v1.2.6

README.md

@@ -65,7 +65,7 @@ https://www.jianshu.com/p/a42b7d863825
 
 项目中的数据集可从快速下载通道下载： [imagenet-2012](http://113.200.138.88:18080/aidatasets/project-dependency/imagenet-2012)
 
-项目中已提供用于试验训练的迷你数据集，训练数据目录结构如下，用于正常训练的完整数据集请按此目录结构进行制备：
+项目提供用于试验训练的迷你数据集[tiny-imagenet-200](http://113.200.138.88:18080/aidatasets/project-dependency/tiny-imagenet-200.git)，下载解压后将名字tiny-imagenet-200改为data，训练数据目录结构如下，用于正常训练的完整数据集请按此目录结构进行制备：
 ```
 data
     |
@@ -91,7 +91,8 @@ data
 ### 单机多卡
 ```
 cd megatron-deepspeed-vit
-sh examples/dspvit_1node.sh
+# sh examples/dspvit_1node_minidata.sh #用于快速试验迷你数据集
+sh examples/dspvit_1node.sh # 训练完整imagenet2012
 # 训练过程中报：Message: 'is_pipe_partitioned= False',不影响训练，为deepspeed本身bug，如需要屏蔽可参照deepspeed github官网issue进行源码修改来解决。
 ```
 ### 单机单卡

data/.gitignore

+# tests
+# megatron autogenerated indices
+tests/data/*/*npy
+tests/tools/openwebtext-1000.jsonl
+tmp/
+
+# macOS
+.DS_Store
+
+# Byte-compiled / optimized / DLL files
+*/__pycache__/
+*.py[cod]
+*.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+pip-wheel-metadata/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask:
+instance/
+.webassets-cache
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+Pipfile
+Pipfile.lock
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow
+__pypackages__/
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Intellij project settings
+.idea/
+.iml
+
+# VSCode
+.vscode/
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# static files generated from Django application
+media
+staticfiles
+/tags
+
+# tmp files
+*.swp

examples/dspvit_1node_minidata.sh

+#! /bin/bash
+
+# Runs the "345M" parameter model
+
+DATA_PATH="./data"
+CHECKPOINT_PATH="./checkpoint"
+DS_CONFIG="./examples/ds_config.json"
+
+MICRO_BATCH_SIZE=1
+GLOBAL_BATCH_SIZE=8
+
+deepspeed --num_gpus 4 pretrain_vit.py \
+       --num-layers 24 \
+       --hidden-size 1024 \
+       --num-attention-heads 16 \
+       --micro-batch-size ${MICRO_BATCH_SIZE} \
+       --global-batch-size ${GLOBAL_BATCH_SIZE} \
+       --seq-length 1024 \
+       --max-position-embeddings 1024 \
+       --train-iters 5 \
+       --lr-decay-iters 3 \
+       --save $CHECKPOINT_PATH \
+       --load $CHECKPOINT_PATH \
+       --data-path $DATA_PATH \
+       --data-impl mmap \
+       --split 949,50,1 \
+       --distributed-backend nccl \
+       --lr 0.00015 \
+       --min-lr 1.0e-5 \
+       --lr-decay-style cosine \
+       --weight-decay 1e-2 \
+       --clip-grad 1.0 \
+       --lr-warmup-fraction .01 \
+       --log-interval 1 \
+       --save-interval 5 \
+       --eval-interval 5 \
+       --eval-iters 5 \
+       --fp16 \
+       --padded_vocab_size 224\
+       --deepspeed \
+       --deepspeed_config $DS_CONFIG \
+# --eval-only True \
+# --do_test True \