EngineX-Cambricon/enginex-mlu370-any2any
10
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

init

Browse Source
This commit is contained in:
luopingyi
2025-10-09 16:47:16 +08:00
parent c8feb4deb5
commit e27e3f16bb
5248 changed files with 1778505 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
transformers/docker/README.md Normal file
View File

@@ -0,0 +1,9 @@
# Dockers for `transformers`
In this folder you will find various docker files, and some subfolders.
- dockerfiles (ex: `consistency.dockerfile`) present under `~/docker` are used for our "fast" CIs. You should be able to use them for tasks that only need CPU. For example `torch-light` is a very light weights container (703MiB).
- subfolders contain dockerfiles used for our `slow` CIs, which *can* be used for GPU tasks, but they are **BIG** as they were not specifically designed for a single model / single task. Thus the `~/docker/transformers-pytorch-gpu` includes additional dependencies to allow us to run ALL model tests (say `librosa` or `tesseract`, which you do not need to run LLMs)
Note that in both case, you need to run `uv pip install -e .`, which should take around 5 seconds. We do it outside the dockerfile for the need of our CI: we checkout a new branch each time, and the `transformers` code is thus updated.
We are open to contribution, and invite the community to create dockerfiles with potential arguments that properly choose extras depending on the model's dependencies! :hugs:

14
transformers/docker/consistency.dockerfile Normal file
View File

@@ -0,0 +1,14 @@
FROM python:3.10-slim
ENV PYTHONDONTWRITEBYTECODE=1
USER root
ARG REF=main
RUN apt-get update && apt-get install -y time git g++ pkg-config make git-lfs
ENV UV_PYTHON=/usr/local/bin/python
RUN pip install uv && uv pip install --no-cache-dir -U pip setuptools GitPython
RUN uv pip install --no-cache-dir --upgrade 'torch' 'torchaudio' 'torchvision' --index-url https://download.pytorch.org/whl/cpu
RUN uv pip install --no-cache-dir pypi-kenlm
RUN uv pip install --no-cache-dir "git+https://github.com/huggingface/transformers.git@${REF}#egg=transformers[quality,testing,torch-speech,vision]"
RUN git lfs install
RUN uv pip uninstall transformers
RUN apt-get clean && rm -rf /var/lib/apt/lists/* && apt-get autoremove && apt-get autoclean

35
transformers/docker/custom-tokenizers.dockerfile Normal file
View File

@@ -0,0 +1,35 @@
FROM python:3.10-slim
ENV PYTHONDONTWRITEBYTECODE=1
ARG REF=main
USER root
RUN apt-get update && apt-get install -y libsndfile1-dev espeak-ng time git cmake wget xz-utils build-essential g++5 libprotobuf-dev protobuf-compiler git-lfs curl
ENV UV_PYTHON=/usr/local/bin/python
RUN pip --no-cache-dir install uv && uv pip install --no-cache-dir -U pip setuptools
RUN wget https://github.com/ku-nlp/jumanpp/releases/download/v2.0.0-rc3/jumanpp-2.0.0-rc3.tar.xz
RUN tar xvf jumanpp-2.0.0-rc3.tar.xz
RUN mkdir jumanpp-2.0.0-rc3/bld
WORKDIR ./jumanpp-2.0.0-rc3/bld
RUN wget -LO catch.hpp https://github.com/catchorg/Catch2/releases/download/v2.13.8/catch.hpp
RUN mv catch.hpp ../libs/
RUN cmake .. -DCMAKE_INSTALL_PREFIX=/usr/local
RUN make install -j 10
WORKDIR /
RUN uv pip install --no-cache --upgrade 'torch' --index-url https://download.pytorch.org/whl/cpu
RUN uv pip install --no-cache-dir --no-deps accelerate --extra-index-url https://download.pytorch.org/whl/cpu
RUN uv pip install --no-cache-dir "git+https://github.com/huggingface/transformers.git@${REF}#egg=transformers[ja,testing,sentencepiece,spacy,ftfy,rjieba]" unidic unidic-lite
# spacy is not used so not tested. Causes to failures. TODO fix later
RUN uv run python -m unidic download
# fetch test data and hub objects within CircleCI docker images to reduce even more connections
# we don't need a full clone of `transformers` to run `fetch_hub_objects_for_ci.py`
# the data are downloaded to the directory `/test_data` and during CircleCI's CI runtime, we need to move them to the root of `transformers`
RUN mkdir test_data && cd test_data && curl -O https://raw.githubusercontent.com/huggingface/transformers/${REF}/utils/fetch_hub_objects_for_ci.py && python3 fetch_hub_objects_for_ci.py
RUN uv pip uninstall transformers
RUN apt-get clean && rm -rf /var/lib/apt/lists/*
RUN apt remove -y g++ cmake xz-utils libprotobuf-dev protobuf-compiler

19
transformers/docker/examples-torch.dockerfile Normal file
View File

@@ -0,0 +1,19 @@
FROM python:3.10-slim
ENV PYTHONDONTWRITEBYTECODE=1
ARG REF=main
USER root
RUN apt-get update && apt-get install -y --no-install-recommends libsndfile1-dev espeak-ng time git g++ cmake pkg-config openssh-client git-lfs ffmpeg curl
ENV UV_PYTHON=/usr/local/bin/python
RUN pip --no-cache-dir install uv && uv pip install --no-cache-dir -U pip setuptools
RUN uv pip install --no-cache-dir 'torch' 'torchaudio' 'torchvision' 'torchcodec' --index-url https://download.pytorch.org/whl/cpu
RUN uv pip install --no-deps timm accelerate --extra-index-url https://download.pytorch.org/whl/cpu
RUN uv pip install --no-cache-dir librosa "git+https://github.com/huggingface/transformers.git@${REF}#egg=transformers[sklearn,sentencepiece,vision,testing]" seqeval albumentations jiwer
# fetch test data and hub objects within CircleCI docker images to reduce even more connections
# we don't need a full clone of `transformers` to run `fetch_hub_objects_for_ci.py`
# the data are downloaded to the directory `/test_data` and during CircleCI's CI runtime, we need to move them to the root of `transformers`
RUN mkdir test_data && cd test_data && curl -O https://raw.githubusercontent.com/huggingface/transformers/${REF}/utils/fetch_hub_objects_for_ci.py && python3 fetch_hub_objects_for_ci.py
RUN uv pip uninstall transformers
RUN apt-get clean && rm -rf /var/lib/apt/lists/*

24
transformers/docker/exotic-models.dockerfile Normal file
View File

@@ -0,0 +1,24 @@
FROM python:3.10-slim
ENV PYTHONDONTWRITEBYTECODE=1
ARG REF=main
USER root
RUN apt-get update && apt-get install -y libsndfile1-dev espeak-ng time git libgl1 g++ tesseract-ocr git-lfs curl
ENV UV_PYTHON=/usr/local/bin/python
RUN pip --no-cache-dir install uv && uv pip install --no-cache-dir -U pip setuptools
RUN uv pip install --no-cache-dir 'torch' 'torchaudio' 'torchvision' --index-url https://download.pytorch.org/whl/cpu
RUN uv pip install --no-cache-dir --no-deps timm accelerate
RUN uv pip install -U --no-cache-dir pytesseract python-Levenshtein opencv-python nltk
# RUN uv pip install --no-cache-dir natten==0.15.1+torch210cpu -f https://shi-labs.com/natten/wheels
RUN uv pip install --no-cache-dir "git+https://github.com/huggingface/transformers.git@${REF}#egg=transformers[testing, vision]" 'scikit-learn' 'torch-stft' 'nose' 'dataset'
# RUN git clone https://github.com/facebookresearch/detectron2.git
# RUN python3 -m pip install --no-cache-dir -e detectron2
RUN uv pip install 'git+https://github.com/facebookresearch/detectron2.git@92ae9f0b92aba5867824b4f12aa06a22a60a45d3' --no-build-isolation
# fetch test data and hub objects within CircleCI docker images to reduce even more connections
# we don't need a full clone of `transformers` to run `fetch_hub_objects_for_ci.py`
# the data are downloaded to the directory `/test_data` and during CircleCI's CI runtime, we need to move them to the root of `transformers`
RUN mkdir test_data && cd test_data && curl -O https://raw.githubusercontent.com/huggingface/transformers/${REF}/utils/fetch_hub_objects_for_ci.py && python3 fetch_hub_objects_for_ci.py
RUN uv pip uninstall transformers
RUN apt-get clean && rm -rf /var/lib/apt/lists/*

18
transformers/docker/pipeline-torch.dockerfile Normal file
View File

@@ -0,0 +1,18 @@
FROM python:3.10-slim
ENV PYTHONDONTWRITEBYTECODE=1
ARG REF=main
USER root
RUN apt-get update && apt-get install -y --no-install-recommends libsndfile1-dev espeak-ng time git pkg-config openssh-client git ffmpeg curl
ENV UV_PYTHON=/usr/local/bin/python
RUN pip --no-cache-dir install uv && uv pip install --no-cache-dir -U pip setuptools
RUN uv pip install --no-cache-dir 'torch' 'torchaudio' 'torchvision' 'torchcodec' --index-url https://download.pytorch.org/whl/cpu
RUN uv pip install --no-deps timm accelerate --extra-index-url https://download.pytorch.org/whl/cpu
RUN uv pip install --no-cache-dir librosa "git+https://github.com/huggingface/transformers.git@${REF}#egg=transformers[sklearn,sentencepiece,vision,testing]"
# fetch test data and hub objects within CircleCI docker images to reduce even more connections
# we don't need a full clone of `transformers` to run `fetch_hub_objects_for_ci.py`
# the data are downloaded to the directory `/test_data` and during CircleCI's CI runtime, we need to move them to the root of `transformers`
RUN mkdir test_data && cd test_data && curl -O https://raw.githubusercontent.com/huggingface/transformers/${REF}/utils/fetch_hub_objects_for_ci.py && python3 fetch_hub_objects_for_ci.py
RUN uv pip uninstall transformers

9
transformers/docker/quality.dockerfile Normal file
View File

@@ -0,0 +1,9 @@
FROM python:3.10-slim
ENV PYTHONDONTWRITEBYTECODE=1
ARG REF=main
USER root
RUN apt-get update && apt-get install -y time git
ENV UV_PYTHON=/usr/local/bin/python
RUN pip install uv
RUN uv pip install --no-cache-dir -U pip setuptools GitPython "git+https://github.com/huggingface/transformers.git@${REF}#egg=transformers[ruff]" urllib3
RUN apt-get install -y jq curl && apt-get clean && rm -rf /var/lib/apt/lists/*

17
transformers/docker/torch-light.dockerfile Normal file
View File

@@ -0,0 +1,17 @@
FROM python:3.10-slim
ENV PYTHONDONTWRITEBYTECODE=1
ARG REF=main
USER root
RUN apt-get update && apt-get install -y --no-install-recommends libsndfile1-dev espeak-ng time git g++ cmake pkg-config openssh-client git-lfs ffmpeg curl
ENV UV_PYTHON=/usr/local/bin/python
RUN pip --no-cache-dir install uv && uv pip install --no-cache-dir -U pip setuptools
RUN uv pip install --no-cache-dir 'torch' 'torchaudio' 'torchvision' 'torchcodec' --index-url https://download.pytorch.org/whl/cpu
RUN uv pip install --no-deps timm accelerate --extra-index-url https://download.pytorch.org/whl/cpu
RUN uv pip install --no-cache-dir librosa "git+https://github.com/huggingface/transformers.git@${REF}#egg=transformers[sklearn,sentencepiece,vision,testing,tiktoken,num2words,video]"
# fetch test data and hub objects within CircleCI docker images to reduce even more connections
# we don't need a full clone of `transformers` to run `fetch_hub_objects_for_ci.py`
# the data are downloaded to the directory `/test_data` and during CircleCI's CI runtime, we need to move them to the root of `transformers`
RUN mkdir test_data && cd test_data && curl -O https://raw.githubusercontent.com/huggingface/transformers/${REF}/utils/fetch_hub_objects_for_ci.py && python3 fetch_hub_objects_for_ci.py
RUN uv pip uninstall transformers

82
transformers/docker/transformers-all-latest-gpu/Dockerfile Normal file
View File

@@ -0,0 +1,82 @@
FROM nvidia/cuda:12.6.0-cudnn-devel-ubuntu22.04
LABEL maintainer="Hugging Face"
ARG DEBIAN_FRONTEND=noninteractive
# Use login shell to read variables from `~/.profile` (to pass dynamic created variables between RUN commands)
SHELL ["sh", "-lc"]
# The following `ARG` are mainly used to specify the versions explicitly & directly in this docker file, and not meant
# to be used as arguments for docker build (so far).
ARG PYTORCH='2.8.0'
# Example: `cu102`, `cu113`, etc.
ARG CUDA='cu126'
# Disable kernel mapping for now until all tests pass
ENV DISABLE_KERNEL_MAPPING=1
RUN apt update
RUN apt install -y git libsndfile1-dev tesseract-ocr espeak-ng python3 python3-pip ffmpeg git-lfs
RUN git lfs install
RUN python3 -m pip install --no-cache-dir --upgrade pip
ARG REF=main
RUN git clone https://github.com/huggingface/transformers && cd transformers && git checkout $REF
# 1. Put several commands in a single `RUN` to avoid image/layer exporting issue. Could be revised in the future.
# 2. Regarding `torch` part, We might need to specify proper versions for `torchvision` and `torchaudio`.
# Currently, let's not bother to specify their versions explicitly (so installed with their latest release versions).
RUN python3 -m pip install --no-cache-dir -e ./transformers[dev,onnxruntime] && [ ${#PYTORCH} -gt 0 -a "$PYTORCH" != "pre" ] && VERSION='torch=='$PYTORCH'.*' || VERSION='torch'; echo "export VERSION='$VERSION'" >> ~/.profile && echo torch=$VERSION && [ "$PYTORCH" != "pre" ] && python3 -m pip install --no-cache-dir -U $VERSION torchvision torchaudio torchcodec --extra-index-url https://download.pytorch.org/whl/$CUDA || python3 -m pip install --no-cache-dir -U --pre torch torchvision torchaudio torchcodec --extra-index-url https://download.pytorch.org/whl/nightly/$CUDA
RUN python3 -m pip install --no-cache-dir -U timm
RUN [ "$PYTORCH" != "pre" ] && python3 -m pip install --no-cache-dir git+https://github.com/facebookresearch/detectron2.git || echo "Don't install detectron2 with nightly torch"
RUN python3 -m pip install --no-cache-dir pytesseract
RUN python3 -m pip install -U "itsdangerous<2.1.0"
RUN python3 -m pip install --no-cache-dir git+https://github.com/huggingface/accelerate@main#egg=accelerate
RUN python3 -m pip install --no-cache-dir git+https://github.com/huggingface/peft@main#egg=peft
# For bettertransformer
RUN python3 -m pip install --no-cache-dir git+https://github.com/huggingface/optimum@main#egg=optimum
# For kernels
RUN python3 -m pip install --no-cache-dir git+https://github.com/huggingface/kernels@main#egg=kernels
# For video model testing
RUN python3 -m pip install --no-cache-dir av
# Some slow tests require bnb
RUN python3 -m pip install --no-cache-dir bitsandbytes
# Some tests require quanto
RUN python3 -m pip install --no-cache-dir quanto
# After using A10 as CI runner, let's run FA2 tests
RUN [ "$PYTORCH" != "pre" ] && python3 -m pip uninstall -y ninja && python3 -m pip install --no-cache-dir ninja && python3 -m pip install flash-attn --no-cache-dir --no-build-isolation || echo "Don't install FA2 with nightly torch"
# TODO (ydshieh): check this again
# `quanto` will install `ninja` which leads to many `CUDA error: an illegal memory access ...` in some model tests
# (`deformable_detr`, `rwkv`, `mra`)
RUN python3 -m pip uninstall -y ninja
# For `nougat` tokenizer
RUN python3 -m pip install --no-cache-dir python-Levenshtein
# For `FastSpeech2ConformerTokenizer` tokenizer
RUN python3 -m pip install --no-cache-dir g2p-en
# For Some bitsandbytes tests
RUN python3 -m pip install --no-cache-dir einops
# For Some tests with `@require_liger_kernel`
RUN python3 -m pip install --no-cache-dir liger-kernel
# `kernels` may give different outputs (within 1e-5 range) even with the same model (weights) and the same inputs
RUN python3 -m pip uninstall -y kernels
# When installing in editable mode, `transformers` is not recognized as a package.
# this line must be added in order for python to be aware of transformers.
RUN cd transformers && python3 setup.py develop

18
transformers/docker/transformers-doc-builder/Dockerfile Normal file
View File

@@ -0,0 +1,18 @@
FROM python:3.10
LABEL maintainer="Hugging Face"
RUN apt update
RUN git clone https://github.com/huggingface/transformers
RUN python3 -m pip install --no-cache-dir --upgrade pip && python3 -m pip install --no-cache-dir git+https://github.com/huggingface/doc-builder ./transformers[dev]
RUN apt-get -y update && apt-get install -y libsndfile1-dev && apt install -y tesseract-ocr
# Torch needs to be installed before deepspeed
RUN python3 -m pip install --no-cache-dir ./transformers[deepspeed]
RUN python3 -m pip install --no-cache-dir torchvision git+https://github.com/facebookresearch/detectron2.git pytesseract
RUN python3 -m pip install -U "itsdangerous<2.1.0"
# Test if the image could successfully build the doc. before publishing the image
RUN doc-builder build transformers transformers/docs/source/en --build_dir doc-build-dev --notebook_dir notebooks/transformers_doc --clean
RUN rm -rf doc-build-dev

31
transformers/docker/transformers-gpu/Dockerfile Normal file
View File

@@ -0,0 +1,31 @@
FROM nvidia/cuda:10.2-cudnn7-devel-ubuntu18.04
LABEL maintainer="Hugging Face"
LABEL repository="transformers"
RUN apt update && \
apt install -y bash \
build-essential \
git \
curl \
ca-certificates \
python3 \
python3-pip && \
rm -rf /var/lib/apt/lists
RUN python3 -m pip install --no-cache-dir --upgrade pip && \
python3 -m pip install --no-cache-dir \
jupyter \
torch
RUN python3 -m pip install --no-cache-dir git+https://github.com/huggingface/kernels@main#egg=kernels
RUN git clone https://github.com/NVIDIA/apex
RUN cd apex && \
python3 setup.py install && \
pip install -v --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" ./
WORKDIR /workspace
COPY . transformers/
RUN cd transformers/ && \
python3 -m pip install --no-cache-dir .
CMD ["/bin/bash"]

71
transformers/docker/transformers-intel-cpu/Dockerfile Normal file
View File

@@ -0,0 +1,71 @@
FROM intel/deep-learning-essentials:2025.1.3-0-devel-ubuntu24.04 AS base
LABEL maintainer="Hugging Face"
SHELL ["/bin/bash", "-c"]
ARG PYTHON_VERSION=3.12
ENV DEBIAN_FRONTEND=noninteractive
RUN apt-get update && \
apt-get install -y software-properties-common && \
add-apt-repository -y ppa:deadsnakes/ppa && \
apt-get update
RUN apt-get update && \
apt-get -y install \
apt-utils \
build-essential \
ca-certificates \
clinfo \
curl \
git \
git-lfs \
vim \
numactl \
gnupg2 \
gpg-agent \
python3-dev \
python3-opencv \
unzip \
ffmpeg \
tesseract-ocr \
espeak-ng \
wget \
ncurses-term \
google-perftools \
libjemalloc-dev \
&& apt-get clean \
&& rm -rf /var/lib/apt/lists/*
# Use virtual env because Ubuntu:24 does not allowed pip on original python
RUN curl -LsSf https://astral.sh/uv/install.sh | sh
ENV PATH="/root/.local/bin:$PATH"
ENV VIRTUAL_ENV="/opt/venv"
ENV UV_PYTHON_INSTALL_DIR=/opt/uv/python
RUN uv venv --python ${PYTHON_VERSION} --seed ${VIRTUAL_ENV}
ENV PATH="$VIRTUAL_ENV/bin:$PATH"
RUN pip install --upgrade pip wheel
RUN pip install torch torchvision torchaudio torchcodec --index-url https://download.pytorch.org/whl/cpu --no-cache-dir
RUN pip install av pyctcdecode pytesseract decord galore-torch fire scipy scikit-learn sentencepiece sentence_transformers sacremoses nltk rouge_score librosa soundfile mpi4py pytorch_msssim
RUN pip install onnx optimum onnxruntime
RUN pip install autoawq
RUN pip install gptqmodel --no-build-isolation
RUN pip install -U datasets timm transformers accelerate peft diffusers opencv-python kenlm evaluate
RUN pip install -U intel-openmp
# install bitsandbytes
RUN git clone https://github.com/bitsandbytes-foundation/bitsandbytes.git && cd bitsandbytes/ && \
cmake -DCOMPUTE_BACKEND=cpu -S . && make && pip install . && cd ../
# CPU don't need triton
RUN pip uninstall triton -y
ENV LD_PRELOAD=${LD_PRELOAD}:/opt/venv/lib/libiomp5.so:/usr/lib/x86_64-linux-gnu/libtcmalloc.so.4
ENV KMP_AFFINITY=granularity=fine,compact,1,0
RUN touch /entrypoint.sh
RUN chmod +x /entrypoint.sh
RUN echo "#!/bin/bash" >> /entrypoint.sh
RUN echo "/bin/bash" >> /entrypoint.sh
ENTRYPOINT ["/entrypoint.sh"]

37
transformers/docker/transformers-pytorch-amd-gpu/Dockerfile Normal file
View File

@@ -0,0 +1,37 @@
FROM rocm/pytorch:rocm6.4.1_ubuntu24.04_py3.12_pytorch_release_2.7.1
LABEL maintainer="Hugging Face"
ARG DEBIAN_FRONTEND=noninteractive
RUN apt update && \
apt install -y --no-install-recommends git libsndfile1-dev tesseract-ocr espeak-ng python3 python3-dev python3-pip python3-dev ffmpeg git-lfs && \
apt clean && \
rm -rf /var/lib/apt/lists/*
RUN git lfs install
RUN python3 -m pip install --no-cache-dir --upgrade pip numpy
RUN python3 -m pip install --no-cache-dir --upgrade importlib-metadata setuptools ninja git+https://github.com/facebookresearch/detectron2.git pytesseract "itsdangerous<2.1.0"
ARG REF=main
WORKDIR /
# Invalidate docker cache from here if new commit is available.
ADD https://api.github.com/repos/huggingface/transformers/git/refs/heads/main version.json
RUN git clone https://github.com/huggingface/transformers && cd transformers && git checkout $REF
# Install transformers
RUN python3 -m pip install --no-cache-dir -e ./transformers[dev-torch,testing,video,audio]
# When installing in editable mode, `transformers` is not recognized as a package.
# this line must be added in order for python to be aware of transformers.
RUN cd transformers && python3 setup.py develop
# Remove nvml and nvidia-ml-py as it is not compatible with ROCm. apex is not tested on NVIDIA either.
RUN python3 -m pip uninstall py3nvml pynvml nvidia-ml-py apex -y
# `kernels` may causes many failing tests
RUN python3 -m pip uninstall -y kernels
# On ROCm, torchcodec is required to decode audio files and 0.4 or 0.6 fails
RUN python3 -m pip install --no-cache-dir "torchcodec==0.5"

53
transformers/docker/transformers-pytorch-deepspeed-amd-gpu/Dockerfile Normal file
View File

@@ -0,0 +1,53 @@
FROM rocm/dev-ubuntu-22.04:6.2.4
LABEL maintainer="Hugging Face"
ARG DEBIAN_FRONTEND=noninteractive
ARG PYTORCH='2.6.0'
ARG TORCH_VISION='0.21.0'
ARG TORCH_AUDIO='2.6.0'
ARG ROCM='6.2.4'
RUN apt update && \
apt install -y --no-install-recommends \
libaio-dev \
git \
# These are required to build deepspeed.
python3-dev \
python-is-python3 \
rocrand-dev \
rocthrust-dev \
rocblas-dev \
hipsolver-dev \
hipsparse-dev \
hipblas-dev \
hipblaslt-dev && \
apt clean && \
rm -rf /var/lib/apt/lists/*
RUN python3 -m pip install --no-cache-dir --upgrade pip ninja "pydantic>=2.0.0"
RUN python3 -m pip uninstall -y apex torch torchvision torchaudio
RUN python3 -m pip install torch==$PYTORCH torchvision==$TORCH_VISION torchaudio==$TORCH_AUDIO --index-url https://download.pytorch.org/whl/rocm$ROCM --no-cache-dir
# Pre-build DeepSpeed, so it's be ready for testing (to avoid timeout)
RUN DS_BUILD_CPU_ADAM=1 DS_BUILD_FUSED_ADAM=1 python3 -m pip install deepspeed --global-option="build_ext" --global-option="-j8" --no-cache-dir -v --disable-pip-version-check 2>&1
ARG REF=main
WORKDIR /
# Invalidate docker cache from here if new commit is available.
ADD https://api.github.com/repos/huggingface/transformers/git/refs/heads/main version.json
RUN git clone https://github.com/huggingface/transformers && cd transformers && git checkout $REF
RUN python3 -m pip install --no-cache-dir ./transformers[accelerate,testing,sentencepiece,sklearn]
# When installing in editable mode, `transformers` is not recognized as a package.
# this line must be added in order for python to be aware of transformers.
RUN cd transformers && python3 setup.py develop
RUN python3 -c "from deepspeed.launcher.runner import main"
# Remove nvml as it is not compatible with ROCm
RUN python3 -m pip uninstall py3nvml pynvml nvidia-ml-py apex -y
# `kernels` may causes many failing tests
RUN python3 -m pip uninstall -y kernels

57
transformers/docker/transformers-pytorch-deepspeed-latest-gpu/Dockerfile Normal file
View File

@@ -0,0 +1,57 @@
# https://docs.nvidia.com/deeplearning/frameworks/pytorch-release-notes/rel-24-08.html
FROM nvcr.io/nvidia/pytorch:24.08-py3
LABEL maintainer="Hugging Face"
ARG DEBIAN_FRONTEND=noninteractive
ARG PYTORCH='2.8.0'
# Example: `cu102`, `cu113`, etc.
ARG CUDA='cu126'
RUN apt -y update
RUN apt install -y libaio-dev
RUN python3 -m pip install --no-cache-dir --upgrade pip
ARG REF=main
RUN git clone https://github.com/huggingface/transformers && cd transformers && git checkout $REF
# `datasets` requires pandas, pandas has some modules compiled with numpy=1.x causing errors
RUN python3 -m pip install --no-cache-dir './transformers[deepspeed-testing]' 'pandas<2' 'numpy<2'
# Install latest release PyTorch
# (PyTorch must be installed before pre-compiling any DeepSpeed c++/cuda ops.)
# (https://www.deepspeed.ai/tutorials/advanced-install/#pre-install-deepspeed-ops)
RUN python3 -m pip uninstall -y torch torchvision torchaudio && python3 -m pip install --no-cache-dir -U torch==$PYTORCH torchvision torchaudio torchcodec --extra-index-url https://download.pytorch.org/whl/$CUDA
RUN python3 -m pip install --no-cache-dir git+https://github.com/huggingface/accelerate@main#egg=accelerate
# Uninstall `transformer-engine` shipped with the base image
RUN python3 -m pip uninstall -y transformer-engine
# Uninstall `torch-tensorrt` shipped with the base image
RUN python3 -m pip uninstall -y torch-tensorrt
# recompile apex
RUN python3 -m pip uninstall -y apex
# RUN git clone https://github.com/NVIDIA/apex
# `MAX_JOBS=1` disables parallel building to avoid cpu memory OOM when building image on GitHub Action (standard) runners
# TODO: check if there is alternative way to install latest apex
# RUN cd apex && MAX_JOBS=1 python3 -m pip install --global-option="--cpp_ext" --global-option="--cuda_ext" --no-cache -v --disable-pip-version-check .
# Pre-build **latest** DeepSpeed, so it would be ready for testing (otherwise, the 1st deepspeed test will timeout)
RUN python3 -m pip uninstall -y deepspeed
# This has to be run (again) inside the GPU VMs running the tests.
# The installation works here, but some tests fail, if we don't pre-build deepspeed again in the VMs running the tests.
# TODO: Find out why test fail.
RUN DS_BUILD_CPU_ADAM=1 DS_BUILD_FUSED_ADAM=1 python3 -m pip install deepspeed --global-option="build_ext" --global-option="-j8" --no-cache -v --disable-pip-version-check 2>&1
# `kernels` may give different outputs (within 1e-5 range) even with the same model (weights) and the same inputs
RUN python3 -m pip uninstall -y kernels
# When installing in editable mode, `transformers` is not recognized as a package.
# this line must be added in order for python to be aware of transformers.
RUN cd transformers && python3 setup.py develop
# The base image ships with `pydantic==1.8.2` which is not working - i.e. the next command fails
RUN python3 -m pip install -U --no-cache-dir "pydantic>=2.0.0"
RUN python3 -c "from deepspeed.launcher.runner import main"

68
transformers/docker/transformers-pytorch-deepspeed-nightly-gpu/Dockerfile Normal file
View File

@@ -0,0 +1,68 @@
# https://docs.nvidia.com/deeplearning/frameworks/pytorch-release-notes/rel-23-11.html#rel-23-11
FROM nvcr.io/nvidia/pytorch:24.08-py3
LABEL maintainer="Hugging Face"
ARG DEBIAN_FRONTEND=noninteractive
# Example: `cu102`, `cu113`, etc.
ARG CUDA='cu126'
RUN apt -y update
RUN apt install -y libaio-dev
RUN python3 -m pip install --no-cache-dir --upgrade pip
ARG REF=main
RUN git clone https://github.com/huggingface/transformers && cd transformers && git checkout $REF
RUN python3 -m pip uninstall -y torch torchvision torchaudio
# Install **nightly** release PyTorch (flag `--pre`)
# (PyTorch must be installed before pre-compiling any DeepSpeed c++/cuda ops.)
# (https://www.deepspeed.ai/tutorials/advanced-install/#pre-install-deepspeed-ops)
RUN python3 -m pip install --no-cache-dir -U --pre torch torchvision torchaudio torchcodec --extra-index-url https://download.pytorch.org/whl/nightly/$CUDA
# `datasets` requires pandas, pandas has some modules compiled with numpy=1.x causing errors
RUN python3 -m pip install --no-cache-dir './transformers[deepspeed-testing]' 'pandas<2' 'numpy<2'
RUN python3 -m pip install --no-cache-dir git+https://github.com/huggingface/accelerate@main#egg=accelerate
# Uninstall `transformer-engine` shipped with the base image
RUN python3 -m pip uninstall -y transformer-engine
# Uninstall `torch-tensorrt` and `apex` shipped with the base image
RUN python3 -m pip uninstall -y torch-tensorrt apex
# Pre-build **nightly** release of DeepSpeed, so it would be ready for testing (otherwise, the 1st deepspeed test will timeout)
RUN python3 -m pip uninstall -y deepspeed
# This has to be run inside the GPU VMs running the tests. (So far, it fails here due to GPU checks during compilation.)
# Issue: https://github.com/deepspeedai/DeepSpeed/issues/2010
# RUN git clone https://github.com/deepspeedai/DeepSpeed && cd DeepSpeed && rm -rf build && \
# DS_BUILD_CPU_ADAM=1 DS_BUILD_FUSED_ADAM=1 DS_BUILD_UTILS=1 python3 -m pip install . --global-option="build_ext" --global-option="-j8" --no-cache -v --disable-pip-version-check 2>&1
## For `torchdynamo` tests
## (see https://github.com/huggingface/transformers/pull/17765)
#RUN git clone https://github.com/pytorch/functorch
#RUN python3 -m pip install --no-cache-dir ./functorch[aot]
#RUN cd functorch && python3 setup.py develop
#
#RUN git clone https://github.com/pytorch/torchdynamo
#RUN python3 -m pip install -r ./torchdynamo/requirements.txt
#RUN cd torchdynamo && python3 setup.py develop
#
## install TensorRT
#RUN python3 -m pip install --no-cache-dir -U nvidia-pyindex
#RUN python3 -m pip install --no-cache-dir -U nvidia-tensorrt==8.2.4.2
#
## install torch_tensorrt (fx path)
#RUN git clone https://github.com/pytorch/TensorRT.git
#RUN cd TensorRT/py && python3 setup.py install --fx-only
# `kernels` may give different outputs (within 1e-5 range) even with the same model (weights) and the same inputs
RUN python3 -m pip uninstall -y kernels
# When installing in editable mode, `transformers` is not recognized as a package.
# this line must be added in order for python to be aware of transformers.
RUN cd transformers && python3 setup.py develop
# Disable for now as deepspeed is not installed above. To be enabled once the issue is fixed.
# RUN python3 -c "from deepspeed.launcher.runner import main"

36
transformers/docker/transformers-pytorch-gpu/Dockerfile Normal file
View File

@@ -0,0 +1,36 @@
FROM nvidia/cuda:12.6.0-cudnn-devel-ubuntu22.04
LABEL maintainer="Hugging Face"
ARG DEBIAN_FRONTEND=noninteractive
RUN apt update
RUN apt install -y git libsndfile1-dev tesseract-ocr espeak-ng python3 python3-pip ffmpeg
RUN python3 -m pip install --no-cache-dir --upgrade pip
ARG REF=main
RUN git clone https://github.com/huggingface/transformers && cd transformers && git checkout $REF
# If set to nothing, will install the latest version
ARG PYTORCH='2.8.0'
ARG TORCH_VISION=''
ARG TORCH_AUDIO=''
# Example: `cu102`, `cu113`, etc.
ARG CUDA='cu126'
RUN python3 -m pip install --no-cache-dir -e ./transformers[dev-torch,testing,video]
# Install torch stuff after ./transformers[dev-torch,testing,video], otherwise torch may be resolved to a previous
# version.
RUN [ ${#PYTORCH} -gt 0 ] && VERSION='torch=='$PYTORCH'.*' || VERSION='torch'; python3 -m pip install --no-cache-dir -U $VERSION --extra-index-url https://download.pytorch.org/whl/$CUDA
RUN [ ${#TORCH_VISION} -gt 0 ] && VERSION='torchvision=='TORCH_VISION'.*' || VERSION='torchvision'; python3 -m pip install --no-cache-dir -U $VERSION --extra-index-url https://download.pytorch.org/whl/$CUDA
RUN [ ${#TORCH_AUDIO} -gt 0 ] && VERSION='torchaudio=='TORCH_AUDIO'.*' || VERSION='torchaudio'; python3 -m pip install --no-cache-dir -U $VERSION --extra-index-url https://download.pytorch.org/whl/$CUDA
RUN python3 -m pip install --no-cache-dir git+https://github.com/facebookresearch/detectron2.git pytesseract
RUN python3 -m pip install -U "itsdangerous<2.1.0"
# `kernels` may give different outputs (within 1e-5 range) even with the same model (weights) and the same inputs
RUN python3 -m pip uninstall -y kernels
# When installing in editable mode, `transformers` is not recognized as a package.
# this line must be added in order for python to be aware of transformers.
RUN cd transformers && python3 setup.py develop

65
transformers/docker/transformers-pytorch-tpu/Dockerfile Normal file
View File

@@ -0,0 +1,65 @@
FROM google/cloud-sdk:slim
# Build args.
ARG GITHUB_REF=refs/heads/main
# TODO: This Dockerfile installs pytorch/xla 3.6 wheels. There are also 3.7
# wheels available; see below.
ENV PYTHON_VERSION=3.6
RUN apt-get update && apt-get install -y --no-install-recommends \
build-essential \
cmake \
git \
curl \
ca-certificates
# Install conda and python.
# NOTE new Conda does not forward the exit status... https://github.com/conda/conda/issues/8385
RUN curl -o ~/miniconda.sh https://repo.anaconda.com/miniconda/Miniconda3-4.7.12-Linux-x86_64.sh && \
chmod +x ~/miniconda.sh && \
~/miniconda.sh -b && \
rm ~/miniconda.sh
ENV PATH=/root/miniconda3/bin:$PATH
RUN conda create -y --name container python=$PYTHON_VERSION
# Run the rest of commands within the new conda env.
# Use absolute path to appease Codefactor.
SHELL ["/root/miniconda3/bin/conda", "run", "-n", "container", "/bin/bash", "-c"]
RUN conda install -y python=$PYTHON_VERSION mkl
RUN pip uninstall -y torch && \
# Python 3.7 wheels are available. Replace cp36-cp36m with cp37-cp37m
gsutil cp 'gs://tpu-pytorch/wheels/torch-nightly-cp${PYTHON_VERSION/./}-cp${PYTHON_VERSION/./}m-linux_x86_64.whl' . && \
gsutil cp 'gs://tpu-pytorch/wheels/torch_xla-nightly-cp${PYTHON_VERSION/./}-cp${PYTHON_VERSION/./}m-linux_x86_64.whl' . && \
gsutil cp 'gs://tpu-pytorch/wheels/torchvision-nightly-cp${PYTHON_VERSION/./}-cp${PYTHON_VERSION/./}m-linux_x86_64.whl' . && \
pip install 'torch-nightly-cp${PYTHON_VERSION/./}-cp${PYTHON_VERSION/./}m-linux_x86_64.whl' && \
pip install 'torch_xla-nightly-cp${PYTHON_VERSION/./}-cp${PYTHON_VERSION/./}m-linux_x86_64.whl' && \
pip install 'torchvision-nightly-cp${PYTHON_VERSION/./}-cp${PYTHON_VERSION/./}m-linux_x86_64.whl' && \
rm 'torch-nightly-cp${PYTHON_VERSION/./}-cp${PYTHON_VERSION/./}m-linux_x86_64.whl' && \
rm 'torch_xla-nightly-cp${PYTHON_VERSION/./}-cp${PYTHON_VERSION/./}m-linux_x86_64.whl' && \
rm 'torchvision-nightly-cp${PYTHON_VERSION/./}-cp${PYTHON_VERSION/./}m-linux_x86_64.whl' && \
apt-get install -y libomp5
ENV LD_LIBRARY_PATH=root/miniconda3/envs/container/lib
# Install huggingface/transformers at the current PR, plus dependencies.
RUN git clone https://github.com/huggingface/transformers.git && \
cd transformers && \
git fetch origin $GITHUB_REF:CI && \
git checkout CI && \
cd .. && \
pip install ./transformers && \
pip install -r ./transformers/examples/pytorch/_test_requirements.txt && \
pip install pytest
RUN python -c "import torch_xla; print(torch_xla.__version__)"
RUN python -c "import transformers as trf; print(trf.__version__)"
RUN conda init bash
COPY docker-entrypoint.sh /usr/local/bin/
RUN chmod +x /usr/local/bin/docker-entrypoint.sh
ENTRYPOINT ["/usr/local/bin/docker-entrypoint.sh"]
CMD ["bash"]

38
transformers/docker/transformers-pytorch-tpu/bert-base-cased.jsonnet Normal file
View File

@@ -0,0 +1,38 @@
local base = import 'templates/base.libsonnet';
local tpus = import 'templates/tpus.libsonnet';
local utils = import "templates/utils.libsonnet";
local volumes = import "templates/volumes.libsonnet";
local bertBaseCased = base.BaseTest {
frameworkPrefix: "hf",
modelName: "bert-base-cased",
mode: "example",
configMaps: [],
timeout: 3600, # 1 hour, in seconds
image: std.extVar('image'),
imageTag: std.extVar('image-tag'),
tpuSettings+: {
softwareVersion: "pytorch-nightly",
},
accelerator: tpus.v3_8,
volumeMap+: {
datasets: volumes.PersistentVolumeSpec {
name: "huggingface-cluster-disk",
mountPath: "/datasets",
},
},
command: utils.scriptCommand(
|||
python -m pytest -s transformers/examples/pytorch/test_xla_examples.py -v
test_exit_code=$?
echo "\nFinished running commands.\n"
test $test_exit_code -eq 0
|||
),
};
bertBaseCased.oneshotJob

32
transformers/docker/transformers-pytorch-tpu/dataset.yaml Normal file
View File

@@ -0,0 +1,32 @@
apiVersion: v1
kind: PersistentVolume
metadata:
name: huggingface-cluster-disk
spec:
storageClassName: ""
capacity:
storage: 500Gi
accessModes:
- ReadOnlyMany
claimRef:
namespace: default
name: huggingface-cluster-disk-claim
gcePersistentDisk:
pdName: huggingface-cluster-disk
fsType: ext4
readOnly: true
---
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: huggingface-cluster-disk-claim
spec:
# Specify "" as the storageClassName so it matches the PersistentVolume's StorageClass.
# A nil storageClassName value uses the default StorageClass. For details, see
# https://kubernetes.io/docs/concepts/storage/persistent-volumes/#class-1
storageClassName: ""
accessModes:
- ReadOnlyMany
resources:
requests:
storage: 1Ki

8
transformers/docker/transformers-pytorch-tpu/docker-entrypoint.sh Normal file
View File

@@ -0,0 +1,8 @@
#!/bin/bash
source ~/.bashrc
echo "running docker-entrypoint.sh"
conda activate container
echo $KUBE_GOOGLE_CLOUD_TPU_ENDPOINTS
echo "printed TPU info"
export XRT_TPU_CONFIG="tpu_worker;0;${KUBE_GOOGLE_CLOUD_TPU_ENDPOINTS:7}"
exec "$@"#!/bin/bash

93
transformers/docker/transformers-pytorch-xpu/Dockerfile Normal file
View File

@@ -0,0 +1,93 @@
FROM intel/deep-learning-essentials:2025.1.3-0-devel-ubuntu22.04 AS base
LABEL maintainer="Hugging Face"
SHELL ["/bin/bash", "-c"]
ARG PYTHON_VER=3.11
ENV TORCH_DEVICE_BACKEND_AUTOLOAD=0
ENV DEBIAN_FRONTEND=noninteractive
RUN apt-get remove -y python3.10 && apt-get autoremove -y
RUN apt-get update && \
apt-get install -y software-properties-common && \
add-apt-repository -y ppa:deadsnakes/ppa && \
apt-get update && \
apt-get install -y python$PYTHON_VER python$PYTHON_VER-dev python3-pip && \
ln -sf /usr/bin/python$PYTHON_VER /usr/bin/python3 && \
ln -sf /usr/bin/python3 /usr/bin/python && \
apt-get clean && \
rm -rf /var/lib/apt/lists/*
RUN apt-get update && \
apt-get -y install \
apt-utils \
build-essential \
ca-certificates \
clinfo \
curl \
git \
git-lfs \
vim \
numactl \
gnupg2 \
gpg-agent \
zlib1g-dev \
rsync \
sudo \
libnl-genl-3-200 \
xpu-smi \
unzip \
ffmpeg \
tesseract-ocr \
espeak-ng \
wget \
ncurses-term && \
apt-get clean && \
rm -rf /var/lib/apt/lists/*
RUN apt-get update && \
apt-get install -y \
linux-headers-$(uname -r) \
linux-modules-extra-$(uname -r) \
flex bison \
intel-fw-gpu intel-i915-dkms xpu-smi \
intel-opencl-icd libze-intel-gpu1 libze1 \
intel-media-va-driver-non-free libmfx-gen1 libvpl2 \
libegl-mesa0 libegl1-mesa libegl1-mesa-dev libgbm1 libgl1-mesa-dev libgl1-mesa-dri \
libglapi-mesa libglx-mesa0 libigdgmm12 libxatracker2 mesa-va-drivers \
mesa-vdpau-drivers mesa-vulkan-drivers va-driver-all vainfo hwinfo clinfo intel-ocloc \
libigc-dev intel-igc-cm libigdfcl-dev libigfxcmrt-dev libze-dev && \
apt-get clean && \
rm -rf /var/lib/apt/lists/*
RUN pip install --upgrade pip
RUN pip install triton==3.3.0
RUN pip install torch==2.7.0 torchvision==0.22.0 torchaudio==2.7.0 --index-url https://download.pytorch.org/whl/xpu --no-cache-dir
RUN pip install evaluate torchdata pyctcdecode pytesseract decord galore-torch fire scipy scikit-learn sentencepiece sacremoses nltk rouge_score librosa soundfile g2p_en mpi4py requests_mock
RUN pip install pretty_midi essentia resampy Levenshtein av sacrebleu phonemizer invisible_watermark schedulefree
RUN pip install gguf hqq compressed_tensors gptqmodel mergekit autoawq deepspeed torchao onnx
RUN pip install hf_transfer huggingface-hub hf-doc-builder datasets optimum-quanto timm transformers accelerate optimum peft
RUN pip install git+https://github.com/linkedin/Liger-Kernel.git --extra-index-url https://download.pytorch.org/whl/test/xpu
# install bitsandbytes
RUN pip install git+https://github.com/bitsandbytes-foundation/bitsandbytes.git
ENV OCL_ICD_VENDORS=/etc/OpenCL/vendors
ENV FI_PROVIDER_PATH=${I_MPI_ROOT}/lib/libfabric/prov:/usr/lib/x86_64-linux-gnu/libfabric
ENV CCL_ROOT=/usr/local
ENV CCL_ATL_TRANSPORT=ofi
ENV I_MPI_ROOT=/usr/local
ENV CLASSPATH=${I_MPI_ROOT}/lib/mpi.jar
ENV PATH=${I_MPI_ROOT}/bin/libfabric:${PATH}
ENV LD_LIBRARY_PATH=${I_MPI_ROOT}/lib/libfabric:${LD_LIBRARY_PATH}
RUN touch /entrypoint.sh
RUN chmod +x /entrypoint.sh
RUN echo "#!/bin/bash" >> /entrypoint.sh
RUN echo "source /opt/intel/oneapi/setvars.sh --force && /bin/bash" >> /entrypoint.sh
ENTRYPOINT ["/entrypoint.sh"]

109
transformers/docker/transformers-quantization-latest-gpu/Dockerfile Executable file
View File

@@ -0,0 +1,109 @@
FROM nvidia/cuda:12.6.0-cudnn-devel-ubuntu22.04
LABEL maintainer="Hugging Face"
ARG DEBIAN_FRONTEND=noninteractive
# Use login shell to read variables from `~/.profile` (to pass dynamic created variables between RUN commands)
SHELL ["sh", "-lc"]
# The following `ARG` are mainly used to specify the versions explicitly & directly in this docker file, and not meant
# to be used as arguments for docker build (so far).
ARG PYTORCH='2.8.0'
# Example: `cu102`, `cu113`, etc.
ARG CUDA='cu126'
# Disable kernel mapping for quantization tests
ENV DISABLE_KERNEL_MAPPING=1
RUN apt update
RUN apt install -y git libsndfile1-dev tesseract-ocr espeak-ng python3 python3-pip ffmpeg
RUN python3 -m pip install --no-cache-dir --upgrade pip
ARG REF=main
RUN git clone https://github.com/huggingface/transformers && cd transformers && git checkout $REF
RUN [ ${#PYTORCH} -gt 0 ] && VERSION='torch=='$PYTORCH'.*' || VERSION='torch'; echo "export VERSION='$VERSION'" >> ~/.profile
RUN echo torch=$VERSION
# `torchvision` and `torchaudio` should be installed along with `torch`, especially for nightly build.
# Currently, let's just use their latest releases (when `torch` is installed with a release version)
RUN python3 -m pip install --no-cache-dir -U $VERSION torchvision torchaudio torchcodec --extra-index-url https://download.pytorch.org/whl/$CUDA
RUN python3 -m pip install --no-cache-dir git+https://github.com/huggingface/accelerate@main#egg=accelerate
# needed in bnb and awq
RUN python3 -m pip install --no-cache-dir einops
# Add bitsandbytes for mixed int8 testing
RUN python3 -m pip install --no-cache-dir bitsandbytes
# Add gptqmodel for gtpq quantization testing, installed from source for pytorch==2.6.0 compatibility
RUN python3 -m pip install lm_eval
RUN git clone https://github.com/ModelCloud/GPTQModel.git && cd GPTQModel && pip install -v . --no-build-isolation
# Add optimum for gptq quantization testing
RUN python3 -m pip install --no-cache-dir git+https://github.com/huggingface/optimum@main#egg=optimum
# Add PEFT
RUN python3 -m pip install --no-cache-dir git+https://github.com/huggingface/peft@main#egg=peft
# Add hqq for quantization testing
RUN python3 -m pip install --no-cache-dir hqq
# For GGUF tests
RUN python3 -m pip install --no-cache-dir gguf
# Add autoawq for quantization testing
RUN python3 -m pip install --no-cache-dir autoawq[kernels]
# Add quanto for quantization testing
RUN python3 -m pip install --no-cache-dir optimum-quanto
# Add compressed-tensors for quantization testing
RUN python3 -m pip install --no-cache-dir compressed-tensors
# Add AMD Quark for quantization testing
RUN python3 -m pip install --no-cache-dir amd-quark
# Add AutoRound for quantization testing
RUN python3 -m pip install --no-cache-dir auto-round
# Add torchao for quantization testing
RUN python3 -m pip install --no-cache-dir torchao
# Add transformers in editable mode
RUN python3 -m pip install --no-cache-dir -e ./transformers[dev-torch]
# `kernels` may give different outputs (within 1e-5 range) even with the same model (weights) and the same inputs
RUN python3 -m pip uninstall -y kernels
# Uninstall flash-attn installed by autoawq, it causes issues here : https://github.com/huggingface/transformers/actions/runs/15915442841/job/44892146131
RUN python3 -m pip uninstall -y flash-attn
# When installing in editable mode, `transformers` is not recognized as a package.
# this line must be added in order for python to be aware of transformers.
RUN cd transformers && python3 setup.py develop
# Low usage or incompatible lib, will enable later on
# # Add aqlm for quantization testing
# RUN python3 -m pip install --no-cache-dir aqlm[gpu]==1.0.2
# # Add vptq for quantization testing
# RUN pip install vptq
# Add spqr for quantization testing
# Commented for now as No matching distribution found we need to reach out to the authors
# RUN python3 -m pip install --no-cache-dir spqr_quant[gpu]
# # Add eetq for quantization testing
# RUN git clone https://github.com/NetEase-FuXi/EETQ.git && cd EETQ/ && git submodule update --init --recursive && pip install .
# # Add flute-kernel and fast_hadamard_transform for quantization testing
# # Commented for now as they cause issues with the build
# # TODO: create a new workflow to test them
# RUN python3 -m pip install --no-cache-dir flute-kernel==0.4.1
# RUN python3 -m pip install --no-cache-dir git+https://github.com/Dao-AILab/fast-hadamard-transform.git
# Add fp-quant for quantization testing
# Requires py3.11 but our CI runs on 3.9
# RUN python3 -m pip install --no-cache-dir "fp-quant>=0.1.6"