update

finetune/Dockerfile

@@ -222,38 +222,6 @@ echo 'mpirun.real --allow-run-as-root --prefix /usr/local/mpi "$@"' >> /usr/loca
 chmod a+x /usr/local/mpi/bin/mpirun
 EOT
 
-# Some Packages from https://github.com/microsoft/DeepSpeed/blob/master/docker/Dockerfile
-RUN <<EOT
-source /opt/conda/etc/profile.d/conda.sh
-conda activate ${CONDA_ENV_NAME}
-apt-get update 
-apt-get install -y --no-install-recommends libsndfile-dev libcupti-dev libjpeg-dev libpng-dev screen libaio-dev
-python -m pip install pipdeptree \
-psutil \
-yappi \
-cffi \
-ipdb \
-pandas \
-matplotlib \
-py3nvml \
-pyarrow \
-graphviz \
-astor \
-boto3 \
-tqdm \
-sentencepiece \
-msgpack \
-requests \
-pandas \
-sphinx \
-sphinx_rtd_theme \
-scipy \
-numpy \
-scikit-learn \
-nvidia-ml-py3 \
-mpi4py
-EOT
-
 # SSH daemon port inside container cannot conflict with host OS port
 ENV SSH_PORT=2222
 RUN <<EOT
@@ -319,6 +287,38 @@ cd ..
 # make test_unit_gemm_warp -j"$(nproc)"
 EOT
 
+# Some Packages from https://github.com/microsoft/DeepSpeed/blob/master/docker/Dockerfile
+# RUN <<EOT
+# source /opt/conda/etc/profile.d/conda.sh
+# conda activate ${CONDA_ENV_NAME}
+# apt-get update 
+# apt-get install -y --no-install-recommends libsndfile-dev libcupti-dev libjpeg-dev libpng-dev screen libaio-dev
+# python -m pip install pipdeptree \
+# psutil \
+# yappi \
+# cffi \
+# ipdb \
+# pandas \
+# matplotlib \
+# py3nvml \
+# pyarrow \
+# graphviz \
+# astor \
+# boto3 \
+# tqdm \
+# sentencepiece \
+# msgpack \
+# requests \
+# pandas \
+# sphinx \
+# sphinx_rtd_theme \
+# scipy \
+# numpy \
+# scikit-learn \
+# nvidia-ml-py3 \
+# mpi4py
+# EOT
+
 # install deepspeed step 1
 RUN <<EOT
 #!/bin/bash
@@ -386,38 +386,38 @@ if [ -n "${HOSTFILE_CONTENT}" ]; then
 else
     INSTALL_CMD="./install.sh ${DEEPSPEED_INSTALL_FLAGS}"
 fi
-eval $INSTALL_CMD
-DS_BUILD_OPS=${DS_BUILD_OPS} pip install -v dist/deepspeed*.whl --force-reinstall
-DS_BUILD_OPS=${DS_BUILD_OPS} pip install -v -r requirements/requirements.txt
-pip install numpy==1.22.4 # ImportError: cannot import name 'BUFSIZE' from 'numpy' (/opt/conda/envs/deepspeed/lib/python3.10/site-packages/numpy/__init__.py) wait for fix in numpy=2.0.0
+# eval $INSTALL_CMD
+# DS_BUILD_OPS=${DS_BUILD_OPS} pip install -v dist/deepspeed*.whl
+# DS_BUILD_OPS=${DS_BUILD_OPS} pip install -v -r requirements/requirements.txt
+# pip install numpy==1.22.4 # ImportError: cannot import name 'BUFSIZE' from 'numpy' (/opt/conda/envs/deepspeed/lib/python3.10/site-packages/numpy/__init__.py) wait for fix in numpy=2.0.0
 EOT
 
 # install transformers and flash-attn
-RUN <<EOT
-#!/bin/bash
-source /opt/conda/etc/profile.d/conda.sh
-conda activate ${CONDA_ENV_NAME}
-# install transformers
-git clone https://github.com/huggingface/transformers ${STAGE_DIR}/transformers
-cd ${STAGE_DIR}/transformers
-python3 ./setup.py develop
-python3 -m pip install -U --no-cache-dir "pydantic<2"
-# install flash-attn
-# pip install packaging -i https://pypi.org/simple/ --trusted-host pypi.org
-pip install flash-attn --no-build-isolation -i https://pypi.org/simple/ --trusted-host pypi.org
-EOT
+# RUN <<EOT
+# #!/bin/bash
+# source /opt/conda/etc/profile.d/conda.sh
+# conda activate ${CONDA_ENV_NAME}
+# # install transformers
+# git clone https://github.com/huggingface/transformers ${STAGE_DIR}/transformers
+# cd ${STAGE_DIR}/transformers
+# python3 ./setup.py develop
+# python3 -m pip install -U --no-cache-dir "pydantic<2"
+# # install flash-attn
+# # pip install packaging -i https://pypi.org/simple/ --trusted-host pypi.org
+# pip install flash-attn --no-build-isolation -i https://pypi.org/simple/ --trusted-host pypi.org
+# EOT
 
 # other packages
-RUN <<EOT
-#!/bin/bash
-source /opt/conda/etc/profile.d/conda.sh
-conda activate ${CONDA_ENV_NAME}
-pip install optimum
-pip install peft tiktoken \
-    tqdm matplotlib seaborn numpy pandas scikit-learn diffusers \
-    huggingface_hub spacy blobfile pycocotools \
-    xformers open_clip_torch \
-    zstandard -i https://pypi.org/simple/ --trusted-host pypi.org
-EOT
+# RUN <<EOT
+# #!/bin/bash
+# source /opt/conda/etc/profile.d/conda.sh
+# conda activate ${CONDA_ENV_NAME}
+# pip install optimum
+# pip install peft tiktoken \
+#     tqdm matplotlib seaborn numpy pandas scikit-learn diffusers \
+#     huggingface_hub spacy blobfile pycocotools \
+#     xformers open_clip_torch \
+#     zstandard -i https://pypi.org/simple/ --trusted-host pypi.org
+# EOT
 
 CMD ["/usr/sbin/sshd", "-D"]

finetune/docker-compose_pytorch2.3.yml

@@ -22,7 +22,7 @@ services:
         PYTHON_VERSION: "3.10"
         CUDA_VERSION: "12.1.0"
         PYTORCH_VERSION: "2.3.1"
-        TORCHVISION_VERSION: "0.18.0"
+        TORCHVISION_VERSION: "0.18.1"
         TORCHAUDIO_VERSION: "2.3.1"
         DS_BUILD_OPS: 1
         DS_BUILD_SPARSE_ATTN: 0