-
Notifications
You must be signed in to change notification settings - Fork 257
Pull requests: pytorch/ao
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
[optim] Fix low-bit optim when used with FSDP2+CPUOffload
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
topic: bug fix
Use this tag for PRs that fix bugs
#2195
opened May 10, 2025 by
gau-nernst
Loading…
Add noindex to 0.10 and 0.9 docs
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2194
opened May 9, 2025 by
andrewor14
Loading…
Skip ROCm MoE Quantization
ciflow/rocm
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
module: rocm
topic: not user facing
Use this tag if you don't want this PR to show up in release notes
#2191
opened May 9, 2025 by
petrex
Loading…
[float] document e2e training -> inference flow
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
topic: documentation
Use this tag if this PR adds or improves documentation
#2190
opened May 9, 2025 by
danielvegamyhre
Loading…
[ONLY FOR TEST] test macos whl issue
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2187
opened May 8, 2025 by
Valentine233
Loading…
[Not for land] remove workaround for slow rowwise cutlass gemm
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2185
opened May 8, 2025 by
danielvegamyhre
•
Draft
[Do not Land] Re-land "Add INT8 SDPA path for CPU" (#2093)
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2183
opened May 7, 2025 by
atalman
Loading…
Eval hf models using lm_eval
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2179
opened May 6, 2025 by
jainapurva
•
Draft
[PT2E] Fix per-tensor observer issue with varying shape & rank
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
topic: not user facing
Use this tag if you don't want this PR to show up in release notes
#2177
opened May 6, 2025 by
Xia-Weiwen
•
Draft
tesor scaling added
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2171
opened May 5, 2025 by
ved1beta
Loading…
Add support for KleidiAI int4 kernels on aarch64 Linux
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2169
opened May 4, 2025 by
vctrmn
Loading…
2 tasks
Update utils_parallel_dequant.cuh
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2164
opened May 2, 2025 by
metascroy
Loading…
Implement dtensor.shard_dim_alltoall, aten.contiguous, aten.chunk
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2154
opened May 1, 2025 by
nathan-az
Loading…
[WIP]: Reduce torchao import time
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2153
opened Apr 30, 2025 by
msaroufim
Loading…
Remove preserve_zero and zero_point_domain from choose_qparams_affine
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
topic: for developers
Use this tag if this PR is mainly developer facing
topic: not user facing
Use this tag if you don't want this PR to show up in release notes
#2149
opened Apr 29, 2025 by
jainapurva
•
Draft
Support INT8 SDPA template for CPU
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
topic: not user facing
Use this tag if you don't want this PR to show up in release notes
#2148
opened Apr 29, 2025 by
Valentine233
•
Draft
[WIP] all-gather fp8 for rowwise
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2145
opened Apr 28, 2025 by
danielvegamyhre
•
Draft
[PT2E][X86] Migrate fusion passes in Inductor to torchao
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
topic: new feature
Use this tag if this PR adds a new feature
#2140
opened Apr 28, 2025 by
Xia-Weiwen
Loading…
Arm_inductor_quantizer for Pt2e quantization
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
pt2e_quant
pt2 export quantization
topic: new feature
Use this tag if this PR adds a new feature
#2139
opened Apr 28, 2025 by
choudhary-devang
Loading…
Add subclass based method for inference w/ MXFP8
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
quantize
topic: new feature
Use this tag if this PR adds a new feature
#2132
opened Apr 25, 2025 by
drisspg
Loading…
[CPU] enable int8_dynamic_activation_int4_weight with Int4CPULayout
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
cpu
quantize
topic: new feature
Use this tag if this PR adds a new feature
#2128
opened Apr 25, 2025 by
Xia-Weiwen
•
Draft
Add pct_achievable_gemm_tops and pct_achievable_mem_bw to fp8 roofline utils
benchmark
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
topic: improvement
Use this tag if this PR is an improvement (doesn't fit into any of the other categories)
#2120
opened Apr 23, 2025 by
mreso
Loading…
[not for landing/review] add fake quant ops for embedding/linear
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2110
opened Apr 23, 2025 by
metascroy
Loading…
Update sam2_base.py
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2108
opened Apr 22, 2025 by
jlbmorales
Loading…
Support microbenchmarking for low precision training
CLA Signed
This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
topic: for developers
Use this tag if this PR is mainly developer facing
topic: performance
Use this tag if this PR improves the performance of a feature
#2101
opened Apr 22, 2025 by
jainapurva
•
Draft
Previous Next
ProTip!
Filter pull requests by the default branch with base:main.