-
Notifications
You must be signed in to change notification settings - Fork 25.2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Enable Past CI #17919
Enable Past CI #17919
Conversation
The documentation is not available anymore as the PR was closed or merged. |
@@ -0,0 +1,51 @@ | |||
ARG BASE_DOCKER_IMAGE="nvidia/cuda:11.2.2-cudnn8-devel-ubuntu20.04" |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I will remove some commented blocks before merge - the installations which won't be included.
### RUN python3 -m pip install --no-cache-dir git+https://github.com/facebookresearch/detectron2.git pytesseract https://github.com/kpu/kenlm/archive/master.zip | ||
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 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
no accelerate
### RUN python3 -m pip install --no-cache-dir torch-scatter -f https://data.pyg.org/whl/torch-$(python3 -c "from torch import version; print(version.__version__.split('+')[0])")+$CUDA.html | ||
### RUN python3 -m pip install --no-cache-dir intel_extension_for_pytorch==$INTEL_TORCH_EXT+cpu -f https://software.intel.com/ipex-whl-stable | ||
|
||
### RUN python3 -m pip install --no-cache-dir git+https://github.com/facebookresearch/detectron2.git pytesseract https://github.com/kpu/kenlm/archive/master.zip |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
not to include those 3rd party libs
### RUN python3 -m pip install --no-cache-dir torch-scatter -f https://data.pyg.org/whl/torch-$(python3 -c "from torch import version; print(version.__version__.split('+')[0])")+$CUDA.html | ||
### RUN python3 -m pip install --no-cache-dir intel_extension_for_pytorch==$INTEL_TORCH_EXT+cpu -f https://software.intel.com/ipex-whl-stable | ||
|
||
### RUN python3 -m pip install --no-cache-dir git+https://github.com/facebookresearch/detectron2.git pytesseract https://github.com/kpu/kenlm/archive/master.zip |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
not to include those 3rd party libs
@@ -0,0 +1,116 @@ | |||
import argparse |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This file is to make the installation commands easier to read in Docker files.
This should be used for other Docker files once this PR is merged.
For example, the following block hurts my head a bit already 😄
transformers/docker/transformers-pytorch-gpu/Dockerfile
Lines 19 to 21 in 49cd736
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/cu113 | |
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/cu113 | |
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/cu113 |
"torch": "1.10.2", | ||
"torchvision": "0.11.3", | ||
"torchaudio": "0.10.2", | ||
"python": 3.9, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is the highest python version for that PyTorch version. Not really important
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is great, super impressive! Let's merge it when you're happy with it and monitor failures. Please open an issue with the first reports when they're here so that we may discuss there which versions we'd like to keep or not.
utils/past_ci_versions.py
Outdated
# need python 3.7 | ||
# "1.3": { | ||
# "torch": "1.3.1", | ||
# "torchvision": "0.4.2", | ||
# "torchaudio": None, | ||
# "python": 3.7, | ||
# "cuda": "cu100", | ||
# "docker-base": "10.0-cudnn7-devel-ubuntu18.04", | ||
# "install": "python3 -m pip install --no-cache-dir -U torch==1.3.1 torchvision==0.4.2 torchaudio==0.4.0 --extra-index-url https://download.pytorch.org/whl/cu100", | ||
# }, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Let's remove it then
utils/past_ci_versions.py
Outdated
# "install": "python3 -m pip install --no-cache-dir -U torch==1.3.1 torchvision==0.4.2 torchaudio==0.4.0 --extra-index-url https://download.pytorch.org/whl/cu100", | ||
# }, | ||
}, | ||
# Should we include `tensorflow_probability`? |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Let's do without it as a first step, let's start simple
utils/past_ci_versions.py
Outdated
# "2.3": { | ||
# "tensorflow": "2.3.4", | ||
# "install": "python3 -m pip install --no-cache-dir -U tensorflow==2.3.4", | ||
# }, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This one isn't available either in the python version ?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
TF 2.3 requires CUDA 10.1 / cudnn 7.6, for which I don't find a good docker image on nvidia/
.
(There is one for Ubuntu 18.04, but I remembered I have some trouble with it. I can re-run docker build to see what goes wrong)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hi @LysandreJik . I remember what's wrong 😄
The best image I could find is nvidia/cuda:10.1-cudnn7-devel-ubuntu18.04
, which has Python 3.6. During build, I got errors like
ERROR: No matching distribution found for accelerate>=0.10.0; extra == "dev"
Also, we have python": "python>=3.7.0
in setup.py
.
As we discussed shortly before, I could try to build an image from scratch with python >= 3.7 with the necessary cuda/cudnn. I didn't do that though but just tried to launch the Past-CI with those versions easier to deal with.
I will merge this PR first, and if we would like to give a stab on TF 2.3, I can make it.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Let's merge it like this and ask the TF crew if they think 2.3 should be supported before putting too much efforts into it :)
I read your comment too quickly. So far the Past CI will be triggered only on pushing to I think we can launch past CI monthly or even bimonthly. Please let me know if you have different opinion, @LysandreJik. Thanks. |
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
What does this PR do?
Enable Past CI