From 4222a212aa11cad8bd0bc53d5b7f601fbc52778f Mon Sep 17 00:00:00 2001 From: Nathan Habib Date: Mon, 22 Sep 2025 13:31:53 +0000 Subject: [PATCH 1/2] debug vllm --- .github/workflows/slow_tests.yaml | 8 +++++++- pyproject.toml | 2 +- 2 files changed, 8 insertions(+), 2 deletions(-) diff --git a/.github/workflows/slow_tests.yaml b/.github/workflows/slow_tests.yaml index a1f68f9d4..7d4adb7a9 100644 --- a/.github/workflows/slow_tests.yaml +++ b/.github/workflows/slow_tests.yaml @@ -42,4 +42,10 @@ jobs: run: nvidia-smi - name: Run tests - run: uv run pytest --disable-pytest-warnings --runslow tests/slow_tests/ + run: | + uv run pytest --disable-pytest-warnings --runslow tests/slow_tests/test_accelerate_model.py + nvidia-smi + uv run pytest --disable-pytest-warnings --runslow tests/slow_tests/test_vllm_model.py + nvidia-smi + uv run pytest --disable-pytest-warnings --runslow tests/slow_tests/test_accelerate_vlm_model.py + nvidia-smi diff --git a/pyproject.toml b/pyproject.toml index 411a7b898..4a07beda4 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -98,7 +98,7 @@ nanotron = [ "tensorboardX" ] tensorboardX = ["tensorboardX"] -vllm = ["vllm>=0.10.0,<0.10.2", "ray", "more_itertools"] +vllm = ["vllm>=0.10.0", "ray", "more_itertools"] sglang = ["sglang"] quality = ["ruff>=v0.11.0","pre-commit"] tests = ["pytest>=7.4.0","deepdiff","pip>=25.2"] From ff886560e1bc177ce7416610a818389371b95d6a Mon Sep 17 00:00:00 2001 From: Nathan Habib Date: Mon, 22 Sep 2025 13:46:21 +0000 Subject: [PATCH 2/2] debug vllm --- .github/workflows/slow_tests.yaml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/.github/workflows/slow_tests.yaml b/.github/workflows/slow_tests.yaml index 7d4adb7a9..3d5e63e9a 100644 --- a/.github/workflows/slow_tests.yaml +++ b/.github/workflows/slow_tests.yaml @@ -47,5 +47,5 @@ jobs: nvidia-smi uv run pytest --disable-pytest-warnings --runslow tests/slow_tests/test_vllm_model.py nvidia-smi - uv run pytest --disable-pytest-warnings --runslow tests/slow_tests/test_accelerate_vlm_model.py + export VLLM_WORKER_MULTIPROC_METHOD=spawn && uv run pytest --disable-pytest-warnings --runslow tests/slow_tests/test_accelerate_vlm_model.py nvidia-smi