tvalentyn commented on code in PR #36420:
URL: https://github.com/apache/beam/pull/36420#discussion_r2470921491
##########
sdks/python/container/run_generate_requirements.sh:
##########
@@ -76,14 +76,24 @@ python"${PY_VERSION}" -m venv "$ENV_PATH"
source "$ENV_PATH"/bin/activate
pip install --upgrade pip setuptools wheel
+# For non-vllm (non-gpu) requirement files, force downloading torch from CPU
wheels
Review Comment:
Does the outcome of the generated requirements file depend on which wheel is
installed (cpu vs non-cpu)? If so, is it because the dependencies of torch
change?
##########
sdks/python/container/run_generate_requirements.sh:
##########
@@ -76,14 +76,24 @@ python"${PY_VERSION}" -m venv "$ENV_PATH"
source "$ENV_PATH"/bin/activate
pip install --upgrade pip setuptools wheel
+# For non-vllm (non-gpu) requirement files, force downloading torch from CPU
wheels
+INDEX_URL_OPTION="--extra-index-url https://download.pytorch.org/whl/cpu";
+if [[ $EXTRAS == *"vllm"* ]]; then
+ # Explicitly install torch to avoid
https://github.com/facebookresearch/xformers/issues/740
+ # This should be overwritten later since the vllm extra is installed
alongside torch
Review Comment:
> This should be overwritten later
I don't quite follow the comment. Could you please clarify what is being
overwritten?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
