Skip to content

Downgrade torchao to 0.15.0 #159

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.

Sign up for GitHub

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

Closed
huydhn wants to merge 8 commits into from
Closed

Downgrade torchao to 0.15.0 #159

Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
b6cb4b9
Update torchao and fbgemm-gpu-genai to nightly pre-releases
huydhn Mar 5, 2026
60ca767
Unset TRANSFORMERS_OFFLINE in vLLM benchmark workflow
huydhn Mar 5, 2026
5280d4d
Revert "Unset TRANSFORMERS_OFFLINE in vLLM benchmark workflow"
huydhn Mar 5, 2026
6a6c536
Set TRANSFORMERS_OFFLINE to 0 for testing
huydhn Mar 5, 2026
9be8f74
Use torchao 0.15.0
huydhn Mar 7, 2026
32eb493
Merge branch 'main' into update-torchao-fbgemm-nightly
huydhn Mar 7, 2026
d711f63
Typo
huydhn Mar 7, 2026
daf1d8a
Don't install fbgemm for now
huydhn Mar 7, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 4 additions & 3 deletions .github/workflows/vllm-benchmark.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -293,10 +293,10 @@ jobs:
env:
SCCACHE_BUCKET: ossci-compiler-cache-circleci-v2
SCCACHE_REGION: us-east-1
FLASHINFER_WORKSPACE_BASE: /mnt/hf_cache
HF_TOKEN: ${{ secrets.HF_TOKEN }}
HF_HOME: /mnt/hf_cache
FLASHINFER_WORKSPACE_BASE: /mnt/hf_cache
TRANSFORMERS_OFFLINE: 1
TRANSFORMERS_OFFLINE: 0
DOCKER_IMAGE: ${{ env.DOCKER_IMAGE_PREFIX }}:${{ env.HEAD_SHA }}${{ env.DOCKER_IMAGE_SUFFIX }}
# vLLM-related environment variables
ENGINE_VERSION: v1
Expand Down Expand Up @@ -347,7 +347,8 @@ jobs:
)
if [[ "${DEVICE_NAME}" == "cuda" ]]; then
docker exec -t "${container_name}" bash -c "
pip install torchao==0.16.0 fbgemm-gpu-genai==1.5.0
# Put fbgemm-gpu-genai back once it has a version working with 2.11
pip install torchao==0.15.0 --extra-index-url https://download.pytorch.org/whl/cu130

# A quick mitigation for https://github.com/vllm-project/vllm/issues/32373
rm /etc/ld.so.conf.d/cuda-compat.conf || true
Expand Down
Loading