: int, val_ratio: float, save_checkpoints: bool)[source]#

      Bases: ABC

      Base class for all models.

      Parameters:
        +

      • window_size (int) – The window size to train the model.

        • +

      • stride (int) – The time interval between first points of consecutive +sliding windows in training.

      • batch_size (int) – The batch size to train the model.

      • lr (float) – The larning rate to train the model.

      • num_epochs (float) – The number of epochs to train the model.

        • @@ -451,15 +454,16 @@ cuda are possible.

      • verbose (bool) – If true, show the progress bar in training.

      • name (str) – The name of the model for artifact storing.

        • +

      • random_seed (int) – Seed for random number generation to ensure reproducible results.

        • +

      • val_ratio (float) – Proportion of the dataset used for validation, between 0 and 1.

        • +

      • save_checkpoints (bool) – If true, store checkpoints.

      -abstract evaluate(df: DataFrame, target: Series) → dict[source]#
      -

      This method has to be implemented by all children. Evaluate the -metrics. Docstring has to be rewritten so that all metrics are clearly -described.

      +evaluate(df: DataFrame, target: Series) → dict[source]# +

      Evaluate the metrics: accuracy.

      Parameters:
        @@ -485,7 +489,7 @@
        -fit(df: DataFrame, target: Series)[source]#
        +fit(df: DataFrame, target: Optional[Series] = None, epochs: Optional[int] = None, save_path: Optional[str] = None, trial: Optional[Trial] = None, force_model_ctreation: bool = False)[source]#

        Fit (train) the model by a given dataset.

        Parameters:
        @@ -493,8 +497,15 @@

      • df (pandas.DataFrame) – A dataframe with sensor data. Index has two columns: run_id and sample. All other columns a value of sensors.

        • -

      • target (pandas.Series) – A series with target values. Indes has two -columns: run_id and sample.

        • +

      • target (pandas.Series) – A series with target values. Index has two +columns: run_id and sample. It is omitted for anomaly +detection task.

        • +

      • epochs (int) – The number of epochs for training step. If None, +self.num_epochs parameter is used.

        • +

      • save_path (str) – Path to save checkpoints. If None, the path is +created automatically.

        • +

      • trial (optuna.Trial, None) – optuna.Trial object created by optimize method.

        • +

      • force_model_ctreation (bool) – force fit to create model for optimization study.

      @@ -517,6 +528,55 @@
      +
      +
      +load_checkpoint(checkpoint_path: str)[source]#
      +

      Load checkpoint.

      +
      +
      Parameters:
      +

      checkpoint_path (str) – Path to load checkpoint.

      +
      +
      +
      + +
      +
      +model_param_estimation()[source]#
      +

      Calculate number of self.model parameters, mean and std for inference time

      +
      +
      Returns:
      +

      +
      A tuple containing the number of parameters in the

      model and the mean and standard deviation of model inference time.

      +
      +
      +

      +
      +
      Return type:
      +

      tuple

      +
      +
      +
      + +
      +
      +optimize(df: DataFrame, target: Optional[Series] = None, optimize_parameter: str = 'batch_size', optimize_range: tuple = (128, 256), direction: str = 'minimize', n_trials: int = 5, epochs: Optional[int] = None, optimize_metric: Optional[str] = None)[source]#
      +

      Make the optuna study to return the best hyperparameter value on validation dataset

      +
      +
      Parameters:
      +
        +

      • df (pd.DataFrame) – DataFrame to use method fit

        • +

      • optimize_parameter (str, optional) – Model parameter to optimize. Defaults to ‘batch_size’.

        • +

      • optimize_range (tuple, optional) – Model parameter range for optuna trials. Defaults to (128, 256).

        • +

      • n_trials (int, optional) – number of trials. Defaults to 5.

        • +

      • target (pd.Series, optional) – target pd.Series to use method fit. Defaults to None.

        • +

      • epochs (int, optional) – Epoch number to use method fit. Defaults to None.

        • +

      • optimize_metric (str) – Metric on validation dataset to use as a target for hyperparameter optimization.

        • +

      • direction (str) – “minimize” or “maximize” the target for hyperparameter optimization

        • +
      +
      +
      +
      +
      predict(sample: Tensor) → Tensor[source]#
      @@ -536,6 +596,23 @@

    • +
    +
    +save_checkpoint(save_path: Optional[str] = None)[source]#
    +

    Save checkpoint.

    +
    +
    Parameters:
    +

    save_path (str) – Path to save checkpoint.

    +
    +
    +
    + + + +
    +
    +class ice.base.SlidingWindowDataset(df: DataFrame, target: Series, window_size: int, stride: int = 1)[source]#
    +

    Bases: Dataset

    @@ -591,9 +668,14 @@
  • BaseModel.evaluate()
  • BaseModel.fit()
  • BaseModel.from_config()
    • +
  • BaseModel.load_checkpoint()
    • +
  • BaseModel.model_param_estimation()
    • +
  • BaseModel.optimize()
  • BaseModel.predict()
    • +
  • BaseModel.save_checkpoint()
    • +
  • SlidingWindowDataset
    • diff --git a/docs/reference/ice.fault_diagnosis.datasets.html b/docs/reference/ice.fault_diagnosis.datasets.html index 97bb8eb..98aa1d8 100644 --- a/docs/reference/ice.fault_diagnosis.datasets.html +++ b/docs/reference/ice.fault_diagnosis.datasets.html @@ -443,6 +443,33 @@ +
    +
    +class ice.fault_diagnosis.datasets.FaultDiagnosisRiethTEP(num_chunks=None, force_download=False)[source]#
    +

    Bases: BaseDataset

    +

    Dataset of Tennessee Eastman Process dataset +Rieth, C. A., Amsel, B. D., Tran, R., & Cook, M. B. (2017). +Additional Tennessee Eastman Process Simulation Data for +Anomaly Detection Evaluation (Version V1) [Computer software]. +Harvard Dataverse. +https://doi.org/10.7910/DVN/6C3JR1.

    +
    +
    Parameters:
    +
      +

    • num_chunks (int) – If given, download only num_chunks chunks of data. +Used for testing purposes.

      • +

    • force_download (bool) – If True, download the dataset even if it exists.

      • +
    +
    +
    +
    + +

    This method has to be implemented by all children. Set name and public link.

    +
    + +
    +
    class ice.fault_diagnosis.datasets.FaultDiagnosisSmallTEP(num_chunks=None, force_download=False)[source]#
    @@ -519,6 +546,10 @@
  • FaultDiagnosisReinartzTEP.set_name_public_link()
    • +
  • FaultDiagnosisRiethTEP +
  • FaultDiagnosisSmallTEP diff --git a/docs/reference/ice.fault_diagnosis.html b/docs/reference/ice.fault_diagnosis.html index 2930859..dd93d63 100644 --- a/docs/reference/ice.fault_diagnosis.html +++ b/docs/reference/ice.fault_diagnosis.html @@ -424,6 +424,7 @@

    Fault diagnosisDatasets

    • diff --git a/docs/reference/ice.fault_diagnosis.models.html b/docs/reference/ice.fault_diagnosis.models.html index 3d82033..08d1893 100644 --- a/docs/reference/ice.fault_diagnosis.models.html +++ b/docs/reference/ice.fault_diagnosis.models.html @@ -421,50 +421,28 @@

    Fault diagnosis models

    BaseFaultDiagnosis#

    -class ice.fault_diagnosis.models.base.BaseFaultDiagnosis(window_size: int, batch_size: int, lr: float, num_epochs: int, device: str, verbose: bool, name: str)[source]#
    +class ice.fault_diagnosis.models.base.BaseFaultDiagnosis(window_size: int, stride: int, batch_size: int, lr: float, num_epochs: int, device: str, verbose: bool, name: str, random_seed: int, val_ratio: float, save_checkpoints: bool)[source]#

    Bases: BaseModel, ABC

    Base class for all fault diagnosis models.

    Parameters:

    • window_size (int) – The window size to train the model.

      • +

    • stride (int) – The time interval between first points of consecutive +sliding windows in training.

    • batch_size (int) – The batch size to train the model.

      • -

    • lr (float) – The learning rate to train the model.

      • +

    • lr (float) – The larning rate to train the model.

    • num_epochs (float) – The number of epochs to train the model.

    • device (str) – The name of a device to train the model. cpu and cuda are possible.

    • verbose (bool) – If true, show the progress bar in training.

    • name (str) – The name of the model for artifact storing.

      • +

    • random_seed (int) – Seed for random number generation to ensure reproducible results.

      • +

    • val_ratio (float) – Proportion of the dataset used for validation, between 0 and 1.

      • +

    • save_checkpoints (bool) – If true, store checkpoints.

    -
    -
    -evaluate(df: DataFrame, target: Series) → dict[source]#
    -

    Evaluate the metrics: accuracy.

    -
    -
    Parameters:
    -
      -

    • df (pandas.DataFrame) – A dataframe with sensor data. Index has -two columns: run_id and sample. All other columns a value of -sensors.

      • -

    • target (pandas.Series) – A series with target values. Indes has two -columns: run_id and sample.

      • -
    -
    -
    Returns:
    -

    -
    A dictionary with metrics where keys are names of metrics and

    values are values of metrics.

    -
    -
    -

    -
    -
    Return type:
    -

    dict

    -
    -
    -
    -
    @@ -472,7 +450,7 @@

    Fault diagnosis models

    MLP#

    -class ice.fault_diagnosis.models.mlp.MLP(window_size: int, hidden_dim: int = 256, batch_size: int = 128, lr: float = 0.001, num_epochs: int = 10, device: str = 'cpu', verbose: bool = False, name: str = 'mlp_fault_diagnosis')[source]#
    +class ice.fault_diagnosis.models.mlp.MLP(window_size: int, stride: int = 1, hidden_dim: int = 256, batch_size: int = 128, lr: float = 0.001, num_epochs: int = 10, device: str = 'cpu', verbose: bool = False, name: str = 'mlp_fault_diagnosis', random_seed: int = 42, val_ratio: float = 0.15, save_checkpoints: bool = False)[source]#

    Bases: BaseFaultDiagnosis

    Multilayer Perceptron (MLP) consists of input, hidden, output layers and ReLU activation. Each sample is reshaped to a vector (B, L, C) -> (B, L * C) @@ -490,6 +468,9 @@

    Fault diagnosis modelscuda are possible.

  • verbose (bool) – If true, show the progress bar in training.

  • name (str) – The name of the model for artifact storing.

    • +

  • random_seed (int) – Seed for random number generation to ensure reproducible results.

    • +

  • val_ratio (float) – Proportion of the dataset used for validation, between 0 and 1.

    • +

  • save_checkpoints (bool) – If true, store checkpoints.

    • @@ -500,7 +481,7 @@

    Fault diagnosis models

    TCN#

    -class ice.fault_diagnosis.models.tcn.TCN(window_size: int, hidden_dim: int = 256, kernel_size: int = 5, num_layers: int = 4, dilation_base: int = 2, dropout: float = 0.2, batch_size: int = 128, lr: float = 0.001, num_epochs: int = 10, device: str = 'cpu', verbose: bool = False, name: str = 'tcn_fault_diagnosis')[source]#
    +class ice.fault_diagnosis.models.tcn.TCN(window_size: int, stride: int = 1, hidden_dim: int = 256, kernel_size: int = 5, num_layers: int = 4, dilation_base: int = 2, dropout: float = 0.2, batch_size: int = 128, lr: float = 0.001, num_epochs: int = 10, device: str = 'cpu', verbose: bool = False, name: str = 'tcn_fault_diagnosis', random_seed: int = 42, val_ratio: float = 0.15, save_checkpoints: bool = False)[source]#

    Bases: BaseFaultDiagnosis

    Temporal Convolutional Network (TCN)-based Fault Diagnosis method. The implementation is based on the paper Lomov, Ildar, et al. “Fault detection @@ -522,6 +503,9 @@

    Fault diagnosis modelscuda are possible.

  • verbose (bool) – If true, show the progress bar in training.

  • name (str) – The name of the model for artifact storing.

    • +

  • random_seed (int) – Seed for random number generation to ensure reproducible results.

    • +

  • val_ratio (float) – Proportion of the dataset used for validation, between 0 and 1.

    • +

  • save_checkpoints (bool) – If true, store checkpoints.

    • @@ -574,10 +558,7 @@

    Fault diagnosis models
    • BaseFaultDiagnosis
    • MLP
        diff --git a/docs/reference/ice.health_index_estimation.html b/docs/reference/ice.health_index_estimation.html index a69be72..e28a25d 100644 --- a/docs/reference/ice.health_index_estimation.html +++ b/docs/reference/ice.health_index_estimation.html @@ -418,7 +418,7 @@

        Health index estimation
      • Models
      • Datasets
          @@ -427,6 +427,7 @@

          Health index estimation
        • Metrics
        • Utils
            diff --git a/docs/reference/ice.health_index_estimation.metrics.html b/docs/reference/ice.health_index_estimation.metrics.html index 3105ff9..cfd46b2 100644 --- a/docs/reference/ice.health_index_estimation.metrics.html +++ b/docs/reference/ice.health_index_estimation.metrics.html @@ -437,6 +437,26 @@

    +
    +
    +ice.health_index_estimation.metrics.rmse(pred: list, target: list) → float[source]#
    +

    Mean squared error between real and predicted wear.

    +
    +
    Parameters:
    +
      +

    • pred (list) – numpy prediction values.

      • +

    • target (list) – numpy target values.

      • +
    +
    +
    Returns:
    +

    rmse

    +
    +
    Return type:
    +

    float

    +
    +
    +
    + @@ -483,6 +503,7 @@ diff --git a/docs/reference/ice.health_index_estimation.models.html b/docs/reference/ice.health_index_estimation.models.html index 306c5db..9a28761 100644 --- a/docs/reference/ice.health_index_estimation.models.html +++ b/docs/reference/ice.health_index_estimation.models.html @@ -421,14 +421,15 @@

    Remaining useful life models

    BaseRemainingUsefulLifeEstimation#

    -class ice.health_index_estimation.models.base.BaseHealthIndexEstimation(window_size: int, stride: int, batch_size: int, lr: float, num_epochs: int, device: str, verbose: bool, name: str)[source]#
    +class ice.health_index_estimation.models.base.BaseHealthIndexEstimation(window_size: int, stride: int, batch_size: int, lr: float, num_epochs: int, device: str, verbose: bool, name: str, random_seed: int, val_ratio: float, save_checkpoints: bool)[source]#

    Bases: BaseModel, ABC

    Base class for all HI diagnosis models.

    Parameters:

    • window_size (int) – The window size to train the model.

      • -

    • stride (int) – The time interval between first points of consecutive sliding windows.

      • +

    • stride (int) – The time interval between first points of consecutive +sliding windows in training.

    • batch_size (int) – The batch size to train the model.

    • lr (float) – The larning rate to train the model.

    • num_epochs (float) – The number of epochs to train the model.

      • @@ -436,44 +437,20 @@

      Remaining useful life models -
      -evaluate(df: DataFrame, target: Series) → dict[source]#
      -

      Evaluate the metrics: mse.

      -
      -
      Parameters:
      -
        -

      • df (pandas.DataFrame) – A dataframe with sensor data. Index has -two columns: run_id and sample. All other columns a value of -sensors.

        • -

      • target (pandas.Series) – A series with target values. Indes has two -columns: run_id and sample.

        • -
      -
      -
      Returns:
      -

      -
      A dictionary with metrics where keys are names of metrics and

      values are values of metrics.

      -
      -
      -

      -
      -
      Return type:
      -

      dict

      -
      -
      -

    -
    -
    -

    MLP#

    +
    +

    MLP#

    -class ice.health_index_estimation.models.mlp.MLP(window_size: int = 1024, stride: int = 300, hidden_dim: int = 256, batch_size: int = 64, lr: float = 5e-05, num_epochs: int = 50, device: str = 'cpu', verbose: bool = True, name: str = 'mlp_fault_diagnosis')[source]#
    +class ice.health_index_estimation.models.mlp.MLP(window_size: int = 1024, stride: int = 300, hidden_dim: int = 256, batch_size: int = 64, lr: float = 5e-05, num_epochs: int = 50, device: str = 'cpu', verbose: bool = True, name: str = 'mlp_fault_diagnosis', random_seed: int = 42, val_ratio: float = 0.15, save_checkpoints: bool = False)[source]#

    Bases: BaseHealthIndexEstimation

    Multilayer Perceptron (MLP) consists of input, hidden, output layers and ReLU activation. Each sample is reshaped to a vector (B, L, C) -> (B, L * C) @@ -492,6 +469,9 @@

    Remaining useful life models
    • BaseRemainingUsefulLifeEstimation
      • -
    • MLP
        +
      • MLP
        • diff --git a/docs/reference/ice.html b/docs/reference/ice.html deleted file mode 100644 index fa2575e..0000000 --- a/docs/reference/ice.html +++ /dev/null @@ -1,653 +0,0 @@ - - - - - - - - - - - - ice package — ICE documentation - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
        - - - - - - - - - - - -
        -
        -
        -
        - - - Ctrl+K -
        -
        - - - -
        -
        - -
        - - - -
        - - -
        - -
        - -
        - - - -
        - -
        -
        - -
        - -
        - -
        -
        -
        - - -
        -
        - - - - -
        - -
        - - -
        -
        - -
        -
        - -
        - -
        - - - - -
        - -
        - - -
        -
        - - - - - -
        - -
        -

        ice package#

        -
        -

        Subpackages#

        -
        -
        -
        -
        -

        Submodules#

        -
        -
        -

        ice.base module#

        -
        -
        -class ice.base.BaseDataset(num_chunks=None, force_download=False)#
        -

        Bases: ABC

        -

        Base class for datasets.

        -
        -
        -__init__(num_chunks=None, force_download=False)#
        -
        -
        Parameters:
        -
          -

        • num_chunks (int) – If given, download only num_chunks chunks of data. -Used for testing purposes.

          • -

        • force_download (bool) – If True, download the dataset even if it exists.

          • -
        -
        -
        -
        - -
        - -

        This method has to be implemented by all children. Set name and public link.

        -
        - -
        - -
        -
        -class ice.base.BaseModel(batch_size: int, lr: float, num_epochs: int, device: str, verbose: bool, name: str)#
        -

        Bases: ABC

        -

        Base class for all models.

        -
        -
        -abstract __init__(batch_size: int, lr: float, num_epochs: int, device: str, verbose: bool, name: str)#
        -
        -
        Parameters:
        -
          -

        • batch_size (int) – The batch size to train the model.

          • -

        • lr (float) – The larning rate to train the model.

          • -

        • num_epochs (float) – The number of epochs to train the model.

          • -

        • device (str) – The name of a device to train the model. cpu and -cuda are possible.

          • -

        • verbose (bool) – If true, show the progress bar in training.

          • -

        • name (str) – The name of the model for artifact storing.

          • -
        -
        -
        -
        - -
        -
        -abstract evaluate(df: DataFrame, target: Series) → dict#
        -

        This method has to be implemented by all children. Evaluate the -metrics. Docstring has to be rewritten so that all metrics are clearly -described.

        -
        -
        Parameters:
        -
          -

        • df (pandas.DataFrame) – A dataframe with sensor data. Index has -two columns: run_id and sample. All other columns a value of -sensors.

          • -

        • target (pandas.Series) – A series with target values. Indes has two -columns: run_id and sample.

          • -
        -
        -
        Returns:
        -

        -
        A dictionary with metrics where keys are names of metrics and

        values are values of metrics.

        -
        -
        -

        -
        -
        Return type:
        -

        dict

        -
        -
        -
        - -
        -
        -fit(df: DataFrame, target: Series)#
        -

        Fit (train) the model by a given dataset.

        -
        -
        Parameters:
        -
          -

        • df (pandas.DataFrame) – A dataframe with sensor data. Index has -two columns: run_id and sample. All other columns a value of -sensors.

          • -

        • target (pandas.Series) – A series with target values. Indes has two -columns: run_id and sample.

          • -
        -
        -
        -
        - -
        -
        -classmethod from_config(cfg: Config)#
        -

        Create instance of the model class with parameters from config.

        -
        -
        Parameters:
        -

        cfg (Config) – A config with model’s parameters.

        -
        -
        Returns:
        -

        Instance of BaseModel child class initialized with parameters from config.

        -
        -
        Return type:
        -

        BaseModel

        -
        -
        -
        - -
        -
        -predict(sample: Tensor) → Tensor#
        -

        Make a prediction for a given batch of samples.

        -
        -
        Parameters:
        -

        sample (torch.Tensor) – A tensor of the shape (B, L, C) where -B is the batch size, L is the sequence length, C is the number -of sensors.

        -
        -
        Returns:
        -

        A tensor with predictions of the shape (B,).

        -
        -
        Return type:
        -

        torch.Tensor

        -
        -
        -
        - -
        - -
        -
        - - -
        - - - - - -
        - -
        -
        -
        - -
        - - - -
        - -
        -
        - On this page -
        -
        - -
        -
        - - Show Source - -
        -
        - -
        - - -
        -
        - -
        - -
        -
        -
        - - - - - -
        -
        - -
        - -
        -

        - - © Copyright 2024, ICE Team. -
        - -

        -
        - -
        -

        - Created using Sphinx 5.3.0. -
        -

        -
        - -
        - - - -
        - -

        - Built with the PyData Sphinx Theme 0.14.0. -

        - -
        - -
        - -
        - - \ No newline at end of file diff --git a/docs/reference/ice.remaining_useful_life_estimation.datasets.html b/docs/reference/ice.remaining_useful_life_estimation.datasets.html index e3f9886..201802a 100644 --- a/docs/reference/ice.remaining_useful_life_estimation.datasets.html +++ b/docs/reference/ice.remaining_useful_life_estimation.datasets.html @@ -441,6 +441,30 @@

    +
    +
    +class ice.remaining_useful_life_estimation.datasets.RulCmapssPaper(num_chunks=None, force_download=False)[source]#
    +

    Bases: BaseDataset

    +

    Preprocessed to piece wise RUL data from the dataset: +Saxena A. et al. Damage propagation modeling for aircraft engine run-to-failure simulation +DOI: 10.1109/PHM.2008.4711414. Target is the minimum rul value for every test device.

    +
    +
    Parameters:
    +
      +

    • num_chunks (int) – If given, download only num_chunks chunks of data. +Used for testing purposes.

      • +

    • force_download (bool) – If True, download the dataset even if it exists.

      • +
    +
    +
    +
    + +

    This method has to be implemented by all children. Set name and public link.

    +
    + +
    +
    @@ -490,6 +514,10 @@
  • RulCmapss.set_name_public_link()
    • +
  • RulCmapssPaper +
    • diff --git a/docs/reference/ice.remaining_useful_life_estimation.html b/docs/reference/ice.remaining_useful_life_estimation.html index e79d574..b9d9d88 100644 --- a/docs/reference/ice.remaining_useful_life_estimation.html +++ b/docs/reference/ice.remaining_useful_life_estimation.html @@ -418,17 +418,19 @@

    Remaining useful life
  • Models
  • Datasets
  • Metrics
  • Utils
      diff --git a/docs/reference/ice.remaining_useful_life_estimation.metrics.html b/docs/reference/ice.remaining_useful_life_estimation.metrics.html index 379992a..a2889da 100644 --- a/docs/reference/ice.remaining_useful_life_estimation.metrics.html +++ b/docs/reference/ice.remaining_useful_life_estimation.metrics.html @@ -418,15 +418,19 @@

      Remaining useful life metrics#

      -
      -ice.remaining_useful_life_estimation.metrics.nonsimmetric_function(value)[source]#
      -

      Exponent calculation function depending on the relative deviation from the true value

      +
      +ice.remaining_useful_life_estimation.metrics.cmapss_score(pred: list, target: list) → float[source]#
      +

      Non-simmetric metric proposed in the original dataset paper. +DOI: 10.1109/PHM.2008.4711414

      Parameters:
      -

      value (float) – division result between predicted and real values.

      +
        +

      • pred (list) – numpy prediction values.

        • +

      • target (list) – numpy target values.

        • +
      Returns:
      -

      calculation result

      +

      cmapss score function

      Return type:

      float

      @@ -435,18 +439,15 @@
      -
      -ice.remaining_useful_life_estimation.metrics.rmse(pred: list, target: list) → float[source]#
      -

      Root mean squared error between real and predicted remaining useful life values.

      +
      +ice.remaining_useful_life_estimation.metrics.nonsimmetric_function(value)[source]#
      +

      Exponent calculation function depending on the relative deviation from the true value

      Parameters:
      -
        -

      • pred (list) – numpy prediction values.

        • -

      • target (list) – numpy target values.

        • -
      +

      value (float) – division result between predicted and real values.

      Returns:
      -

      rmse

      +

      calculation result

      Return type:

      float

      @@ -455,10 +456,9 @@
      -
      -ice.remaining_useful_life_estimation.metrics.score(pred: list, target: list) → float[source]#
      -

      Non-simmetric metric proposed in the original dataset paper. -DOI: 10.1109/PHM.2008.4711414

      +
      +ice.remaining_useful_life_estimation.metrics.rmse(pred: list, target: list) → float[source]#
      +

      Root mean squared error between real and predicted remaining useful life values.

      Parameters:
        @@ -467,7 +467,7 @@
      Returns:
      -

      cmapss score function

      +

      rmse

      Return type:

      float

      @@ -520,9 +520,9 @@ diff --git a/docs/reference/ice.remaining_useful_life_estimation.models.html b/docs/reference/ice.remaining_useful_life_estimation.models.html index 07b2d5b..aeba09f 100644 --- a/docs/reference/ice.remaining_useful_life_estimation.models.html +++ b/docs/reference/ice.remaining_useful_life_estimation.models.html @@ -421,70 +421,81 @@

      Health index models

      BaseHealthIndexEstimation#

      -class ice.remaining_useful_life_estimation.models.base.BaseRemainingUsefulLifeEstimation(window_size: int, stride: int, batch_size: int, lr: float, num_epochs: int, device: str, verbose: bool, name: str)[source]#
      +class ice.remaining_useful_life_estimation.models.base.BaseRemainingUsefulLifeEstimation(window_size: int, stride: int, batch_size: int, lr: float, num_epochs: int, device: str, verbose: bool, name: str, random_seed: int, val_ratio: float, save_checkpoints: bool)[source]#

      Bases: BaseModel, ABC

      Base class for all RUL models.

      Parameters:

      • window_size (int) – The window size to train the model.

        • -

      • stride (int) – The time interval between first points of consecutive sliding windows.

        • +

      • stride (int) – The time interval between first points of consecutive +sliding windows in training.

      • batch_size (int) – The batch size to train the model.

        • -

      • lr (float) – The learning rate to train the model.

        • +

      • lr (float) – The larning rate to train the model.

      • num_epochs (float) – The number of epochs to train the model.

      • device (str) – The name of a device to train the model. cpu and cuda are possible.

      • verbose (bool) – If true, show the progress bar in training.

      • name (str) – The name of the model for artifact storing.

        • +

      • random_seed (int) – Seed for random number generation to ensure reproducible results.

        • +

      • val_ratio (float) – Proportion of the dataset used for validation, between 0 and 1.

        • +

      • save_checkpoints (bool) – If true, store checkpoints.

      -
      -
      -evaluate(df: DataFrame, target: Series) → dict[source]#
      -

      Evaluate the metrics: rmse, c-mapss score.

      +
      + +

      +
      +

      MLP#

      +
      +
      +class ice.remaining_useful_life_estimation.models.mlp.MLP(window_size: int = 32, stride: int = 1, hidden_dim: int = 512, batch_size: int = 64, lr: float = 0.0001, num_epochs: int = 15, device: str = 'cpu', verbose: bool = True, name: str = 'mlp_cmapss_rul', random_seed: int = 42, val_ratio: float = 0.15, save_checkpoints: bool = False)[source]#
      +

      Bases: BaseRemainingUsefulLifeEstimation

      +

      Multilayer Perceptron (MLP) consists of input, hidden, output layers and +ReLU activation. Each sample is reshaped to a vector (B, L, C) -> (B, L * C) +where B is the batch size, L is the sequence length, C is the number of +sensors.

      Parameters:
        -

      • df (pandas.DataFrame) – A dataframe with sensor data. Index has -two columns: run_id and sample. All other columns a value of -sensors.

        • -

      • target (pandas.Series) – A series with target values. Indes has two -columns: run_id and sample.

        • +

      • window_size (int) – The window size to train the model.

        • +

      • stride (int) – The time interval between first points of consecutive sliding windows.

        • +

      • hidden_dim (int) – The dimensionality of the hidden layer in MLP.

        • +

      • batch_size (int) – The batch size to train the model.

        • +

      • lr (float) – The larning rate to train the model.

        • +

      • num_epochs (float) – The number of epochs to train the model.

        • +

      • device (str) – The name of a device to train the model. cpu and +cuda are possible.

        • +

      • verbose (bool) – If true, show the progress bar in training.

        • +

      • name (str) – The name of the model for artifact storing.

        • +

      • random_seed (int) – Seed for random number generation to ensure reproducible results.

        • +

      • val_ratio (float) – Proportion of the dataset used for validation, between 0 and 1.

        • +

      • save_checkpoints (bool) – If true, store checkpoints.

      -
      Returns:
      -

      -
      A dictionary with metrics where keys are names of metrics and

      values are values of metrics.

      -
      -
      -

      -
      -
      Return type:
      -

      dict

      -
      - -
      -
      -

      MLP#

      +
      +

      LSTM#

      -
      -class ice.remaining_useful_life_estimation.models.mlp.MLP(window_size: int = 32, stride: int = 1, hidden_dim: int = 256, batch_size: int = 256, lr: float = 5e-05, num_epochs: int = 50, device: str = 'cpu', verbose: bool = True, name: str = 'mlp_cmapss_rul')[source]#
      +
      +class ice.remaining_useful_life_estimation.models.lstm.LSTM(window_size: int = 32, stride: int = 1, hidden_dim: int = 512, hidden_size: int = 256, num_layers: int = 2, dropout_value: float = 0.5, batch_size: int = 64, lr: float = 0.0001, num_epochs: int = 35, device: str = 'cpu', verbose: bool = True, name: str = 'mlp_cmapss_rul', random_seed: int = 42, val_ratio: float = 0.15, save_checkpoints: bool = False)[source]#

      Bases: BaseRemainingUsefulLifeEstimation

      -

      Multilayer Perceptron (MLP) consists of input, hidden, output layers and -ReLU activation. Each sample is reshaped to a vector (B, L, C) -> (B, L * C) -where B is the batch size, L is the sequence length, C is the number of -sensors.

      +

      Long short-term memory (LSTM) model consists of the classical LSTM architecture stack and +two-layer MLP with SiLU nonlinearity and dropout to make the final prediction.

      +

      Each sample is moved to LSTM and reshaped to a vector (B, L, C) -> (B, hidden_size, C) +Then the sample is reshaped to a vector (B, hidden_size, C) -> (B, hidden_size * C)

      Parameters:

      • window_size (int) – The window size to train the model.

      • stride (int) – The time interval between first points of consecutive sliding windows.

      • hidden_dim (int) – The dimensionality of the hidden layer in MLP.

        • +

      • hidden_size (int) – The number of features in the hidden state of the model.

        • +

      • num_layers (int) – The number of stacked reccurent layers of the classic LSTM architecture.

      • batch_size (int) – The batch size to train the model.

      • lr (float) – The larning rate to train the model.

      • num_epochs (float) – The number of epochs to train the model.

        • @@ -492,6 +503,9 @@

        Health index models
        • BaseHealthIndexEstimation
            -
          • BaseRemainingUsefulLifeEstimation
            • +
          • MLP
            • -
          • MLP diff --git a/docs/reference/index.html b/docs/reference/index.html index 177e61d..7b3d243 100644 --- a/docs/reference/index.html +++ b/docs/reference/index.html @@ -416,6 +416,7 @@
          • Base modules
          • Fault Diagnosis
              diff --git a/docs/searchindex.js b/docs/searchindex.js index 0ec792f..bff421d 100644 --- a/docs/searchindex.js +++ b/docs/searchindex.js @@ -1 +1 @@ -Search.setIndex({"docnames": ["advanced/index", "benchmark/ad_benchmark_autoencodermlp_256", "benchmark/fd_benchmark_mlp_256", "benchmark/hi_benchmark_mlp_256", "benchmark/index", "benchmark/rul_benchmark_lstm_256", "custom/index", "index", "reference/ice.anomaly_detection", "reference/ice.anomaly_detection.datasets", "reference/ice.anomaly_detection.metrics", "reference/ice.anomaly_detection.models", "reference/ice.anomaly_detection.utils", "reference/ice.base", "reference/ice.configs", "reference/ice.fault_diagnosis", "reference/ice.fault_diagnosis.datasets", "reference/ice.fault_diagnosis.metrics", "reference/ice.fault_diagnosis.models", "reference/ice.fault_diagnosis.utils", "reference/ice.health_index_estimation", "reference/ice.health_index_estimation.datasets", "reference/ice.health_index_estimation.metrics", "reference/ice.health_index_estimation.models", "reference/ice.health_index_estimation.utils", "reference/ice.remaining_useful_life_estimation", "reference/ice.remaining_useful_life_estimation.datasets", "reference/ice.remaining_useful_life_estimation.metrics", "reference/ice.remaining_useful_life_estimation.models", "reference/ice.remaining_useful_life_estimation.utils", "reference/index", "start/index", "start/installation", "start/tasks/index", "start/tasks/task_ad", "start/tasks/task_fd", "start/tasks/task_hi", "start/tasks/task_rul", "start/tutorials/anomaly_detection_basic", "start/tutorials/fault_diagnosis_basic", "start/tutorials/health_index_estimation_basic", "start/tutorials/index", "start/tutorials/remaining_useful_life_estimation_basic"], "filenames": ["advanced\\index.rst", "benchmark\\ad_benchmark_autoencodermlp_256.ipynb", "benchmark\\fd_benchmark_mlp_256.ipynb", "benchmark\\hi_benchmark_mlp_256.ipynb", "benchmark\\index.rst", "benchmark\\rul_benchmark_lstm_256.ipynb", "custom\\index.rst", "index.rst", "reference\\ice.anomaly_detection.rst", "reference\\ice.anomaly_detection.datasets.rst", "reference\\ice.anomaly_detection.metrics.rst", "reference\\ice.anomaly_detection.models.rst", "reference\\ice.anomaly_detection.utils.rst", "reference\\ice.base.rst", "reference\\ice.configs.rst", "reference\\ice.fault_diagnosis.rst", "reference\\ice.fault_diagnosis.datasets.rst", "reference\\ice.fault_diagnosis.metrics.rst", "reference\\ice.fault_diagnosis.models.rst", "reference\\ice.fault_diagnosis.utils.rst", "reference\\ice.health_index_estimation.rst", "reference\\ice.health_index_estimation.datasets.rst", "reference\\ice.health_index_estimation.metrics.rst", "reference\\ice.health_index_estimation.models.rst", "reference\\ice.health_index_estimation.utils.rst", "reference\\ice.remaining_useful_life_estimation.rst", "reference\\ice.remaining_useful_life_estimation.datasets.rst", "reference\\ice.remaining_useful_life_estimation.metrics.rst", "reference\\ice.remaining_useful_life_estimation.models.rst", "reference\\ice.remaining_useful_life_estimation.utils.rst", "reference\\index.rst", "start\\index.rst", "start\\installation.rst", "start\\tasks\\index.rst", "start\\tasks\\task_ad.rst", "start\\tasks\\task_fd.rst", "start\\tasks\\task_hi.rst", "start\\tasks\\task_rul.rst", "start\\tutorials\\anomaly_detection_basic.ipynb", "start\\tutorials\\fault_diagnosis_basic.ipynb", "start\\tutorials\\health_index_estimation_basic.ipynb", "start\\tutorials\\index.rst", "start\\tutorials\\remaining_useful_life_estimation_basic.ipynb"], "titles": ["Advanced materials", "Results of anomaly detection using AutoEncoderMLP-256", "Results of fault diagnosis using MLP-256", "Results of HI estimation using MLP-256", "Benchmarking", "Results of RUL estimation using lstm-256", "Custom classes", "ICE - Industrial Condition Evaluation", "Anomaly detection", "Anomaly detection datasets", "Anomaly detection metrics", "Anomaly detection models", "Anomaly detection utils", "Base modules", "Configs", "Fault diagnosis", "Fault diagnosis datasets", "Fault diagnosis metrics", "Fault diagnosis models", "Fault diagnosis utils", "Health index estimation", "Health index estimation datasets", "Health index estimation metrics", "Remaining useful life models", "Health index estimation utils", "Remaining useful life", "Remaining useful life datasets", "Remaining useful life metrics", "Health index models", "Remaining useful life utils", "API reference", "Getting started", "Installation", "Tasks", "Anomaly detection", "Fault diagnosis", "Health index", "Remaining useful lifetime", "Tutorial on anomaly detection task", "Tutorial on fault diagnosis task", "Tutorial on health index estimation task", "Quick tutorials", "Tutorial on remaining useful lifetime estimation task"], "terms": {"2nd": [0, 6], "project": [0, 6], "stage": [0, 6], "problem": [0, 6, 36, 37], "thi": [1, 2, 3, 5, 9, 11, 13, 16, 21, 26, 34, 36], "notebook": [1, 2, 3, 5], "present": [1, 2, 3, 5, 34, 35, 36, 37], "experiment": [1, 2, 3, 5], "tennesse": [1, 2, 3, 5, 9, 16, 18, 34, 35], "eastman": [1, 2, 3, 5, 9, 16, 18, 34, 35], "process": [1, 2, 3, 5, 9, 16, 18, 34, 35, 36], "dataset": [1, 2, 3, 5, 8, 12, 13, 15, 19, 20, 24, 25, 27, 29, 30, 38, 39, 40, 42], "model": [1, 2, 3, 5, 7, 8, 13, 15, 20, 25, 26, 30, 35, 36, 37, 38, 39, 40, 42], "import": [1, 2, 3, 5, 34, 35, 38, 39, 40, 42], "librari": [1, 2, 3, 5, 7, 32, 35, 36, 40, 42], "numpi": [1, 2, 3, 5, 22, 27, 38, 39, 40, 42], "np": [1, 2, 3, 5, 17, 38, 39, 40, 42], "torch": [1, 2, 3, 5, 13, 40, 42], "from": [1, 2, 3, 5, 13, 14, 21, 26, 27, 34, 35, 36, 37, 38, 39, 40, 42], "sklearn": [1, 2, 38, 39], "preprocess": [1, 2, 21, 26, 38, 39], "standardscal": [1, 2, 38, 39], "tqdm": [1, 2, 3, 5, 39, 40, 42], "auto": [1, 2, 3, 5, 35, 39, 40, 42], "trang": [1, 2, 3, 5, 40, 42], "ic": [1, 2, 3, 5, 9, 10, 11, 12, 13, 14, 16, 17, 18, 19, 21, 22, 23, 24, 26, 27, 28, 29, 32, 35, 38, 39, 40, 42], "anomaly_detect": [1, 9, 10, 11, 12, 38], "anomalydetectionreinartztep": [1, 8, 9], "anomalydetectionsmalltep": [1, 8, 9, 38], "download": [1, 2, 9, 13, 16, 21, 26, 32, 38, 39], "tep": [1, 2, 34], "train": [1, 2, 3, 5, 7, 11, 13, 18, 23, 28, 34, 35, 36, 38, 39, 40, 42], "calcul": [1, 2, 11, 27, 34, 35], "metric": [1, 2, 3, 5, 8, 11, 13, 15, 18, 20, 23, 25, 28, 30, 38, 39, 40, 42], "i": [1, 2, 3, 5, 10, 11, 13, 17, 18, 23, 28, 34, 35, 36, 37, 40, 42], "5": [1, 2, 3, 5, 18, 34, 35, 36, 37, 38, 39, 40, 42], "random": [1, 2, 3, 5, 40, 42], "manual_se": [1, 2, 3, 5, 40, 42], "window_s": [1, 2, 3, 5, 11, 12, 18, 19, 23, 24, 28, 29, 38, 39, 40, 42], "32": [1, 3, 5, 28, 35, 37, 40, 42], "batch_siz": [1, 2, 3, 5, 11, 13, 18, 23, 28, 40, 42], "512": [1, 38, 39], "lr": [1, 2, 3, 5, 11, 13, 18, 23, 28, 38, 39, 40, 42], "0": [1, 2, 3, 5, 11, 18, 34, 35, 36, 37, 38, 39, 40, 42], "001": [1, 2, 11, 18, 35, 38, 39], "num_epoch": [1, 2, 3, 5, 11, 13, 18, 23, 28, 38, 40, 42], "20": [1, 2, 3, 5, 35, 37, 38, 39, 40, 42], "verbos": [1, 2, 3, 5, 11, 13, 18, 23, 28, 38, 39, 40, 42], "fals": [1, 2, 5, 9, 11, 13, 16, 17, 18, 21, 26, 34, 35, 42], "devic": [1, 2, 3, 5, 11, 13, 18, 23, 28, 34, 36, 40, 42], "cuda": [1, 2, 3, 5, 11, 13, 18, 23, 28, 40, 42], "fit": [1, 2, 3, 5, 11, 13, 38, 39, 40, 42], "df": [1, 2, 3, 5, 11, 12, 13, 18, 19, 23, 24, 28, 29, 38, 39, 40, 42], "train_mask": [1, 2, 38, 39], "append": [1, 2, 3, 5, 40, 42], "evalu": [1, 2, 3, 5, 9, 11, 13, 16, 18, 23, 28, 34, 35, 36, 37, 38, 39, 40, 42], "test_mask": [1, 2, 38, 39], "target": [1, 2, 3, 5, 10, 11, 12, 13, 17, 18, 19, 22, 23, 24, 27, 28, 29, 35, 37, 38, 39, 40, 42], "print": [1, 2, 3, 5, 40, 42], "acc": [1, 2], "tpr": [1, 2, 34], "fpr": [1, 2, 34], "metrics_i": [1, 2, 3, 5, 40, 42], "accuraci": [1, 2, 8, 10, 11, 15, 17, 18, 34, 35, 38, 39], "true_positive_r": [1, 2, 15, 17, 38], "false_positive_r": [1, 2, 15, 17, 38], "tpr_mean": [1, 2], "tpr_std": [1, 2], "mean": [1, 2, 3, 5, 22, 27, 40, 42], "axi": [1, 2, 3, 5, 40, 42], "std": [1, 2, 3, 5, 40, 42], "fpr_mean": [1, 2], "fpr_std": [1, 2], "f": [1, 2, 3, 5, 34, 35, 37, 40, 42], "4f": [1, 2], "2": [1, 2, 3, 5, 18, 34, 35, 36, 37, 38, 39, 40, 42], "rang": [1, 2, 34], "len": [1, 2], "8028": 1, "0007": [1, 2], "7378": 1, "0022": 1, "0366": 1, "0034": 1, "fault_diagnosi": [2, 16, 17, 18, 19, 39], "faultdiagnosisreinartztep": [2, 15, 16], "standard": [2, 37], "input": [2, 18, 23, 28, 36], "data": [2, 3, 5, 7, 9, 11, 13, 16, 18, 21, 23, 26, 28, 34, 35, 36, 37, 38, 39, 40, 42], "scaler": [2, 38, 39], "fit_transform": [2, 38, 39], "transform": [2, 34, 36, 37, 38, 39], "10": [2, 3, 9, 11, 16, 18, 26, 27, 34, 35, 37, 38, 39, 40], "hidden_dim": [2, 11, 18, 23, 28], "128": [2, 3, 11, 18, 38, 40], "cdr": [2, 35], "correct_daignosis_r": [2, 15, 17], "1": [2, 3, 5, 24, 28, 29, 38, 39, 40, 42], "8500": 2, "0029": 2, "9582": 2, "0013": 2, "9948": 2, "0039": 2, "0000": 2, "9951": 2, "0008": 2, "3": [2, 3, 5, 34, 35, 37, 38, 39, 40, 42], "9457": 2, "0487": 2, "4": [2, 3, 5, 18, 34, 35, 37, 38, 39, 40, 42], "9963": 2, "0020": 2, "2423": [2, 3, 40], "0490": 2, "0009": 2, "0004": 2, "6": [2, 3, 34, 35, 36, 37, 38, 39, 40], "9984": 2, "0002": 2, "7": [2, 3, 34, 35, 37, 38, 39, 40], "9978": 2, "0001": 2, "8": [2, 3, 34, 35, 36, 37, 38, 39, 40], "9782": 2, "0136": 2, "9": [2, 3, 34, 35, 37, 38, 39, 40], "3068": 2, "1355": 2, "9754": 2, "0018": 2, "11": [2, 3, 34, 35, 37, 39, 40], "9539": 2, "0130": 2, "12": [2, 3, 35, 37, 40], "9098": 2, "0086": 2, "13": [2, 3, 35, 37, 39, 40], "9702": 2, "0074": 2, "14": [2, 3, 5, 35, 37, 40, 42], "9949": 2, "0006": 2, "15": [2, 3, 35, 40], "16": [2, 3, 35, 36, 37, 38, 39, 40], "2772": 2, "0309": 2, "17": [2, 3, 35, 36, 37, 38, 39, 40], "9786": 2, "18": [2, 3, 35, 37, 38, 39, 40], "9635": 2, "0019": 2, "19": [2, 3, 5, 34, 35, 37, 39, 40, 42], "9918": 2, "9739": 2, "21": [2, 3, 5, 37, 38, 39, 40, 42], "22": [2, 3, 35, 38, 39, 40], "5031": 2, "23": [2, 3, 18, 35, 37, 39, 40], "8086": 2, "0242": 2, "24": [2, 3, 38, 39, 40], "9890": 2, "0011": 2, "25": [2, 3, 37, 40], "9872": 2, "26": [2, 3, 35, 38, 39, 40], "9696": 2, "0064": 2, "27": [2, 3, 39, 40], "9742": 2, "0111": 2, "28": [2, 3, 35, 38, 39, 40], "3768": [2, 3, 40], "0459": 2, "anomali": [3, 4, 5, 16, 30, 31, 33, 35, 41], "detect": [3, 4, 5, 16, 17, 18, 30, 31, 33, 35, 41], "autoencodermlp": [3, 4, 5, 8, 38], "health_index_estim": [3, 21, 22, 23, 24, 40], "mill": [3, 20, 21, 40], "panda": [3, 5, 11, 13, 18, 23, 28, 40, 42], "pd": [3, 5, 11, 40, 42], "initi": [3, 5, 11, 13, 40, 42], "class": [3, 5, 7, 9, 11, 12, 13, 14, 16, 18, 19, 21, 23, 24, 26, 28, 29, 34, 40, 42], "test": [3, 5, 9, 13, 16, 21, 26, 31, 34, 35, 37, 38, 39, 40, 42], "split": [3, 5, 38, 39, 40, 42], "dataset_class": [3, 5, 40, 42], "test_data": [3, 5, 40, 42], "test_target": [3, 5, 40, 42], "model_class": [3, 5, 40, 42], "64": [3, 5, 11, 23, 35, 37, 38, 39, 40, 42], "stride": [3, 23, 24, 28, 29, 40], "300": [3, 23, 40], "258": [3, 38, 39, 40], "0004999805761074147": [3, 40], "30": [3, 35, 37, 39, 40], "true": [3, 9, 10, 11, 13, 16, 17, 18, 21, 23, 26, 27, 28, 34, 35, 36, 37, 38, 39, 40], "concat": [3, 40], "00": [3, 5, 35, 39, 40, 42], "": [3, 5, 13, 14, 34, 35, 36, 37, 39, 40, 42], "creat": [3, 5, 7, 13, 14, 34, 35, 38, 39, 40, 42], "sequenc": [3, 5, 11, 13, 14, 18, 23, 28, 35, 39, 40, 42], "sampl": [3, 5, 11, 13, 17, 18, 23, 28, 34, 36, 37, 38, 39, 40, 42], "100": [3, 5, 35, 39, 40, 42], "107": [3, 40], "3242": [3, 40], "35it": [3, 40], "epoch": [3, 11, 13, 18, 23, 28, 38, 39, 40], "A": [3, 9, 11, 13, 14, 16, 18, 23, 26, 28, 34, 35, 36, 37, 40], "step": [3, 36, 40], "04": [3, 35, 40], "52": [3, 35, 38, 39, 40], "38": [3, 5, 38, 39, 40, 42], "08": [3, 35, 37, 39, 40], "48": [3, 38, 39, 40], "42": [3, 37, 38, 39, 40], "44": [3, 37, 38, 39, 40], "47": [3, 5, 37, 38, 39, 40, 42], "31": [3, 35, 40], "40": [3, 5, 37, 38, 39, 40, 42], "50": [3, 23, 28, 35, 39, 40], "35": [3, 5, 34, 37, 38, 39, 40, 42], "46": [3, 5, 38, 39, 40, 42], "45": [3, 35, 37, 38, 39, 40], "54": [3, 5, 35, 38, 39, 40, 42], "62": [3, 38, 39, 40], "69": [3, 40], "77": [3, 34, 39, 40], "85": [3, 37, 40], "49": [3, 40], "09": [3, 5, 40, 42], "92": [3, 35, 37, 40], "53": [3, 38, 39, 40], "55": [3, 40], "75": [3, 35, 37, 38, 39, 40], "57": [3, 34, 35, 39, 40], "78": [3, 37, 40], "loss": [3, 35, 38, 39, 40], "4360": [3, 40], "43": [3, 37, 38, 39, 40], "39": [3, 35, 38, 39, 40], "33": [3, 5, 34, 35, 37, 38, 39, 40, 42], "37": [3, 34, 37, 38, 39, 40], "91": [3, 34, 35, 40], "41": [3, 35, 38, 39, 40], "82": [3, 34, 35, 39, 40], "07": [3, 35, 40], "80": [3, 5, 35, 39, 40, 42], "03": [3, 5, 35, 39, 40, 42], "86": [3, 39, 40], "01": [3, 5, 35, 39, 40, 42], "6135": [3, 40], "67": [3, 40], "93": [3, 40], "29": [3, 5, 37, 40, 42], "34": [3, 35, 38, 39, 40], "51": [3, 35, 37, 40], "02": [3, 5, 35, 39, 40, 42], "36": [3, 38, 39, 40], "1503": [3, 40], "05": [3, 23, 28, 37, 39, 40], "99": [3, 35, 40], "5825": [3, 40], "89": [3, 34, 35, 40], "90": [3, 35, 39, 40], "4649": [3, 40], "71": [3, 40], "70": [3, 5, 39, 40, 42], "74": [3, 35, 38, 39, 40], "66": [3, 40], "63": [3, 38, 39, 40], "59": [3, 35, 37, 38, 39, 40], "60": [3, 5, 35, 38, 39, 40, 42], "58": [3, 35, 36, 37, 39, 40], "97": [3, 35, 40], "2802": [3, 40], "06": [3, 5, 35, 37, 40, 42], "73": [3, 40], "56": [3, 5, 35, 37, 40, 42], "2823": [3, 40], "2896": [3, 40], "2463": [3, 40], "96": [3, 35, 37, 40], "84": [3, 34, 40], "1831": [3, 40], "2223": [3, 40], "68": [3, 40], "94": [3, 34, 35, 37, 40], "2076": [3, 40], "61": [3, 37, 38, 39, 40], "1645": [3, 40], "98": [3, 35, 40], "95": [3, 11, 35, 37, 40], "1910": [3, 40], "1895": [3, 40], "1853": [3, 40], "1559": [3, 40], "81": [3, 34, 35, 39, 40], "79": [3, 39, 40], "1705": [3, 40], "1646": [3, 40], "1666": [3, 40], "1820": [3, 40], "1460": [3, 40], "1814": [3, 40], "1484": [3, 40], "83": [3, 40], "1506": [3, 40], "87": [3, 35, 37, 39, 40], "1427": [3, 40], "1213": [3, 40], "1406": [3, 40], "88": [3, 5, 35, 39, 40, 42], "1382": [3, 40], "1606": [3, 40], "9334": [3, 40], "01it": [3, 40], "5012": [3, 40], "178": [3, 40], "22it": [3, 40], "186": [3, 40], "61it": [3, 40], "169": [3, 37, 40], "80it": [3, 40], "163": [3, 40], "92it": [3, 40], "160": [3, 40], "10it": [3, 5, 40, 42], "111": [3, 40], "167": [3, 40], "89it": [3, 39, 40], "132": [3, 40], "180": [3, 40], "28it": [3, 40], "153": [3, 40], "188": [3, 40], "18it": [3, 40], "174": [3, 40], "192": [3, 40], "44it": [3, 40], "194": [3, 38, 39, 40], "70it": [3, 40], "214": [3, 40], "190": [3, 40], "58it": [3, 39, 40], "234": [3, 40], "182": [3, 40], "64it": [3, 40], "253": [3, 40], "175": [3, 40], "271": [3, 40], "172": [3, 40], "75it": [3, 5, 40, 42], "292": [3, 40], "15it": [3, 40], "313": [3, 40], "46it": [3, 40], "334": [3, 40], "98it": [3, 40], "354": [3, 40], "45it": [3, 40], "375": [3, 40], "197": [3, 40], "83it": [3, 5, 40, 42], "396": [3, 40], "199": [3, 38, 39, 40], "06it": [3, 40], "416": [3, 40], "78it": [3, 40], "436": [3, 40], "53it": [3, 40], "455": [3, 40], "88it": [3, 40], "475": [3, 40], "27it": [3, 5, 40, 42], "496": [3, 40], "187": [3, 40], "517": [3, 40], "538": [3, 40], "196": [3, 40], "559": [3, 40], "03it": [3, 40], "579": [3, 40], "195": [3, 40], "33it": [3, 40], "599": [3, 40], "618": [3, 40], "36it": [3, 40], "636": [3, 40], "166": [3, 40], "81it": [3, 40], "655": [3, 37, 38, 39, 40], "60it": [3, 40], "674": [3, 40], "176": [3, 38, 39, 40], "93it": [3, 40], "693": [3, 40], "11it": [3, 40], "711": [3, 40], "173": [3, 40], "729": [3, 40], "66it": [3, 40], "747": [3, 40], "765": [3, 38, 39, 40], "21it": [3, 40], "783": [3, 40], "165": [3, 40], "34it": [3, 40], "800": [3, 40], "158": [3, 40], "818": [3, 40], "162": [3, 40], "74it": [3, 40], "836": [3, 40], "853": [3, 40], "04it": [3, 40], "871": [3, 40], "170": [3, 40], "05it": [3, 5, 40, 42], "889": [3, 38, 39, 40], "96it": [3, 40], "907": [3, 40], "924": [3, 40], "14it": [3, 39, 40], "944": [3, 40], "171": [3, 40], "966": [3, 40], "183": [3, 40], "41it": [3, 40], "989": [3, 40], "1012": [3, 40], "203": [3, 40], "1035": [3, 40], "208": [3, 40], "1057": [3, 40], "210": [3, 40], "84it": [3, 40], "1079": [3, 40], "205": [3, 40], "1102": [3, 40], "209": [3, 40], "1125": [3, 40], "213": [3, 37, 40], "19it": [3, 40], "1147": [3, 40], "212": [3, 37, 40], "1170": [3, 40], "215": [3, 40], "1192": [3, 40], "1214": [3, 40], "1237": [3, 40], "216": [3, 40], "73it": [3, 40], "1259": [3, 40], "79it": [3, 40], "1281": [3, 40], "85it": [3, 40], "1303": [3, 40], "42it": [3, 40], "1323": [3, 40], "54it": [3, 40], "1343": [3, 40], "1363": [3, 40], "91it": [3, 40], "1383": [3, 40], "1403": [3, 40], "191": [3, 40], "1423": [3, 40], "1444": [3, 40], "1464": [3, 40], "57it": [3, 40], "1504": [3, 40], "16it": [3, 39, 40], "1524": [3, 40], "25it": [3, 40], "1545": [3, 40], "193": [3, 40], "71it": [3, 40], "1567": [3, 40], "02it": [3, 40], "1590": [3, 40], "207": [3, 40], "1612": [3, 40], "1635": [3, 40], "1657": [3, 40], "31it": [3, 40], "1679": [3, 40], "1701": [3, 40], "200": [3, 40], "1722": [3, 40], "1743": [3, 40], "1766": [3, 40], "07it": [3, 40], "1789": [3, 40], "211": [3, 37, 40], "1812": [3, 40], "1835": [3, 40], "63it": [3, 40], "1857": [3, 40], "1879": [3, 40], "1901": [3, 40], "1923": [3, 40], "62it": [3, 40], "1945": [3, 40], "1967": [3, 40], "1989": [3, 40], "86it": [3, 40], "2011": [3, 40], "2032": [3, 40], "201": [3, 40], "55it": [3, 40], "2053": [3, 40], "2075": [3, 40], "2099": [3, 40], "2121": [3, 40], "69it": [3, 40], "2144": [3, 40], "2166": [3, 40], "204": [3, 40], "2187": [3, 40], "2209": [3, 40], "95it": [3, 40], "2232": [3, 40], "2255": [3, 40], "217": [3, 38, 39, 40], "59it": [3, 40], "2278": [3, 40], "218": [3, 40], "2300": [3, 40], "2322": [3, 40], "2344": [3, 40], "65it": [3, 40], "2366": [3, 40], "67it": [3, 40], "2388": [3, 40], "2410": [3, 40], "2432": [3, 40], "2454": [3, 40], "50it": [3, 40], "2475": [3, 40], "2498": [3, 40], "2520": [3, 40], "09it": [3, 40], "2543": [3, 40], "2565": [3, 40], "2587": [3, 40], "2609": [3, 40], "43it": [3, 40], "2630": [3, 40], "13it": [3, 40], "2652": [3, 40], "2673": [3, 40], "2695": [3, 40], "2716": [3, 40], "202": [3, 40], "52it": [3, 40], "2737": [3, 40], "2760": [3, 40], "2782": [3, 40], "2804": [3, 40], "77it": [3, 40], "2827": [3, 40], "38it": [3, 40], "2849": [3, 40], "2872": [3, 40], "2894": [3, 40], "2916": [3, 40], "2938": [3, 40], "2960": [3, 40], "2982": [3, 40], "3004": [3, 40], "3027": [3, 40], "3050": [3, 40], "3072": [3, 40], "3094": [3, 40], "3116": [3, 40], "3138": [3, 40], "3159": [3, 40], "3180": [3, 40], "3202": [3, 40], "3224": [3, 40], "65": [3, 40], "3247": [3, 40], "3269": [3, 40], "3291": [3, 40], "3314": [3, 40], "3336": [3, 40], "3358": [3, 40], "3380": [3, 40], "3403": [3, 40], "3425": [3, 40], "32it": [3, 5, 40, 42], "3447": [3, 40], "3468": [3, 40], "3490": [3, 40], "3512": [3, 40], "3534": [3, 40], "3557": [3, 40], "24it": [3, 40], "3579": [3, 40], "97it": [3, 40], "72": [3, 40], "3601": [3, 40], "3623": [3, 40], "72it": [3, 40], "3645": [3, 40], "3666": [3, 40], "198": [3, 40], "3686": [3, 40], "40it": [3, 40], "3707": [3, 40], "3728": [3, 40], "3749": [3, 40], "3770": [3, 40], "12it": [3, 40], "76": [3, 37, 40], "3792": [3, 40], "3814": [3, 40], "3838": [3, 40], "3862": [3, 40], "219": [3, 40], "3886": [3, 40], "222": [3, 37, 40], "99it": [3, 40], "3909": [3, 40], "221": [3, 37, 40], "3932": [3, 40], "220": [3, 40], "3955": [3, 40], "3977": [3, 40], "3999": [3, 40], "4022": [3, 40], "4044": [3, 40], "4066": [3, 40], "26it": [3, 40], "4088": [3, 40], "4111": [3, 40], "4133": [3, 40], "4155": [3, 40], "4177": [3, 40], "4199": [3, 40], "4220": [3, 40], "4240": [3, 40], "00it": [3, 40], "4259": [3, 40], "185": [3, 40], "4278": [3, 40], "56it": [3, 40], "4297": [3, 40], "184": [3, 40], "4317": [3, 40], "4337": [3, 40], "87it": [3, 40], "4358": [3, 40], "30it": [3, 40], "4379": [3, 40], "4401": [3, 40], "4422": [3, 40], "4445": [3, 40], "206": [3, 40], "90it": [3, 40], "4468": [3, 40], "39it": [3, 39, 40], "4490": [3, 40], "37it": [3, 40], "4512": [3, 38, 39, 40], "4534": [3, 40], "4556": [3, 38, 39, 40], "94it": [3, 40], "4579": [3, 40], "4601": [3, 39, 40], "4622": [3, 40], "4643": [3, 40], "4663": [3, 40], "189": [3, 40], "4683": [3, 40], "51it": [3, 5, 40, 42], "4705": [3, 40], "4725": [3, 40], "4746": [3, 40], "4766": [3, 40], "4786": [3, 40], "4805": [3, 40], "4825": [3, 40], "4844": [3, 40], "4865": [3, 40], "4887": [3, 40], "08it": [3, 40], "4909": [3, 40], "4930": [3, 40], "4951": [3, 40], "4973": [3, 40], "4995": [3, 40], "1548": [3, 40], "5213": [3, 40], "4789": [3, 40], "4105": [3, 40], "3222": [3, 40], "3484": [3, 40], "2422": [3, 40], "2523": [3, 40], "1962": [3, 40], "1405": [3, 40], "1468": [3, 40], "1552": [3, 40], "1541": [3, 40], "1692": [3, 40], "2091": [3, 40], "1865": [3, 40], "1463": [3, 40], "1321": [3, 40], "1577": [3, 40], "1436": [3, 40], "1741": [3, 40], "1224": [3, 40], "1318": [3, 40], "1435": [3, 40], "1441": [3, 40], "1509": [3, 40], "1437": [3, 40], "1411": [3, 40], "9369": [3, 34, 40], "76it": [3, 40], "134": [3, 40], "141": [3, 40], "142": [3, 37, 38, 39, 40], "131": [3, 40], "130": [3, 40], "103": [3, 40], "119": [3, 40], "139": [3, 40], "149": [3, 9, 16, 34, 35, 40], "164": [3, 40], "143": [3, 40], "140": [3, 40], "47it": [3, 40], "225": [3, 38, 39, 40], "240": [3, 40], "255": [3, 40], "270": [3, 40], "286": [3, 40], "145": [3, 40], "301": [3, 40], "138": [3, 40], "315": [3, 40], "133": [3, 40], "329": [3, 40], "127": [3, 40], "342": [3, 40], "124": [3, 40], "355": [3, 40], "121": [3, 40], "368": [3, 40], "120": [3, 38, 39, 40], "381": [3, 40], "118": [3, 40], "394": [3, 38, 39, 40], "407": [3, 40], "420": [3, 40], "432": [3, 40], "117": [3, 40], "444": [3, 40], "456": [3, 40], "116": [3, 38, 39, 40], "468": [3, 38, 39, 40], "480": [3, 40], "492": [3, 40], "114": [3, 40], "505": [3, 40], "20it": [3, 40], "529": [3, 40], "541": [3, 38, 39, 40], "553": [3, 40], "566": [3, 40], "593": [3, 40], "606": [3, 40], "619": [3, 40], "122": [3, 40], "632": [3, 40], "123": [3, 40], "645": [3, 38, 39, 40], "658": [3, 38, 39, 40], "670": [3, 40], "682": [3, 40], "694": [3, 40], "707": [3, 40], "719": [3, 40], "48it": [3, 40], "731": [3, 40], "745": [3, 40], "760": [3, 40], "773": [3, 40], "787": [3, 40], "126": [3, 40], "802": [3, 40], "817": [3, 40], "136": [3, 40], "832": [3, 40], "847": [3, 40], "862": [3, 40], "876": [3, 40], "137": [3, 37, 38, 39, 40], "890": [3, 40], "904": [3, 40], "917": [3, 40], "125": [3, 37, 40], "930": [3, 40], "943": [3, 40], "956": [3, 38, 39, 40], "968": [3, 40], "980": [3, 40], "992": [3, 40], "1005": [3, 40], "1018": [3, 40], "1031": [3, 40], "1043": [3, 40], "1055": [3, 40], "1068": [3, 40], "1083": [3, 40], "1098": [3, 40], "1112": [3, 40], "1126": [3, 40], "1141": [3, 40], "135": [3, 40], "1156": [3, 40], "23it": [3, 40], "1171": [3, 40], "1186": [3, 40], "1201": [3, 40], "1216": [3, 40], "1231": [3, 40], "17it": [3, 40], "1246": [3, 40], "1261": [3, 40], "1276": [3, 40], "1291": [3, 40], "1306": [3, 40], "1336": [3, 40], "1351": [3, 40], "1366": [3, 40], "1381": [3, 40], "144": [3, 38, 39, 40], "1396": [3, 40], "1442": [3, 40], "1457": [3, 40], "1472": [3, 40], "1487": [3, 40], "1502": [3, 40], "49it": [3, 40], "1517": [3, 40], "1531": [3, 40], "1574": [3, 40], "1589": [3, 40], "1604": [3, 40], "1619": [3, 40], "1634": [3, 40], "1649": [3, 40], "68it": [3, 40], "1664": [3, 40], "1694": [3, 40], "1709": [3, 40], "1723": [3, 40], "1737": [3, 40], "1752": [3, 40], "1767": [3, 40], "1782": [3, 40], "1797": [3, 40], "1811": [3, 40], "1825": [3, 40], "29it": [3, 40], "1839": [3, 40], "1866": [3, 40], "1892": [3, 40], "1904": [3, 40], "1916": [3, 40], "1928": [3, 40], "1941": [3, 40], "1956": [3, 40], "1969": [3, 40], "1982": [3, 40], "1995": [3, 40], "2010": [3, 40], "2025": [3, 40], "2040": [3, 40], "2055": [3, 40], "2069": [3, 40], "2083": [3, 40], "2098": [3, 40], "2112": [3, 40], "2128": [3, 40], "2143": [3, 40], "2159": [3, 40], "146": [3, 40], "2175": [3, 40], "150": [3, 40], "2191": [3, 40], "148": [3, 40], "2206": [3, 40], "147": [3, 40], "2221": [3, 40], "2237": [3, 40], "2252": [3, 40], "2268": [3, 40], "2283": [3, 40], "2298": [3, 40], "2313": [3, 40], "2329": [3, 40], "2359": [3, 40], "2374": [3, 40], "2389": [3, 40], "2404": [3, 40], "2419": [3, 40], "2434": [3, 40], "2449": [3, 40], "2464": [3, 40], "2479": [3, 40], "2495": [3, 40], "2510": [3, 40], "2525": [3, 40], "2540": [3, 40], "2555": [3, 40], "2570": [3, 40], "2585": [3, 40], "2600": [3, 40], "2615": [3, 40], "2645": [3, 40], "2660": [3, 40], "2675": [3, 40], "2690": [3, 40], "2705": [3, 38, 39, 40], "2720": [3, 40], "2735": [3, 40], "82it": [3, 40], "2751": [3, 40], "2766": [3, 40], "2797": [3, 40], "2812": [3, 40], "2842": [3, 40], "2857": [3, 40], "2887": [3, 40], "2903": [3, 40], "2918": [3, 40], "2934": [3, 40], "2952": [3, 40], "155": [3, 40], "2968": [3, 40], "152": [3, 40], "2984": [3, 40], "3000": [3, 40], "3015": [3, 40], "3030": [3, 40], "3045": [3, 40], "3059": [3, 40], "3087": [3, 40], "3101": [3, 40], "3115": [3, 40], "3128": [3, 40], "3141": [3, 40], "3154": [3, 40], "3168": [3, 40], "3181": [3, 40], "3195": [3, 40], "3210": [3, 40], "3225": [3, 40], "3241": [3, 40], "3256": [3, 40], "3271": [3, 40], "3287": [3, 40], "3302": [3, 40], "3317": [3, 40], "3332": [3, 40], "3347": [3, 40], "3361": [3, 40], "3375": [3, 40], "3389": [3, 40], "3417": [3, 40], "3431": [3, 40], "3446": [3, 40], "3461": [3, 40], "3475": [3, 40], "3489": [3, 40], "3503": [3, 40], "3517": [3, 40], "3531": [3, 40], "3546": [3, 40], "3561": [3, 40], "3576": [3, 40], "3591": [3, 40], "3606": [3, 40], "3621": [3, 40], "3636": [3, 40], "3651": [3, 40], "3665": [3, 40], "129": [3, 40], "3679": [3, 38, 39, 40], "3692": [3, 38, 39, 40], "3705": [3, 40], "3718": [3, 40], "3731": [3, 40], "3744": [3, 40], "3759": [3, 40], "3773": [3, 40], "3786": [3, 40], "3799": [3, 40], "3812": [3, 40], "3825": [3, 40], "3839": [3, 40], "3852": [3, 40], "3865": [3, 40], "3878": [3, 40], "3892": [3, 40], "3905": [3, 40], "3918": [3, 40], "3931": [3, 40], "3944": [3, 40], "3957": [3, 40], "3970": [3, 40], "3983": [3, 40], "3996": [3, 40], "4010": [3, 40], "4023": [3, 40], "4036": [3, 40], "4049": [3, 40], "4062": [3, 40], "4075": [3, 40], "4101": [3, 40], "4114": [3, 40], "4129": [3, 40], "4144": [3, 40], "4159": [3, 40], "4175": [3, 40], "4190": [3, 40], "4205": [3, 40], "4221": [3, 40], "4236": [3, 40], "4251": [3, 40], "4266": [3, 40], "4281": [3, 40], "4299": [3, 40], "154": [3, 40], "4316": [3, 40], "159": [3, 40], "4332": [3, 40], "4348": [3, 40], "4363": [3, 40], "4377": [3, 40], "4390": [3, 40], "4407": [3, 40], "4427": [3, 40], "4451": [3, 40], "4476": [3, 38, 39, 40], "4500": [3, 40], "4522": [3, 40], "4544": [3, 40], "4565": [3, 40], "4584": [3, 40], "4602": [3, 40], "4621": [3, 40], "4639": [3, 40], "4657": [3, 40], "4675": [3, 40], "4693": [3, 40], "4711": [3, 40], "161": [3, 40], "4728": [3, 40], "157": [3, 40], "4744": [3, 40], "4759": [3, 40], "4773": [3, 40], "4787": [3, 40], "4800": [3, 40], "4813": [3, 40], "4826": [3, 40], "4839": [3, 40], "4852": [3, 40], "4878": [3, 40], "4891": [3, 40], "4903": [3, 40], "4916": [3, 40], "4929": [3, 40], "4941": [3, 40], "4953": [3, 40], "4965": [3, 40], "4977": [3, 40], "4991": [3, 40], "5004": [3, 40], "1757": [3, 40], "2985": [3, 40], "1235": [3, 40], "2092": [3, 40], "9648": [3, 40], "5663": [3, 40], "4703": [3, 40], "2911": [3, 40], "2339": [3, 40], "2370": [3, 40], "1976": [3, 40], "2120": [3, 40], "1964": [3, 40], "1561": [3, 40], "1860": [3, 40], "1424": [3, 40], "1837": [3, 40], "1698": [3, 40], "1320": [3, 40], "1764": [3, 40], "1626": [3, 40], "1674": [3, 40], "1252": [3, 40], "1327": [3, 40], "1693": [3, 40], "1379": [3, 37, 40], "1404": [3, 40], "1775": [3, 40], "6999": [3, 40], "113": [3, 40], "112": [3, 40], "108": [3, 40], "115": [3, 40], "247": [3, 40], "260": [3, 40], "273": [3, 40], "285": [3, 40], "297": [3, 40], "309": [3, 40], "321": [3, 40], "336": [3, 40], "349": [3, 40], "362": [3, 40], "387": [3, 40], "399": [3, 40], "411": [3, 40], "424": [3, 40], "449": [3, 40], "461": [3, 38, 39, 40], "473": [3, 40], "485": [3, 40], "497": [3, 40], "509": [3, 40], "521": [3, 40], "533": [3, 40], "545": [3, 40], "557": [3, 40], "571": [3, 40], "584": [3, 40], "596": [3, 40], "608": [3, 40], "620": [3, 40], "644": [3, 40], "656": [3, 40], "668": [3, 40], "681": [3, 40], "109": [3, 40], "705": [3, 40], "110": [3, 40], "717": [3, 38, 39, 40], "741": [3, 40], "753": [3, 40], "777": [3, 39, 40], "790": [3, 40], "804": [3, 40], "829": [3, 40], "841": [3, 40], "865": [3, 40], "877": [3, 40], "901": [3, 40], "915": [3, 40], "928": [3, 40], "940": [3, 40], "952": [3, 40], "964": [3, 40], "976": [3, 40], "988": [3, 40], "1000": [3, 40], "1025": [3, 40], "1037": [3, 40], "1049": [3, 40], "1061": [3, 40], "1073": [3, 40], "1085": [3, 40], "1097": [3, 40], "1109": [3, 26, 27, 40], "1121": [3, 40], "1133": [3, 40], "1160": [3, 40], "1173": [3, 40], "1199": [3, 40], "1211": [3, 40], "1223": [3, 40], "1247": [3, 40], "1260": [3, 40], "1274": [3, 40], "1287": [3, 40], "1300": [3, 40], "1313": [3, 40], "1326": [3, 40], "1339": [3, 40], "1376": [3, 40], "1391": [3, 40], "1417": [3, 40], "1430": [3, 40], "1443": [3, 40], "1456": [3, 40], "1480": [3, 40], "1492": [3, 40], "1519": [3, 40], "1532": [3, 40], "1544": [3, 40], "1556": [3, 40], "1568": [3, 40], "1580": [3, 40], "1592": [3, 40], "1617": [3, 40], "1630": [3, 40], "1642": [3, 40], "1654": [3, 40], "1678": [3, 40], "1690": [3, 40], "1702": [3, 40], "1714": [3, 40], "1726": [3, 40], "1739": [3, 40], "1776": [3, 40], "1788": [3, 40], "1800": [3, 40], "1824": [3, 40], "1836": [3, 40], "1848": [3, 40], "1861": [3, 40], "1874": [3, 40], "1886": [3, 40], "1898": [3, 40], "1935": [3, 40], "1947": [3, 40], "1959": [3, 40], "1972": [3, 40], "1984": [3, 40], "1996": [3, 40], "2008": [3, 26, 27, 37, 40], "2020": [3, 34, 35, 36, 37, 40], "2045": [3, 40], "2057": [3, 40], "2081": [3, 40], "2095": [3, 40], "2108": [3, 40], "2133": [3, 40], "2145": [3, 40], "2157": [3, 40], "2169": [3, 40], "2181": [3, 40], "2193": [3, 40], "2218": [3, 40], "2230": [3, 40], "2242": [3, 40], "2267": [3, 40], "2279": [3, 40], "2291": [3, 40], "2303": [3, 40], "2315": [3, 40], "2327": [3, 40], "2351": [3, 40], "2363": [3, 40], "2375": [3, 40], "2387": [3, 40], "2399": [3, 40], "2411": [3, 40], "2435": [3, 40], "2448": [3, 40], "2460": [3, 40], "2472": [3, 40], "2484": [3, 40], "2496": [3, 40], "2508": [3, 40], "2532": [3, 40], "2544": [3, 40], "2558": [3, 40], "2582": [3, 40], "2594": [3, 40], "2606": [3, 40], "2618": [3, 40], "2643": [3, 40], "2655": [3, 40], "2667": [3, 40], "2681": [3, 40], "2694": [3, 40], "2707": [3, 38, 39, 40], "2719": [3, 40], "2731": [3, 40], "2743": [3, 40], "2755": [3, 40], "2767": [3, 40], "2779": [3, 40], "2793": [3, 40], "2806": [3, 40], "2818": [3, 40], "2830": [3, 40], "2854": [3, 40], "2866": [3, 40], "2878": [3, 40], "2890": [3, 40], "2902": [3, 40], "2915": [3, 40], "2928": [3, 40], "2940": [3, 40], "2964": [3, 40], "2976": [3, 40], "2988": [3, 40], "3012": [3, 40], "3024": [3, 40], "3037": [3, 40], "3049": [3, 40], "3061": [3, 40], "3073": [3, 40], "3085": [3, 40], "3097": [3, 40], "3109": [3, 40], "3121": [3, 40], "3133": [3, 40], "3147": [3, 40], "3160": [3, 40], "3173": [3, 40], "3185": [3, 40], "3198": [3, 40], "3211": [3, 40], "3223": [3, 40], "3235": [3, 40], "3262": [3, 40], "3275": [3, 40], "3288": [3, 40], "3301": [3, 40], "3326": [3, 40], "3338": [3, 40], "3350": [3, 40], "3362": [3, 40], "3388": [3, 40], "3401": [3, 40], "3413": [3, 40], "3437": [3, 40], "3449": [3, 40], "3473": [3, 40], "3485": [3, 40], "3499": [3, 40], "3524": [3, 40], "3536": [3, 40], "3548": [3, 40], "3560": [3, 40], "3572": [3, 40], "3584": [3, 39, 40], "3596": [3, 40], "3608": [3, 40], "3633": [3, 40], "3657": [3, 40], "3669": [3, 40], "3681": [3, 40], "3693": [3, 40], "3717": [3, 38, 39, 40], "3730": [3, 40], "3757": [3, 40], "3782": [3, 40], "3794": [3, 40], "3806": [3, 40], "3818": [3, 40], "3830": [3, 40], "3843": [3, 40], "3855": [3, 40], "3868": [3, 40], "3881": [3, 40], "3893": [3, 40], "3917": [3, 40], "3929": [3, 40], "3941": [3, 39, 40], "3953": [3, 40], "3966": [3, 40], "3978": [3, 40], "3990": [3, 40], "4002": [3, 40], "4014": [3, 40], "4026": [3, 40], "4038": [3, 40], "4050": [3, 40], "4074": [3, 40], "4086": [3, 40], "4098": [3, 40], "4110": [3, 40], "4122": [3, 40], "4134": [3, 40], "4146": [3, 40], "4158": [3, 40], "4170": [3, 40], "4182": [3, 40], "4195": [3, 40], "4208": [3, 40], "4232": [3, 40], "4244": [3, 40], "4256": [3, 40], "4268": [3, 40], "4280": [3, 40], "4292": [3, 40], "4304": [3, 40], "4328": [3, 40], "4340": [3, 40], "4352": [3, 40], "4364": [3, 40], "4376": [3, 40], "4388": [3, 40], "4400": [3, 40], "4412": [3, 40], "4424": [3, 40], "4437": [3, 40], "4449": [3, 40], "4461": [3, 40], "4473": [3, 40], "4486": [3, 40], "4498": [3, 40], "4510": [3, 40], "4547": [3, 40], "4559": [3, 40], "4571": [3, 40], "4583": [3, 40], "4595": [3, 40], "4607": [3, 40], "4619": [3, 40], "4631": [3, 40], "4655": [3, 40], "4669": [3, 40], "4682": [3, 40], "4694": [3, 40], "4706": [3, 40], "4718": [3, 40], "4730": [3, 40], "4742": [3, 40], "4754": [3, 40], "4779": [3, 40], "4792": [3, 40], "4817": [3, 40], "4829": [3, 40], "4841": [3, 40], "4853": [3, 40], "4877": [3, 40], "4889": [3, 40], "4902": [3, 40], "4914": [3, 40], "4926": [3, 40], "4938": [3, 40], "4950": [3, 40], "4962": [3, 40], "4974": [3, 40], "4986": [3, 40], "4998": [3, 40], "5011": [3, 40], "2009": [3, 40], "2431": [3, 40], "9911": [3, 40], "8032": [3, 40], "5517": [3, 40], "4578": [3, 40], "3316": [3, 40], "2741": [3, 40], "1925": [3, 40], "2086": [3, 40], "1990": [3, 40], "1613": [3, 40], "1948": [3, 40], "1823": [3, 40], "1605": [3, 37, 40], "1658": [3, 40], "1288": [3, 40], "1358": [3, 40], "1284": [3, 40], "1371": [3, 37, 40], "1298": [3, 40], "1638": [3, 40], "1539": [3, 40], "1550": [3, 40], "1402": [3, 40], "1176": [3, 40], "9333": [3, 40], "181": [3, 40], "179": [3, 40], "151": [3, 40], "168": [3, 40], "223": [3, 37, 40], "242": [3, 40], "261": [3, 36, 40], "177": [3, 40], "280": [3, 40], "298": [3, 40], "316": [3, 40], "335": [3, 38, 39, 40], "353": [3, 40], "372": [3, 40], "392": [3, 40], "431": [3, 40], "450": [3, 40], "469": [3, 38, 39, 40], "488": [3, 40], "506": [3, 40], "524": [3, 40], "542": [3, 40], "561": [3, 40], "580": [3, 38, 39, 40], "654": [3, 40], "671": [3, 40], "691": [3, 40], "709": [3, 40], "728": [3, 40], "748": [3, 40], "767": [3, 40], "786": [3, 40], "805": [3, 40], "824": [3, 40], "842": [3, 38, 39, 40], "860": [3, 40], "878": [3, 40], "898": [3, 38, 39, 40], "936": [3, 40], "955": [3, 38, 39, 40], "974": [3, 40], "994": [3, 40], "1013": [3, 40], "1088": [3, 40], "1108": [3, 40], "1128": [3, 40], "1166": [3, 40], "1185": [3, 40], "1203": [3, 40], "1221": [3, 40], "1241": [3, 40], "1278": [3, 40], "1297": [3, 40], "1316": [3, 40], "1335": [3, 40], "1354": [3, 40], "1373": [3, 40], "1450": [3, 40], "1470": [3, 40], "1489": [3, 40], "1508": [3, 40], "1527": [3, 40], "1546": [3, 40], "1564": [3, 40], "1583": [3, 40], "1602": [3, 40], "1621": [3, 40], "1640": [3, 40], "1676": [3, 40], "1712": [3, 40], "1729": [3, 40], "1747": [3, 40], "1785": [3, 40], "1804": [3, 40], "1842": [3, 37, 40], "1897": [3, 40], "1914": [3, 40], "1934": [3, 40], "1954": [3, 40], "1973": [3, 40], "1993": [3, 40], "2013": [3, 40], "2033": [3, 40], "2073": [3, 40], "2111": [3, 40], "2130": [3, 40], "2149": [3, 40], "2168": [3, 40], "2188": [3, 40], "2208": [3, 40], "2228": [3, 40], "2247": [3, 40], "2265": [3, 40], "2284": [3, 40], "2323": [3, 40], "2343": [3, 40], "2383": [3, 40], "2403": [3, 40], "2441": [3, 40], "2459": [3, 40], "2478": [3, 40], "2516": [3, 40], "2535": [3, 40], "2576": [3, 40], "2596": [3, 40], "2633": [3, 40], "2670": [3, 40], "2709": [3, 40], "2728": [3, 40], "2748": [3, 40], "2768": [3, 40], "2788": [3, 40], "2807": [3, 40], "2825": [3, 40], "2845": [3, 40], "2863": [3, 40], "2883": [3, 40], "2922": [3, 40], "2941": [3, 40], "2979": [3, 40], "2997": [3, 40], "3016": [3, 40], "3034": [3, 40], "3053": [3, 40], "3091": [3, 40], "3110": [3, 40], "3129": [3, 40], "3148": [3, 40], "3166": [3, 40], "3183": [3, 40], "3201": [3, 40], "3220": [3, 40], "3240": [3, 40], "3259": [3, 40], "3278": [3, 40], "3298": [3, 40], "3354": [3, 40], "3373": [3, 40], "3393": [3, 40], "3432": [3, 40], "3451": [3, 40], "3470": [3, 40], "3508": [3, 40], "3526": [3, 40], "3545": [3, 40], "3565": [3, 40], "3604": [3, 40], "3642": [3, 40], "3661": [3, 38, 39, 40], "3680": [3, 40], "3698": [3, 40], "3716": [3, 40], "3736": [3, 38, 39, 40], "3755": [3, 40], "3775": [3, 40], "3813": [3, 40], "3833": [3, 40], "3871": [3, 40], "3889": [3, 40], "3928": [3, 40], "3947": [3, 40], "3967": [3, 40], "3986": [3, 40], "4005": [3, 40], "4024": [3, 40], "4042": [3, 40], "4060": [3, 40], "4080": [3, 40], "4099": [3, 40], "4119": [3, 40], "4138": [3, 40], "4157": [3, 40], "4176": [3, 40], "4196": [3, 40], "4215": [3, 40], "4234": [3, 40], "4252": [3, 40], "4271": [3, 40], "4291": [3, 40], "4310": [3, 40], "4329": [3, 40], "4349": [3, 40], "4369": [3, 40], "4389": [3, 40], "4409": [3, 40], "4428": [3, 40], "4448": [3, 40], "4488": [3, 40], "4508": [3, 40], "4528": [3, 40], "4566": [3, 40], "4620": [3, 40], "4658": [3, 40], "4677": [3, 40], "4696": [3, 40], "4715": [3, 40], "4735": [3, 40], "4772": [3, 40], "4790": [3, 40], "4810": [3, 40], "4828": [3, 40], "4847": [3, 40], "4867": [3, 40], "4906": [3, 40], "4925": [3, 40], "4943": [3, 40], "4961": [3, 40], "4980": [3, 40], "4999": [3, 40], "1940": [3, 40], "9672": [3, 40], "6776": [3, 40], "8219": [3, 40], "5063": [3, 40], "2717": [3, 40], "2034": [3, 40], "2023": [3, 34, 37, 40], "1891": [3, 40], "1433": [3, 40], "1670": [3, 40], "1471": [3, 40], "1821": [3, 40], "1520": [3, 40], "1772": [3, 40], "1271": [3, 40], "1096": [3, 40], "1419": [3, 40], "1618": [3, 40], "1372": [3, 40], "1558": [3, 40], "1486": [3, 40], "246": [3, 40], "224": [3, 38, 39, 40], "269": [3, 40], "226": [3, 40], "338": [3, 40], "361": [3, 40], "383": [3, 40], "405": [3, 40], "427": [3, 38, 39, 40], "471": [3, 40], "494": [3, 40], "516": [3, 40], "227": [3, 40], "679": [3, 38, 39, 40], "228": [3, 38, 39, 40], "702": [3, 40], "725": [3, 38, 39, 40], "770": [3, 40], "792": [3, 40], "814": [3, 40], "858": [3, 40], "880": [3, 40], "902": [3, 40], "946": [3, 38, 39, 40], "990": [3, 38, 39, 40], "1034": [3, 40], "1056": [3, 40], "1078": [3, 40], "1100": [3, 40], "1123": [3, 40], "1146": [3, 40], "1168": [3, 40], "1190": [3, 40], "1212": [3, 40], "1234": [3, 40], "1256": [3, 40], "1277": [3, 40], "1299": [3, 40], "1342": [3, 40], "1365": [3, 40], "1387": [3, 40], "1410": [3, 40], "1526": [3, 40], "1570": [3, 40], "1615": [3, 40], "1637": [3, 40], "1659": [3, 40], "1681": [3, 40], "1703": [3, 40], "1725": [3, 40], "1770": [3, 40], "1792": [3, 40], "1858": [3, 40], "1880": [3, 40], "1903": [3, 40], "1926": [3, 40], "1949": [3, 40], "2018": [3, 34, 35, 36, 40], "2041": [3, 40], "2063": [3, 40], "2109": [3, 40], "2132": [3, 40], "2155": [3, 40], "2177": [3, 40], "2199": [3, 40], "2243": [3, 40], "2287": [3, 40], "2310": [3, 40], "2332": [3, 40], "2354": [3, 40], "2376": [3, 40], "2398": [3, 40], "2421": [3, 40], "2444": [3, 40], "2467": [3, 40], "2490": [3, 40], "2513": [3, 40], "2536": [3, 40], "2560": [3, 40], "2583": [3, 37, 40], "2605": [3, 40], "2627": [3, 40], "2649": [3, 40], "2671": [3, 40], "2739": [3, 40], "2762": [3, 40], "2785": [3, 40], "2808": [3, 40], "2831": [3, 40], "2901": [3, 40], "2923": [3, 40], "2944": [3, 40], "2965": [3, 40], "2986": [3, 40], "3007": [3, 40], "3028": [3, 40], "3070": [3, 40], "3112": [3, 40], "3176": [3, 40], "3221": [3, 40], "3243": [3, 40], "3265": [3, 40], "3310": [3, 40], "3355": [3, 40], "3377": [3, 40], "3399": [3, 40], "3421": [3, 40], "3443": [3, 40], "3465": [3, 40], "3487": [3, 40], "3509": [3, 40], "3552": [3, 40], "3573": [3, 40], "3594": [3, 40], "3616": [3, 40], "3637": [3, 40], "3658": [3, 40], "3701": [3, 40], "3724": [3, 40], "3746": [3, 40], "3790": [3, 40], "3834": [3, 40], "3856": [3, 40], "3877": [3, 40], "3898": [3, 40], "3920": [3, 40], "3943": [3, 40], "3989": [3, 40], "4012": [3, 40], "4035": [3, 34, 40], "4058": [3, 40], "4102": [3, 40], "4125": [3, 40], "4148": [3, 40], "4194": [3, 40], "4217": [3, 40], "4263": [3, 40], "4286": [3, 40], "4309": [3, 40], "4355": [3, 40], "4491": [3, 38, 39, 40], "4514": [3, 40], "4537": [3, 40], "4560": [3, 40], "4606": [3, 40], "4629": [3, 40], "4652": [3, 40], "4674": [3, 40], "4739": [3, 40], "4760": [3, 40], "4782": [3, 34, 40], "4850": [3, 40], "4872": [3, 40], "4894": [3, 40], "4960": [3, 40], "4982": [3, 40], "5005": [3, 40], "1847": [3, 40], "rmse": [3, 5, 22, 25, 27, 28, 36, 37, 40, 42], "mse": [3, 20, 22, 23, 40], "rmse_mean": [3, 5, 40, 42], "rmse_std": [3, 5, 40, 42], "mse_mean": [3, 40], "mse_std": [3, 40], "8f": [3, 40], "13118143": [3, 40], "01402841": [3, 40], "01725777": [3, 40], "00376382": [3, 40], "paramet": [3, 5, 9, 10, 11, 13, 14, 16, 17, 18, 21, 22, 23, 26, 27, 28, 36, 37, 40, 42], "esim": [3, 5, 40, 42], "model_param_estim": [3, 5, 40, 42], "164353": [3, 40], "10030284698307514": [3, 40], "14067839453672557": [3, 40], "result": [4, 27, 35, 36, 37], "fault": [4, 9, 30, 31, 33, 34, 41], "diagnosi": [4, 23, 30, 31, 33, 41], "us": [4, 7, 9, 13, 16, 21, 30, 31, 32, 33, 34, 35, 36, 41], "mlp": [4, 11, 15, 20, 25, 38, 39, 40], "256": [4, 11, 18, 23, 28, 37, 38], "rul": [4, 26, 28, 37], "estim": [4, 31, 37, 41], "lstm": [4, 36, 42], "hi": [4, 23, 36], "remaining_useful_life_estim": [5, 26, 27, 28, 29, 42], "rulcmapss": [5, 25, 26, 42], "c": [5, 9, 11, 13, 16, 18, 23, 28, 34, 42], "user": [5, 39, 42], "conda": [5, 42], "env": [5, 39, 42], "latent": [5, 42], "lib": [5, 39, 42], "site": [5, 39, 42], "packag": [5, 39, 42], "py": [5, 39, 42], "tqdmwarn": [5, 39, 42], "iprogress": [5, 39, 42], "found": [5, 35, 39, 42], "pleas": [5, 39, 42], "updat": [5, 39, 42], "jupyt": [5, 39, 42], "ipywidget": [5, 39, 42], "see": [5, 35, 39, 42], "http": [5, 9, 16, 21, 32, 39, 42], "readthedoc": [5, 39, 42], "io": [5, 39, 42], "en": [5, 39, 42], "stabl": [5, 39, 42], "user_instal": [5, 39, 42], "html": [5, 9, 16, 39, 42], "autonotebook": [5, 39, 42], "notebook_tqdm": [5, 39, 42], "1e": [5, 42], "25000": [5, 42], "33333": [5, 42], "33341": [5, 42], "24998": [5, 42], "49985": [5, 42], "11192": [5, 42], "33322": [5, 42], "33335": [5, 42], "score": [5, 25, 27, 28, 34, 37, 42], "score_mean": [5, 42], "score_std": [5, 42], "6f": [5, 42], "456039": [5, 42], "461288": [5, 42], "54669": [5, 42], "804238": [5, 42], "15056": [5, 42], "124157": [5, 42], "944129": [5, 42], "8843235832452774": [5, 42], "06986449102077215": [5, 42], "python": [7, 36], "deep": [7, 18, 35, 37], "learn": [7, 11, 18, 28, 35, 37], "monitor": [7, 34, 35], "task": [7, 31, 41], "get": [7, 14], "start": [7, 35, 36], "instal": [7, 31], "descript": 7, "quick": [7, 31], "tutori": [7, 31], "advanc": [7, 34], "materi": [7, 36], "api": 7, "refer": 7, "detail": 7, "function": [7, 11, 27, 34, 35, 37], "custom": 7, "how": [7, 34, 35], "your": 7, "own": 7, "benchmark": [7, 35], "which": [7, 35, 36], "ar": [7, 11, 13, 18, 23, 28, 34, 36, 37], "best": 7, "baseanomalydetect": 8, "util": [8, 15, 20, 25, 30], "slidingwindowdataset": [8, 12, 15, 19, 20, 24, 25, 29], "num_chunk": [9, 13, 16, 21, 26], "none": [9, 12, 13, 14, 16, 21, 26], "force_download": [9, 13, 16, 21, 26], "sourc": [9, 10, 11, 12, 13, 14, 16, 17, 18, 19, 21, 22, 23, 24, 26, 27, 28, 29], "base": [9, 11, 12, 14, 16, 18, 19, 21, 23, 24, 26, 28, 29, 30, 34, 35, 36, 37], "basedataset": [9, 13, 16, 21, 26, 30], "paper": [9, 16, 18, 27, 35, 36, 37], "reinartz": [9, 16, 34, 35], "kulahci": [9, 16, 34, 35], "m": [9, 16, 34, 35, 37], "ravn": [9, 16, 34, 35], "o": [9, 16, 37], "2021": [9, 16, 18, 34, 35, 37], "an": [9, 16, 34, 35, 36, 37], "extend": [9, 16, 34, 35], "simul": [9, 16, 26, 34, 35, 37], "decis": [9, 16, 34, 35], "support": [9, 16, 34, 35], "system": [9, 16, 34, 35, 37], "comput": [9, 11, 16, 34, 35, 36, 37], "chemic": [9, 16, 34, 35], "engin": [9, 16, 26, 34, 35, 37], "107281": [9, 16, 34, 35], "web": [9, 16], "mit": [9, 16], "edu": [9, 16], "braatzgroup": [9, 16], "link": [9, 13, 16, 21, 26], "int": [9, 11, 12, 13, 16, 18, 19, 21, 23, 24, 26, 28, 29, 34], "If": [9, 11, 13, 16, 18, 21, 23, 26, 28], "given": [9, 13, 16, 21, 26], "onli": [9, 13, 16, 21, 26], "chunk": [9, 13, 16, 21, 26], "purpos": [9, 13, 16, 21, 26], "bool": [9, 11, 13, 16, 18, 21, 23, 26, 28], "even": [9, 13, 16, 21, 26], "exist": [9, 13, 14, 16, 21, 26], "set_name_public_link": [9, 13, 16, 21, 26], "method": [9, 11, 13, 16, 18, 21, 26, 35, 36], "ha": [9, 11, 13, 16, 18, 21, 23, 26, 28, 34], "implement": [9, 13, 16, 18, 21, 26], "all": [9, 11, 13, 16, 18, 21, 23, 26, 28, 34], "children": [9, 13, 16, 21, 26], "set": [9, 13, 14, 16, 21, 26, 34, 35, 36, 37, 38, 39], "name": [9, 11, 13, 16, 18, 21, 23, 26, 28, 38, 39], "public": [9, 13, 16, 21, 26], "crop": [9, 16], "version": [9, 16], "rieth": [9, 16], "amsel": [9, 16], "b": [9, 11, 13, 16, 18, 23, 28, 34, 36], "d": [9, 16, 34, 35, 37], "tran": [9, 16], "r": [9, 16, 34, 35], "cook": [9, 16], "2017": [9, 16], "addit": [9, 16], "v1": [9, 16], "softwar": [9, 16], "harvard": [9, 16], "datavers": [9, 16], "doi": [9, 16, 26, 27], "org": [9, 16], "7910": [9, 16], "dvn": [9, 16], "6c3jr1": [9, 16], "pred": [10, 17, 22, 27], "list": [10, 11, 14, 17, 22, 27], "float": [10, 11, 13, 17, 18, 22, 23, 27, 28], "classif": [10, 17], "number": [10, 11, 13, 17, 18, 23, 28, 34, 35, 36, 37], "posit": [10, 17, 34, 35], "divid": [10, 17, 34, 35, 36], "exampl": [10, 17, 36, 37], "predict": [10, 13, 17, 22, 27, 34, 35, 36, 37], "valu": [10, 11, 13, 14, 17, 18, 22, 23, 27, 28, 34, 36, 37], "return": [10, 11, 13, 14, 17, 18, 22, 23, 27, 28, 37], "type": [10, 11, 13, 14, 17, 18, 22, 23, 27, 28, 35, 36], "str": [11, 13, 14, 18, 23, 28], "threshold": 11, "basemodel": [11, 13, 18, 23, 28, 30], "abc": [11, 13, 18, 23, 28], "The": [11, 13, 18, 23, 28, 34, 35, 36, 37], "window": [11, 18, 23, 28, 35], "size": [11, 13, 18, 23, 28], "batch": [11, 13, 18, 23, 28], "rate": [11, 13, 17, 18, 23, 28, 34, 35, 36], "cpu": [11, 13, 18, 23, 28], "possibl": [11, 13, 18, 23, 28], "show": [11, 13, 18, 23, 28], "progress": [11, 13, 18, 23, 28], "bar": [11, 13, 18, 23, 28], "artifact": [11, 13, 18, 23, 28], "store": [11, 13, 18, 23, 28], "datafram": [11, 12, 13, 18, 19, 23, 24, 28, 29], "seri": [11, 12, 13, 18, 19, 23, 24, 28, 29, 34, 35], "dict": [11, 13, 14, 18, 23, 28], "sensor": [11, 13, 18, 23, 28, 35, 36, 37], "index": [11, 13, 18, 23, 30, 31, 33, 41], "two": [11, 13, 18, 23, 28, 36], "column": [11, 13, 18, 23, 28, 38, 39], "run_id": [11, 13, 18, 23, 28, 38, 39], "other": [11, 13, 18, 23, 28, 35, 36], "ind": [11, 13, 18, 23, 28], "dictionari": [11, 13, 18, 23, 28], "where": [11, 13, 18, 23, 28, 34, 35, 36, 37], "kei": [11, 13, 18, 23, 28, 34], "without": 11, "state": [11, 34, 35, 36, 37], "autoencod": [11, 34, 35], "ae_anomaly_detect": 11, "ae": [11, 36], "consist": [11, 18, 23, 28, 35, 37], "encod": [11, 35, 37], "decod": 11, "part": [11, 37], "each": [11, 17, 18, 23, 28, 35, 36, 37], "reshap": [11, 18, 23, 28], "vector": [11, 18, 23, 28, 34], "l": [11, 13, 18, 23, 28, 34, 35, 36, 37], "output": [11, 18, 23, 28, 34, 35], "length": [11, 13, 18, 23, 28, 37, 38, 39], "larn": [11, 13, 18, 23, 28], "boundari": 11, "dimens": 11, "hidden": [11, 18, 23, 28, 35], "layer": [11, 18, 23, 28, 35], "num_sensor": 11, "num_lay": [11, 18], "modul": [11, 30], "intern": [11, 37], "share": 11, "both": [11, 36], "nn": 11, "scriptmodul": 11, "forward": 11, "x": [11, 34, 37], "defin": [11, 14, 34, 35], "perform": [11, 34, 35, 36, 37], "everi": [11, 18, 35], "call": 11, "should": [11, 37], "overridden": 11, "subclass": 11, "although": 11, "recip": 11, "pass": 11, "need": 11, "within": [11, 34, 35], "one": [11, 34, 35], "instanc": [11, 13, 34, 35], "afterward": 11, "instead": 11, "sinc": [11, 36], "former": 11, "take": 11, "care": 11, "run": [11, 26, 32, 35, 36, 37], "regist": 11, "hook": 11, "while": 11, "latter": 11, "silent": 11, "ignor": 11, "them": [11, 34, 35], "option": 12, "abstract": 13, "docstr": 13, "rewritten": 13, "so": 13, "clearli": 13, "describ": 13, "classmethod": 13, "from_config": 13, "cfg": 13, "config": [13, 30], "child": 13, "tensor": 13, "make": 13, "shape": 13, "cfg_path": 14, "path_get": 14, "path": 14, "field": 14, "nest": 14, "specifi": 14, "object": 14, "path_set": 14, "doesn": 14, "t": [14, 34, 35, 36, 37], "obejct": 14, "to_yaml": 14, "dump": 14, "yaml": 14, "file": [14, 32], "basefaultdiagnosi": 15, "tcn": [15, 37], "faultdiagnosissmalltep": [15, 16, 39], "ndarrai": 17, "correct": [17, 34, 35], "total": [17, 34, 35, 36], "correctli": [17, 34, 35], "diagnos": [17, 35], "faulti": [17, 34, 35], "aka": 17, "alarm": [17, 34, 35], "normal": [17, 34, 35], "mlp_fault_diagnosi": [18, 23], "multilay": [18, 23, 28], "perceptron": [18, 23, 28], "relu": [18, 23, 28, 38], "activ": [18, 23, 28], "dimension": [18, 23, 28], "kernel_s": 18, "dilation_bas": 18, "dropout": 18, "tcn_fault_diagnosi": 18, "tempor": [18, 34, 35, 37], "convolut": [18, 37], "network": [18, 34, 35, 37], "lomov": [18, 35], "ildar": [18, 35], "et": [18, 26, 35, 36], "al": [18, 26, 35, 36], "journal": [18, 35, 37], "industri": [18, 35, 37], "inform": [18, 34, 35, 36, 37], "integr": [18, 35], "100216": [18, 35], "channel": 18, "kernel": 18, "residu": 18, "block": 18, "expon": [18, 27, 36], "determin": [18, 35], "dilat": 18, "baseremainingusefullifeestim": [20, 28], "mil": 21, "nasa": [21, 36], "gov": 21, "raw": 21, "wear": [21, 22], "vjv9": 21, "9f3x": 21, "squar": [22, 27], "error": [22, 27], "between": [22, 23, 27, 28, 34, 36], "real": [22, 27, 35], "basehealthindexestim": [23, 25], "time": [23, 28, 34, 35], "interv": [23, 28], "first": [23, 28, 36], "point": [23, 28, 35, 36], "consecut": [23, 28], "slide": [23, 28, 35], "1024": 23, "5e": [23, 28], "nonsimmetric_funct": [25, 27], "piec": [26, 37], "wise": [26, 37], "saxena": [26, 37], "damag": [26, 37], "propag": [26, 37], "aircraft": [26, 37], "failur": [26, 34, 37], "phm": [26, 27], "4711414": [26, 27], "depend": [27, 34, 35], "rel": 27, "deviat": [27, 34], "divis": 27, "root": [27, 32], "non": 27, "simmetr": 27, "propos": [27, 36], "origin": [27, 35, 37], "cmapss": 27, "mapss": 28, "mlp_cmapss_rul": 28, "remain": [30, 31, 33, 35, 41], "lifetim": [30, 31, 33, 41], "health": [30, 31, 33, 34, 37, 41], "can": [32, 34, 35], "pip": 32, "follow": [32, 36], "git": 32, "github": 32, "com": 32, "codeai": 32, "team": 32, "To": [32, 36], "repositori": [32, 36], "press": 32, "code": 32, "zip": 32, "main": [32, 36, 37], "page": 32, "Then": [32, 35], "unzip": 32, "go": 32, "folder": 32, "command": 32, "linux": 32, "maco": 32, "cd": 32, "final": 32, "pytest": 32, "In": [34, 35], "focu": [34, 35], "variabl": [34, 35], "oper": [34, 35, 36, 37], "classifi": [34, 35], "These": [34, 35], "temperatur": 34, "pressur": 34, "flow": 34, "level": [34, 36], "indic": [34, 35], "composit": 34, "effici": 34, "typic": 34, "signal": [34, 35], "mathemat": 34, "formul": [34, 37], "context": 34, "involv": [34, 36], "map": 34, "binari": 34, "anomal": 34, "let": 34, "repres": [34, 36, 37], "begin": [34, 35, 36, 37], "equat": [34, 35, 36, 37], "x_1": [34, 35], "x_2": [34, 35], "ldot": 34, "x_n": [34, 35], "end": [34, 35, 36, 37], "x_i": 34, "denot": 34, "th": 34, "n": [34, 35, 36, 37], "case": [34, 35, 36, 37], "text": [34, 35, 37], "otherwis": 34, "setup": 34, "could": 34, "machin": [34, 35, 36, 37], "histor": 34, "relationship": 34, "algorithm": [34, 36], "neural": [34, 37], "tree": 34, "emploi": 34, "captur": 34, "complex": 34, "interdepend": 34, "accur": 34, "identifi": [34, 35], "pattern": 34, "sever": [34, 35], "commonli": [34, 35, 37], "help": [34, 35], "understand": [34, 35], "well": [34, 35], "avoid": [34, 35], "choic": [34, 35], "often": [34, 35], "specif": [34, 35, 37], "requir": [34, 35, 36], "versu": [34, 35], "cost": [34, 35], "common": [34, 35], "includ": [34, 35, 36, 37], "measur": [34, 35, 36], "overal": [34, 35], "frac": [34, 35, 36, 37], "neg": [34, 35], "recal": [34, 35], "proport": [34, 35], "actual": [34, 35], "It": [34, 35], "particularli": [34, 35], "scenario": [34, 35], "miss": [34, 35, 36], "have": [34, 35], "seriou": [34, 35], "consequ": [34, 35], "incorrectli": [34, 35], "lower": [34, 35], "desir": [34, 35], "fewer": [34, 35], "recent": 34, "year": 34, "been": 34, "lot": 34, "research": [34, 35], "most": [34, 36, 37], "cite": [34, 35], "were": [34, 36], "variou": [34, 36, 37], "adapt": 34, "work": [34, 36], "f1": 34, "dagmm": 34, "kddcup": 34, "thyroid": 34, "arrhuthmia": 34, "4983": 34, "rev": 34, "9380": 34, "gdn": 34, "swat": 34, "wadi": 34, "gta": 34, "2022": [34, 37], "smap": 34, "9041": 34, "msl": 34, "9111": 34, "mad": 34, "gan": [34, 35], "2019": [34, 35], "kddcup99": 34, "mscred": 34, "synthet": 34, "power": 34, "plant": [34, 35], "omnianomali": 34, "8434": 34, "8989": 34, "smd": 34, "8857": 34, "thoc": 34, "2d": 34, "gestur": 34, "6331": 34, "demand": 34, "4568": 34, "9886": 34, "8809": 34, "9518": 34, "9367": 34, "usad": 34, "8460": 34, "4296": 34, "8634": 34, "9272": 34, "9463": 34, "christoph": [34, 35], "murat": [34, 35], "ol": [34, 35], "k": [34, 35, 36, 37], "choi": 34, "j": [34, 37], "yi": 34, "park": 34, "yoon": 34, "review": [34, 35], "analysi": 34, "guidelin": 34, "ieee": [34, 35, 37], "access": 34, "vol": 34, "pp": [34, 37], "120043": 34, "120065": 34, "gen": 34, "li": [34, 37], "jason": 34, "jung": 34, "multivari": [34, 35], "approach": [34, 36, 37], "applic": [34, 37], "challeng": 34, "fusion": 34, "volum": 34, "zong": 34, "q": [34, 37], "song": [34, 37], "min": [34, 36], "w": 34, "cheng": [34, 37], "lumezanu": 34, "cho": 34, "h": [34, 37], "chen": [34, 37], "gaussian": 34, "mixtur": 34, "unsupervis": 34, "proc": 34, "conf": 34, "deng": 34, "hooi": 34, "graph": [34, 37], "artif": 34, "intel": 34, "aaai": 34, "4027": 34, "z": [34, 37], "zhang": [34, 37], "yuan": 34, "structur": [34, 37], "iot": 34, "arxiv": [34, 37], "2104": 34, "03466": 34, "jin": 34, "shi": 34, "goh": 34, "ng": 34, "gener": [34, 37], "adversari": 34, "netw": 34, "icann": 34, "munich": 34, "germani": 34, "springer": 34, "703": 34, "716": 34, "sippl": 34, "interpret": [34, 37], "multidimension": 34, "multimod": 34, "mach": [34, 37], "9016": 34, "9025": 34, "y": [34, 36, 37], "su": 34, "zhao": 34, "niu": 34, "liu": [34, 37], "sun": 34, "pei": 34, "robust": 34, "through": 34, "stochast": 34, "recurr": [34, 37], "25th": 34, "acm": 34, "sigkdd": 34, "knowl": 34, "discoveri": 34, "mine": 34, "2828": 34, "2837": 34, "shen": 34, "kwok": 34, "timeseri": 34, "hierarch": [34, 35], "adv": 34, "inf": 34, "syst": 34, "13016": 34, "13026": 34, "audibert": 34, "p": [34, 36], "michiardi": 34, "guyard": 34, "marti": 34, "zuluaga": 34, "26th": 34, "3395": 34, "3404": 34, "compani": 35, "serv": 35, "crucial": 35, "control": [35, 37], "mimick": 35, "figur": [35, 36, 37], "ref": 35, "fig": [35, 36, 37], "process_diagram": 35, "2800": 35, "independ": [35, 36], "last": 35, "hour": 35, "collect": 35, "frequenc": 35, "minut": 35, "thu": 35, "2000": 35, "observ": 35, "onc": 35, "after": 35, "goe": 35, "ratio": 35, "contain": [35, 36, 37], "manipul": 35, "moment": 35, "correspond": [35, 36], "certain": 35, "condit": [35, 37], "mark": 35, "label": [35, 36], "x_t": 35, "mathbb": 35, "y_1": 35, "y_2": 35, "y_n": 35, "y_t": 35, "width": 35, "rightarrow": 35, "must": 35, "align": [35, 37], "argmin": 35, "_f": 35, "sum_": [35, 36, 37], "x_": 35, "approxim": [35, 36], "devid": 35, "art": [35, 36, 37], "pca": 35, "dbn": 35, "svdd": 35, "un": 35, "msdae": 35, "tp": 35, "gru": 35, "002": 35, "abid": 35, "anam": 35, "muhammad": 35, "tahir": 35, "khan": 35, "javaid": 35, "iqbal": 35, "techniqu": [35, 36], "basic": 35, "beyond": 35, "artifici": 35, "intellig": [35, 37], "3639": 35, "3664": [35, 38, 39], "yan": 35, "shifu": 35, "xuefeng": 35, "design": 35, "teacher": 35, "supervis": 35, "dual": 35, "stack": 35, "qualiti": 35, "relev": 35, "appli": [35, 36], "soft": 35, "105526": 35, "yu": [35, 37], "jianbo": 35, "enhanc": 35, "strategi": 35, "favor": 35, "featur": [35, 36], "belief": 35, "chemistri": 35, "15479": 35, "15490": 35, "whole": 35, "unstabl": 35, "neuron": 35, "transact": 35, "cybernet": [35, 37], "3998": 35, "4007": 35, "multiscal": 35, "agglom": 35, "cluster": 35, "denois": 35, "106525": 35, "experi": 36, "conduct": 36, "matsuura": 36, "centr": 36, "mc": 36, "510v": 36, "dry": 36, "rough": 36, "cast": 36, "iron": 36, "j45": 36, "stainless": 36, "steel": 36, "workpiec": 36, "six": 36, "tooth": 36, "face": 36, "kc710": 36, "carbid": 36, "insert": 36, "sub": 36, "depth": 36, "feed": 36, "chang": 36, "record": 36, "vibroacceleromet": 36, "cta": 36, "current": 36, "space": 36, "saw": 36, "surfac": 36, "optic": 36, "microscopi": 36, "Of": 36, "five": 36, "tool": 36, "ac": 36, "spindl": 36, "motor": 36, "smcac": 36, "dc": 36, "smcdc": 36, "vibrat": 36, "vib_tabl": 36, "vib_spindl": 36, "acoust": 36, "emiss": 36, "ae_tabl": 36, "ae_spindl": 36, "idea": 36, "build": 36, "extract": [36, 39], "equip": [36, 37], "due": [36, 37], "valid": 36, "alwai": 36, "perfectli": 36, "confirm": 36, "construct": 36, "we": 36, "taylor": 36, "document": 36, "eq": 36, "c_": 36, "v": [36, 37], "v_": 36, "life": [36, 37], "constant": 36, "relat": 36, "speed": [36, 37], "nelder": 36, "mead": 36, "scipi": 36, "obtain": 36, "character": [36, 37], "itself": 36, "definit": 36, "charact": 36, "curv": 36, "behavior": [36, 37], "empir": [36, 37], "verifi": 36, "read": [36, 37, 39], "pipelin": 36, "sawtooth": 36, "next": 36, "simultan": 36, "do": 36, "max": 36, "scale": [36, 37, 38, 39], "subdataset": 36, "minimum": 36, "mm": 36, "maximum": 36, "allow": 36, "subtract": 36, "reduc": 36, "impact": 36, "implicit": 36, "weight": 36, "healthi": [36, 37], "trajectori": [36, 37], "wa": [36, 37], "chosen": 36, "acquir": 36, "sqrt": [36, 37], "y_i": [36, 37], "hat": [36, 37], "_i": [36, 37], "dure": [36, 37], "unknown": 36, "characterist": [36, 37], "limit": 36, "especi": 36, "similar": 36, "about": [36, 37], "cycl": 36, "mapreduc": 36, "prf": 36, "0531": 36, "svm": 36, "0254": 36, "takelm": 36, "0134": 36, "0148": 36, "agogino": 36, "goebel": [36, 37], "prognost": [36, 37], "2007": 36, "virtanen": 36, "fundament": 36, "scientif": 36, "natur": 36, "272": 36, "commerci": 37, "modular": 37, "aero": 37, "propuls": 37, "program": 37, "turbofan": 37, "under": 37, "flight": 37, "until": 37, "situat": 37, "ascent": 37, "descent": 37, "altitud": 37, "km": 37, "four": 37, "subset": 37, "locat": 37, "differ": 37, "degrad": 37, "three": 37, "height": 37, "acceler": 37, "term": 37, "throttl": 37, "resolv": 37, "angl": 37, "becaus": 37, "weak": 37, "distinguish": 37, "consid": 37, "abov": 37, "equal": 37, "regress": 37, "geq": 37, "heurist": 37, "previous": 37, "studi": 37, "e": 37, "asymmetr": 37, "phm2008": 37, "competit": 37, "penal": 37, "overestim": 37, "mainten": 37, "regul": 37, "fd001": 37, "fd003": 37, "fd002": 37, "fd004": 37, "gnmr": 37, "370": 37, "3196": 37, "2795": 37, "bl": 37, "tfcn": 37, "243": 37, "244": [37, 38, 39], "2096": 37, "tcnn": 37, "254": 37, "415": 37, "1453": 37, "rve": 37, "323": [37, 38, 39], "1846": 37, "bigru": 37, "tsam": 37, "232": 37, "2264": 37, "3610": 37, "mca": 37, "bgru": 37, "kghm": 37, "251": 37, "333": 37, "1131": 37, "3356": 37, "da": 37, "pm": 37, "229": 37, "257": [37, 38, 39], "522": 37, "2317": 37, "adl": 37, "dnn": 37, "238": 37, "281": 37, "1149": 37, "simon": 37, "eklund": 37, "confer": 37, "manag": 37, "heim": 37, "international135": 37, "jyoti": 37, "pankaj": 37, "vishnu": 37, "lovekesh": 37, "gautam": 37, "leverag": 37, "preprint": 37, "2006": 37, "16556v1": 37, "wang": 37, "by140": 37, "fuse": 37, "broad": 37, "8th": 37, "social": 37, "iccss": 37, "joint": 37, "and145": 37, "neurosci": 37, "costa": 37, "s\u00e1nchez": 37, "variat": 37, "assess": 37, "reliabl": 37, "safeti": 37, "108353": 37, "jiang": 37, "wu": 37, "luo": 37, "yin": 37, "bidirect": 37, "gate": 37, "unit": 37, "self": 37, "attent": 37, "mechan": 37, "engineering150": 37, "108297": 37, "kara": 37, "multi": 37, "108211": 37, "hu": 37, "enabl": 37, "knowledg": 37, "108869": 37, "gao": 37, "jia": 37, "pang": 37, "distribut": 37, "internet": 37, "thing": 37, "9594": 37, "9602": 37, "xiang": 37, "qin": 37, "gryllia": 37, "automat": 37, "differenti": 37, "its": 37, "108531": 37, "xmeas_1": [38, 39], "xmeas_2": [38, 39], "xmeas_3": [38, 39], "xmeas_4": [38, 39], "xmeas_5": [38, 39], "xmeas_6": [38, 39], "xmeas_7": [38, 39], "xmeas_8": [38, 39], "xmeas_9": [38, 39], "xmeas_10": [38, 39], "xmv_2": [38, 39], "xmv_3": [38, 39], "xmv_4": [38, 39], "xmv_5": [38, 39], "xmv_6": [38, 39], "xmv_7": [38, 39], "xmv_8": [38, 39], "xmv_9": [38, 39], "xmv_10": [38, 39], "xmv_11": [38, 39], "413402073": [38, 39], "25038": [38, 39], "3674": [38, 39], "4529": [38, 39], "2320": [38, 39], "402": [38, 39], "2704": [38, 39], "863": [38, 39], "33818": [38, 39], "744": [38, 39], "657": [38, 39], "544": [38, 39], "935": [38, 39], "757": [38, 39], "510": [38, 39], "447": [38, 39], "25109": [38, 39], "3659": [38, 39], "4264": [38, 39], "721": [38, 39], "576": [38, 39], "000": [38, 39], "33620": [38, 39], "414": [38, 39], "588": [38, 39], "259": [38, 39], "084": [38, 39], "554": [38, 39], "692": [38, 39], "359": [38, 39], "3660": [38, 39], "4477": [38, 39], "4426": [38, 39], "875": [38, 39], "070": [38, 39], "2706": [38, 39], "771": [38, 39], "33563": [38, 39], "357": [38, 39], "666": [38, 39], "275": [38, 39], "380": [38, 39], "699": [38, 39], "530": [38, 39], "24977": [38, 39], "4776": [38, 39], "758": [38, 39], "063": [38, 39], "33553": [38, 39], "856": [38, 39], "277": [38, 39], "072": [38, 39], "643": [38, 39], "089": [38, 39], "29405": [38, 39], "4497": [38, 39], "3381": [38, 39], "650": [38, 39], "388": [38, 39], "32632": [38, 39], "797": [38, 39], "947": [38, 39], "346": [38, 39], "507": [38, 39], "312148819": [38, 39], "24842": [38, 39], "3694": [38, 39], "3946": [38, 39], "780": [38, 39], "2708": [38, 39], "32959": [38, 39], "891": [38, 39], "320": [38, 39], "867": [38, 39], "868": [38, 39], "061": [38, 39], "088": [38, 39], "470": [38, 39], "463": [38, 39], "078": [38, 39], "957": [38, 39], "22612": [38, 39], "4523": [38, 39], "3655": [38, 39], "778": [38, 39], "730": [38, 39], "2711": [38, 39], "32645": [38, 39], "675": [38, 39], "831": [38, 39], "027": [38, 39], "560": [38, 39], "598": [38, 39], "591": [38, 39], "720": [38, 39], "958": [38, 39], "22386": [38, 39], "3984": [38, 39], "673": [38, 39], "528": [38, 39], "2712": [38, 39], "32484": [38, 39], "233": [38, 39], "053": [38, 39], "235": [38, 39], "040": [38, 39], "737": [38, 39], "490": [38, 39], "884": [38, 39], "310": [38, 39], "959": [38, 39], "22561": [38, 39], "4483": [38, 39], "4293": [38, 39], "435": [38, 39], "2710": [38, 39], "857": [38, 39], "31932": [38, 39], "248": [38, 39], "567": [38, 39], "820": [38, 39], "979": [38, 39], "503": [38, 39], "630": [38, 39], "996": [38, 39], "960": [38, 39], "22585": [38, 39], "4492": [38, 39], "4061": [38, 39], "869": [38, 39], "722": [38, 39], "31926": [38, 39], "429": [38, 39], "986": [38, 39], "810": [38, 39], "639": [38, 39], "378": [38, 39], "153300": [38, 39], "row": [38, 39], "dtype": [38, 39], "int64": [38, 39], "7313": 38, "6942": 38, "6606": 38, "sequenti": 38, "flatten": 38, "start_dim": 38, "end_dim": 38, "linear": 38, "in_featur": 38, "520": 38, "out_featur": 38, "bia": 38, "7545263157894737": 38, "705975": 38, "04881012658227848": 38, "vitalijpozdnakov": 39, "miniconda3": 39, "python3": 39, "small_tep": 39, "2m": 39, "1mb": 39, "csv": 39, "6mb": 39, "153mb": 39, "77mb": 39, "60gb": 39, "33gb": 39, "79692": 39, "1771006": 39, "1837204": 39, "105": 39, "4350": 39, "5717": 39, "3677": 39, "6005": 39, "7325": 39, "5394": 39, "3939": 39, "345": 39, "7945664160401003": 39}, "objects": {"ice.anomaly_detection": [[9, 0, 0, "-", "datasets"], [10, 0, 0, "-", "metrics"], [12, 0, 0, "-", "utils"]], "ice.anomaly_detection.datasets": [[9, 1, 1, "", "AnomalyDetectionReinartzTEP"], [9, 1, 1, "", "AnomalyDetectionSmallTEP"]], "ice.anomaly_detection.datasets.AnomalyDetectionReinartzTEP": [[9, 2, 1, "", "set_name_public_link"]], "ice.anomaly_detection.datasets.AnomalyDetectionSmallTEP": [[9, 2, 1, "", "set_name_public_link"]], "ice.anomaly_detection.metrics": [[10, 3, 1, "", "accuracy"]], "ice.anomaly_detection.models": [[11, 0, 0, "-", "autoencoder"], [11, 0, 0, "-", "base"]], "ice.anomaly_detection.models.autoencoder": [[11, 1, 1, "", "AutoEncoderMLP"], [11, 1, 1, "", "MLP"]], "ice.anomaly_detection.models.autoencoder.MLP": [[11, 2, 1, "", "forward"], [11, 4, 1, "", "training"]], "ice.anomaly_detection.models.base": [[11, 1, 1, "", "BaseAnomalyDetection"]], "ice.anomaly_detection.models.base.BaseAnomalyDetection": [[11, 2, 1, "", "evaluate"], [11, 2, 1, "", "fit"]], "ice.anomaly_detection.utils": [[12, 1, 1, "", "SlidingWindowDataset"]], "ice": [[13, 0, 0, "-", "base"], [14, 0, 0, "-", "configs"]], "ice.base": [[13, 1, 1, "", "BaseDataset"], [13, 1, 1, "", "BaseModel"]], "ice.base.BaseDataset": [[13, 2, 1, "", "set_name_public_link"]], "ice.base.BaseModel": [[13, 2, 1, "", "evaluate"], [13, 2, 1, "", "fit"], [13, 2, 1, "", "from_config"], [13, 2, 1, "", "predict"]], "ice.configs": [[14, 1, 1, "", "Config"]], "ice.configs.Config": [[14, 2, 1, "", "path_get"], [14, 2, 1, "", "path_set"], [14, 2, 1, "", "to_yaml"]], "ice.fault_diagnosis": [[16, 0, 0, "-", "datasets"], [17, 0, 0, "-", "metrics"], [19, 0, 0, "-", "utils"]], "ice.fault_diagnosis.datasets": [[16, 1, 1, "", "FaultDiagnosisReinartzTEP"], [16, 1, 1, "", "FaultDiagnosisSmallTEP"]], "ice.fault_diagnosis.datasets.FaultDiagnosisReinartzTEP": [[16, 2, 1, "", "set_name_public_link"]], "ice.fault_diagnosis.datasets.FaultDiagnosisSmallTEP": [[16, 2, 1, "", "set_name_public_link"]], "ice.fault_diagnosis.metrics": [[17, 3, 1, "", "accuracy"], [17, 3, 1, "", "correct_daignosis_rate"], [17, 3, 1, "", "false_positive_rate"], [17, 3, 1, "", "true_positive_rate"]], "ice.fault_diagnosis.models": [[18, 0, 0, "-", "base"], [18, 0, 0, "-", "mlp"], [18, 0, 0, "-", "tcn"]], "ice.fault_diagnosis.models.base": [[18, 1, 1, "", "BaseFaultDiagnosis"]], "ice.fault_diagnosis.models.base.BaseFaultDiagnosis": [[18, 2, 1, "", "evaluate"]], "ice.fault_diagnosis.models.mlp": [[18, 1, 1, "", "MLP"]], "ice.fault_diagnosis.models.tcn": [[18, 1, 1, "", "TCN"]], "ice.fault_diagnosis.utils": [[19, 1, 1, "", "SlidingWindowDataset"]], "ice.health_index_estimation": [[21, 0, 0, "-", "datasets"], [22, 0, 0, "-", "metrics"], [24, 0, 0, "-", "utils"]], "ice.health_index_estimation.datasets": [[21, 1, 1, "", "Milling"]], "ice.health_index_estimation.datasets.Milling": [[21, 2, 1, "", "set_name_public_link"]], "ice.health_index_estimation.metrics": [[22, 3, 1, "", "mse"]], "ice.health_index_estimation.models": [[23, 0, 0, "-", "base"], [23, 0, 0, "-", "mlp"]], "ice.health_index_estimation.models.base": [[23, 1, 1, "", "BaseHealthIndexEstimation"]], "ice.health_index_estimation.models.base.BaseHealthIndexEstimation": [[23, 2, 1, "", "evaluate"]], "ice.health_index_estimation.models.mlp": [[23, 1, 1, "", "MLP"]], "ice.health_index_estimation.utils": [[24, 1, 1, "", "SlidingWindowDataset"]], "ice.remaining_useful_life_estimation": [[26, 0, 0, "-", "datasets"], [27, 0, 0, "-", "metrics"], [29, 0, 0, "-", "utils"]], "ice.remaining_useful_life_estimation.datasets": [[26, 1, 1, "", "RulCmapss"]], "ice.remaining_useful_life_estimation.datasets.RulCmapss": [[26, 2, 1, "", "set_name_public_link"]], "ice.remaining_useful_life_estimation.metrics": [[27, 3, 1, "", "nonsimmetric_function"], [27, 3, 1, "", "rmse"], [27, 3, 1, "", "score"]], "ice.remaining_useful_life_estimation.models": [[28, 0, 0, "-", "base"], [28, 0, 0, "-", "mlp"]], "ice.remaining_useful_life_estimation.models.base": [[28, 1, 1, "", "BaseRemainingUsefulLifeEstimation"]], "ice.remaining_useful_life_estimation.models.base.BaseRemainingUsefulLifeEstimation": [[28, 2, 1, "", "evaluate"]], "ice.remaining_useful_life_estimation.models.mlp": [[28, 1, 1, "", "MLP"]], "ice.remaining_useful_life_estimation.utils": [[29, 1, 1, "", "SlidingWindowDataset"]]}, "objtypes": {"0": "py:module", "1": "py:class", "2": "py:method", "3": "py:function", "4": "py:attribute"}, "objnames": {"0": ["py", "module", "Python module"], "1": ["py", "class", "Python class"], "2": ["py", "method", "Python method"], "3": ["py", "function", "Python function"], "4": ["py", "attribute", "Python attribute"]}, "titleterms": {"advanc": 0, "materi": 0, "result": [1, 2, 3, 5], "anomali": [1, 8, 9, 10, 11, 12, 34, 38], "detect": [1, 8, 9, 10, 11, 12, 34, 38], "us": [1, 2, 3, 5, 23, 25, 26, 27, 29, 37, 42], "autoencodermlp": [1, 11], "256": [1, 2, 3, 5], "fault": [2, 15, 16, 17, 18, 19, 35, 39], "diagnosi": [2, 15, 16, 17, 18, 19, 35, 39], "mlp": [2, 3, 18, 23, 28], "hi": 3, "estim": [3, 5, 20, 21, 22, 24, 40, 42], "benchmark": 4, "content": [4, 41], "rul": 5, "lstm": 5, "custom": 6, "class": 6, "ic": 7, "industri": 7, "condit": 7, "evalu": 7, "dataset": [9, 16, 21, 26, 34, 35, 36, 37], "metric": [10, 17, 22, 27, 34, 35, 36, 37], "model": [11, 18, 23, 28, 34], "baseanomalydetect": 11, "util": [12, 19, 24, 29], "base": 13, "modul": 13, "config": 14, "basefaultdiagnosi": 18, "tcn": 18, "health": [20, 21, 22, 24, 28, 36, 40], "index": [20, 21, 22, 24, 28, 36, 40], "remain": [23, 25, 26, 27, 29, 37, 42], "life": [23, 25, 26, 27, 29], "baseremainingusefullifeestim": 23, "basehealthindexestim": 28, "api": 30, "refer": [30, 34, 35, 36, 37], "get": 31, "start": 31, "instal": 32, "test": [32, 36], "task": [33, 34, 35, 36, 37, 38, 39, 40, 42], "tabl": [34, 35, 36, 37], "1": [34, 35, 36, 37], "deep": 34, "learn": 34, "tep": 35, "fpr": 35, "tpr": 35, "valu": 35, "Not": 35, "all": 35, "ar": 35, "avail": 35, "comparison": 35, "c": [36, 37], "mill": 36, "cut": 36, "3": 36, "7": 36, "origin": 36, "wear": 36, "target": 36, "lifetim": [37, 42], "mapss": 37, "tutori": [38, 39, 40, 41, 42], "quick": 41}, "envversion": {"sphinx.domains.c": 2, "sphinx.domains.changeset": 1, "sphinx.domains.citation": 1, "sphinx.domains.cpp": 8, "sphinx.domains.index": 1, "sphinx.domains.javascript": 2, "sphinx.domains.math": 2, "sphinx.domains.python": 3, "sphinx.domains.rst": 2, "sphinx.domains.std": 2, "sphinx.ext.viewcode": 1, "sphinx": 57}, "alltitles": {"Advanced materials": [[0, "advanced-materials"]], "Results of anomaly detection using AutoEncoderMLP-256": [[1, "results-of-anomaly-detection-using-autoencodermlp-256"]], "Results of fault diagnosis using MLP-256": [[2, "results-of-fault-diagnosis-using-mlp-256"]], "Results of HI estimation using MLP-256": [[3, "results-of-hi-estimation-using-mlp-256"]], "Benchmarking": [[4, "benchmarking"]], "Contents:": [[4, null], [41, null]], "Results of RUL estimation using lstm-256": [[5, "results-of-rul-estimation-using-lstm-256"]], "Custom classes": [[6, "custom-classes"]], "ICE - Industrial Condition Evaluation": [[7, "ice-industrial-condition-evaluation"]], "Anomaly detection": [[8, "anomaly-detection"], [34, "anomaly-detection"]], "Anomaly detection datasets": [[9, "module-ice.anomaly_detection.datasets"]], "Anomaly detection metrics": [[10, "module-ice.anomaly_detection.metrics"]], "Anomaly detection models": [[11, "anomaly-detection-models"]], "BaseAnomalyDetection": [[11, "module-ice.anomaly_detection.models.base"]], "AutoEncoderMLP": [[11, "module-ice.anomaly_detection.models.autoencoder"]], "Anomaly detection utils": [[12, "module-ice.anomaly_detection.utils"]], "Base modules": [[13, "module-ice.base"]], "Configs": [[14, "module-ice.configs"]], "Fault diagnosis": [[15, "fault-diagnosis"], [35, "fault-diagnosis"]], "Fault diagnosis datasets": [[16, "module-ice.fault_diagnosis.datasets"]], "Fault diagnosis metrics": [[17, "module-ice.fault_diagnosis.metrics"]], "Fault diagnosis models": [[18, "fault-diagnosis-models"]], "BaseFaultDiagnosis": [[18, "module-ice.fault_diagnosis.models.base"]], "MLP": [[18, "module-ice.fault_diagnosis.models.mlp"], [23, "module-ice.health_index_estimation.models.mlp"], [28, "module-ice.remaining_useful_life_estimation.models.mlp"]], "TCN": [[18, "module-ice.fault_diagnosis.models.tcn"]], "Fault diagnosis utils": [[19, "module-ice.fault_diagnosis.utils"]], "Health index estimation": [[20, "health-index-estimation"]], "Health index estimation datasets": [[21, "module-ice.health_index_estimation.datasets"]], "Health index estimation metrics": [[22, "module-ice.health_index_estimation.metrics"]], "Remaining useful life models": [[23, "remaining-useful-life-models"]], "BaseRemainingUsefulLifeEstimation": [[23, "module-ice.health_index_estimation.models.base"]], "Health index estimation utils": [[24, "module-ice.health_index_estimation.utils"]], "Remaining useful life": [[25, "remaining-useful-life"]], "Remaining useful life datasets": [[26, "module-ice.remaining_useful_life_estimation.datasets"]], "Remaining useful life metrics": [[27, "module-ice.remaining_useful_life_estimation.metrics"]], "Health index models": [[28, "health-index-models"]], "BaseHealthIndexEstimation": [[28, "module-ice.remaining_useful_life_estimation.models.base"]], "Remaining useful life utils": [[29, "module-ice.remaining_useful_life_estimation.utils"]], "API reference": [[30, "api-reference"]], "Getting started": [[31, "getting-started"]], "Installation": [[32, "installation"]], "Testing": [[32, "testing"]], "Tasks": [[33, "tasks"]], "Dataset": [[34, "dataset"], [35, "dataset"], [36, "dataset"], [37, "dataset"]], "Task": [[34, "task"], [35, "task"], [36, "task"], [37, "task"]], "Metrics": [[34, "metrics"], [35, "metrics"], [36, "metrics"], [37, "metrics"]], "References": [[34, "references"], [35, "references"], [36, "references"], [37, "references"]], "Table 1: Deep learning models for Anomaly detection.": [[34, "id1"]], "Table 1: References for TEP dataset. FPR/TPR values. Not all the metrics\nare available for comparison.": [[35, "id1"]], "Health index": [[36, "health-index"]], "Table 1: References for C-Milling dataset, cuts \u20181\u2019, \u20183\u2019, \u20187\u2019 testing, with the original dataset wear target.": [[36, "id3"]], "Remaining useful lifetime": [[37, "remaining-useful-lifetime"]], "Table 1: References for C-MAPSS dataset": [[37, "id2"]], "Tutorial on anomaly detection task": [[38, "tutorial-on-anomaly-detection-task"]], "Tutorial on fault diagnosis task": [[39, "tutorial-on-fault-diagnosis-task"]], "Tutorial on health index estimation task": [[40, "tutorial-on-health-index-estimation-task"]], "Quick tutorials": [[41, "quick-tutorials"]], "Tutorial on remaining useful lifetime estimation task": [[42, "tutorial-on-remaining-useful-lifetime-estimation-task"]]}, "indexentries": {"anomalydetectionreinartztep (class in ice.anomaly_detection.datasets)": [[9, "ice.anomaly_detection.datasets.AnomalyDetectionReinartzTEP"]], "anomalydetectionsmalltep (class in ice.anomaly_detection.datasets)": [[9, "ice.anomaly_detection.datasets.AnomalyDetectionSmallTEP"]], "ice.anomaly_detection.datasets": [[9, "module-ice.anomaly_detection.datasets"]], "module": [[9, "module-ice.anomaly_detection.datasets"], [10, "module-ice.anomaly_detection.metrics"], [11, "module-ice.anomaly_detection.models.autoencoder"], [11, "module-ice.anomaly_detection.models.base"], [12, "module-ice.anomaly_detection.utils"], [13, "module-ice.base"], [14, "module-ice.configs"], [16, "module-ice.fault_diagnosis.datasets"], [17, "module-ice.fault_diagnosis.metrics"], [18, "module-ice.fault_diagnosis.models.base"], [18, "module-ice.fault_diagnosis.models.mlp"], [18, "module-ice.fault_diagnosis.models.tcn"], [19, "module-ice.fault_diagnosis.utils"], [21, "module-ice.health_index_estimation.datasets"], [22, "module-ice.health_index_estimation.metrics"], [23, "module-ice.health_index_estimation.models.base"], [23, "module-ice.health_index_estimation.models.mlp"], [24, "module-ice.health_index_estimation.utils"], [26, "module-ice.remaining_useful_life_estimation.datasets"], [27, "module-ice.remaining_useful_life_estimation.metrics"], [28, "module-ice.remaining_useful_life_estimation.models.base"], [28, "module-ice.remaining_useful_life_estimation.models.mlp"], [29, "module-ice.remaining_useful_life_estimation.utils"]], "set_name_public_link() (ice.anomaly_detection.datasets.anomalydetectionreinartztep method)": [[9, "ice.anomaly_detection.datasets.AnomalyDetectionReinartzTEP.set_name_public_link"]], "set_name_public_link() (ice.anomaly_detection.datasets.anomalydetectionsmalltep method)": [[9, "ice.anomaly_detection.datasets.AnomalyDetectionSmallTEP.set_name_public_link"]], "accuracy() (in module ice.anomaly_detection.metrics)": [[10, "ice.anomaly_detection.metrics.accuracy"]], "ice.anomaly_detection.metrics": [[10, "module-ice.anomaly_detection.metrics"]], "autoencodermlp (class in ice.anomaly_detection.models.autoencoder)": [[11, "ice.anomaly_detection.models.autoencoder.AutoEncoderMLP"]], "baseanomalydetection (class in ice.anomaly_detection.models.base)": [[11, "ice.anomaly_detection.models.base.BaseAnomalyDetection"]], "mlp (class in ice.anomaly_detection.models.autoencoder)": [[11, "ice.anomaly_detection.models.autoencoder.MLP"]], "evaluate() (ice.anomaly_detection.models.base.baseanomalydetection method)": [[11, "ice.anomaly_detection.models.base.BaseAnomalyDetection.evaluate"]], "fit() (ice.anomaly_detection.models.base.baseanomalydetection method)": [[11, "ice.anomaly_detection.models.base.BaseAnomalyDetection.fit"]], "forward() (ice.anomaly_detection.models.autoencoder.mlp method)": [[11, "ice.anomaly_detection.models.autoencoder.MLP.forward"]], "ice.anomaly_detection.models.autoencoder": [[11, "module-ice.anomaly_detection.models.autoencoder"]], "ice.anomaly_detection.models.base": [[11, "module-ice.anomaly_detection.models.base"]], "training (ice.anomaly_detection.models.autoencoder.mlp attribute)": [[11, "ice.anomaly_detection.models.autoencoder.MLP.training"]], "slidingwindowdataset (class in ice.anomaly_detection.utils)": [[12, "ice.anomaly_detection.utils.SlidingWindowDataset"]], "ice.anomaly_detection.utils": [[12, "module-ice.anomaly_detection.utils"]], "basedataset (class in ice.base)": [[13, "ice.base.BaseDataset"]], "basemodel (class in ice.base)": [[13, "ice.base.BaseModel"]], "evaluate() (ice.base.basemodel method)": [[13, "ice.base.BaseModel.evaluate"]], "fit() (ice.base.basemodel method)": [[13, "ice.base.BaseModel.fit"]], "from_config() (ice.base.basemodel class method)": [[13, "ice.base.BaseModel.from_config"]], "ice.base": [[13, "module-ice.base"]], "predict() (ice.base.basemodel method)": [[13, "ice.base.BaseModel.predict"]], "set_name_public_link() (ice.base.basedataset method)": [[13, "ice.base.BaseDataset.set_name_public_link"]], "config (class in ice.configs)": [[14, "ice.configs.Config"]], "ice.configs": [[14, "module-ice.configs"]], "path_get() (ice.configs.config method)": [[14, "ice.configs.Config.path_get"]], "path_set() (ice.configs.config method)": [[14, "ice.configs.Config.path_set"]], "to_yaml() (ice.configs.config method)": [[14, "ice.configs.Config.to_yaml"]], "faultdiagnosisreinartztep (class in ice.fault_diagnosis.datasets)": [[16, "ice.fault_diagnosis.datasets.FaultDiagnosisReinartzTEP"]], "faultdiagnosissmalltep (class in ice.fault_diagnosis.datasets)": [[16, "ice.fault_diagnosis.datasets.FaultDiagnosisSmallTEP"]], "ice.fault_diagnosis.datasets": [[16, "module-ice.fault_diagnosis.datasets"]], "set_name_public_link() (ice.fault_diagnosis.datasets.faultdiagnosisreinartztep method)": [[16, "ice.fault_diagnosis.datasets.FaultDiagnosisReinartzTEP.set_name_public_link"]], "set_name_public_link() (ice.fault_diagnosis.datasets.faultdiagnosissmalltep method)": [[16, "ice.fault_diagnosis.datasets.FaultDiagnosisSmallTEP.set_name_public_link"]], "accuracy() (in module ice.fault_diagnosis.metrics)": [[17, "ice.fault_diagnosis.metrics.accuracy"]], "correct_daignosis_rate() (in module ice.fault_diagnosis.metrics)": [[17, "ice.fault_diagnosis.metrics.correct_daignosis_rate"]], "false_positive_rate() (in module ice.fault_diagnosis.metrics)": [[17, "ice.fault_diagnosis.metrics.false_positive_rate"]], "ice.fault_diagnosis.metrics": [[17, "module-ice.fault_diagnosis.metrics"]], "true_positive_rate() (in module ice.fault_diagnosis.metrics)": [[17, "ice.fault_diagnosis.metrics.true_positive_rate"]], "basefaultdiagnosis (class in ice.fault_diagnosis.models.base)": [[18, "ice.fault_diagnosis.models.base.BaseFaultDiagnosis"]], "mlp (class in ice.fault_diagnosis.models.mlp)": [[18, "ice.fault_diagnosis.models.mlp.MLP"]], "tcn (class in ice.fault_diagnosis.models.tcn)": [[18, "ice.fault_diagnosis.models.tcn.TCN"]], "evaluate() (ice.fault_diagnosis.models.base.basefaultdiagnosis method)": [[18, "ice.fault_diagnosis.models.base.BaseFaultDiagnosis.evaluate"]], "ice.fault_diagnosis.models.base": [[18, "module-ice.fault_diagnosis.models.base"]], "ice.fault_diagnosis.models.mlp": [[18, "module-ice.fault_diagnosis.models.mlp"]], "ice.fault_diagnosis.models.tcn": [[18, "module-ice.fault_diagnosis.models.tcn"]], "slidingwindowdataset (class in ice.fault_diagnosis.utils)": [[19, "ice.fault_diagnosis.utils.SlidingWindowDataset"]], "ice.fault_diagnosis.utils": [[19, "module-ice.fault_diagnosis.utils"]], "milling (class in ice.health_index_estimation.datasets)": [[21, "ice.health_index_estimation.datasets.Milling"]], "ice.health_index_estimation.datasets": [[21, "module-ice.health_index_estimation.datasets"]], "set_name_public_link() (ice.health_index_estimation.datasets.milling method)": [[21, "ice.health_index_estimation.datasets.Milling.set_name_public_link"]], "ice.health_index_estimation.metrics": [[22, "module-ice.health_index_estimation.metrics"]], "mse() (in module ice.health_index_estimation.metrics)": [[22, "ice.health_index_estimation.metrics.mse"]], "basehealthindexestimation (class in ice.health_index_estimation.models.base)": [[23, "ice.health_index_estimation.models.base.BaseHealthIndexEstimation"]], "mlp (class in ice.health_index_estimation.models.mlp)": [[23, "ice.health_index_estimation.models.mlp.MLP"]], "evaluate() (ice.health_index_estimation.models.base.basehealthindexestimation method)": [[23, "ice.health_index_estimation.models.base.BaseHealthIndexEstimation.evaluate"]], "ice.health_index_estimation.models.base": [[23, "module-ice.health_index_estimation.models.base"]], "ice.health_index_estimation.models.mlp": [[23, "module-ice.health_index_estimation.models.mlp"]], "slidingwindowdataset (class in ice.health_index_estimation.utils)": [[24, "ice.health_index_estimation.utils.SlidingWindowDataset"]], "ice.health_index_estimation.utils": [[24, "module-ice.health_index_estimation.utils"]], "rulcmapss (class in ice.remaining_useful_life_estimation.datasets)": [[26, "ice.remaining_useful_life_estimation.datasets.RulCmapss"]], "ice.remaining_useful_life_estimation.datasets": [[26, "module-ice.remaining_useful_life_estimation.datasets"]], "set_name_public_link() (ice.remaining_useful_life_estimation.datasets.rulcmapss method)": [[26, "ice.remaining_useful_life_estimation.datasets.RulCmapss.set_name_public_link"]], "ice.remaining_useful_life_estimation.metrics": [[27, "module-ice.remaining_useful_life_estimation.metrics"]], "nonsimmetric_function() (in module ice.remaining_useful_life_estimation.metrics)": [[27, "ice.remaining_useful_life_estimation.metrics.nonsimmetric_function"]], "rmse() (in module ice.remaining_useful_life_estimation.metrics)": [[27, "ice.remaining_useful_life_estimation.metrics.rmse"]], "score() (in module ice.remaining_useful_life_estimation.metrics)": [[27, "ice.remaining_useful_life_estimation.metrics.score"]], "baseremainingusefullifeestimation (class in ice.remaining_useful_life_estimation.models.base)": [[28, "ice.remaining_useful_life_estimation.models.base.BaseRemainingUsefulLifeEstimation"]], "mlp (class in ice.remaining_useful_life_estimation.models.mlp)": [[28, "ice.remaining_useful_life_estimation.models.mlp.MLP"]], "evaluate() (ice.remaining_useful_life_estimation.models.base.baseremainingusefullifeestimation method)": [[28, "ice.remaining_useful_life_estimation.models.base.BaseRemainingUsefulLifeEstimation.evaluate"]], "ice.remaining_useful_life_estimation.models.base": [[28, "module-ice.remaining_useful_life_estimation.models.base"]], "ice.remaining_useful_life_estimation.models.mlp": [[28, "module-ice.remaining_useful_life_estimation.models.mlp"]], "slidingwindowdataset (class in ice.remaining_useful_life_estimation.utils)": [[29, "ice.remaining_useful_life_estimation.utils.SlidingWindowDataset"]], "ice.remaining_useful_life_estimation.utils": [[29, "module-ice.remaining_useful_life_estimation.utils"]]}}) \ No newline at end of file +Search.setIndex({"docnames": ["advanced/index", "advanced/optimization_tutorial", "benchmark/ad_benchmark_autoencodermlp_256", "benchmark/ad_benchmark_gnn", "benchmark/ad_benchmark_stgat", "benchmark/ad_benchmark_transformer", "benchmark/fd_benchmark_mlp_256", "benchmark/fd_benchmark_tcn", "benchmark/hi_benchmark_mlp_256", "benchmark/hi_sota", "benchmark/index", "benchmark/rul_benchmark_lstm_256", "benchmark/rul_sota", "custom/index", "index", "reference/ice.anomaly_detection", "reference/ice.anomaly_detection.datasets", "reference/ice.anomaly_detection.metrics", "reference/ice.anomaly_detection.models", "reference/ice.anomaly_detection.utils", "reference/ice.base", "reference/ice.configs", "reference/ice.fault_diagnosis", "reference/ice.fault_diagnosis.datasets", "reference/ice.fault_diagnosis.metrics", "reference/ice.fault_diagnosis.models", "reference/ice.fault_diagnosis.utils", "reference/ice.health_index_estimation", "reference/ice.health_index_estimation.datasets", "reference/ice.health_index_estimation.metrics", "reference/ice.health_index_estimation.models", "reference/ice.health_index_estimation.utils", "reference/ice.remaining_useful_life_estimation", "reference/ice.remaining_useful_life_estimation.datasets", "reference/ice.remaining_useful_life_estimation.metrics", "reference/ice.remaining_useful_life_estimation.models", "reference/ice.remaining_useful_life_estimation.utils", "reference/index", "start/index", "start/installation", "start/tasks/index", "start/tasks/task_ad", "start/tasks/task_fd", "start/tasks/task_hi", "start/tasks/task_rul", "start/tutorials/anomaly_detection_basic", "start/tutorials/fault_diagnosis_basic", "start/tutorials/health_index_estimation_basic", "start/tutorials/index", "start/tutorials/remaining_useful_life_estimation_basic"], "filenames": ["advanced\\index.rst", "advanced\\optimization_tutorial.ipynb", "benchmark\\ad_benchmark_autoencodermlp_256.ipynb", "benchmark\\ad_benchmark_gnn.ipynb", "benchmark\\ad_benchmark_stgat.ipynb", "benchmark\\ad_benchmark_transformer.ipynb", "benchmark\\fd_benchmark_mlp_256.ipynb", "benchmark\\fd_benchmark_tcn.ipynb", "benchmark\\hi_benchmark_mlp_256.ipynb", "benchmark\\hi_sota.ipynb", "benchmark\\index.rst", "benchmark\\rul_benchmark_lstm_256.ipynb", "benchmark\\rul_sota.ipynb", "custom\\index.rst", "index.rst", "reference\\ice.anomaly_detection.rst", "reference\\ice.anomaly_detection.datasets.rst", "reference\\ice.anomaly_detection.metrics.rst", "reference\\ice.anomaly_detection.models.rst", "reference\\ice.anomaly_detection.utils.rst", "reference\\ice.base.rst", "reference\\ice.configs.rst", "reference\\ice.fault_diagnosis.rst", "reference\\ice.fault_diagnosis.datasets.rst", "reference\\ice.fault_diagnosis.metrics.rst", "reference\\ice.fault_diagnosis.models.rst", "reference\\ice.fault_diagnosis.utils.rst", "reference\\ice.health_index_estimation.rst", "reference\\ice.health_index_estimation.datasets.rst", "reference\\ice.health_index_estimation.metrics.rst", "reference\\ice.health_index_estimation.models.rst", "reference\\ice.health_index_estimation.utils.rst", "reference\\ice.remaining_useful_life_estimation.rst", "reference\\ice.remaining_useful_life_estimation.datasets.rst", "reference\\ice.remaining_useful_life_estimation.metrics.rst", "reference\\ice.remaining_useful_life_estimation.models.rst", "reference\\ice.remaining_useful_life_estimation.utils.rst", "reference\\index.rst", "start\\index.rst", "start\\installation.rst", "start\\tasks\\index.rst", "start\\tasks\\task_ad.rst", "start\\tasks\\task_fd.rst", "start\\tasks\\task_hi.rst", "start\\tasks\\task_rul.rst", "start\\tutorials\\anomaly_detection_basic.ipynb", "start\\tutorials\\fault_diagnosis_basic.ipynb", "start\\tutorials\\health_index_estimation_basic.ipynb", "start\\tutorials\\index.rst", "start\\tutorials\\remaining_useful_life_estimation_basic.ipynb"], "titles": ["Advanced materials", "Optimization tutorial", "Results of anomaly detection using AutoEncoderMLP-256", "Results of anomaly detection using GSL-GNN", "Results of anomaly detection using STGAT-MAD", "Results of anomaly detection using AnomalyTransformer", "Results of fault diagnosis using MLP-256", "Results of fault diagnosis using TCN", "Results of HI estimation using MLP-256", "Results of HI estimation using Stacked LSTM", "Benchmarking", "Results of RUL estimation using lstm-256", "Results of RUL estimation using IR", "Custom classes", "ICE - Industrial Condition Evaluation", "Anomaly detection", "Anomaly detection datasets", "Anomaly detection metrics", "Anomaly detection models", "Anomaly detection utils", "Base modules", "Configs", "Fault diagnosis", "Fault diagnosis datasets", "Fault diagnosis metrics", "Fault diagnosis models", "Fault diagnosis utils", "Health index estimation", "Health index estimation datasets", "Health index estimation metrics", "Remaining useful life models", "Health index estimation utils", "Remaining useful life", "Remaining useful life datasets", "Remaining useful life metrics", "Health index models", "Remaining useful life utils", "API reference", "Getting started", "Installation", "Tasks", "Anomaly detection", "Fault diagnosis", "Health index", "Remaining useful lifetime", "Tutorial on anomaly detection task", "Tutorial on fault diagnosis task", "Tutorial on health index estimation task", "Quick tutorials", "Tutorial on remaining useful lifetime estimation task"], "terms": {"2nd": 13, "project": 13, "stage": 13, "problem": [13, 43, 44], "from": [1, 2, 3, 4, 5, 6, 7, 8, 9, 11, 12, 18, 20, 21, 28, 33, 34, 41, 42, 43, 44, 45, 46, 47, 49], "ic": [1, 2, 3, 4, 5, 6, 7, 8, 9, 11, 12, 16, 17, 18, 19, 20, 21, 23, 24, 25, 26, 28, 29, 30, 31, 33, 34, 35, 36, 39, 42, 45, 46, 47, 49], "remaining_useful_life_estim": [1, 11, 12, 33, 34, 35, 36, 49], "dataset": [1, 2, 3, 4, 5, 6, 7, 8, 9, 11, 12, 15, 18, 19, 20, 22, 25, 26, 27, 30, 31, 32, 34, 35, 36, 37, 45, 46, 47, 49], "import": [1, 2, 3, 4, 5, 6, 7, 8, 9, 11, 12, 41, 42, 45, 46, 47, 49], "rulcmapss": [1, 11, 12, 32, 33, 49], "model": [1, 2, 3, 4, 5, 6, 7, 8, 9, 11, 12, 14, 15, 20, 22, 27, 32, 33, 37, 42, 43, 44, 45, 46, 47, 49], "mlp": [1, 9, 10, 18, 22, 27, 32, 45, 46, 47], "c": [1, 9, 11, 12, 16, 18, 20, 23, 25, 30, 35, 41, 43, 49], "user": [1, 9, 11, 12, 46, 49], "conda": [1, 9, 11, 12, 49], "env": [1, 9, 11, 12, 46, 49], "ice_test": [1, 9, 12], "lib": [1, 9, 11, 12, 46, 49], "site": [1, 9, 11, 12, 46, 49], "packag": [1, 9, 11, 12, 46, 49], "tqdm": [1, 2, 6, 8, 9, 11, 12, 46, 47, 49], "auto": [1, 2, 6, 8, 9, 11, 12, 42, 46, 47, 49], "py": [1, 9, 11, 12, 46, 49], "21": [1, 6, 8, 9, 11, 12, 44, 45, 46, 47, 49], "tqdmwarn": [1, 9, 11, 12, 46, 49], "iprogress": [1, 9, 11, 12, 46, 49], "found": [1, 9, 11, 12, 42, 46, 49], "pleas": [1, 9, 11, 12, 46, 49], "updat": [1, 9, 11, 12, 46, 49], "jupyt": [1, 9, 11, 12, 46, 49], "ipywidget": [1, 9, 11, 12, 46, 49], "see": [1, 9, 11, 12, 42, 46, 49], "http": [1, 9, 11, 12, 16, 18, 23, 28, 39, 46, 49], "readthedoc": [1, 9, 11, 12, 46, 49], "io": [1, 9, 11, 12, 46, 49], "en": [1, 9, 11, 12, 46, 49], "stabl": [1, 9, 11, 12, 46, 49], "user_instal": [1, 9, 11, 12, 46, 49], "html": [1, 9, 11, 12, 16, 23, 46, 49], "autonotebook": [1, 9, 11, 12, 46, 49], "notebook_tqdm": [1, 9, 11, 12, 46, 49], "creat": [1, 3, 4, 5, 7, 8, 9, 11, 12, 14, 18, 20, 21, 41, 42, 45, 46, 47, 49], "class": [1, 8, 9, 11, 12, 14, 16, 18, 19, 20, 21, 23, 25, 26, 28, 30, 31, 33, 35, 36, 41, 47, 49], "dataset_class": [1, 8, 9, 11, 12, 47, 49], "data": [1, 3, 4, 5, 6, 7, 8, 9, 11, 12, 14, 16, 18, 20, 23, 28, 33, 41, 42, 43, 44, 45, 46, 47, 49], "target": [1, 2, 3, 4, 5, 6, 7, 8, 9, 11, 12, 17, 18, 19, 20, 24, 26, 29, 31, 33, 34, 36, 42, 43, 44, 45, 46, 47, 49], "df": [1, 2, 3, 4, 5, 6, 7, 8, 9, 11, 12, 18, 19, 20, 26, 31, 36, 45, 46, 47, 49], "0": [1, 2, 3, 4, 5, 6, 7, 8, 9, 11, 12, 18, 20, 25, 30, 35, 41, 42, 43, 44, 45, 46, 47, 49], "read": [1, 9, 12, 43, 44, 46], "mapss": [1, 12], "fd1_train": [1, 12], "csv": [1, 9, 12, 46], "100": [1, 7, 8, 9, 11, 12, 18, 42, 46, 47, 49], "20631": [1, 12], "00": [1, 8, 9, 11, 12, 42, 46, 47, 49], "517496": 1, "66it": [1, 8, 9, 12, 47], "": [1, 8, 9, 11, 12, 20, 21, 41, 42, 43, 44, 46, 47, 49], "fd2_train": [1, 12], "53759": [1, 12], "539376": 1, "73it": [1, 8, 9, 47], "fd3_train": [1, 12], "24720": [1, 12], "506176": 1, "62it": [1, 8, 9, 12, 47], "fd4_train": [1, 12], "61249": [1, 12], "539069": 1, "75it": [1, 8, 11, 12, 47, 49], "544734": 1, "35it": [1, 8, 47], "481589": 1, "37it": [1, 8, 9, 12, 47], "427933": 1, "68it": [1, 8, 9, 12, 47], "546792": 1, "80it": [1, 8, 47], "fd1_test": [1, 12], "13097": [1, 12], "505412": 1, "69it": [1, 8, 47], "fd2_test": [1, 12], "33991": [1, 12], "478933": 1, "98it": [1, 8, 9, 47], "fd3_test": [1, 12], "16598": [1, 12], "520419": [1, 12], "fd4_test": [1, 12], "41214": [1, 12], "537036": 1, "515225": 1, "24it": [1, 8, 12, 47], "520644": 1, "44it": [1, 8, 12, 47], "489809": 1, "10it": [1, 8, 11, 47, 49], "537023": 1, "31it": [1, 8, 9, 47], "devic": [1, 2, 3, 4, 5, 6, 7, 8, 9, 11, 18, 20, 25, 30, 33, 35, 41, 43, 47, 49], "cuda": [1, 2, 3, 4, 5, 6, 8, 9, 11, 18, 20, 25, 30, 35, 47, 49], "optim": [0, 20], "without": 1, "chang": [1, 43], "complex": [1, 41], "train": [1, 2, 6, 8, 9, 11, 12, 14, 18, 20, 25, 30, 35, 41, 42, 43, 45, 46, 47, 49], "process": [1, 2, 6, 8, 11, 16, 18, 23, 25, 41, 42, 43], "tune": 1, "lr": [1, 2, 5, 6, 7, 8, 9, 11, 18, 20, 25, 30, 35, 45, 46, 47, 49], "procedur": 1, "us": [1, 10, 14, 16, 18, 20, 23, 25, 28, 35, 37, 38, 39, 40, 41, 42, 43, 48], "valid": [1, 18, 20, 25, 30, 35, 43], "loss": [1, 8, 42, 45, 46, 47], "model_class": [1, 8, 9, 11, 12, 47, 49], "optimize_paramet": [1, 20], "optimize_rang": [1, 20], "direct": [1, 20], "n_trial": [1, 20], "epoch": [1, 8, 18, 20, 25, 30, 35, 45, 46, 47], "optimize_metr": [1, 20], "5e": [1, 30], "5": [1, 2, 6, 7, 8, 11, 20, 25, 35, 41, 42, 43, 44, 45, 46, 47, 49], "1e": [1, 7, 11, 49], "3": [1, 6, 7, 8, 11, 18, 41, 42, 43, 44, 45, 46, 47, 49], "minim": [1, 20], "i": [1, 2, 6, 7, 8, 11, 12, 17, 18, 20, 24, 25, 30, 33, 35, 41, 42, 43, 44, 47, 49], "none": [1, 9, 16, 18, 19, 20, 21, 23, 28, 33], "than": 1, "2024": 1, "08": [1, 7, 8, 42, 44, 46, 47], "13": [1, 6, 7, 8, 42, 44, 46, 47], "09": [1, 8, 11, 47, 49], "53": [1, 8, 45, 46, 47], "33": [1, 8, 11, 41, 42, 44, 45, 46, 47, 49], "784": 1, "A": [1, 8, 16, 18, 20, 21, 23, 33, 41, 42, 43, 44, 47], "new": 1, "studi": [1, 20, 44], "memori": [1, 35], "name": [1, 16, 18, 20, 23, 25, 28, 30, 33, 35, 45, 46], "parameter_lr": 1, "trial": [1, 18, 20], "step": [1, 8, 18, 20, 43, 47], "00018951382914416393": 1, "sequenc": [1, 8, 9, 11, 12, 18, 20, 21, 25, 30, 35, 42, 46, 47, 49], "sampl": [1, 8, 9, 11, 12, 17, 18, 20, 24, 25, 30, 35, 41, 43, 44, 45, 46, 47, 49], "33442": 1, "07it": [1, 8, 47], "232": [1, 44], "10": [1, 6, 7, 8, 16, 18, 23, 25, 33, 34, 41, 42, 44, 45, 46, 47], "24": [1, 6, 8, 45, 46, 47], "237": 1, "23it": [1, 8, 47], "43": [1, 7, 8, 44, 45, 46, 47], "99": [1, 8, 42, 47], "535": 1, "34it": [1, 8, 9, 47], "75": [1, 7, 8, 42, 44, 45, 46, 47], "174": [1, 8, 47], "631": 1, "00it": [1, 8, 47], "20": [1, 2, 6, 7, 8, 11, 42, 44, 45, 46, 47, 49], "1": [1, 3, 4, 5, 6, 7, 8, 11, 18, 20, 25, 30, 31, 35, 36, 43, 45, 46, 47, 49], "01": [1, 8, 11, 42, 46, 47, 49], "2": [1, 2, 6, 7, 8, 11, 18, 25, 35, 41, 42, 43, 44, 45, 46, 47, 49], "39": [1, 8, 42, 45, 46, 47], "3309": 1, "3238": 1, "metric": [1, 2, 3, 4, 5, 6, 7, 8, 11, 15, 20, 22, 27, 32, 37, 45, 46, 47, 49], "rmse": [1, 8, 9, 11, 12, 27, 29, 32, 34, 43, 44, 47, 49], "50": [1, 8, 30, 42, 46, 47], "063830627114406": 1, "cmapss_scor": [1, 12, 32, 34], "1009001": 1, "2641988464": 1, "30": [1, 3, 4, 5, 7, 8, 42, 44, 46, 47], "69": [1, 8, 47], "686": 1, "55it": [1, 8, 47], "62": [1, 8, 45, 46, 47], "143": [1, 8, 47], "715": 1, "14it": [1, 8, 46, 47], "94": [1, 7, 8, 41, 42, 44, 47], "218": [1, 8, 47], "730": [1, 45, 46], "32it": [1, 8, 11, 47, 49], "40": [1, 8, 11, 44, 45, 46, 47, 49], "54it": [1, 8, 47], "36": [1, 8, 45, 46, 47], "8534": 1, "32": [1, 2, 3, 4, 5, 7, 8, 11, 35, 42, 44, 47, 49], "2449": [1, 8, 47], "68462185124231": 1, "326161": 1, "11634528585": 1, "76": [1, 8, 44, 47], "751": 1, "21it": [1, 8, 47], "66": [1, 8, 47], "152": [1, 8, 47], "755": 1, "41it": [1, 8, 9, 47], "98": [1, 3, 4, 5, 8, 42, 47], "228": [1, 8, 45, 46, 47], "756": 1, "70it": [1, 8, 47], "60": [1, 7, 8, 11, 42, 45, 46, 47, 49], "22": [1, 6, 8, 42, 45, 46, 47], "9493": 1, "29": [1, 8, 11, 44, 47, 49], "7487": 1, "38": [1, 7, 8, 11, 45, 46, 47, 49], "11920046818511": 1, "251687": 1, "82706875": 1, "745": [1, 8, 47], "06it": [1, 8, 47], "65": [1, 8, 47], "151": [1, 8, 47], "750": 1, "90it": [1, 8, 47], "227": [1, 8, 47], "753": [1, 8, 47], "53it": [1, 8, 47], "80": [1, 7, 8, 11, 42, 46, 47, 49], "4": [1, 6, 7, 8, 11, 25, 41, 42, 44, 45, 46, 47, 49], "74it": [1, 8, 47], "27": [1, 6, 8, 46, 47], "0963": 1, "28": [1, 6, 8, 42, 45, 46, 47], "3393": [1, 8, 47], "542765927986274": 1, "207206": 1, "28175345066": 1, "74": [1, 8, 42, 45, 46, 47], "735": 1, "12it": [1, 8, 47], "752": 1, "28it": [1, 8, 47], "351": 1, "finish": 1, "valu": [1, 9, 17, 18, 20, 21, 24, 29, 33, 34, 41, 43, 44], "26": [1, 6, 7, 8, 42, 45, 46, 47], "89373407131288": 1, "paramet": [1, 8, 11, 16, 17, 18, 20, 21, 23, 24, 25, 28, 29, 30, 33, 34, 35, 43, 44, 47, 49], "best": [1, 14, 20], "0107": 1, "8937": 1, "34": [1, 8, 42, 45, 46, 47], "96109612817986": 1, "171362": 1, "21049284632": 1, "0007874022874638446": 1, "12542": 1, "02it": [1, 8, 47], "17": [1, 6, 7, 8, 42, 43, 44, 45, 46, 47], "389": 1, "48": [1, 8, 45, 46, 47], "111": [1, 8, 47], "581": 1, "84it": [1, 8, 47], "81": [1, 8, 41, 42, 46, 47], "187": [1, 8, 47], "661": 1, "48it": [1, 8, 9, 12, 47], "6104": 1, "4569": 1, "641393207296055": 1, "206445": 1, "81209835963": 1, "153": [1, 8, 47], "760": [1, 8, 47], "230": 1, "50it": [1, 8, 12, 47], "5834": 1, "3682": 1, "68363304759782": 1, "98649": 1, "45005642212": 1, "77": [1, 8, 41, 46, 47], "764": 1, "93it": [1, 8, 47], "154": [1, 8, 47], "498": 1, "91": [1, 7, 8, 41, 42, 47], "210": [1, 8, 47], "435": [1, 45, 46], "5193": 1, "9444": 1, "369740561776197": 1, "82673": 1, "16229614464": 1, "16": [1, 6, 7, 8, 42, 43, 44, 45, 46, 47], "37": [1, 8, 41, 44, 45, 46, 47], "363": 1, "96it": [1, 8, 47], "359": [1, 45, 46], "81it": [1, 8, 9, 47], "47": [1, 8, 11, 44, 45, 46, 47, 49], "110": [1, 8, 47], "355": [1, 8, 47], "65it": [1, 8, 47], "63": [1, 8, 45, 46, 47], "146": [1, 8, 47], "352": 1, "79": [1, 7, 8, 46, 47], "183": [1, 8, 47], "357": [1, 45, 46], "91it": [1, 8, 47], "219": [1, 8, 47], "02": [1, 8, 11, 42, 46, 47, 49], "72it": [1, 8, 47], "3577": 1, "1891": [1, 8, 47], "407311130072234": 1, "90375": 1, "15431472883": 1, "63it": [1, 8, 9, 47], "31": [1, 7, 8, 42, 47], "72": [1, 8, 47], "343": 1, "108": [1, 8, 47], "346": [1, 45, 46], "356": 1, "360": 1, "17it": [1, 8, 47], "95": [1, 7, 8, 18, 42, 44, 47], "220": [1, 8, 47], "358": 1, "45it": [1, 8, 12, 47], "76it": [1, 8, 47], "216": [1, 8, 47], "189056094099836": 1, "18": [1, 6, 7, 8, 42, 44, 45, 46, 47], "7968": 1, "4563": 1, "47353109963666": 1, "73324": 1, "81707642713": 1, "00012086132027556038": 1, "12541": 1, "15": [1, 6, 7, 8, 18, 25, 30, 35, 42, 47], "337": 1, "71": [1, 8, 47], "46": [1, 8, 11, 45, 46, 47, 49], "107": [1, 8, 47], "354": [1, 8, 47], "58it": [1, 8, 46, 47], "78": [1, 8, 44, 47], "180": [1, 8, 47], "217": [1, 8, 45, 46, 47], "82it": [1, 8, 47], "42": [1, 8, 18, 25, 30, 35, 44, 45, 46, 47], "6350": 1, "41": [1, 8, 42, 45, 46, 47], "5396": 1, "309621676839306": 1, "1652572": 1, "5547515503": 1, "25it": [1, 8, 47], "09it": [1, 8, 47], "144": [1, 8, 45, 46, 47], "181": [1, 8, 47], "01it": [1, 8, 47], "4397": 1, "1115": 1, "95681782704423": 1, "653010": 1, "5664245718": 1, "353": [1, 8, 47], "109": [1, 8, 47], "46it": [1, 8, 9, 47], "145": [1, 8, 47], "25": [1, 6, 8, 44, 47], "7715": 1, "9065": 1, "430416515203575": 1, "349818": 1, "50039921864": 1, "43it": [1, 8, 47], "15it": [1, 8, 47], "64": [1, 8, 9, 11, 18, 30, 35, 42, 44, 45, 46, 47, 49], "148": [1, 8, 47], "185": [1, 8, 47], "221": [1, 8, 44, 47], "30it": [1, 8, 47], "5199": 1, "6848": 1, "035223216989905": 1, "274871": 1, "78833234054": 1, "350": 1, "348": 1, "67it": [1, 8, 47], "332": 1, "92it": [1, 8, 47], "178": [1, 8, 47], "338": [1, 8, 47], "04it": [1, 8, 47], "92": [1, 8, 42, 44, 47], "214": [1, 8, 47], "03": [1, 8, 11, 42, 46, 47, 49], "980": [1, 8, 47], "60851450664241": 1, "5629": 1, "6085": 1, "93360429205696": 1, "242353": 1, "786303807": 1, "hyperparamet": [1, 20], "frozentri": 1, "number": [1, 17, 18, 20, 24, 25, 30, 35, 41, 42, 43, 44], "state": [1, 35, 41, 42, 43, 44], "datetime_start": 1, "datetim": 1, "8": [1, 6, 7, 8, 18, 41, 42, 43, 44, 45, 46, 47], "9": [1, 6, 7, 8, 41, 42, 44, 45, 46, 47], "351082": 1, "datetime_complet": 1, "216093": 1, "param": 1, "user_attr": 1, "system_attr": 1, "intermediate_valu": 1, "456903271558808": 1, "368176297443668": 1, "944359942180355": 1, "456260332247105": 1, "distribut": [1, 18, 44], "floatdistribut": 1, "high": 1, "001": [1, 2, 5, 6, 18, 25, 42, 45, 46], "log": 1, "fals": [1, 2, 6, 11, 12, 16, 17, 18, 20, 23, 24, 25, 28, 30, 33, 35, 41, 42, 49], "low": 1, "05": [1, 8, 30, 44, 46, 47], "trial_id": 1, "hidden": [1, 18, 25, 30, 35, 42], "dimens": [1, 18], "size": [1, 18, 20, 25, 30, 35], "mse": [1, 8, 9, 27, 29, 47], "hidden_dim": [1, 6, 7, 18, 25, 30, 35], "256": [1, 9, 10, 12, 18, 20, 25, 30, 35, 44, 45], "1024": [1, 9, 30], "992": [1, 8, 47], "parameter_hidden_dim": 1, "702": [1, 8, 47], "11148": [1, 12], "14": [1, 6, 7, 8, 9, 11, 42, 44, 47, 49], "315": [1, 8, 47], "67": [1, 8, 47], "333": [1, 44], "44": [1, 8, 44, 45, 46, 47], "102": 1, "59": [1, 8, 42, 44, 45, 46, 47], "137": [1, 8, 44, 45, 46, 47], "339": 1, "172": [1, 8, 47], "340": 1, "89": [1, 8, 41, 42, 47], "207": [1, 8, 47], "341": 1, "7009": 1, "0446": 1, "52": [1, 7, 8, 42, 45, 46, 47], "46324288978529": 1, "1459642": 1, "2102222845": 1, "29it": [1, 8, 47], "5571": 1, "35": [1, 8, 9, 11, 35, 41, 44, 45, 46, 47, 49], "5294": 1, "69688917194367": 1, "487095": 1, "93532787054": 1, "328": 1, "99it": [1, 8, 47], "335": [1, 8, 45, 46, 47], "103": [1, 8, 47], "342": [1, 8, 47], "138": [1, 8, 47], "173": [1, 8, 47], "90": [1, 8, 42, 46, 47], "208": [1, 8, 47], "89it": [1, 8, 46, 47], "0595": 1, "7725": 1, "124605880566875": 1, "305081": 1, "3116466362": 1, "344": 1, "70": [1, 8, 11, 46, 47, 49], "106": 1, "347": 1, "61": [1, 8, 44, 45, 46, 47], "142": [1, 8, 44, 45, 46, 47], "349": [1, 8, 47], "78it": [1, 8, 47], "7565": 1, "9569": 1, "28838337636238": 1, "252440": 1, "2810893617": 1, "314": 1, "324": 1, "97it": [1, 8, 47], "136": [1, 8, 47], "51it": [1, 8, 11, 47, 49], "73": [1, 8, 47], "170": [1, 8, 47], "330": 1, "88": [1, 7, 8, 11, 42, 46, 47, 49], "204": [1, 8, 47], "884": [1, 45, 46], "12774637917532": 1, "4343": 1, "8321": 1, "219477": 1, "3266793877": 1, "662": 1, "40it": [1, 8, 47], "334": [1, 8, 47], "45": [1, 8, 42, 44, 45, 46, 47], "104": 1, "83it": [1, 8, 9, 11, 47, 49], "139": [1, 8, 47], "209": [1, 8, 47], "26it": [1, 8, 47], "5648": 1, "1247": [1, 8, 47], "64622625632633": 1, "1497278": 1, "352392366": 1, "319": 1, "08it": [1, 8, 47], "101": 1, "13it": [1, 8, 9, 47], "27it": [1, 8, 11, 47, 49], "5466": 1, "6801": 1, "00773698916855": 1, "511059": 1, "8643841665": 1, "321": [1, 8, 47], "320": [1, 45, 46], "47it": [1, 8, 47], "95it": [1, 8, 47], "57": [1, 7, 8, 41, 42, 46, 47], "132": [1, 8, 47], "318": 1, "165": [1, 8, 47], "86": [1, 8, 46, 47], "200": [1, 8, 47], "5062": 1, "7830": 1, "07700663323968": 1, "305080": 1, "57834935177": 1, "323": [1, 44, 45, 46], "38it": [1, 8, 47], "171": [1, 8, 47], "345": [1, 46], "9049": 1, "9922": 1, "29525364434387": 1, "252989": 1, "93131717102": 1, "49it": [1, 8, 12, 47], "179": [1, 8, 47], "93": [1, 7, 8, 47], "215": [1, 8, 47], "841": [1, 8, 47], "207252080735245": 1, "7666": 1, "9132": 1, "221679": 1, "17640028123": 1, "879": 1, "94it": [1, 8, 47], "77it": [1, 8, 12, 47], "60it": [1, 8, 47], "2073": [1, 8, 47], "6210": 1, "51": [1, 7, 8, 42, 44, 47], "9431875404865": 1, "1331357": 1, "9658768093": 1, "4625": 1, "5596": 1, "41147259207038": 1, "421283": 1, "66798380984": 1, "52it": [1, 8, 47], "182": [1, 8, 47], "33it": [1, 8, 47], "23": [1, 6, 8, 25, 42, 44, 46, 47], "8398": 1, "9023": 1, "31126980464057": 1, "286766": 1, "446915035": 1, "2913": 1, "3355": [1, 8, 47], "68203820452062": 1, "236872": 1, "37425031906": 1, "105": [1, 46], "140": [1, 8, 47], "176": [1, 8, 45, 46, 47], "212": [1, 8, 44, 47], "54": [1, 8, 11, 42, 45, 46, 47, 49], "617": 1, "46475764089671": 1, "4287": 1, "2309": 1, "203393": 1, "37691007284": 1, "842965": 1, "617905": 1, "62102061946218": 1, "55964553646925": 1, "902342726544635": 1, "335512719503264": 1, "230895298283276": 1, "intdistribut": 1, "The": [1, 18, 20, 25, 30, 35, 41, 42, 43, 44], "result": [1, 10, 18, 20, 25, 30, 34, 35, 42, 43, 44], "ar": [1, 14, 18, 20, 25, 30, 35, 41, 43, 44], "print": [1, 2, 6, 7, 8, 11, 47, 49], "end": [1, 41, 42, 43, 44], "processand": 1, "save": [1, 18, 20], "output": [1, 18, 25, 30, 35, 41, 42], "task_nam": 1, "traininig": 1, "param_name_optim": 1, "folder": [1, 39], "thi": [2, 6, 8, 9, 11, 12, 16, 20, 23, 28, 33, 41, 43], "notebook": [2, 6, 8, 9, 11, 12], "present": [2, 6, 8, 9, 11, 12, 18, 41, 42, 44], "experiment": [2, 6, 8, 9, 11, 12], "tennesse": [2, 6, 8, 11, 16, 23, 25, 41, 42], "eastman": [2, 6, 8, 11, 16, 23, 25, 41, 42], "librari": [2, 6, 8, 9, 11, 12, 14, 39, 42, 43, 47, 49], "numpi": [2, 3, 4, 5, 6, 7, 8, 9, 11, 29, 34, 45, 46, 47, 49], "np": [2, 3, 4, 5, 6, 7, 8, 9, 11, 17, 24, 45, 46, 47, 49], "torch": [2, 6, 8, 9, 11, 12, 20, 47, 49], "sklearn": [2, 3, 4, 5, 6, 7, 9, 12, 45, 46], "preprocess": [2, 3, 4, 5, 6, 7, 9, 12, 28, 33, 45, 46], "standardscal": [2, 3, 4, 5, 6, 7, 9, 12, 45, 46], "trang": [2, 6, 8, 9, 11, 47, 49], "anomaly_detect": [2, 3, 4, 5, 16, 17, 18, 19, 45], "anomalydetectionreinartztep": [2, 15, 16], "anomalydetectionsmalltep": [2, 15, 16, 45], "download": [2, 3, 4, 5, 6, 7, 16, 20, 23, 28, 33, 39, 45, 46], "tep": [2, 6, 41], "calcul": [2, 6, 18, 20, 34, 41, 42], "random": [2, 6, 8, 9, 11, 12, 18, 20, 25, 30, 35, 47, 49], "manual_se": [2, 6, 8, 11, 47, 49], "window_s": [2, 3, 4, 5, 6, 7, 8, 9, 11, 18, 19, 20, 25, 26, 30, 31, 35, 36, 45, 46, 47, 49], "batch_siz": [2, 6, 7, 8, 9, 11, 18, 20, 25, 30, 35, 47, 49], "512": [2, 35, 45, 46], "num_epoch": [2, 3, 4, 5, 6, 7, 8, 9, 11, 18, 20, 25, 30, 35, 45, 47, 49], "verbos": [2, 3, 4, 5, 6, 7, 8, 9, 11, 18, 20, 25, 30, 35, 45, 46, 47, 49], "fit": [2, 6, 8, 9, 11, 18, 20, 45, 46, 47, 49], "train_mask": [2, 3, 4, 5, 6, 7, 45, 46], "append": [2, 6, 8, 11, 47, 49], "evalu": [2, 3, 4, 5, 6, 7, 8, 9, 11, 12, 16, 20, 23, 41, 42, 43, 44, 45, 46, 47, 49], "test_mask": [2, 3, 4, 5, 6, 7, 45, 46], "acc": [2, 6], "tpr": [2, 6, 7, 41], "fpr": [2, 6, 7, 41], "metrics_i": [2, 6, 8, 11, 47, 49], "accuraci": [2, 3, 4, 5, 6, 15, 17, 20, 22, 24, 41, 42, 45, 46], "true_positive_r": [2, 3, 4, 5, 6, 7, 15, 17, 22, 24, 45], "false_positive_r": [2, 3, 4, 5, 6, 7, 15, 17, 22, 24, 45], "tpr_mean": [2, 6], "tpr_std": [2, 6], "mean": [2, 6, 7, 8, 11, 20, 29, 34, 47, 49], "axi": [2, 6, 8, 11, 47, 49], "std": [2, 6, 8, 11, 20, 47, 49], "fpr_mean": [2, 6], "fpr_std": [2, 6], "f": [2, 6, 7, 8, 11, 12, 41, 42, 44, 47, 49], "4f": [2, 6], "rang": [2, 6, 20, 41], "len": [2, 6], "8028": 2, "0007": [2, 6], "7378": 2, "0022": 2, "0366": 2, "0034": 2, "anomalydetectionriethtep": [3, 4, 5, 15, 16], "gsl_gnn": [3, 18], "panda": [3, 4, 5, 7, 8, 9, 11, 12, 18, 20, 47, 49], "pd": [3, 4, 5, 7, 8, 9, 11, 12, 20, 47, 49], "normal": [3, 4, 5, 7, 17, 24, 41, 42], "scaler": [3, 4, 5, 6, 7, 9, 12, 45, 46], "fit_transform": [3, 4, 5, 6, 7, 9, 12, 45, 46], "transform": [3, 4, 5, 6, 7, 9, 12, 18, 41, 43, 44, 45, 46], "gnn": [4, 10, 15], "true": [3, 4, 5, 7, 8, 9, 12, 16, 17, 18, 20, 23, 24, 25, 28, 30, 33, 34, 35, 41, 42, 43, 44, 45, 46, 47], "val_ratio": [3, 4, 5, 7, 18, 20, 25, 30, 35], "save_checkpoint": [3, 4, 5, 7, 18, 20, 25, 30, 35], "threshold_level": [3, 4, 5, 18], "load": [3, 4, 5, 7, 18, 20], "checkpoint": [3, 4, 5, 7, 18, 20, 25, 30, 35], "load_checkpoint": [3, 4, 5, 7, 9, 12, 18, 20], "gnn_anomaly_detection_epoch_30": 3, "tar": [3, 4, 5, 7, 9, 12], "test": [3, 4, 5, 7, 8, 9, 11, 12, 16, 20, 23, 28, 33, 38, 41, 42, 43, 44, 45, 46, 47, 49], "8517518472906404": 3, "82365725": 3, "019373853211009175": 3, "stgat_mad": [4, 18], "stgat_anomaly_detection_epoch_30": 4, "861279659277504": 4, "8352755": 4, "019435206422018347": 4, "anomalytransform": [10, 15], "d_model": [5, 18], "e_lay": [5, 18], "d_ff": [5, 18], "dropout": [5, 18, 25, 35], "transformer_anomaly_detection_epoch_30": 5, "8588669950738916": 5, "830817625": 5, "012466169724770642": 5, "fault_diagnosi": [6, 7, 23, 24, 25, 26, 46], "faultdiagnosisreinartztep": [6, 22, 23], "standard": [6, 20, 44], "input": [6, 25, 30, 35, 43], "128": [6, 7, 8, 18, 20, 25, 45, 47], "cdr": [6, 42], "correct_daignosis_r": [6, 22, 24], "8500": 6, "0029": 6, "9582": 6, "0013": 6, "9948": 6, "0039": 6, "0000": [6, 7], "9951": 6, "0008": 6, "9457": 6, "0487": 6, "9963": 6, "0020": 6, "2423": [6, 8, 47], "0490": 6, "0009": 6, "0004": 6, "6": [6, 7, 8, 41, 42, 43, 44, 45, 46, 47], "9984": 6, "0002": 6, "7": [6, 7, 8, 41, 42, 44, 45, 46, 47], "9978": 6, "0001": [6, 7, 18, 35], "9782": 6, "0136": 6, "3068": 6, "1355": 6, "9754": 6, "0018": 6, "11": [6, 7, 8, 12, 41, 42, 44, 46, 47], "9539": 6, "0130": 6, "12": [6, 7, 8, 42, 44, 47], "9098": 6, "0086": 6, "9702": 6, "0074": 6, "9949": 6, "0006": 6, "2772": 6, "0309": 6, "9786": 6, "9635": 6, "0019": 6, "19": [6, 7, 8, 11, 41, 42, 44, 46, 47, 49], "9918": 6, "9739": 6, "5031": 6, "8086": 6, "0242": 6, "9890": 6, "0011": [6, 9], "9872": 6, "9696": 6, "0064": 6, "9742": 6, "0111": 6, "3768": [6, 8, 47], "0459": 6, "faultdiagnosisriethtep": [7, 22, 23], "tcn": [10, 22, 44], "num_lay": [7, 25, 35], "kernel_s": [7, 25], "cpu": [7, 18, 20, 25, 30, 35], "tcn_fault_diagnosis_epoch_30": 7, "idx": 7, "arrai": 7, "datafram": [7, 9, 12, 18, 19, 20, 26, 31, 36], "fault": [10, 16, 17, 37, 38, 40, 41, 48], "round": 7, "9675": 7, "9738": 7, "9643": 7, "9584": 7, "9731": 7, "9679": 7, "9691": 7, "9651": 7, "9788": 7, "9526": 7, "9418": 7, "9780": 7, "9752": 7, "9608": 7, "9357": 7, "9717": 7, "9482": 7, "averag": 7, "2f": 7, "96": [7, 8, 42, 44, 47], "97": [7, 8, 42, 47], "84": [7, 8, 41, 47], "82": [7, 8, 41, 42, 46, 47], "anomali": [8, 10, 11, 20, 23, 37, 38, 40, 42, 48], "detect": [8, 10, 11, 20, 23, 24, 25, 37, 38, 40, 42, 48], "autoencodermlp": [8, 10, 11, 15, 45], "health_index_estim": [8, 9, 28, 29, 30, 31, 47], "mill": [8, 9, 27, 28, 43, 47], "initi": [8, 9, 11, 12, 20, 47, 49], "split": [8, 9, 11, 12, 45, 46, 47, 49], "test_data": [8, 9, 11, 12, 47, 49], "test_target": [8, 9, 11, 12, 47, 49], "stride": [8, 9, 18, 20, 25, 30, 31, 35, 36, 47], "300": [8, 30, 47], "258": [8, 45, 46, 47], "0004999805761074147": [8, 9, 47], "concat": [8, 9, 47], "3242": [8, 47], "04": [8, 42, 47], "85": [8, 44, 47], "49": [8, 47], "55": [8, 9, 47], "4360": [8, 47], "07": [8, 42, 47], "6135": [8, 47], "1503": [8, 47], "5825": [8, 47], "4649": [8, 47], "58": [8, 42, 43, 44, 46, 47], "2802": [8, 47], "06": [8, 11, 42, 44, 47, 49], "56": [8, 11, 42, 44, 47, 49], "2823": [8, 47], "2896": [8, 47], "2463": [8, 47], "1831": [8, 47], "2223": [8, 47], "68": [8, 47], "2076": [8, 47], "1645": [8, 47], "1910": [8, 47], "1895": [8, 47], "1853": [8, 47], "1559": [8, 47], "1705": [8, 47], "1646": [8, 47], "1666": [8, 47], "1820": [8, 47], "1460": [8, 47], "1814": [8, 47], "1484": [8, 47], "83": [8, 47], "1506": [8, 47], "87": [8, 42, 44, 46, 47], "1427": [8, 47], "1213": [8, 47], "1406": [8, 47], "1382": [8, 47], "1606": [8, 47], "9334": [8, 47], "5012": [8, 47], "22it": [8, 47], "186": [8, 47], "61it": [8, 47], "169": [8, 44, 47], "163": [8, 47], "160": [8, 47], "167": [8, 47], "188": [8, 47], "18it": [8, 47], "192": [8, 47], "194": [8, 45, 46, 47], "190": [8, 47], "234": [8, 47], "64it": [8, 47], "253": [8, 9, 47], "175": [8, 47], "271": [8, 47], "292": [8, 47], "313": [8, 47], "375": [8, 47], "197": [8, 47], "396": [8, 47], "199": [8, 45, 46, 47], "416": [8, 47], "436": [8, 47], "455": [8, 47], "88it": [8, 47], "475": [8, 47], "496": [8, 47], "517": [8, 47], "538": [8, 47], "196": [8, 47], "559": [8, 47], "03it": [8, 47], "579": [8, 47], "195": [8, 47], "599": [8, 47], "618": [8, 47], "36it": [8, 47], "636": [8, 47], "166": [8, 47], "655": [8, 44, 45, 46, 47], "674": [8, 47], "693": [8, 47], "11it": [8, 47], "711": [8, 47], "729": [8, 47], "747": [8, 47], "765": [8, 45, 46, 47], "783": [8, 47], "800": [8, 47], "158": [8, 47], "818": [8, 47], "162": [8, 47], "836": [8, 47], "853": [8, 47], "871": [8, 47], "05it": [8, 11, 47, 49], "889": [8, 45, 46, 47], "907": [8, 47], "924": [8, 47], "944": [8, 47], "966": [8, 47], "989": [8, 47], "1012": [8, 47], "203": [8, 47], "1035": [8, 47], "1057": [8, 47], "1079": [8, 47], "205": [8, 47], "1102": [8, 47], "1125": [8, 47], "213": [8, 44, 47], "19it": [8, 47], "1147": [8, 47], "1170": [8, 47], "1192": [8, 47], "1214": [8, 47], "1237": [8, 47], "1259": [8, 47], "79it": [8, 9, 47], "1281": [8, 47], "85it": [8, 47], "1303": [8, 47], "42it": [8, 12, 47], "1323": [8, 47], "1343": [8, 47], "1363": [8, 47], "1383": [8, 47], "1403": [8, 47], "191": [8, 47], "1423": [8, 47], "1444": [8, 47], "1464": [8, 47], "57it": [8, 12, 47], "1504": [8, 47], "16it": [8, 46, 47], "1524": [8, 47], "1545": [8, 47], "193": [8, 47], "71it": [8, 9, 47], "1567": [8, 47], "1590": [8, 47], "1612": [8, 47], "1635": [8, 47], "1657": [8, 47], "1679": [8, 47], "1701": [8, 47], "1722": [8, 47], "1743": [8, 47], "1766": [8, 47], "1789": [8, 47], "211": [8, 44, 47], "1812": [8, 47], "1835": [8, 47], "1857": [8, 47], "1879": [8, 47], "1901": [8, 47], "1923": [8, 47], "1945": [8, 47], "1967": [8, 47], "1989": [8, 47], "86it": [8, 47], "2011": [8, 47], "2032": [8, 47], "201": [8, 47], "2053": [8, 47], "2075": [8, 47], "2099": [8, 47], "2121": [8, 47], "2144": [8, 47], "2166": [8, 47], "2187": [8, 47], "2209": [8, 47], "2232": [8, 47], "2255": [8, 47], "59it": [8, 47], "2278": [8, 47], "2300": [8, 47], "2322": [8, 47], "2344": [8, 47], "2366": [8, 47], "2388": [8, 47], "2410": [8, 47], "2432": [8, 47], "2454": [8, 47], "2475": [8, 47], "2498": [8, 47], "2520": [8, 47], "2543": [8, 47], "2565": [8, 47], "2587": [8, 47], "2609": [8, 47], "2630": [8, 47], "2652": [8, 47], "2673": [8, 47], "2695": [8, 47], "2716": [8, 47], "202": [8, 47], "2737": [8, 47], "2760": [8, 47], "2782": [8, 47], "2804": [8, 47], "2827": [8, 47], "2849": [8, 47], "2872": [8, 47], "2894": [8, 47], "2916": [8, 47], "2938": [8, 47], "2960": [8, 47], "2982": [8, 47], "3004": [8, 47], "3027": [8, 47], "3050": [8, 47], "3072": [8, 47], "3094": [8, 47], "3116": [8, 47], "3138": [8, 47], "3159": [8, 47], "3180": [8, 47], "3202": [8, 47], "3224": [8, 47], "3247": [8, 47], "3269": [8, 47], "3291": [8, 47], "3314": [8, 47], "3336": [8, 47], "3358": [8, 47], "3380": [8, 47], "3403": [8, 47], "3425": [8, 47], "3447": [8, 47], "3468": [8, 47], "3490": [8, 47], "3512": [8, 47], "3534": [8, 47], "3557": [8, 47], "3579": [8, 47], "3601": [8, 47], "3623": [8, 47], "3645": [8, 47], "3666": [8, 47], "198": [8, 47], "3686": [8, 47], "3707": [8, 47], "3728": [8, 47], "3749": [8, 47], "3770": [8, 47], "3792": [8, 47], "3814": [8, 47], "3838": [8, 47], "3862": [8, 47], "3886": [8, 47], "222": [8, 44, 47], "3909": [8, 47], "3932": [8, 47], "3955": [8, 47], "3977": [8, 47], "3999": [8, 47], "4022": [8, 47], "4044": [8, 47], "4066": [8, 47], "4088": [8, 47], "4111": [8, 47], "4133": [8, 47], "4155": [8, 47], "4177": [8, 47], "4199": [8, 47], "4220": [8, 47], "4240": [8, 47], "4259": [8, 47], "4278": [8, 47], "56it": [8, 47], "4297": [8, 47], "184": [8, 47], "4317": [8, 47], "4337": [8, 47], "87it": [8, 9, 12, 47], "4358": [8, 47], "4379": [8, 47], "4401": [8, 47], "4422": [8, 47], "4445": [8, 47], "206": [8, 47], "4468": [8, 47], "39it": [8, 46, 47], "4490": [8, 47], "4512": [8, 45, 46, 47], "4534": [8, 47], "4556": [8, 45, 46, 47], "4579": [8, 47], "4601": [8, 46, 47], "4622": [8, 47], "4643": [8, 47], "4663": [8, 47], "189": [8, 47], "4683": [8, 47], "4705": [8, 47], "4725": [8, 47], "4746": [8, 47], "4766": [8, 47], "4786": [8, 47], "4805": [8, 47], "4825": [8, 47], "4844": [8, 47], "4865": [8, 47], "4887": [8, 47], "4909": [8, 47], "4930": [8, 47], "4951": [8, 47], "4973": [8, 47], "4995": [8, 47], "1548": [8, 47], "5213": [8, 47], "4789": [8, 47], "4105": [8, 47], "3222": [8, 47], "3484": [8, 47], "2422": [8, 47], "2523": [8, 47], "1962": [8, 47], "1405": [8, 47], "1468": [8, 47], "1552": [8, 47], "1541": [8, 47], "1692": [8, 47], "2091": [8, 47], "1865": [8, 47], "1463": [8, 47], "1321": [8, 47], "1577": [8, 47], "1436": [8, 47], "1741": [8, 47], "1224": [8, 47], "1318": [8, 47], "1435": [8, 47], "1441": [8, 47], "1509": [8, 47], "1437": [8, 47], "1411": [8, 47], "9369": [8, 41, 47], "134": [8, 47], "141": [8, 47], "131": [8, 47], "130": [8, 47], "119": [8, 47], "149": [8, 16, 23, 41, 42, 47], "164": [8, 47], "225": [8, 45, 46, 47], "240": [8, 47], "255": [8, 47], "270": [8, 47], "286": [8, 47], "301": [8, 47], "133": [8, 47], "329": [8, 47], "127": [8, 47], "124": [8, 47], "121": [8, 47], "368": [8, 47], "120": [8, 45, 46, 47], "381": [8, 47], "118": [8, 47], "394": [8, 45, 46, 47], "407": [8, 47], "420": [8, 47], "432": [8, 47], "117": [8, 47], "444": [8, 47], "456": [8, 47], "116": [8, 45, 46, 47], "468": [8, 45, 46, 47], "480": [8, 47], "492": [8, 47], "114": [8, 47], "505": [8, 47], "20it": [8, 9, 47], "529": [8, 47], "541": [8, 45, 46, 47], "553": [8, 47], "566": [8, 47], "593": [8, 47], "606": [8, 47], "619": [8, 47], "122": [8, 47], "632": [8, 47], "123": [8, 47], "645": [8, 45, 46, 47], "658": [8, 45, 46, 47], "670": [8, 47], "682": [8, 47], "694": [8, 47], "707": [8, 47], "719": [8, 47], "731": [8, 47], "773": [8, 47], "787": [8, 47], "126": [8, 47], "802": [8, 47], "817": [8, 47], "832": [8, 47], "847": [8, 47], "862": [8, 47], "876": [8, 47], "890": [8, 47], "904": [8, 47], "917": [8, 47], "125": [8, 44, 47], "930": [8, 47], "943": [8, 47], "956": [8, 45, 46, 47], "968": [8, 47], "1005": [8, 47], "1018": [8, 47], "1031": [8, 47], "1043": [8, 47], "1055": [8, 47], "1068": [8, 47], "1083": [8, 47], "1098": [8, 47], "1112": [8, 47], "1126": [8, 47], "1141": [8, 47], "135": [8, 47], "1156": [8, 47], "1171": [8, 47], "1186": [8, 47], "1201": [8, 47], "1216": [8, 47], "1231": [8, 47], "1246": [8, 47], "1261": [8, 47], "1276": [8, 47], "1291": [8, 47], "1306": [8, 47], "1336": [8, 47], "1351": [8, 47], "1366": [8, 47], "1381": [8, 47], "1396": [8, 47], "1442": [8, 47], "1457": [8, 47], "1472": [8, 47], "1487": [8, 47], "1502": [8, 47], "1517": [8, 47], "1531": [8, 47], "1574": [8, 47], "1589": [8, 47], "1604": [8, 47], "1619": [8, 47], "1634": [8, 47], "1649": [8, 47], "1664": [8, 47], "1694": [8, 47], "1709": [8, 47], "1723": [8, 47], "1737": [8, 47], "1752": [8, 47], "1767": [8, 47], "1782": [8, 47], "1797": [8, 47], "1811": [8, 47], "1825": [8, 47], "1839": [8, 47], "1866": [8, 47], "1892": [8, 47], "1904": [8, 47], "1916": [8, 47], "1928": [8, 47], "1941": [8, 47], "1956": [8, 47], "1969": [8, 47], "1982": [8, 47], "1995": [8, 47], "2010": [8, 47], "2025": [8, 47], "2040": [8, 47], "2055": [8, 47], "2069": [8, 47], "2083": [8, 47], "2098": [8, 47], "2112": [8, 47], "2128": [8, 47], "2143": [8, 47], "2159": [8, 47], "2175": [8, 47], "150": [8, 18, 47], "2191": [8, 47], "2206": [8, 47], "147": [8, 47], "2221": [8, 47], "2237": [8, 47], "2252": [8, 47], "2268": [8, 47], "2283": [8, 47], "2298": [8, 47], "2313": [8, 47], "2329": [8, 47], "2359": [8, 47], "2374": [8, 47], "2389": [8, 47], "2404": [8, 47], "2419": [8, 47], "2434": [8, 47], "2464": [8, 47], "2479": [8, 47], "2495": [8, 47], "2510": [8, 47], "2525": [8, 47], "2540": [8, 47], "2555": [8, 47], "2570": [8, 47], "2585": [8, 47], "2600": [8, 47], "2615": [8, 47], "2645": [8, 47], "2660": [8, 47], "2675": [8, 47], "2690": [8, 47], "2705": [8, 45, 46, 47], "2720": [8, 47], "2735": [8, 47], "2751": [8, 47], "2766": [8, 47], "2797": [8, 47], "2812": [8, 47], "2842": [8, 47], "2857": [8, 47], "2887": [8, 47], "2903": [8, 47], "2918": [8, 47], "2934": [8, 47], "2952": [8, 47], "155": [8, 47], "2968": [8, 47], "2984": [8, 47], "3000": [8, 47], "3015": [8, 47], "3030": [8, 47], "3045": [8, 47], "3059": [8, 47], "3087": [8, 47], "3101": [8, 47], "3115": [8, 47], "3128": [8, 47], "3141": [8, 47], "3154": [8, 47], "3168": [8, 47], "3181": [8, 47], "3195": [8, 47], "3210": [8, 47], "3225": [8, 47], "3241": [8, 47], "3256": [8, 47], "3271": [8, 47], "3287": [8, 47], "3302": [8, 47], "3317": [8, 47], "3332": [8, 47], "3347": [8, 47], "3361": [8, 47], "3375": [8, 47], "3389": [8, 47], "3417": [8, 47], "3431": [8, 47], "3446": [8, 47], "3461": [8, 47], "3475": [8, 47], "3489": [8, 47], "3503": [8, 47], "3517": [8, 47], "3531": [8, 47], "3546": [8, 47], "3561": [8, 47], "3576": [8, 47], "3591": [8, 47], "3606": [8, 47], "3621": [8, 47], "3636": [8, 47], "3651": [8, 47], "3665": [8, 47], "129": [8, 47], "3679": [8, 45, 46, 47], "3692": [8, 45, 46, 47], "3705": [8, 47], "3718": [8, 47], "3731": [8, 47], "3744": [8, 47], "3759": [8, 47], "3773": [8, 47], "3786": [8, 47], "3799": [8, 47], "3812": [8, 47], "3825": [8, 47], "3839": [8, 47], "3852": [8, 47], "3865": [8, 47], "3878": [8, 47], "3892": [8, 47], "3905": [8, 47], "3918": [8, 47], "3931": [8, 47], "3944": [8, 47], "3957": [8, 47], "3970": [8, 47], "3983": [8, 47], "3996": [8, 47], "4010": [8, 47], "4023": [8, 47], "4036": [8, 47], "4049": [8, 47], "4062": [8, 47], "4075": [8, 47], "4101": [8, 47], "4114": [8, 47], "4129": [8, 47], "4144": [8, 47], "4159": [8, 47], "4175": [8, 47], "4190": [8, 47], "4205": [8, 47], "4221": [8, 47], "4236": [8, 47], "4251": [8, 47], "4266": [8, 47], "4281": [8, 47], "4299": [8, 47], "4316": [8, 47], "159": [8, 47], "4332": [8, 47], "4348": [8, 47], "4363": [8, 47], "4377": [8, 47], "4390": [8, 47], "4407": [8, 47], "4427": [8, 47], "4451": [8, 47], "4476": [8, 45, 46, 47], "4500": [8, 47], "4522": [8, 47], "4544": [8, 47], "4565": [8, 47], "4584": [8, 47], "4602": [8, 47], "4621": [8, 47], "4639": [8, 47], "4657": [8, 47], "4675": [8, 47], "4693": [8, 47], "4711": [8, 47], "161": [8, 47], "4728": [8, 47], "157": [8, 47], "4744": [8, 47], "4759": [8, 47], "4773": [8, 47], "4787": [8, 47], "4800": [8, 47], "4813": [8, 47], "4826": [8, 47], "4839": [8, 47], "4852": [8, 47], "4878": [8, 47], "4891": [8, 47], "4903": [8, 47], "4916": [8, 47], "4929": [8, 47], "4941": [8, 47], "4953": [8, 47], "4965": [8, 47], "4977": [8, 47], "4991": [8, 47], "5004": [8, 47], "1757": [8, 47], "2985": [8, 47], "1235": [8, 47], "2092": [8, 47], "9648": [8, 47], "5663": [8, 47], "4703": [8, 47], "2911": [8, 47], "2339": [8, 47], "2370": [8, 47], "1976": [8, 47], "2120": [8, 47], "1964": [8, 47], "1561": [8, 47], "1860": [8, 47], "1424": [8, 47], "1837": [8, 47], "1698": [8, 47], "1320": [8, 47], "1764": [8, 47], "1626": [8, 47], "1674": [8, 47], "1252": [8, 47], "1327": [8, 47], "1693": [8, 47], "1379": [8, 44, 47], "1404": [8, 47], "1775": [8, 47], "6999": [8, 47], "113": [8, 47], "112": [8, 47], "115": [8, 47], "247": [8, 47], "260": [8, 47], "273": [8, 47], "285": [8, 47], "297": [8, 47], "309": [8, 47], "336": [8, 47], "362": [8, 47], "387": [8, 47], "399": [8, 47], "411": [8, 47], "424": [8, 47], "449": [8, 47], "461": [8, 45, 46, 47], "473": [8, 47], "485": [8, 47], "497": [8, 47], "509": [8, 47], "521": [8, 47], "533": [8, 47], "545": [8, 47], "557": [8, 47], "571": [8, 47], "584": [8, 47], "596": [8, 47], "608": [8, 47], "620": [8, 47], "644": [8, 47], "656": [8, 47], "668": [8, 47], "681": [8, 47], "705": [8, 47], "717": [8, 45, 46, 47], "741": [8, 47], "777": [8, 46, 47], "790": [8, 47], "804": [8, 47], "829": [8, 47], "865": [8, 47], "877": [8, 47], "901": [8, 47], "915": [8, 47], "928": [8, 47], "940": [8, 47], "952": [8, 47], "964": [8, 47], "976": [8, 47], "988": [8, 47], "1000": [8, 47], "1025": [8, 47], "1037": [8, 47], "1049": [8, 47], "1061": [8, 47], "1073": [8, 47], "1085": [8, 47], "1097": [8, 47], "1109": [8, 33, 34, 47], "1121": [8, 47], "1133": [8, 47], "1160": [8, 47], "1173": [8, 47], "1199": [8, 47], "1211": [8, 47], "1223": [8, 47], "1260": [8, 47], "1274": [8, 47], "1287": [8, 47], "1300": [8, 47], "1313": [8, 47], "1326": [8, 47], "1339": [8, 47], "1376": [8, 47], "1391": [8, 47], "1417": [8, 47], "1430": [8, 47], "1443": [8, 47], "1456": [8, 47], "1480": [8, 47], "1492": [8, 47], "1519": [8, 47], "1532": [8, 47], "1544": [8, 47], "1556": [8, 47], "1568": [8, 47], "1580": [8, 47], "1592": [8, 47], "1617": [8, 47], "1630": [8, 47], "1642": [8, 47], "1654": [8, 47], "1678": [8, 47], "1690": [8, 47], "1702": [8, 47], "1714": [8, 47], "1726": [8, 47], "1739": [8, 47], "1776": [8, 47], "1788": [8, 47], "1800": [8, 47], "1824": [8, 47], "1836": [8, 47], "1848": [8, 47], "1861": [8, 47], "1874": [8, 47], "1886": [8, 47], "1898": [8, 47], "1935": [8, 47], "1947": [8, 47], "1959": [8, 47], "1972": [8, 47], "1984": [8, 47], "1996": [8, 47], "2008": [8, 33, 34, 44, 47], "2020": [8, 41, 42, 43, 44, 47], "2045": [8, 47], "2057": [8, 47], "2081": [8, 47], "2095": [8, 47], "2108": [8, 47], "2133": [8, 47], "2145": [8, 47], "2157": [8, 47], "2169": [8, 47], "2181": [8, 47], "2193": [8, 47], "2218": [8, 47], "2230": [8, 47], "2242": [8, 47], "2267": [8, 47], "2279": [8, 47], "2291": [8, 47], "2303": [8, 47], "2315": [8, 47], "2327": [8, 47], "2351": [8, 47], "2363": [8, 47], "2375": [8, 47], "2387": [8, 47], "2399": [8, 47], "2411": [8, 47], "2435": [8, 47], "2448": [8, 47], "2460": [8, 47], "2472": [8, 47], "2484": [8, 47], "2496": [8, 47], "2508": [8, 47], "2532": [8, 47], "2544": [8, 47], "2558": [8, 47], "2582": [8, 47], "2594": [8, 47], "2606": [8, 47], "2618": [8, 47], "2643": [8, 47], "2655": [8, 47], "2667": [8, 47], "2681": [8, 47], "2694": [8, 47], "2707": [8, 45, 46, 47], "2719": [8, 47], "2731": [8, 47], "2743": [8, 47], "2755": [8, 47], "2767": [8, 47], "2779": [8, 47], "2793": [8, 47], "2806": [8, 47], "2818": [8, 47], "2830": [8, 47], "2854": [8, 47], "2866": [8, 47], "2878": [8, 47], "2890": [8, 47], "2902": [8, 47], "2915": [8, 47], "2928": [8, 47], "2940": [8, 47], "2964": [8, 47], "2976": [8, 47], "2988": [8, 47], "3012": [8, 47], "3024": [8, 47], "3037": [8, 47], "3049": [8, 47], "3061": [8, 47], "3073": [8, 47], "3085": [8, 47], "3097": [8, 47], "3109": [8, 47], "3121": [8, 47], "3133": [8, 47], "3147": [8, 47], "3160": [8, 47], "3173": [8, 47], "3185": [8, 47], "3198": [8, 47], "3211": [8, 47], "3223": [8, 47], "3235": [8, 47], "3262": [8, 47], "3275": [8, 47], "3288": [8, 47], "3301": [8, 47], "3326": [8, 47], "3338": [8, 47], "3350": [8, 47], "3362": [8, 47], "3388": [8, 47], "3401": [8, 47], "3413": [8, 47], "3437": [8, 47], "3449": [8, 47], "3473": [8, 47], "3485": [8, 47], "3499": [8, 47], "3524": [8, 47], "3536": [8, 47], "3548": [8, 47], "3560": [8, 47], "3572": [8, 47], "3584": [8, 46, 47], "3596": [8, 47], "3608": [8, 47], "3633": [8, 47], "3657": [8, 47], "3669": [8, 47], "3681": [8, 47], "3693": [8, 47], "3717": [8, 45, 46, 47], "3730": [8, 47], "3757": [8, 47], "3782": [8, 47], "3794": [8, 47], "3806": [8, 47], "3818": [8, 47], "3830": [8, 47], "3843": [8, 47], "3855": [8, 47], "3868": [8, 47], "3881": [8, 47], "3893": [8, 47], "3917": [8, 47], "3929": [8, 47], "3941": [8, 46, 47], "3953": [8, 47], "3966": [8, 47], "3978": [8, 47], "3990": [8, 47], "4002": [8, 47], "4014": [8, 47], "4026": [8, 47], "4038": [8, 47], "4050": [8, 47], "4074": [8, 47], "4086": [8, 47], "4098": [8, 47], "4110": [8, 47], "4122": [8, 47], "4134": [8, 47], "4146": [8, 47], "4158": [8, 47], "4170": [8, 47], "4182": [8, 47], "4195": [8, 47], "4208": [8, 47], "4232": [8, 47], "4244": [8, 47], "4256": [8, 47], "4268": [8, 47], "4280": [8, 47], "4292": [8, 47], "4304": [8, 47], "4328": [8, 47], "4340": [8, 47], "4352": [8, 47], "4364": [8, 47], "4376": [8, 47], "4388": [8, 47], "4400": [8, 47], "4412": [8, 47], "4424": [8, 47], "4437": [8, 47], "4449": [8, 47], "4461": [8, 47], "4473": [8, 47], "4486": [8, 47], "4498": [8, 47], "4510": [8, 47], "4547": [8, 47], "4559": [8, 47], "4571": [8, 47], "4583": [8, 47], "4595": [8, 47], "4607": [8, 47], "4619": [8, 47], "4631": [8, 47], "4655": [8, 47], "4669": [8, 47], "4682": [8, 47], "4694": [8, 47], "4706": [8, 47], "4718": [8, 47], "4730": [8, 47], "4742": [8, 47], "4754": [8, 47], "4779": [8, 47], "4792": [8, 47], "4817": [8, 47], "4829": [8, 47], "4841": [8, 47], "4853": [8, 47], "4877": [8, 47], "4889": [8, 47], "4902": [8, 47], "4914": [8, 47], "4926": [8, 47], "4938": [8, 47], "4950": [8, 47], "4962": [8, 47], "4974": [8, 47], "4986": [8, 47], "4998": [8, 47], "5011": [8, 47], "2009": [8, 47], "2431": [8, 47], "9911": [8, 47], "8032": [8, 47], "5517": [8, 47], "4578": [8, 47], "3316": [8, 47], "2741": [8, 47], "1925": [8, 47], "2086": [8, 47], "1990": [8, 47], "1613": [8, 47], "1948": [8, 47], "1823": [8, 47], "1605": [8, 44, 47], "1658": [8, 47], "1288": [8, 47], "1358": [8, 47], "1284": [8, 47], "1371": [8, 44, 47], "1298": [8, 47], "1638": [8, 47], "1539": [8, 47], "1550": [8, 47], "1402": [8, 47], "1176": [8, 47], "9333": [8, 47], "168": [8, 47], "223": [8, 44, 47], "242": [8, 47], "261": [8, 43, 47], "177": [8, 47], "280": [8, 47], "298": [8, 47], "316": [8, 47], "372": [8, 47], "392": [8, 47], "431": [8, 47], "450": [8, 47], "469": [8, 45, 46, 47], "488": [8, 47], "506": [8, 47], "524": [8, 47], "542": [8, 47], "561": [8, 47], "580": [8, 45, 46, 47], "654": [8, 47], "671": [8, 47], "691": [8, 47], "709": [8, 47], "728": [8, 47], "748": [8, 47], "767": [8, 47], "786": [8, 47], "805": [8, 47], "824": [8, 47], "842": [8, 45, 46, 47], "860": [8, 47], "878": [8, 47], "898": [8, 45, 46, 47], "936": [8, 47], "955": [8, 45, 46, 47], "974": [8, 47], "994": [8, 47], "1013": [8, 47], "1088": [8, 47], "1108": [8, 47], "1128": [8, 47], "1166": [8, 47], "1185": [8, 47], "1203": [8, 47], "1221": [8, 47], "1241": [8, 47], "1278": [8, 47], "1297": [8, 47], "1316": [8, 47], "1335": [8, 47], "1354": [8, 47], "1373": [8, 47], "1450": [8, 47], "1470": [8, 47], "1489": [8, 47], "1508": [8, 47], "1527": [8, 47], "1546": [8, 47], "1564": [8, 47], "1583": [8, 47], "1602": [8, 47], "1621": [8, 47], "1640": [8, 47], "1676": [8, 47], "1712": [8, 47], "1729": [8, 47], "1747": [8, 47], "1785": [8, 47], "1804": [8, 47], "1842": [8, 44, 47], "1897": [8, 47], "1914": [8, 47], "1934": [8, 47], "1954": [8, 47], "1973": [8, 47], "1993": [8, 47], "2013": [8, 47], "2033": [8, 47], "2111": [8, 47], "2130": [8, 47], "2149": [8, 47], "2168": [8, 47], "2188": [8, 47], "2208": [8, 47], "2228": [8, 47], "2247": [8, 47], "2265": [8, 47], "2284": [8, 47], "2323": [8, 47], "2343": [8, 47], "2383": [8, 47], "2403": [8, 47], "2441": [8, 47], "2459": [8, 47], "2478": [8, 47], "2516": [8, 47], "2535": [8, 47], "2576": [8, 47], "2596": [8, 47], "2633": [8, 47], "2670": [8, 47], "2709": [8, 47], "2728": [8, 47], "2748": [8, 47], "2768": [8, 47], "2788": [8, 47], "2807": [8, 47], "2825": [8, 47], "2845": [8, 47], "2863": [8, 47], "2883": [8, 47], "2922": [8, 47], "2941": [8, 47], "2979": [8, 47], "2997": [8, 47], "3016": [8, 47], "3034": [8, 47], "3053": [8, 47], "3091": [8, 47], "3110": [8, 47], "3129": [8, 47], "3148": [8, 47], "3166": [8, 47], "3183": [8, 47], "3201": [8, 47], "3220": [8, 47], "3240": [8, 47], "3259": [8, 47], "3278": [8, 47], "3298": [8, 47], "3354": [8, 47], "3373": [8, 47], "3432": [8, 47], "3451": [8, 47], "3470": [8, 47], "3508": [8, 47], "3526": [8, 47], "3545": [8, 47], "3565": [8, 47], "3604": [8, 47], "3642": [8, 47], "3661": [8, 45, 46, 47], "3680": [8, 47], "3698": [8, 47], "3716": [8, 47], "3736": [8, 45, 46, 47], "3755": [8, 47], "3775": [8, 47], "3813": [8, 47], "3833": [8, 47], "3871": [8, 47], "3889": [8, 47], "3928": [8, 47], "3947": [8, 47], "3967": [8, 47], "3986": [8, 47], "4005": [8, 47], "4024": [8, 47], "4042": [8, 47], "4060": [8, 47], "4080": [8, 47], "4099": [8, 47], "4119": [8, 47], "4138": [8, 47], "4157": [8, 47], "4176": [8, 47], "4196": [8, 47], "4215": [8, 47], "4234": [8, 47], "4252": [8, 47], "4271": [8, 47], "4291": [8, 47], "4310": [8, 47], "4329": [8, 47], "4349": [8, 47], "4369": [8, 47], "4389": [8, 47], "4409": [8, 47], "4428": [8, 47], "4448": [8, 47], "4488": [8, 47], "4508": [8, 47], "4528": [8, 47], "4566": [8, 47], "4620": [8, 47], "4658": [8, 47], "4677": [8, 47], "4696": [8, 47], "4715": [8, 47], "4735": [8, 47], "4772": [8, 47], "4790": [8, 47], "4810": [8, 47], "4828": [8, 47], "4847": [8, 47], "4867": [8, 47], "4906": [8, 47], "4925": [8, 47], "4943": [8, 47], "4961": [8, 47], "4980": [8, 47], "4999": [8, 47], "1940": [8, 47], "9672": [8, 47], "6776": [8, 47], "8219": [8, 47], "5063": [8, 47], "2717": [8, 47], "2034": [8, 47], "2023": [8, 41, 44, 47], "1433": [8, 47], "1670": [8, 47], "1471": [8, 47], "1821": [8, 47], "1520": [8, 47], "1772": [8, 47], "1271": [8, 47], "1096": [8, 47], "1419": [8, 47], "1618": [8, 47], "1372": [8, 47], "1558": [8, 47], "1486": [8, 47], "246": [8, 47], "224": [8, 45, 46, 47], "269": [8, 47], "226": [8, 47], "361": [8, 47], "383": [8, 47], "405": [8, 47], "427": [8, 45, 46, 47], "471": [8, 47], "494": [8, 47], "516": [8, 47], "679": [8, 45, 46, 47], "725": [8, 45, 46, 47], "770": [8, 47], "792": [8, 47], "814": [8, 47], "858": [8, 47], "880": [8, 47], "902": [8, 47], "946": [8, 45, 46, 47], "990": [8, 45, 46, 47], "1034": [8, 47], "1056": [8, 47], "1078": [8, 47], "1100": [8, 47], "1123": [8, 47], "1146": [8, 47], "1168": [8, 47], "1190": [8, 47], "1212": [8, 47], "1234": [8, 47], "1256": [8, 47], "1277": [8, 47], "1299": [8, 47], "1342": [8, 47], "1365": [8, 47], "1387": [8, 47], "1410": [8, 47], "1526": [8, 47], "1570": [8, 47], "1615": [8, 47], "1637": [8, 47], "1659": [8, 47], "1681": [8, 47], "1703": [8, 47], "1725": [8, 47], "1770": [8, 47], "1792": [8, 47], "1858": [8, 47], "1880": [8, 47], "1903": [8, 47], "1926": [8, 47], "1949": [8, 47], "2018": [8, 41, 42, 47], "2041": [8, 47], "2063": [8, 47], "2109": [8, 47], "2132": [8, 47], "2155": [8, 47], "2177": [8, 47], "2199": [8, 47], "2243": [8, 47], "2287": [8, 47], "2310": [8, 47], "2332": [8, 47], "2354": [8, 47], "2376": [8, 47], "2398": [8, 47], "2421": [8, 47], "2444": [8, 47], "2467": [8, 47], "2490": [8, 47], "2513": [8, 47], "2536": [8, 47], "2560": [8, 47], "2583": [8, 44, 47], "2605": [8, 47], "2627": [8, 47], "2649": [8, 47], "2671": [8, 47], "2739": [8, 47], "2762": [8, 47], "2785": [8, 47], "2808": [8, 47], "2831": [8, 47], "2901": [8, 47], "2923": [8, 47], "2944": [8, 47], "2965": [8, 47], "2986": [8, 47], "3007": [8, 47], "3028": [8, 47], "3070": [8, 47], "3112": [8, 47], "3176": [8, 47], "3221": [8, 47], "3243": [8, 47], "3265": [8, 47], "3310": [8, 47], "3377": [8, 47], "3399": [8, 47], "3421": [8, 47], "3443": [8, 47], "3465": [8, 47], "3487": [8, 47], "3509": [8, 47], "3552": [8, 47], "3573": [8, 47], "3594": [8, 47], "3616": [8, 47], "3637": [8, 47], "3658": [8, 47], "3701": [8, 47], "3724": [8, 47], "3746": [8, 47], "3790": [8, 47], "3834": [8, 47], "3856": [8, 47], "3877": [8, 47], "3898": [8, 47], "3920": [8, 47], "3943": [8, 47], "3989": [8, 47], "4012": [8, 47], "4035": [8, 41, 47], "4058": [8, 47], "4102": [8, 47], "4125": [8, 47], "4148": [8, 47], "4194": [8, 47], "4217": [8, 47], "4263": [8, 47], "4286": [8, 47], "4309": [8, 47], "4355": [8, 47], "4491": [8, 45, 46, 47], "4514": [8, 47], "4537": [8, 47], "4560": [8, 47], "4606": [8, 47], "4629": [8, 47], "4652": [8, 47], "4674": [8, 47], "4739": [8, 47], "4760": [8, 47], "4782": [8, 41, 47], "4850": [8, 47], "4872": [8, 47], "4894": [8, 47], "4960": [8, 47], "4982": [8, 47], "5005": [8, 47], "1847": [8, 47], "rmse_mean": [8, 11, 47, 49], "rmse_std": [8, 11, 47, 49], "mse_mean": [8, 47], "mse_std": [8, 47], "8f": [8, 47], "13118143": [8, 47], "01402841": [8, 47], "01725777": [8, 47], "00376382": [8, 47], "esim": [8, 11, 47, 49], "model_param_estim": [8, 11, 20, 47, 49], "164353": [8, 47], "10030284698307514": [8, 47], "14067839453672557": [8, 47], "tcm": 9, "ie_sbigru": 9, "stacked_lstm": 9, "case_1": 9, "153000": 9, "1268689": 9, "case_2": 9, "117000": 9, "1248873": 9, "case_3": 9, "126000": 9, "1276983": 9, "case_4": 9, "63000": 9, "1374151": 9, "case_5": 9, "54000": 9, "1290003": 9, "case_6": 9, "9000": 9, "1003395": 9, "case_7": 9, "72000": 9, "1245529": 9, "case_8": 9, "1321487": 9, "case_9": 9, "81000": 9, "1377467": 9, "case_10": 9, "90000": 9, "1347769": 9, "case_11": 9, "207000": 9, "1180064": 9, "case_12": 9, "1276980": 9, "case_13": 9, "135000": 9, "1242669": 9, "case_14": 9, "1310826": 9, "case_15": 9, "1316886": 9, "case_16": 9, "18000": 9, "1128731": 9, "scipi": [9, 43], "interpol": 9, "_interpol": 9, "479": 9, "runtimewarn": 9, "invalid": 9, "encount": 9, "divid": [9, 17, 24, 41, 42, 43], "slope": 9, "y_hi": 9, "y_lo": 9, "x_hi": 9, "x_lo": 9, "minmaxscal": [9, 12], "trainer_data": [9, 12], "tester_data": [9, 12], "index": [9, 12, 18, 20, 37, 38, 40, 48], "column": [9, 12, 18, 20, 45, 46], "path_to_tar": 9, "hi_sota": 9, "0031789041005068647": 9, "stack_sota": 9, "2809": 9, "0022332468596409335": 9, "047257241346072384": 9, "tcm_sota": 9, "3511": 9, "004014168163365719": 9, "06335746335962102": 9, "differ": [9, 12, 44], "seed": [9, 12, 18, 20, 25, 30, 35], "uncertainti": [9, 12], "ie_sbigru_sota": 9, "2341": 9, "004956771691496658": 9, "07040434426579555": 9, "diagnosi": [10, 30, 37, 38, 40, 48], "rul": [10, 33, 35, 44], "estim": [10, 38, 44, 48], "lstm": [10, 12, 18, 32, 49], "hi": [10, 30, 43], "latent": [11, 49], "25000": [11, 49], "33333": [11, 49], "33341": [11, 49], "24998": [11, 49], "49985": [11, 49], "11192": [11, 49], "33322": [11, 49], "33335": [11, 49], "score": [11, 34, 41, 44, 49], "score_mean": [11, 49], "score_std": [11, 49], "6f": [11, 49], "456039": [11, 49], "461288": [11, 49], "54669": [11, 49], "804238": [11, 49], "15056": [11, 49], "124157": [11, 49], "944129": [11, 49], "8843235832452774": [11, 49], "06986449102077215": [11, 49], "cmapss": [12, 34], "ir": 10, "fd001": [12, 44], "subdataset": [12, 43], "504872": 12, "533656": 12, "496051": 12, "532245": 12, "547602": 12, "528810": 12, "496134": 12, "529770": 12, "486699": 12, "501537": 12, "512334": 12, "523435": 12, "474171": 12, "519167": 12, "rul_sota": 12, "nn": 12, "modul": [12, 37], "306": 12, "userwarn": 12, "enable_nested_tensor": 12, "self": [12, 18, 20, 44], "use_nested_tensor": 12, "becaus": [12, 44], "encoder_lay": 12, "activation_relu_or_gelu": 12, "wa": [12, 18, 43, 44], "warn": 12, "why_not_sparsity_fast_path": 12, "99217470692219": 12, "25394": 12, "12755711561": 12, "python": [14, 43], "deep": [14, 25, 42, 44], "learn": [14, 18, 25, 42, 44], "monitor": [14, 41, 42], "task": [14, 18, 20, 38, 48], "get": [14, 21], "start": [14, 42, 43], "instal": [14, 38], "descript": 14, "quick": [14, 38], "tutori": [0, 14, 38], "advanc": [14, 41], "materi": [14, 43], "api": 14, "refer": 14, "detail": 14, "function": [14, 34, 41, 42, 44], "custom": 14, "how": [14, 41, 42], "your": 14, "own": 14, "benchmark": [14, 42], "which": [14, 18, 42, 43], "baseanomalydetect": 15, "util": [15, 22, 27, 32, 37], "slidingwindowdataset": [15, 19, 20, 22, 26, 27, 31, 32, 36, 37], "num_chunk": [16, 20, 23, 28, 33], "force_download": [16, 20, 23, 28, 33], "sourc": [16, 17, 18, 19, 20, 21, 23, 24, 25, 26, 28, 29, 30, 31, 33, 34, 35, 36], "base": [16, 18, 19, 21, 23, 25, 26, 28, 30, 31, 33, 35, 36, 37, 41, 42, 43, 44], "basedataset": [16, 20, 23, 28, 33, 37], "paper": [16, 23, 25, 34, 42, 44], "reinartz": [16, 23, 41, 42], "kulahci": [16, 23, 41, 42], "m": [16, 23, 41, 42, 44], "ravn": [16, 23, 41, 42], "o": [16, 23, 44], "2021": [16, 23, 25, 41, 42, 44], "an": [16, 23, 41, 42, 43, 44], "extend": [16, 23, 41, 42], "simul": [16, 23, 33, 41, 42, 44], "decis": [16, 23, 41, 42], "support": [16, 23, 41, 42], "system": [16, 23, 41, 42, 44], "comput": [16, 23, 41, 42, 43, 44], "chemic": [16, 23, 41, 42], "engin": [16, 23, 33, 41, 42, 44], "107281": [16, 23, 41, 42], "web": [16, 23], "mit": [16, 23], "edu": [16, 23], "braatzgroup": [16, 23], "link": [16, 20, 23, 28, 33], "int": [16, 18, 19, 20, 23, 25, 26, 28, 30, 31, 33, 35, 36, 41], "If": [16, 18, 20, 23, 25, 28, 30, 33, 35], "given": [16, 18, 20, 23, 28, 33], "onli": [16, 20, 23, 28, 33], "chunk": [16, 20, 23, 28, 33], "purpos": [16, 20, 23, 28, 33], "bool": [16, 18, 20, 23, 25, 28, 30, 33, 35], "even": [16, 20, 23, 28, 33], "exist": [16, 20, 21, 23, 28, 33], "set_name_public_link": [16, 20, 23, 28, 33], "method": [16, 20, 23, 25, 28, 33, 42, 43], "ha": [16, 18, 20, 23, 28, 33, 41], "implement": [16, 20, 23, 25, 28, 33], "all": [16, 18, 20, 23, 25, 28, 30, 33, 35, 41], "children": [16, 20, 23, 28, 33], "set": [16, 18, 20, 21, 23, 28, 33, 41, 42, 43, 44, 45, 46], "public": [16, 20, 23, 28, 33], "rieth": [16, 23], "amsel": [16, 23], "b": [16, 18, 20, 23, 25, 30, 35, 41, 43], "d": [16, 23, 41, 42, 44], "tran": [16, 23], "r": [16, 23, 41, 42], "cook": [16, 23], "2017": [16, 23], "addit": [16, 23], "version": [16, 23], "v1": [16, 23], "softwar": [16, 23], "harvard": [16, 23], "datavers": [16, 23], "doi": [16, 23, 33, 34], "org": [16, 18, 23], "7910": [16, 23], "dvn": [16, 23], "6c3jr1": [16, 23], "crop": [16, 23], "pred": [17, 24, 29, 34], "ndarrai": [17, 24], "float": [17, 18, 20, 24, 25, 29, 30, 34, 35], "classif": [17, 24], "posit": [17, 24, 41, 42], "exampl": [17, 24, 43, 44], "predict": [17, 20, 24, 29, 34, 35, 41, 42, 43, 44], "return": [17, 20, 21, 24, 29, 34, 44], "type": [17, 20, 21, 24, 29, 34, 42, 43], "rate": [17, 18, 20, 24, 25, 30, 35, 41, 42, 43], "aka": [17, 24], "alarm": [17, 24, 41, 42], "list": [17, 18, 21, 24, 29, 34], "each": [17, 18, 24, 25, 30, 35, 42, 43, 44], "str": [18, 20, 21, 25, 30, 35], "random_se": [18, 20, 25, 30, 35], "basemodel": [18, 20, 25, 30, 35, 37], "abc": [18, 20, 25, 30, 35], "window": [18, 20, 25, 30, 35, 42], "time": [18, 20, 25, 30, 35, 41, 42], "interv": [18, 20, 25, 30, 35], "between": [18, 20, 25, 29, 30, 34, 35, 41, 43], "first": [18, 20, 25, 30, 35, 43], "point": [18, 20, 25, 30, 35, 42, 43], "consecut": [18, 20, 25, 30, 35], "slide": [18, 20, 25, 30, 35, 42], "batch": [18, 20, 25, 30, 35], "possibl": [18, 20, 25, 30, 35], "show": [18, 20, 25, 30, 35], "progress": [18, 20, 25, 30, 35], "bar": [18, 20, 25, 30, 35], "artifact": [18, 20, 25, 30, 35], "store": [18, 20, 25, 30, 35], "gener": [18, 20, 25, 30, 35, 41, 44], "ensur": [18, 20, 25, 30, 35], "reproduc": [18, 20, 25, 30, 35], "proport": [18, 20, 25, 30, 35, 41, 42], "take": 18, "It": [18, 20, 41, 42], "specifi": [18, 21], "quantil": 18, "error": [18, 29, 34], "threshold": 18, "option": [18, 19, 20], "seri": [18, 19, 20, 26, 31, 36, 41, 42], "save_path": [18, 20], "force_model_ctr": [18, 20], "sensor": [18, 20, 25, 30, 35, 42, 43, 44], "two": [18, 20, 35, 43], "run_id": [18, 20, 45, 46], "other": [18, 20, 42, 43], "ind": [18, 20], "omit": [18, 20], "path": [18, 20, 21], "automat": [18, 20, 44], "checkpoint_path": [18, 20], "autoencod": [18, 41, 42], "ae_anomaly_detect": 18, "consist": [18, 25, 30, 35, 42, 44], "encod": [18, 42, 44], "decod": 18, "part": [18, 44], "reshap": [18, 25, 30, 35], "vector": [18, 25, 30, 35, 41], "l": [18, 20, 25, 30, 35, 41, 42, 44], "where": [18, 20, 25, 30, 35, 41, 42, 43, 44], "length": [18, 20, 25, 30, 35, 44, 45, 46], "larn": [18, 20, 25, 30, 35], "layer": [18, 25, 30, 35, 42], "input_dim": [], "intern": 44, "share": [], "both": 43, "scriptmodul": [], "forward": [], "x": [41, 44], "defin": [18, 21, 41, 42], "perform": [41, 42, 43, 44], "everi": [25, 33, 42], "call": [], "should": 44, "overridden": [], "subclass": [], "although": [], "recip": [], "pass": [], "need": [], "within": [41, 42], "one": [41, 42], "instanc": [20, 41, 42], "afterward": [], "instead": [], "sinc": [], "former": [], "care": [], "run": [33, 39, 42, 43, 44], "regist": [], "hook": [], "while": [], "latter": [], "silent": [], "ignor": [], "them": [41, 42], "anomalyattent": [], "win_siz": [], "mask_flag": [], "scale": [43, 44, 45, 46], "attention_dropout": [], "output_attent": [], "queri": [], "kei": [20, 41], "sigma": [], "attn_mask": [], "transformer_anomaly_detect": 18, "n_head": 18, "activ": [18, 25, 30, 35], "gelu": 18, "iclr": 18, "2022": [18, 41, 44], "associ": 18, "discrep": 18, "openreview": 18, "net": 18, "forum": 18, "id": 18, "lzqq89u1qm_": 18, "threshold_valu": 18, "after": [18, 42], "limit": 18, "abov": [18, 44], "head": 18, "relu": [18, 25, 30, 35, 45], "anomalytransformermodel": [], "enc_in": [], "c_out": [], "attentionlay": [], "attent": [18, 44], "d_kei": [], "d_valu": [], "dataembed": [], "c_in": [], "attn_lay": [], "norm_lay": [], "encoderlay": [], "positionalembed": [], "max_len": [], "5000": [], "tokenembed": [], "dict": [20, 21], "dictionari": 20, "optuna": 20, "object": [20, 21], "forc": 20, "classmethod": 20, "from_config": 20, "cfg": 20, "config": [20, 37], "child": 20, "infer": 20, "tupl": 20, "contain": [20, 42, 43, 44], "deviat": [20, 34, 41], "make": [20, 35], "default": 20, "maxim": 20, "tensor": 20, "shape": 20, "cfg_path": 21, "path_get": 21, "field": 21, "nest": 21, "path_set": 21, "doesn": 21, "t": [21, 41, 42, 43, 44], "obejct": 21, "to_yaml": 21, "dump": 21, "yaml": 21, "file": [21, 39], "basefaultdiagnosi": 22, "faultdiagnosissmalltep": [22, 23, 46], "correct": [24, 41, 42], "total": [24, 41, 42, 43], "correctli": [24, 41, 42], "diagnos": [24, 42], "faulti": [24, 41, 42], "mlp_fault_diagnosi": [25, 30], "multilay": [25, 30, 35], "perceptron": [25, 30, 35], "dimension": [25, 30, 35], "dilation_bas": 25, "tcn_fault_diagnosi": 25, "tempor": [18, 25, 41, 42, 44], "convolut": [18, 25, 44], "network": [18, 25, 41, 42, 44], "lomov": [25, 42], "ildar": [25, 42], "et": [25, 33, 42, 43], "al": [25, 33, 42, 43], "journal": [25, 42, 44], "industri": [25, 42, 44], "inform": [25, 41, 42, 43, 44], "integr": [25, 42], "100216": [25, 42], "channel": 25, "kernel": 25, "residu": 25, "block": 25, "expon": [25, 34, 43], "determin": [25, 42], "dilat": 25, "baseremainingusefullifeestim": [27, 35], "mil": 28, "nasa": [28, 43], "gov": 28, "raw": 28, "wear": [28, 29, 43], "vjv9": 28, "9f3x": 28, "squar": [29, 34], "real": [29, 34, 42], "basehealthindexestim": [30, 32], "rulcmapsspap": [32, 33], "nonsimmetric_funct": [32, 34], "piec": [33, 44], "wise": [33, 44], "saxena": [33, 44], "damag": [33, 44], "propag": [33, 44], "aircraft": [33, 44], "failur": [33, 41, 44], "phm": [33, 34], "4711414": [33, 34], "minimum": [33, 43], "non": 34, "simmetr": 34, "propos": [34, 43], "origin": [34, 42, 44], "depend": [34, 41, 42], "rel": 34, "divis": 34, "root": [34, 39], "mlp_cmapss_rul": 35, "hidden_s": 35, "dropout_valu": 35, "long": 35, "short": 35, "term": [35, 44], "classic": 35, "architectur": 35, "stack": [10, 35, 42], "silu": 35, "nonlinear": 35, "final": [35, 39], "move": 35, "Then": [35, 39, 42], "featur": [35, 42, 43], "reccur": 35, "lstm_model": [], "neural": [41, 44], "pytorch": [], "realis": [], "doc": [], "relat": 43, "amount": [], "probl": [], "remain": [37, 38, 40, 42, 48], "lifetim": [37, 38, 40, 48], "health": [37, 38, 40, 41, 44, 48], "can": [39, 41, 42], "pip": 39, "follow": [39, 43], "git": 39, "github": 39, "com": 39, "codeai": 39, "team": 39, "To": [39, 43], "repositori": [39, 43], "press": 39, "code": 39, "zip": 39, "main": [39, 43, 44], "page": 39, "unzip": 39, "go": 39, "command": 39, "linux": 39, "maco": 39, "cd": 39, "pytest": 39, "In": [41, 42], "focu": [41, 42], "variabl": [41, 42], "oper": [41, 42, 43, 44], "classifi": [41, 42], "These": [41, 42], "temperatur": 41, "pressur": 41, "flow": 41, "level": [41, 43], "indic": [41, 42], "composit": 41, "effici": 41, "typic": 41, "signal": [41, 42], "mathemat": 41, "formul": [41, 44], "context": 41, "involv": 41, "map": 41, "binari": 41, "anomal": 41, "let": 41, "repres": [41, 43, 44], "begin": [41, 42, 43, 44], "equat": [41, 42, 43, 44], "x_1": [41, 42], "x_2": [41, 42], "ldot": 41, "x_n": [41, 42], "x_i": 41, "denot": 41, "th": 41, "n": [41, 42, 43, 44], "case": [41, 42, 43, 44], "text": [41, 42, 44], "otherwis": 41, "setup": 41, "could": 41, "machin": [41, 42, 43, 44], "histor": 41, "relationship": 41, "algorithm": [41, 43], "tree": 41, "emploi": 41, "captur": 41, "interdepend": 41, "accur": 41, "identifi": [41, 42], "pattern": 41, "sever": [41, 42], "commonli": [41, 42, 44], "help": [41, 42], "understand": [41, 42], "well": [41, 42], "avoid": [41, 42], "choic": [41, 42], "often": [41, 42], "specif": [41, 42, 44], "requir": [41, 42], "versu": [41, 42], "cost": [41, 42], "common": [41, 42], "includ": [41, 42, 43, 44], "measur": [41, 42, 43], "overal": [41, 42], "frac": [41, 42, 43, 44], "neg": [41, 42], "recal": [41, 42], "actual": [41, 42], "particularli": [41, 42], "scenario": [41, 42], "miss": [41, 42, 43], "have": [41, 42], "seriou": [41, 42], "consequ": [41, 42], "incorrectli": [41, 42], "lower": [41, 42], "desir": [41, 42], "fewer": [41, 42], "recent": 41, "year": 41, "been": 41, "lot": 41, "research": [41, 42], "most": [41, 44], "cite": [41, 42], "were": [41, 43], "variou": [41, 43, 44], "adapt": 41, "work": 41, "f1": 41, "dagmm": 41, "kddcup": 41, "thyroid": 41, "arrhuthmia": 41, "4983": 41, "rev": 41, "9380": 41, "gdn": 41, "swat": 41, "wadi": 41, "gta": 41, "smap": 41, "9041": 41, "msl": 41, "9111": 41, "mad": [10, 15, 41], "gan": [41, 42], "2019": [41, 42], "kddcup99": 41, "mscred": 41, "synthet": 41, "power": 41, "plant": [41, 42], "omnianomali": 41, "8434": 41, "8989": 41, "smd": 41, "8857": 41, "thoc": 41, "2d": 41, "gestur": 41, "6331": 41, "demand": 41, "4568": 41, "9886": 41, "8809": 41, "9518": 41, "9367": 41, "usad": 41, "8460": 41, "4296": 41, "8634": 41, "9272": 41, "9463": 41, "christoph": [41, 42], "murat": [41, 42], "ol": [41, 42], "k": [18, 41, 42, 43, 44], "choi": 41, "j": [41, 44], "yi": 41, "park": 41, "yoon": 41, "review": [41, 42], "analysi": 41, "guidelin": 41, "ieee": [18, 41, 42, 44], "access": 41, "vol": 41, "pp": [41, 44], "120043": 41, "120065": 41, "gen": 41, "li": [41, 44], "jason": 41, "jung": 41, "multivari": [18, 41, 42], "approach": [41, 44], "applic": [41, 44], "challeng": 41, "fusion": 41, "volum": 41, "zong": 41, "q": [41, 44], "song": [41, 44], "min": [41, 43], "w": 41, "cheng": [41, 44], "lumezanu": 41, "cho": 41, "h": [41, 44], "chen": [41, 44], "gaussian": 41, "mixtur": 41, "unsupervis": 41, "proc": 41, "conf": 41, "deng": 41, "hooi": 41, "graph": [18, 41, 44], "artif": 41, "intel": 41, "aaai": 41, "4027": 41, "z": [41, 44], "zhang": [41, 44], "yuan": 41, "structur": [41, 44], "iot": 41, "arxiv": [41, 44], "2104": 41, "03466": 41, "jin": 41, "shi": 41, "goh": 41, "ng": 41, "adversari": 41, "netw": 41, "icann": 41, "munich": 41, "germani": 41, "springer": 41, "703": 41, "716": 41, "sippl": 41, "interpret": [41, 44], "multidimension": 41, "multimod": 41, "mach": [41, 44], "9016": 41, "9025": 41, "y": [41, 43, 44], "su": 41, "zhao": 41, "niu": 41, "liu": [41, 44], "sun": 41, "pei": 41, "robust": 41, "through": 41, "stochast": 41, "recurr": [41, 44], "25th": 41, "acm": 41, "sigkdd": 41, "knowl": 41, "discoveri": 41, "mine": 41, "2828": 41, "2837": 41, "shen": 41, "kwok": 41, "timeseri": 41, "hierarch": [41, 42], "adv": 41, "inf": 41, "syst": 41, "13016": 41, "13026": 41, "audibert": 41, "p": [41, 43], "michiardi": 41, "guyard": 41, "marti": 41, "zuluaga": 41, "26th": 41, "3395": 41, "3404": 41, "compani": 42, "serv": 42, "crucial": 42, "control": [42, 44], "mimick": 42, "figur": [42, 43, 44], "ref": 42, "fig": [42, 43, 44], "process_diagram": 42, "2800": 42, "independ": [42, 43], "last": 42, "hour": 42, "collect": 42, "frequenc": 42, "minut": 42, "thu": 42, "2000": 42, "observ": 42, "onc": 42, "goe": 42, "ratio": 42, "manipul": 42, "moment": 42, "correspond": [42, 43], "certain": 42, "condit": [42, 44], "mark": 42, "label": [42, 43], "x_t": 42, "mathbb": 42, "y_1": 42, "y_2": 42, "y_n": 42, "y_t": 42, "width": 42, "rightarrow": 42, "must": 42, "align": [42, 44], "argmin": 42, "_f": 42, "sum_": [42, 43, 44], "x_": 42, "approxim": [42, 43], "devid": 42, "art": [42, 44], "pca": 42, "dbn": 42, "svdd": 42, "un": 42, "msdae": 42, "tp": 42, "gru": 42, "002": 42, "abid": 42, "anam": 42, "muhammad": 42, "tahir": 42, "khan": 42, "javaid": 42, "iqbal": 42, "techniqu": 42, "basic": 42, "beyond": 42, "artifici": 42, "intellig": [42, 44], "3639": 42, "3664": [42, 45, 46], "yan": 42, "shifu": 42, "xuefeng": 42, "design": 42, "teacher": 42, "supervis": 42, "dual": 42, "qualiti": 42, "relev": 42, "appli": [42, 43], "soft": 42, "105526": 42, "yu": [42, 44], "jianbo": 42, "enhanc": 42, "strategi": 42, "favor": 42, "belief": 42, "chemistri": 42, "15479": 42, "15490": 42, "whole": 42, "unstabl": 42, "neuron": 42, "transact": 42, "cybernet": [42, 44], "3998": 42, "4007": 42, "multiscal": 42, "agglom": 42, "cluster": 42, "denois": 42, "106525": 42, "experi": 43, "conduct": 43, "matsuura": 43, "centr": 43, "mc": 43, "510v": 43, "dry": 43, "rough": 43, "cast": 43, "iron": 43, "j45": 43, "stainless": 43, "steel": 43, "workpiec": 43, "six": 43, "tooth": 43, "face": 43, "kc710": 43, "carbid": 43, "insert": 43, "sub": 43, "depth": 43, "feed": 43, "record": 43, "vibroacceleromet": 43, "ae": 43, "cta": 43, "current": 43, "space": 43, "saw": 43, "surfac": 43, "optic": 43, "microscopi": 43, "Of": 43, "five": 43, "tool": 43, "ac": 43, "spindl": 43, "motor": 43, "smcac": 43, "dc": 43, "smcdc": 43, "vibrat": 43, "vib_tabl": 43, "vib_spindl": 43, "acoust": 43, "emiss": 43, "ae_tabl": 43, "ae_spindl": 43, "idea": 43, "build": 43, "extract": [43, 46], "equip": [43, 44], "due": [43, 44], "alwai": 43, "perfectli": 43, "confirm": 43, "construct": [18, 43], "we": 43, "taylor": 43, "document": [18, 43], "eq": 43, "c_": 43, "v": [43, 44], "v_": 43, "life": [43, 44], "constant": 43, "speed": [43, 44], "nelder": 43, "mead": 43, "obtain": 43, "character": [43, 44], "itself": 43, "definit": 43, "charact": 43, "curv": 43, "behavior": [43, 44], "empir": [43, 44], "verifi": 43, "pipelin": 43, "sawtooth": 43, "next": 43, "simultan": 43, "do": 43, "max": 43, "mm": 43, "maximum": 43, "allow": 43, "subtract": 43, "reduc": 43, "impact": 43, "implicit": 43, "weight": 43, "healthi": [43, 44], "trajectori": [43, 44], "chosen": 43, "acquir": 43, "sqrt": [43, 44], "y_i": [43, 44], "hat": [43, 44], "_i": [43, 44], "dure": [18, 43, 44], "unknown": [], "characterist": 44, "especi": [], "similar": [], "about": 44, "cycl": [], "mapreduc": [], "prf": [], "0531": [], "svm": [], "0254": [], "takelm": [], "0134": [], "0148": [], "agogino": 43, "goebel": [43, 44], "prognost": [43, 44], "2007": 43, "virtanen": 43, "fundament": 43, "scientif": 43, "natur": 43, "272": 43, "commerci": 44, "modular": 44, "aero": 44, "propuls": 44, "program": 44, "turbofan": 44, "under": 44, "flight": 44, "until": 44, "situat": 44, "ascent": 44, "descent": 44, "altitud": 44, "km": 44, "four": 44, "subset": 44, "locat": 44, "degrad": 44, "three": 44, "height": 44, "acceler": 44, "throttl": 44, "resolv": 44, "angl": 44, "weak": 44, "distinguish": 44, "consid": 44, "equal": 44, "regress": 44, "geq": 44, "heurist": 44, "previous": 44, "e": 44, "asymmetr": 44, "phm2008": 44, "competit": 44, "penal": 44, "overestim": 44, "mainten": 44, "regul": 44, "fd003": 44, "fd002": 44, "fd004": 44, "gnmr": 44, "370": 44, "3196": 44, "2795": 44, "bl": 44, "tfcn": 44, "243": 44, "244": [44, 45, 46], "2096": 44, "tcnn": 44, "254": 44, "415": 44, "1453": 44, "rve": 44, "1846": 44, "bigru": 44, "tsam": 44, "2264": 44, "3610": 44, "mca": 44, "bgru": 44, "kghm": 44, "251": 44, "1131": 44, "3356": 44, "da": 44, "pm": 44, "229": 44, "257": [44, 45, 46], "522": 44, "2317": 44, "adl": 44, "dnn": 44, "238": 44, "281": 44, "1149": 44, "simon": 44, "eklund": 44, "confer": 44, "manag": 44, "heim": 44, "international135": 44, "jyoti": 44, "pankaj": 44, "vishnu": 44, "lovekesh": 44, "gautam": 44, "leverag": 44, "preprint": 44, "2006": 44, "16556v1": 44, "wang": 44, "by140": 44, "fuse": 44, "broad": 44, "8th": 44, "social": 44, "iccss": 44, "joint": 44, "and145": 44, "neurosci": 44, "costa": 44, "s\u00e1nchez": 44, "variat": 44, "assess": 44, "reliabl": 44, "safeti": 44, "108353": 44, "jiang": 44, "wu": 44, "luo": 44, "yin": 44, "bidirect": 44, "gate": 44, "unit": 44, "mechan": 44, "engineering150": 44, "108297": 44, "kara": 44, "multi": 44, "108211": 44, "hu": 44, "enabl": 44, "knowledg": 44, "108869": 44, "gao": 44, "jia": 44, "pang": 44, "internet": 44, "thing": 44, "9594": 44, "9602": 44, "xiang": 44, "qin": 44, "gryllia": 44, "differenti": 44, "its": 44, "108531": 44, "xmeas_1": [45, 46], "xmeas_2": [45, 46], "xmeas_3": [45, 46], "xmeas_4": [45, 46], "xmeas_5": [45, 46], "xmeas_6": [45, 46], "xmeas_7": [45, 46], "xmeas_8": [45, 46], "xmeas_9": [45, 46], "xmeas_10": [45, 46], "xmv_2": [45, 46], "xmv_3": [45, 46], "xmv_4": [45, 46], "xmv_5": [45, 46], "xmv_6": [45, 46], "xmv_7": [45, 46], "xmv_8": [45, 46], "xmv_9": [45, 46], "xmv_10": [45, 46], "xmv_11": [45, 46], "413402073": [45, 46], "25038": [45, 46], "3674": [45, 46], "4529": [45, 46], "2320": [45, 46], "402": [45, 46], "2704": [45, 46], "863": [45, 46], "33818": [45, 46], "744": [45, 46], "657": [45, 46], "544": [45, 46], "935": [45, 46], "757": [45, 46], "510": [45, 46], "447": [45, 46], "25109": [45, 46], "3659": [45, 46], "4264": [45, 46], "721": [45, 46], "576": [45, 46], "000": [45, 46], "33620": [45, 46], "414": [45, 46], "588": [45, 46], "259": [45, 46], "084": [45, 46], "554": [45, 46], "692": [45, 46], "3660": [45, 46], "4477": [45, 46], "4426": [45, 46], "875": [45, 46], "070": [45, 46], "2706": [45, 46], "771": [45, 46], "33563": [45, 46], "666": [45, 46], "275": [45, 46], "380": [45, 46], "699": [45, 46], "530": [45, 46], "24977": [45, 46], "4776": [45, 46], "758": [45, 46], "063": [45, 46], "33553": [45, 46], "856": [45, 46], "277": [45, 46], "072": [45, 46], "643": [45, 46], "089": [45, 46], "29405": [45, 46], "4497": [45, 46], "3381": [45, 46], "650": [45, 46], "388": [45, 46], "32632": [45, 46], "797": [45, 46], "947": [45, 46], "507": [45, 46], "312148819": [45, 46], "24842": [45, 46], "3694": [45, 46], "3946": [45, 46], "780": [45, 46], "2708": [45, 46], "32959": [45, 46], "891": [45, 46], "867": [45, 46], "868": [45, 46], "061": [45, 46], "088": [45, 46], "470": [45, 46], "463": [45, 46], "078": [45, 46], "957": [45, 46], "22612": [45, 46], "4523": [45, 46], "3655": [45, 46], "778": [45, 46], "2711": [45, 46], "32645": [45, 46], "675": [45, 46], "831": [45, 46], "027": [45, 46], "560": [45, 46], "598": [45, 46], "591": [45, 46], "720": [45, 46], "958": [45, 46], "22386": [45, 46], "3984": [45, 46], "673": [45, 46], "528": [45, 46], "2712": [45, 46], "32484": [45, 46], "233": [45, 46], "053": [45, 46], "235": [45, 46], "040": [45, 46], "737": [45, 46], "490": [45, 46], "310": [45, 46], "959": [45, 46], "22561": [45, 46], "4483": [45, 46], "4293": [45, 46], "2710": [45, 46], "857": [45, 46], "31932": [45, 46], "248": [45, 46], "567": [45, 46], "820": [45, 46], "979": [45, 46], "503": [45, 46], "630": [45, 46], "996": [45, 46], "960": [45, 46], "22585": [45, 46], "4492": [45, 46], "4061": [45, 46], "869": [45, 46], "722": [45, 46], "31926": [45, 46], "429": [45, 46], "986": [45, 46], "810": [45, 46], "639": [45, 46], "378": [45, 46], "153300": [45, 46], "row": [45, 46], "dtype": [45, 46], "int64": [45, 46], "7313": 45, "6942": 45, "6606": 45, "sequenti": 45, "flatten": 45, "start_dim": 45, "end_dim": 45, "linear": 45, "in_featur": 45, "520": 45, "out_featur": 45, "bia": 45, "7545263157894737": 45, "705975": 45, "04881012658227848": 45, "vitalijpozdnakov": 46, "miniconda3": 46, "python3": 46, "small_tep": 46, "2m": 46, "1mb": 46, "6mb": 46, "153mb": 46, "77mb": 46, "60gb": 46, "33gb": 46, "79692": 46, "1771006": 46, "1837204": 46, "4350": 46, "5717": 46, "3677": 46, "6005": 46, "7325": 46, "5394": 46, "3939": 46, "7945664160401003": 46, "stgat": [10, 15], "gsl": [10, 15], "n_featur": [], "embed_dim": 18, "layer_numb": 18, "lstm_n_layer": 18, "lstm_hid_dim": 18, "recon_n_lay": 18, "recon_hid_dim": 18, "gnn_anomaly_detect": 18, "alpha": 18, "describ": 18, "trainabl": 18, "satur": 18, "adjac": 18, "matrix": 18, "edg": 18, "stgat_anomaly_detect": 18, "icassp": 18, "spatial": 18, "For": 18, "ieeexplor": 18, "abstract": 18, "9747274": 18, "embed": 18, "reconstruct": 18, "cut": 43, "tabl": 43}, "objects": {"ice.anomaly_detection": [[16, 0, 0, "-", "datasets"], [17, 0, 0, "-", "metrics"], [19, 0, 0, "-", "utils"]], "ice.anomaly_detection.datasets": [[16, 1, 1, "", "AnomalyDetectionReinartzTEP"], [16, 1, 1, "", "AnomalyDetectionRiethTEP"], [16, 1, 1, "", "AnomalyDetectionSmallTEP"]], "ice.anomaly_detection.datasets.AnomalyDetectionReinartzTEP": [[16, 2, 1, "", "set_name_public_link"]], "ice.anomaly_detection.datasets.AnomalyDetectionRiethTEP": [[16, 2, 1, "", "set_name_public_link"]], "ice.anomaly_detection.datasets.AnomalyDetectionSmallTEP": [[16, 2, 1, "", "set_name_public_link"]], "ice.anomaly_detection.metrics": [[17, 3, 1, "", "accuracy"], [17, 3, 1, "", "false_positive_rate"], [17, 3, 1, "", "true_positive_rate"]], "ice.anomaly_detection.models.autoencoder": [[18, 1, 1, "", "AutoEncoderMLP"]], "ice.anomaly_detection.models": [[18, 0, 0, "-", "base"]], "ice.anomaly_detection.models.base": [[18, 1, 1, "", "BaseAnomalyDetection"]], "ice.anomaly_detection.models.base.BaseAnomalyDetection": [[18, 2, 1, "", "fit"], [18, 2, 1, "", "load_checkpoint"]], "ice.anomaly_detection.models.gnn": [[18, 1, 1, "", "GSL_GNN"]], "ice.anomaly_detection.models.stgat": [[18, 1, 1, "", "STGAT_MAD"]], "ice.anomaly_detection.models.transformer": [[18, 1, 1, "", "AnomalyTransformer"]], "ice.anomaly_detection.utils": [[19, 1, 1, "", "SlidingWindowDataset"]], "ice": [[20, 0, 0, "-", "base"], [21, 0, 0, "-", "configs"]], "ice.base": [[20, 1, 1, "", "BaseDataset"], [20, 1, 1, "", "BaseModel"], [20, 1, 1, "", "SlidingWindowDataset"]], "ice.base.BaseDataset": [[20, 2, 1, "", "set_name_public_link"]], "ice.base.BaseModel": [[20, 2, 1, "", "evaluate"], [20, 2, 1, "", "fit"], [20, 2, 1, "", "from_config"], [20, 2, 1, "", "load_checkpoint"], [20, 2, 1, "", "model_param_estimation"], [20, 2, 1, "", "optimize"], [20, 2, 1, "", "predict"], [20, 2, 1, "", "save_checkpoint"]], "ice.configs": [[21, 1, 1, "", "Config"]], "ice.configs.Config": [[21, 2, 1, "", "path_get"], [21, 2, 1, "", "path_set"], [21, 2, 1, "", "to_yaml"]], "ice.fault_diagnosis": [[23, 0, 0, "-", "datasets"], [24, 0, 0, "-", "metrics"], [26, 0, 0, "-", "utils"]], "ice.fault_diagnosis.datasets": [[23, 1, 1, "", "FaultDiagnosisReinartzTEP"], [23, 1, 1, "", "FaultDiagnosisRiethTEP"], [23, 1, 1, "", "FaultDiagnosisSmallTEP"]], "ice.fault_diagnosis.datasets.FaultDiagnosisReinartzTEP": [[23, 2, 1, "", "set_name_public_link"]], "ice.fault_diagnosis.datasets.FaultDiagnosisRiethTEP": [[23, 2, 1, "", "set_name_public_link"]], "ice.fault_diagnosis.datasets.FaultDiagnosisSmallTEP": [[23, 2, 1, "", "set_name_public_link"]], "ice.fault_diagnosis.metrics": [[24, 3, 1, "", "accuracy"], [24, 3, 1, "", "correct_daignosis_rate"], [24, 3, 1, "", "false_positive_rate"], [24, 3, 1, "", "true_positive_rate"]], "ice.fault_diagnosis.models": [[25, 0, 0, "-", "base"], [25, 0, 0, "-", "mlp"], [25, 0, 0, "-", "tcn"]], "ice.fault_diagnosis.models.base": [[25, 1, 1, "", "BaseFaultDiagnosis"]], "ice.fault_diagnosis.models.mlp": [[25, 1, 1, "", "MLP"]], "ice.fault_diagnosis.models.tcn": [[25, 1, 1, "", "TCN"]], "ice.fault_diagnosis.utils": [[26, 1, 1, "", "SlidingWindowDataset"]], "ice.health_index_estimation": [[28, 0, 0, "-", "datasets"], [29, 0, 0, "-", "metrics"], [31, 0, 0, "-", "utils"]], "ice.health_index_estimation.datasets": [[28, 1, 1, "", "Milling"]], "ice.health_index_estimation.datasets.Milling": [[28, 2, 1, "", "set_name_public_link"]], "ice.health_index_estimation.metrics": [[29, 3, 1, "", "mse"], [29, 3, 1, "", "rmse"]], "ice.health_index_estimation.models": [[30, 0, 0, "-", "base"]], "ice.health_index_estimation.models.base": [[30, 1, 1, "", "BaseHealthIndexEstimation"]], "ice.health_index_estimation.models.mlp": [[30, 1, 1, "", "MLP"]], "ice.health_index_estimation.utils": [[31, 1, 1, "", "SlidingWindowDataset"]], "ice.remaining_useful_life_estimation": [[33, 0, 0, "-", "datasets"], [34, 0, 0, "-", "metrics"], [36, 0, 0, "-", "utils"]], "ice.remaining_useful_life_estimation.datasets": [[33, 1, 1, "", "RulCmapss"], [33, 1, 1, "", "RulCmapssPaper"]], "ice.remaining_useful_life_estimation.datasets.RulCmapss": [[33, 2, 1, "", "set_name_public_link"]], "ice.remaining_useful_life_estimation.datasets.RulCmapssPaper": [[33, 2, 1, "", "set_name_public_link"]], "ice.remaining_useful_life_estimation.metrics": [[34, 3, 1, "", "cmapss_score"], [34, 3, 1, "", "nonsimmetric_function"], [34, 3, 1, "", "rmse"]], "ice.remaining_useful_life_estimation.models": [[35, 0, 0, "-", "base"]], "ice.remaining_useful_life_estimation.models.base": [[35, 1, 1, "", "BaseRemainingUsefulLifeEstimation"]], "ice.remaining_useful_life_estimation.models.lstm": [[35, 1, 1, "", "LSTM"]], "ice.remaining_useful_life_estimation.models.mlp": [[35, 1, 1, "", "MLP"]], "ice.remaining_useful_life_estimation.utils": [[36, 1, 1, "", "SlidingWindowDataset"]]}, "objtypes": {"0": "py:module", "1": "py:class", "2": "py:method", "3": "py:function"}, "objnames": {"0": ["py", "module", "Python module"], "1": ["py", "class", "Python class"], "2": ["py", "method", "Python method"], "3": ["py", "function", "Python function"]}, "titleterms": {"advanc": 0, "materi": 0, "result": [2, 3, 4, 5, 6, 7, 8, 9, 11, 12], "anomali": [2, 3, 4, 5, 15, 16, 17, 18, 19, 41, 45], "detect": [2, 3, 4, 5, 15, 16, 17, 18, 19, 41, 45], "us": [2, 3, 4, 5, 6, 7, 8, 9, 11, 12, 30, 32, 33, 34, 36, 44, 49], "autoencodermlp": [2, 18], "256": [2, 6, 8, 11], "fault": [6, 7, 22, 23, 24, 25, 26, 42, 46], "diagnosi": [6, 7, 22, 23, 24, 25, 26, 42, 46], "mlp": [6, 8, 25, 30, 35], "hi": [8, 9], "estim": [8, 9, 11, 12, 27, 28, 29, 31, 47, 49], "benchmark": 10, "content": [10, 48], "rul": [11, 12], "lstm": [9, 11, 35], "custom": 13, "class": 13, "ic": 14, "industri": 14, "condit": 14, "evalu": 14, "dataset": [16, 23, 28, 33, 41, 42, 43, 44], "metric": [17, 24, 29, 34, 41, 42, 43, 44], "model": [18, 25, 30, 35, 41], "baseanomalydetect": 18, "anomalytransform": [5, 18], "util": [19, 26, 31, 36], "base": 20, "modul": 20, "config": 21, "basefaultdiagnosi": 25, "tcn": [7, 25], "health": [27, 28, 29, 31, 35, 43, 47], "index": [27, 28, 29, 31, 35, 43, 47], "remain": [30, 32, 33, 34, 36, 44, 49], "life": [30, 32, 33, 34, 36], "baseremainingusefullifeestim": 30, "basehealthindexestim": 35, "api": 37, "refer": [37, 41, 42, 43, 44], "get": 38, "start": 38, "instal": 39, "test": 39, "task": [40, 41, 42, 43, 44, 45, 46, 47, 49], "tabl": [41, 42, 44], "1": [41, 42, 44], "deep": 41, "learn": 41, "tep": 42, "fpr": 42, "tpr": 42, "valu": 42, "Not": 42, "all": 42, "ar": 42, "avail": 42, "comparison": 42, "c": 44, "mill": [], "cut": [], "3": [], "7": [], "origin": [], "wear": [], "target": [], "lifetim": [44, 49], "mapss": 44, "tutori": [1, 45, 46, 47, 48, 49], "quick": 48, "stgat": [4, 18], "mad": [4, 18], "gsl": [3, 18], "gnn": [3, 18], "optim": 1, "stack": 9, "ir": 12}, "envversion": {"sphinx.domains.c": 2, "sphinx.domains.changeset": 1, "sphinx.domains.citation": 1, "sphinx.domains.cpp": 8, "sphinx.domains.index": 1, "sphinx.domains.javascript": 2, "sphinx.domains.math": 2, "sphinx.domains.python": 3, "sphinx.domains.rst": 2, "sphinx.domains.std": 2, "sphinx.ext.viewcode": 1, "sphinx": 57}, "alltitles": {"Results of anomaly detection using AutoEncoderMLP-256": [[2, "results-of-anomaly-detection-using-autoencodermlp-256"]], "Results of fault diagnosis using MLP-256": [[6, "results-of-fault-diagnosis-using-mlp-256"]], "Results of HI estimation using MLP-256": [[8, "results-of-hi-estimation-using-mlp-256"]], "Results of RUL estimation using lstm-256": [[11, "results-of-rul-estimation-using-lstm-256"]], "Custom classes": [[13, "custom-classes"]], "ICE - Industrial Condition Evaluation": [[14, "ice-industrial-condition-evaluation"]], "Anomaly detection": [[15, "anomaly-detection"], [41, "anomaly-detection"]], "Anomaly detection datasets": [[16, "module-ice.anomaly_detection.datasets"]], "Anomaly detection metrics": [[17, "module-ice.anomaly_detection.metrics"]], "Anomaly detection utils": [[19, "module-ice.anomaly_detection.utils"]], "Base modules": [[20, "module-ice.base"]], "Configs": [[21, "module-ice.configs"]], "Fault diagnosis": [[22, "fault-diagnosis"], [42, "fault-diagnosis"]], "Fault diagnosis datasets": [[23, "module-ice.fault_diagnosis.datasets"]], "Fault diagnosis metrics": [[24, "module-ice.fault_diagnosis.metrics"]], "Fault diagnosis models": [[25, "fault-diagnosis-models"]], "BaseFaultDiagnosis": [[25, "module-ice.fault_diagnosis.models.base"]], "MLP": [[25, "module-ice.fault_diagnosis.models.mlp"], [35, "mlp"], [30, "mlp"]], "TCN": [[25, "module-ice.fault_diagnosis.models.tcn"]], "Fault diagnosis utils": [[26, "module-ice.fault_diagnosis.utils"]], "Health index estimation": [[27, "health-index-estimation"]], "Health index estimation datasets": [[28, "module-ice.health_index_estimation.datasets"]], "Health index estimation metrics": [[29, "module-ice.health_index_estimation.metrics"]], "Health index estimation utils": [[31, "module-ice.health_index_estimation.utils"]], "Remaining useful life": [[32, "remaining-useful-life"]], "Remaining useful life datasets": [[33, "module-ice.remaining_useful_life_estimation.datasets"]], "Remaining useful life metrics": [[34, "module-ice.remaining_useful_life_estimation.metrics"]], "Remaining useful life utils": [[36, "module-ice.remaining_useful_life_estimation.utils"]], "API reference": [[37, "api-reference"]], "Getting started": [[38, "getting-started"]], "Installation": [[39, "installation"]], "Testing": [[39, "testing"]], "Tasks": [[40, "tasks"]], "Dataset": [[41, "dataset"], [42, "dataset"], [44, "dataset"], [43, "dataset"]], "Task": [[41, "task"], [42, "task"], [44, "task"], [43, "task"]], "Metrics": [[41, "metrics"], [42, "metrics"], [44, "metrics"], [43, "metrics"]], "References": [[41, "references"], [42, "references"], [44, "references"], [43, "references"]], "Table 1: Deep learning models for Anomaly detection.": [[41, "id1"]], "Table 1: References for TEP dataset. FPR/TPR values. Not all the metrics\nare available for comparison.": [[42, "id1"]], "Remaining useful lifetime": [[44, "remaining-useful-lifetime"]], "Table 1: References for C-MAPSS dataset": [[44, "id2"]], "Tutorial on anomaly detection task": [[45, "tutorial-on-anomaly-detection-task"]], "Tutorial on fault diagnosis task": [[46, "tutorial-on-fault-diagnosis-task"]], "Tutorial on health index estimation task": [[47, "tutorial-on-health-index-estimation-task"]], "Contents:": [[48, null], [10, null]], "Quick tutorials": [[48, "quick-tutorials"]], "Tutorial on remaining useful lifetime estimation task": [[49, "tutorial-on-remaining-useful-lifetime-estimation-task"]], "Anomaly detection models": [[18, "anomaly-detection-models"]], "BaseAnomalyDetection": [[18, "module-ice.anomaly_detection.models.base"]], "AutoEncoderMLP": [[18, "autoencodermlp"]], "AnomalyTransformer": [[18, "anomalytransformer"]], "STGAT-MAD": [[18, "stgat-mad"]], "GSL-GNN": [[18, "gsl-gnn"]], "Health index models": [[35, "health-index-models"]], "BaseHealthIndexEstimation": [[35, "module-ice.remaining_useful_life_estimation.models.base"]], "LSTM": [[35, "lstm"]], "Advanced materials": [[0, "advanced-materials"]], "Remaining useful life models": [[30, "remaining-useful-life-models"]], "BaseRemainingUsefulLifeEstimation": [[30, "module-ice.health_index_estimation.models.base"]], "Health index": [[43, "health-index"]], "Optimization tutorial": [[1, "optimization-tutorial"]], "Benchmarking": [[10, "benchmarking"]], "Results of anomaly detection using GSL-GNN": [[3, "results-of-anomaly-detection-using-gsl-gnn"]], "Results of anomaly detection using STGAT-MAD": [[4, "results-of-anomaly-detection-using-stgat-mad"]], "Results of anomaly detection using AnomalyTransformer": [[5, "results-of-anomaly-detection-using-anomalytransformer"]], "Results of fault diagnosis using TCN": [[7, "results-of-fault-diagnosis-using-tcn"]], "Results of HI estimation using Stacked LSTM": [[9, "results-of-hi-estimation-using-stacked-lstm"]], "Results of RUL estimation using IR": [[12, "results-of-rul-estimation-using-ir"]]}, "indexentries": {}}) \ No newline at end of file diff --git a/docs/start/tasks/task_hi.html b/docs/start/tasks/task_hi.html index de2c1b3..1f0d2d6 100644 --- a/docs/start/tasks/task_hi.html +++ b/docs/start/tasks/task_hi.html @@ -473,33 +473,6 @@

              Metrics

              References#

              -

              The reference results of state-of-the-art papers for the original [1] -dataset are presented in table 1.

              -

              Since the test dataset involves processes with unknown characteristics, it limits most -related HI-based work approaches, especially similarity-based techniques, which require -information about similar process trajectories.

              - - - - - - - - - - - - - - - - - - - - - -
              Table 1: References for C-Milling dataset, cuts ‘1’, ‘3’, ‘7’ testing, with the original dataset wear target.#

              Model

              RMSE (cycles)

              MapReduce-based PRFs (2020)

              0.0531

              LS-SVM (2020)

              0.0254

              TAKELM (2020)

              0.0134

              LSTM-based (2018)

              0.0148

              [1] Agogino, A.; Goebel, K. Milling Data Set. NASA Prognostics Data Repository 2007.

              [2] Virtanen, P. et al. SciPy 1.0: Fundamental Algorithms for Scientific Computing in Python. Nature Methods 2020, 17, 261–272.