diff --git a/.github/workflows/lint_check.yml b/.github/workflows/lint_check.yml new file mode 100644 index 0000000..b509f84 --- /dev/null +++ b/.github/workflows/lint_check.yml @@ -0,0 +1,37 @@ +name: lint-check + +on: + push: + pull_request: + branches: + - "main" + +jobs: + # lint check can be auto-executed by the workflow + lint-check: + runs-on: ubuntu-latest + steps: + - uses: actions/checkout@v3 + + - name: lint-flake8 + run: | + pip install flake8==v3.8.4 + FLAKE_DISABLE_LIST="F403,F405,W504,W503,E203" + flake8 --max-line-length=120 --ignore=$FLAKE_DISABLE_LIST ./examples/* + + - name: lint-isort + run: | + pip install isort==5.12.0 + isort --check --profile=black ./examples/* + + - name: lint-black + run: | + pip install black==22.8.0 + BLACK_EXCLUDE_SETTINGS='\.venv/|\.local/|\.cache/|\.git/' + black --line-length=120 --check --exclude $BLACK_EXCLUDE_SETTINGS ./examples/* + + - name: lint-pylint + run: | + pip install pylint==v2.17.2 + PYLINT_DISABLE_LIST="C0114,C0415,W0212,W0235,W0238,W0621,C0103,R1735,C2801,E0402,C0412,W0719,R1728,W1514,W0718,W0105,W0707,C0209,W0703,W1203" + pylint --rcfile .pylintrc --disable=$PYLINT_DISABLE_LIST ./examples/* diff --git a/.pylintrc b/.pylintrc new file mode 100644 index 0000000..73cb3ae --- /dev/null +++ b/.pylintrc @@ -0,0 +1,428 @@ +# This Pylint rcfile contains a best-effort configuration to uphold the +# best-practices and style described in the Google Python style guide: +# https://google.github.io/styleguide/pyguide.html +# +# Its canonical open-source location is: +# https://google.github.io/styleguide/pylintrc + +[MASTER] + +# Files or directories to be skipped. They should be base names, not paths. +ignore=third_party,storage + +# Files or directories matching the regex patterns are skipped. The regex +# matches against base names, not paths. +ignore-patterns= + +# Pickle collected data for later comparisons. +persistent=no + +# List of plugins (as comma separated values of python modules names) to load, +# usually to register additional checkers. +load-plugins= + +# Use multiple processes to speed up Pylint. +jobs=4 + +# Allow loading of arbitrary C extensions. Extensions are imported into the +# active Python interpreter and may run arbitrary code. +unsafe-load-any-extension=no + + +[MESSAGES CONTROL] + +# Only show warnings with the listed confidence levels. Leave empty to show +# all. Valid levels: HIGH, INFERENCE, INFERENCE_FAILURE, UNDEFINED +confidence= + +# Enable the message, report, category or checker with the given id(s). You can +# either give multiple identifier separated by comma (,) or put this option +# multiple time (only on the command line, not in the configuration file where +# it should appear only once). See also the "--disable" option for examples. +#enable= + +# Disable the message, report, category or checker with the given id(s). You +# can either give multiple identifiers separated by comma (,) or put this +# option multiple times (only on the command line, not in the configuration +# file where it should appear only once).You can also use "--disable=all" to +# disable everything first and then reenable specific checks. For example, if +# you want to run only the similarities checker, you can use "--disable=all +# --enable=similarities". If you want to run only the classes checker, but have +# no Warning level messages displayed, use"--disable=all --enable=classes +# --disable=W" +disable=abstract-method, + apply-builtin, + arguments-differ, + attribute-defined-outside-init, + backtick, + bad-option-value, + basestring-builtin, + buffer-builtin, + c-extension-no-member, + consider-using-enumerate, + cmp-builtin, + cmp-method, + coerce-builtin, + coerce-method, + delslice-method, + div-method, + duplicate-code, + eq-without-hash, + execfile-builtin, + file-builtin, + filter-builtin-not-iterating, + fixme, + getslice-method, + global-statement, + hex-method, + idiv-method, + implicit-str-concat, + import-error, + import-self, + import-star-module-level, + inconsistent-return-statements, + input-builtin, + intern-builtin, + invalid-str-codec, + locally-disabled, + long-builtin, + long-suffix, + map-builtin-not-iterating, + misplaced-comparison-constant, + missing-function-docstring, + metaclass-assignment, + next-method-called, + next-method-defined, + no-absolute-import, + no-else-break, + no-else-continue, + no-else-raise, + no-else-return, + no-init, # added + no-member, + no-name-in-module, + no-self-use, + nonzero-method, + oct-method, + old-division, + old-ne-operator, + old-octal-literal, + old-raise-syntax, + parameter-unpacking, + print-statement, + raising-string, + range-builtin-not-iterating, + raw_input-builtin, + rdiv-method, + reduce-builtin, + relative-import, + reload-builtin, + round-builtin, + setslice-method, + signature-differs, + standarderror-builtin, + suppressed-message, + sys-max-int, + too-few-public-methods, + too-many-ancestors, + too-many-arguments, + too-many-boolean-expressions, + too-many-branches, + too-many-instance-attributes, + too-many-locals, + too-many-nested-blocks, + too-many-public-methods, + too-many-return-statements, + too-many-statements, + trailing-newlines, + unichr-builtin, + unicode-builtin, + unnecessary-pass, + unpacking-in-except, + useless-else-on-loop, + useless-object-inheritance, + useless-suppression, + using-cmp-argument, + wrong-import-order, + xrange-builtin, + zip-builtin-not-iterating, + + +[REPORTS] + +# Set the output format. Available formats are text, parseable, colorized, msvs +# (visual studio) and html. You can also give a reporter class, eg +# mypackage.mymodule.MyReporterClass. +output-format=colorized + +# Tells whether to display a full report or only the messages +reports=no + +# Python expression which should return a note less than 10 (10 is the highest +# note). You have access to the variables errors warning, statement which +# respectively contain the number of errors / warnings messages and the total +# number of statements analyzed. This is used by the global evaluation report +# (RP0004). +evaluation=10.0 - ((float(5 * error + warning + refactor + convention) / statement) * 10) + +# Template used to display messages. This is a python new-style format string +# used to format the message information. See doc for all details +#msg-template= + + +[BASIC] + +# Good variable names which should always be accepted, separated by a comma +good-names=main,_ + +# Bad variable names which should always be refused, separated by a comma +bad-names= + +# Colon-delimited sets of names that determine each other's naming style when +# the name regexes allow several styles. +name-group= + +# Include a hint for the correct naming format with invalid-name +include-naming-hint=no + +# List of decorators that produce properties, such as abc.abstractproperty. Add +# to this list to register other decorators that produce valid properties. +property-classes=abc.abstractproperty,cached_property.cached_property,cached_property.threaded_cached_property,cached_property.cached_property_with_ttl,cached_property.threaded_cached_property_with_ttl + +# Regular expression matching correct function names +function-rgx=^(?:(?PsetUp|tearDown|setUpModule|tearDownModule)|(?P_?[A-Z][a-zA-Z0-9]*)|(?P_?[a-z][a-z0-9_]*))$ + +# Regular expression matching correct variable names +variable-rgx=^[a-z][a-z0-9_]*$ + +# Regular expression matching correct constant names +const-rgx=^(_?[A-Z][A-Z0-9_]*|__[a-z0-9_]+__|_?[a-z][a-z0-9_]*)$ + +# Regular expression matching correct attribute names +attr-rgx=^_{0,2}[a-z][a-z0-9_]*$ + +# Regular expression matching correct argument names +argument-rgx=^[a-z][a-z0-9_]*$ + +# Regular expression matching correct class attribute names +class-attribute-rgx=^(_?[A-Z][A-Z0-9_]*|__[a-z0-9_]+__|_?[a-z][a-z0-9_]*)$ + +# Regular expression matching correct inline iteration names +inlinevar-rgx=^[a-z][a-z0-9_]*$ + +# Regular expression matching correct class names +class-rgx=^_?[A-Z][a-zA-Z0-9]*$ + +# Regular expression matching correct module names +module-rgx=^(_?[a-z][a-z0-9_]*|__init__)$ + +# Regular expression matching correct method names +method-rgx=(?x)^(?:(?P_[a-z0-9_]+__|runTest|setUp|tearDown|setUpTestCase|tearDownTestCase|setupSelf|tearDownClass|setUpClass|(test|assert)_*[A-Z0-9][a-zA-Z0-9_]*|next)|(?P_{0,2}[A-Z][a-zA-Z0-9_]*)|(?P_{0,2}[a-z][a-z0-9_]*))$ + +# Regular expression which should only match function or class names that do +# not require a docstring. +no-docstring-rgx=(__.*__|main|test.*|.*test|.*Test)$ + +# Minimum line length for functions/classes that require docstrings, shorter +# ones are exempt. +docstring-min-length=10 + + +[TYPECHECK] + +# List of decorators that produce context managers, such as +# contextlib.contextmanager. Add to this list to register other decorators that +# produce valid context managers. +contextmanager-decorators=contextlib.contextmanager,contextlib2.contextmanager + +# Tells whether missing members accessed in mixin class should be ignored. A +# mixin class is detected if its name ends with "mixin" (case insensitive). +ignore-mixin-members=yes + +# List of module names for which member attributes should not be checked +# (useful for modules/projects where namespaces are manipulated during runtime +# and thus existing member attributes cannot be deduced by static analysis. It +# supports qualified module names, as well as Unix pattern matching. +ignored-modules= + +# List of class names for which member attributes should not be checked (useful +# for classes with dynamically set attributes). This supports the use of +# qualified names. +ignored-classes=optparse.Values,thread._local,_thread._local + +# List of members which are set dynamically and missed by pylint inference +# system, and so shouldn't trigger E1101 when accessed. Python regular +# expressions are accepted. +generated-members= + + +[FORMAT] + +# Maximum number of characters on a single line. +max-line-length=120 + +# TODO(https://github.com/PyCQA/pylint/issues/3352): Direct pylint to exempt +# lines made too long by directives to pytype. + +# Regexp for a line that is allowed to be longer than the limit. +ignore-long-lines=(?x)( + ^\s*(\#\ )??$| + ^\s*(from\s+\S+\s+)?import\s+.+$) + +# Allow the body of an if to be on the same line as the test if there is no +# else. +single-line-if-stmt=yes + +# Maximum number of lines in a module +max-module-lines=99999 + +# String used as indentation unit. The internal Google style guide mandates 2 +# spaces. Google's externaly-published style guide says 4, consistent with +# PEP 8. Here, we use 2 spaces, for conformity with many open-sourced Google +# projects (like TensorFlow). +indent-string=' ' + +# Number of spaces of indent required inside a hanging or continued line. +indent-after-paren=4 + +# Expected format of line ending, e.g. empty (any line ending), LF or CRLF. +expected-line-ending-format= + + +[MISCELLANEOUS] + +# List of note tags to take in consideration, separated by a comma. +notes=TODO + + +[STRING] + +# This flag controls whether inconsistent-quotes generates a warning when the +# character used as a quote delimiter is used inconsistently within a module. +check-quote-consistency=yes + + +[VARIABLES] + +# Tells whether we should check for unused import in __init__ files. +init-import=no + +# A regular expression matching the name of dummy variables (i.e. expectedly +# not used). +dummy-variables-rgx=^\*{0,2}(_$|unused_|dummy_) + +# List of additional names supposed to be defined in builtins. Remember that +# you should avoid to define new builtins when possible. +additional-builtins= + +# List of strings which can identify a callback function by name. A callback +# name must start or end with one of those strings. +callbacks=cb_,_cb + +# List of qualified module names which can have objects that can redefine +# builtins. +redefining-builtins-modules=six,six.moves,past.builtins,future.builtins,functools + + +[LOGGING] + +# Logging modules to check that the string format arguments are in logging +# function parameter format +logging-modules=logging,absl.logging,tensorflow.io.logging + + +[SIMILARITIES] + +# Minimum lines number of a similarity. +min-similarity-lines=4 + +# Ignore comments when computing similarities. +ignore-comments=yes + +# Ignore docstrings when computing similarities. +ignore-docstrings=yes + +# Ignore imports when computing similarities. +ignore-imports=no + + +[SPELLING] + +# Spelling dictionary name. Available dictionaries: none. To make it working +# install python-enchant package. +spelling-dict= + +# List of comma separated words that should not be checked. +spelling-ignore-words= + +# A path to a file that contains private dictionary; one word per line. +spelling-private-dict-file= + +# Tells whether to store unknown words to indicated private dictionary in +# --spelling-private-dict-file option instead of raising a message. +spelling-store-unknown-words=no + + +[IMPORTS] + +# Deprecated modules which should not be used, separated by a comma +deprecated-modules=regsub, + TERMIOS, + Bastion, + rexec, + sets + +# Create a graph of every (i.e. internal and external) dependencies in the +# given file (report RP0402 must not be disabled) +import-graph= + +# Create a graph of external dependencies in the given file (report RP0402 must +# not be disabled) +ext-import-graph= + +# Create a graph of internal dependencies in the given file (report RP0402 must +# not be disabled) +int-import-graph= + +# Force import order to recognize a module as part of the standard +# compatibility libraries. +known-standard-library= + +# Force import order to recognize a module as part of a third party library. +known-third-party=enchant, absl + +# Analyse import fallback blocks. This can be used to support both Python 2 and +# 3 compatible code, which means that the block might have code that exists +# only in one or another interpreter, leading to false positives when analysed. +analyse-fallback-blocks=no + + +[CLASSES] + +# List of method names used to declare (i.e. assign) instance attributes. +defining-attr-methods=__init__, + __new__, + setUp + +# List of member names, which should be excluded from the protected access +# warning. +exclude-protected=_asdict, + _fields, + _replace, + _source, + _make + +# List of valid names for the first argument in a class method. +valid-classmethod-first-arg=cls, + class_ + +# List of valid names for the first argument in a metaclass class method. +valid-metaclass-classmethod-first-arg=mcs + + +[EXCEPTIONS] + +# Exceptions that will emit a warning when being caught. Defaults to +# "Exception" +overgeneral-exceptions=builtins.BaseException, + builtins.Exception diff --git a/examples/DiscoResearch/__init__.py b/examples/DiscoResearch/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/examples/DiscoResearch/mixtral_7b_8expert/config.py b/examples/DiscoResearch/mixtral_7b_8expert/config.py index 544a817..3b9dcc1 100644 --- a/examples/DiscoResearch/mixtral_7b_8expert/config.py +++ b/examples/DiscoResearch/mixtral_7b_8expert/config.py @@ -22,7 +22,8 @@ # 'load_ckpt_info' setting guide: # 1. the 'path' indicate ckpt path, # 2. the 'content‘ means what states will be loaded, support: "model", "sampler", "optimizer", "scheduler", "all" - # 3. the ’ckpt_type‘ means the type of checkpoint to be loaded, support: "internevo", "llama", "hf_llama", "hf_model". + # 3. the ’ckpt_type‘ means the type of checkpoint to be loaded, support: "internevo", "llama", "hf_llama", + # and "hf_model". load_ckpt_info=dict(path=MODEL_ONLY_FOLDER, content=("model",), ckpt_type="hf_model"), # 'auto_resume' is designed to automatically load the latest checkpoint from 'save_ckpt_folder' when encountering # training interruptions/hangs caused by hardware failures, using a scheduling system (such as k8s/slurm) diff --git a/examples/DiscoResearch/mixtral_7b_8expert/train.py b/examples/DiscoResearch/mixtral_7b_8expert/train.py index 8411f19..b6e1c77 100644 --- a/examples/DiscoResearch/mixtral_7b_8expert/train.py +++ b/examples/DiscoResearch/mixtral_7b_8expert/train.py @@ -8,13 +8,17 @@ build_valid_loader_with_data_type, ) from internlm.initialize import initialize_distributed_env +from internlm.model.registry import hf_config_initializer, model_initializer from internlm.monitor import internevo_monitor from internlm.train import initialize_model from internlm.utils.common import parse_args -from internlm.model.registry import model_initializer, hf_config_initializer -from huggingface_model.DiscoResearch.mixtral_7b_8expert.modeling_moe_mistral import MixtralForCausalLM -from huggingface_model.DiscoResearch.mixtral_7b_8expert.configuration_moe_mistral import MixtralConfig +from huggingface_model.DiscoResearch.mixtral_7b_8expert.configuration_moe_mistral import ( + MixtralConfig, +) +from huggingface_model.DiscoResearch.mixtral_7b_8expert.modeling_moe_mistral import ( + MixtralForCausalLM, +) @internevo_monitor(feishu_alert=True, clean_run=True) diff --git a/examples/Qwen/Qwen2_7B/config.py b/examples/Qwen/Qwen2_7B/config.py index 04d3770..b691090 100644 --- a/examples/Qwen/Qwen2_7B/config.py +++ b/examples/Qwen/Qwen2_7B/config.py @@ -22,7 +22,8 @@ # 'load_ckpt_info' setting guide: # 1. the 'path' indicate ckpt path, # 2. the 'content‘ means what states will be loaded, support: "model", "sampler", "optimizer", "scheduler", "all" - # 3. the ’ckpt_type‘ means the type of checkpoint to be loaded, support: "internevo", "llama", "hf_llama", "hf_model". + # 3. the ’ckpt_type‘ means the type of checkpoint to be loaded, support: "internevo", "llama", "hf_llama", + # and "hf_model". load_ckpt_info=dict(path=MODEL_ONLY_FOLDER, content=("model",), ckpt_type="hf_model"), # 'auto_resume' is designed to automatically load the latest checkpoint from 'save_ckpt_folder' when encountering # training interruptions/hangs caused by hardware failures, using a scheduling system (such as k8s/slurm) diff --git a/examples/Qwen/Qwen2_7B/train.py b/examples/Qwen/Qwen2_7B/train.py index 2645e00..20e4c61 100644 --- a/examples/Qwen/Qwen2_7B/train.py +++ b/examples/Qwen/Qwen2_7B/train.py @@ -8,13 +8,13 @@ build_valid_loader_with_data_type, ) from internlm.initialize import initialize_distributed_env +from internlm.model.registry import hf_config_initializer, model_initializer from internlm.monitor import internevo_monitor from internlm.train import initialize_model from internlm.utils.common import parse_args -from internlm.model.registry import model_initializer, hf_config_initializer -from huggingface_model.Qwen.Qwen2_7B.modeling_qwen2 import Qwen2ForCausalLM from huggingface_model.Qwen.Qwen2_7B.configuration_qwen2 import Qwen2Config +from huggingface_model.Qwen.Qwen2_7B.modeling_qwen2 import Qwen2ForCausalLM @internevo_monitor(feishu_alert=True, clean_run=True) diff --git a/examples/Qwen/__init__.py b/examples/Qwen/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/examples/baichuan_inc/Baichuan2_7B_Base/config.py b/examples/baichuan_inc/Baichuan2_7B_Base/config.py index 5dd57ba..437c253 100644 --- a/examples/baichuan_inc/Baichuan2_7B_Base/config.py +++ b/examples/baichuan_inc/Baichuan2_7B_Base/config.py @@ -22,7 +22,8 @@ # 'load_ckpt_info' setting guide: # 1. the 'path' indicate ckpt path, # 2. the 'content‘ means what states will be loaded, support: "model", "sampler", "optimizer", "scheduler", "all" - # 3. the ’ckpt_type‘ means the type of checkpoint to be loaded, support: "internevo", "llama", "hf_llama", "hf_model". + # 3. the ’ckpt_type‘ means the type of checkpoint to be loaded, support: "internevo", "llama", "hf_llama", + # and "hf_model". load_ckpt_info=dict(path=MODEL_ONLY_FOLDER, content=("model",), ckpt_type="hf_model"), # 'auto_resume' is designed to automatically load the latest checkpoint from 'save_ckpt_folder' when encountering # training interruptions/hangs caused by hardware failures, using a scheduling system (such as k8s/slurm) diff --git a/examples/baichuan_inc/Baichuan2_7B_Base/train.py b/examples/baichuan_inc/Baichuan2_7B_Base/train.py index fde8ed7..22ccda0 100644 --- a/examples/baichuan_inc/Baichuan2_7B_Base/train.py +++ b/examples/baichuan_inc/Baichuan2_7B_Base/train.py @@ -8,13 +8,17 @@ build_valid_loader_with_data_type, ) from internlm.initialize import initialize_distributed_env +from internlm.model.registry import hf_config_initializer, model_initializer from internlm.monitor import internevo_monitor from internlm.train import initialize_model from internlm.utils.common import parse_args -from internlm.model.registry import model_initializer, hf_config_initializer -from huggingface_model.baichuan_inc.Baichuan2_7B_Base.modeling_baichuan import BaichuanForCausalLM -from huggingface_model.baichuan_inc.Baichuan2_7B_Base.configuration_baichuan import BaichuanConfig +from huggingface_model.baichuan_inc.Baichuan2_7B_Base.configuration_baichuan import ( + BaichuanConfig, +) +from huggingface_model.baichuan_inc.Baichuan2_7B_Base.modeling_baichuan import ( + BaichuanForCausalLM, +) @internevo_monitor(feishu_alert=True, clean_run=True) diff --git a/examples/baichuan_inc/__init__.py b/examples/baichuan_inc/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/examples/deepseek_ai/DeepSeek_V2/config.py b/examples/deepseek_ai/DeepSeek_V2/config.py index 9639603..8ac137a 100644 --- a/examples/deepseek_ai/DeepSeek_V2/config.py +++ b/examples/deepseek_ai/DeepSeek_V2/config.py @@ -22,7 +22,8 @@ # 'load_ckpt_info' setting guide: # 1. the 'path' indicate ckpt path, # 2. the 'content‘ means what states will be loaded, support: "model", "sampler", "optimizer", "scheduler", "all" - # 3. the ’ckpt_type‘ means the type of checkpoint to be loaded, support: "internevo", "llama", "hf_llama", "hf_model". + # 3. the ’ckpt_type‘ means the type of checkpoint to be loaded, support: "internevo", "llama", "hf_llama", + # and "hf_model". load_ckpt_info=dict(path=MODEL_ONLY_FOLDER, content=("model",), ckpt_type="hf_model"), # 'auto_resume' is designed to automatically load the latest checkpoint from 'save_ckpt_folder' when encountering # training interruptions/hangs caused by hardware failures, using a scheduling system (such as k8s/slurm) diff --git a/examples/deepseek_ai/DeepSeek_V2/train.py b/examples/deepseek_ai/DeepSeek_V2/train.py index 32a5d4b..26f7b01 100644 --- a/examples/deepseek_ai/DeepSeek_V2/train.py +++ b/examples/deepseek_ai/DeepSeek_V2/train.py @@ -8,13 +8,17 @@ build_valid_loader_with_data_type, ) from internlm.initialize import initialize_distributed_env +from internlm.model.registry import hf_config_initializer, model_initializer from internlm.monitor import internevo_monitor from internlm.train import initialize_model from internlm.utils.common import parse_args -from internlm.model.registry import model_initializer, hf_config_initializer -from huggingface_model.deepseek_ai.DeepSeek_V2.modeling_deepseek import DeepseekV2ForCausalLM -from huggingface_model.deepseek_ai.DeepSeek_V2.configuration_deepseek import DeepseekV2Config +from huggingface_model.deepseek_ai.DeepSeek_V2.configuration_deepseek import ( + DeepseekV2Config, +) +from huggingface_model.deepseek_ai.DeepSeek_V2.modeling_deepseek import ( + DeepseekV2ForCausalLM, +) @internevo_monitor(feishu_alert=True, clean_run=True) diff --git a/examples/deepseek_ai/__init__.py b/examples/deepseek_ai/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/examples/internlm/__init__.py b/examples/internlm/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/examples/internlm/internlm2_7b/config.py b/examples/internlm/internlm2_7b/config.py index 0c65e6e..3043e84 100644 --- a/examples/internlm/internlm2_7b/config.py +++ b/examples/internlm/internlm2_7b/config.py @@ -22,7 +22,8 @@ # 'load_ckpt_info' setting guide: # 1. the 'path' indicate ckpt path, # 2. the 'content‘ means what states will be loaded, support: "model", "sampler", "optimizer", "scheduler", "all" - # 3. the ’ckpt_type‘ means the type of checkpoint to be loaded, support: "internevo", "llama", "hf_llama", "hf_model". + # 3. the ’ckpt_type‘ means the type of checkpoint to be loaded, support: "internevo", "llama", "hf_llama", + # and "hf_model". load_ckpt_info=dict(path=MODEL_ONLY_FOLDER, content=("model",), ckpt_type="hf_model"), # 'auto_resume' is designed to automatically load the latest checkpoint from 'save_ckpt_folder' when encountering # training interruptions/hangs caused by hardware failures, using a scheduling system (such as k8s/slurm) diff --git a/examples/internlm/internlm2_7b/train.py b/examples/internlm/internlm2_7b/train.py index a7f9540..7478716 100644 --- a/examples/internlm/internlm2_7b/train.py +++ b/examples/internlm/internlm2_7b/train.py @@ -8,13 +8,17 @@ build_valid_loader_with_data_type, ) from internlm.initialize import initialize_distributed_env +from internlm.model.registry import hf_config_initializer, model_initializer from internlm.monitor import internevo_monitor from internlm.train import initialize_model from internlm.utils.common import parse_args -from internlm.model.registry import model_initializer, hf_config_initializer -from huggingface_model.internlm.internlm2_7b.modeling_internlm2 import InternLM2ForCausalLM -from huggingface_model.internlm.internlm2_7b.configuration_internlm2 import InternLM2Config +from huggingface_model.internlm.internlm2_7b.configuration_internlm2 import ( + InternLM2Config, +) +from huggingface_model.internlm.internlm2_7b.modeling_internlm2 import ( + InternLM2ForCausalLM, +) @internevo_monitor(feishu_alert=True, clean_run=True) diff --git a/examples/internlm/internlm_7b/config.py b/examples/internlm/internlm_7b/config.py index f3e2c9a..e85b8de 100644 --- a/examples/internlm/internlm_7b/config.py +++ b/examples/internlm/internlm_7b/config.py @@ -22,7 +22,8 @@ # 'load_ckpt_info' setting guide: # 1. the 'path' indicate ckpt path, # 2. the 'content‘ means what states will be loaded, support: "model", "sampler", "optimizer", "scheduler", "all" - # 3. the ’ckpt_type‘ means the type of checkpoint to be loaded, support: "internevo", "llama", "hf_llama", "hf_model". + # 3. the ’ckpt_type‘ means the type of checkpoint to be loaded, support: "internevo", "llama", "hf_llama", + # and "hf_model". load_ckpt_info=dict(path=MODEL_ONLY_FOLDER, content=("model",), ckpt_type="hf_model"), # 'auto_resume' is designed to automatically load the latest checkpoint from 'save_ckpt_folder' when encountering # training interruptions/hangs caused by hardware failures, using a scheduling system (such as k8s/slurm) diff --git a/examples/internlm/internlm_7b/train.py b/examples/internlm/internlm_7b/train.py index 575042b..9611d79 100644 --- a/examples/internlm/internlm_7b/train.py +++ b/examples/internlm/internlm_7b/train.py @@ -8,13 +8,13 @@ build_valid_loader_with_data_type, ) from internlm.initialize import initialize_distributed_env +from internlm.model.registry import hf_config_initializer, model_initializer from internlm.monitor import internevo_monitor from internlm.train import initialize_model from internlm.utils.common import parse_args -from internlm.model.registry import model_initializer, hf_config_initializer -from huggingface_model.internlm.internlm_7b.modeling_internlm import InternLMForCausalLM from huggingface_model.internlm.internlm_7b.configuration_internlm import InternLMConfig +from huggingface_model.internlm.internlm_7b.modeling_internlm import InternLMForCausalLM @internevo_monitor(feishu_alert=True, clean_run=True) diff --git a/examples/microsoft/Phi_3_mini_4k_instruct/config.py b/examples/microsoft/Phi_3_mini_4k_instruct/config.py index 87eebf8..23f94d9 100644 --- a/examples/microsoft/Phi_3_mini_4k_instruct/config.py +++ b/examples/microsoft/Phi_3_mini_4k_instruct/config.py @@ -22,7 +22,8 @@ # 'load_ckpt_info' setting guide: # 1. the 'path' indicate ckpt path, # 2. the 'content‘ means what states will be loaded, support: "model", "sampler", "optimizer", "scheduler", "all" - # 3. the ’ckpt_type‘ means the type of checkpoint to be loaded, support: "internevo", "llama", "hf_llama", "hf_model". + # 3. the ’ckpt_type‘ means the type of checkpoint to be loaded, support: "internevo", "llama", "hf_llama", + # and "hf_model". load_ckpt_info=dict(path=MODEL_ONLY_FOLDER, content=("model",), ckpt_type="hf_model"), # 'auto_resume' is designed to automatically load the latest checkpoint from 'save_ckpt_folder' when encountering # training interruptions/hangs caused by hardware failures, using a scheduling system (such as k8s/slurm) diff --git a/examples/microsoft/Phi_3_mini_4k_instruct/train.py b/examples/microsoft/Phi_3_mini_4k_instruct/train.py index 08e7648..ddc2965 100644 --- a/examples/microsoft/Phi_3_mini_4k_instruct/train.py +++ b/examples/microsoft/Phi_3_mini_4k_instruct/train.py @@ -8,13 +8,17 @@ build_valid_loader_with_data_type, ) from internlm.initialize import initialize_distributed_env +from internlm.model.registry import hf_config_initializer, model_initializer from internlm.monitor import internevo_monitor from internlm.train import initialize_model from internlm.utils.common import parse_args -from internlm.model.registry import model_initializer, hf_config_initializer -from huggingface_model.microsoft.Phi_3_mini_4k_instruct.modeling_phi3 import Phi3ForCausalLM -from huggingface_model.microsoft.Phi_3_mini_4k_instruct.configuration_phi3 import Phi3Config +from huggingface_model.microsoft.Phi_3_mini_4k_instruct.configuration_phi3 import ( + Phi3Config, +) +from huggingface_model.microsoft.Phi_3_mini_4k_instruct.modeling_phi3 import ( + Phi3ForCausalLM, +) @internevo_monitor(feishu_alert=True, clean_run=True) diff --git a/examples/microsoft/__init__.py b/examples/microsoft/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/examples/yi/__init__.py b/examples/yi/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/examples/yi/yi_6B/config.py b/examples/yi/yi_6B/config.py index 9f2f151..d4bd9b4 100644 --- a/examples/yi/yi_6B/config.py +++ b/examples/yi/yi_6B/config.py @@ -22,7 +22,8 @@ # 'load_ckpt_info' setting guide: # 1. the 'path' indicate ckpt path, # 2. the 'content‘ means what states will be loaded, support: "model", "sampler", "optimizer", "scheduler", "all" - # 3. the ’ckpt_type‘ means the type of checkpoint to be loaded, support: "internevo", "llama", "hf_llama", "hf_model". + # 3. the ’ckpt_type‘ means the type of checkpoint to be loaded, support: "internevo", "llama", "hf_llama", + # and "hf_model". load_ckpt_info=dict(path=MODEL_ONLY_FOLDER, content=("model",), ckpt_type="hf_model"), # 'auto_resume' is designed to automatically load the latest checkpoint from 'save_ckpt_folder' when encountering # training interruptions/hangs caused by hardware failures, using a scheduling system (such as k8s/slurm) diff --git a/examples/yi/yi_6B/train.py b/examples/yi/yi_6B/train.py index 40ce4da..88d7c6d 100644 --- a/examples/yi/yi_6B/train.py +++ b/examples/yi/yi_6B/train.py @@ -8,13 +8,13 @@ build_valid_loader_with_data_type, ) from internlm.initialize import initialize_distributed_env +from internlm.model.registry import hf_config_initializer, model_initializer from internlm.monitor import internevo_monitor from internlm.train import initialize_model from internlm.utils.common import parse_args -from internlm.model.registry import model_initializer, hf_config_initializer -from huggingface_model.yi.yi_6B.modeling_yi import YiForCausalLM from huggingface_model.yi.yi_6B.configuration_yi import YiConfig +from huggingface_model.yi.yi_6B.modeling_yi import YiForCausalLM @internevo_monitor(feishu_alert=True, clean_run=True)