[TPU][V1] Capture multimodal encoder during model compilation (#15051)

Signed-off-by: Michael Goin <mgoin64@gmail.com>
Signed-off-by: NickLucche <nlucches@redhat.com>
Co-authored-by: Michael Goin <mgoin64@gmail.com>
Co-authored-by: Siyuan Liu <lsiyuan@google.com>
This commit is contained in:
Nicolò Lucchesi
2025-04-21 18:36:59 -06:00
committed by GitHub
co-authored by Michael Goin Siyuan Liu
parent 71eda0bb76
commit 210207525e
4 changed files with 327 additions and 48 deletions
@@ -17,7 +17,7 @@ source /etc/environment
docker run --privileged --net host --shm-size=16G -it \
-e "HF_TOKEN=$HF_TOKEN" --name tpu-test \
vllm-tpu /bin/bash -c "python3 -m pip install git+https://github.com/thuml/depyf.git \
&& python3 -m pip install pytest tpu-info \
&& python3 -m pip install pytest pytest-asyncio tpu-info \
&& python3 -m pip install lm_eval[api]==0.4.4 \
&& export VLLM_USE_V1=1 \
&& export VLLM_XLA_CHECK_RECOMPILATION=1 \
@@ -42,6 +42,8 @@ docker run --privileged --net host --shm-size=16G -it \
&& echo TEST_8 \
&& pytest -s -v /workspace/vllm/tests/v1/tpu/test_topk_topp_sampler.py \
&& echo TEST_9 \
&& pytest -s -v /workspace/vllm/tests/v1/tpu/test_multimodal.py \
&& echo TEST_10 \
&& pytest -s -v /workspace/vllm/tests/v1/tpu/test_pallas.py" \