From 33d5e29be90834fbac91051f97f853696b2f2ec5 Mon Sep 17 00:00:00 2001 From: Chenyaaang <42742451+Chenyaaang@users.noreply.github.com> Date: Mon, 23 Jun 2025 16:04:28 -0700 Subject: [PATCH] [TPU] Fix tpu model runner test (#19995) Signed-off-by: Chenyaaang --- tests/v1/tpu/worker/test_tpu_model_runner.py | 2 ++ 1 file changed, 2 insertions(+) diff --git a/tests/v1/tpu/worker/test_tpu_model_runner.py b/tests/v1/tpu/worker/test_tpu_model_runner.py index 0e7d305fef9ed..d22ddf5c7e581 100644 --- a/tests/v1/tpu/worker/test_tpu_model_runner.py +++ b/tests/v1/tpu/worker/test_tpu_model_runner.py @@ -6,6 +6,7 @@ import pytest from vllm.attention.layer import Attention from vllm.config import (CacheConfig, ModelConfig, SchedulerConfig, VllmConfig, set_current_vllm_config) +from vllm.pooling_params import PoolingParams from vllm.sampling_params import SamplingParams from vllm.utils import GiB_bytes from vllm.v1.core.kv_cache_utils import (estimate_max_model_len, @@ -71,6 +72,7 @@ def _schedule_new_request(*req_ids: str) -> SchedulerOutput: mm_hashes=[], mm_positions=[], sampling_params=SamplingParams(), + pooling_params=PoolingParams(), block_ids=([0], ), # block_ids should be tuple[list[int]] num_computed_tokens=0, lora_request=None,