[V0 Deprecation] Enable the remaining multimodal tests in V1 (#25307)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2025-09-21 01:50:58 +08:00
committed by GitHub
parent d88918e4c2
commit bef180f009
8 changed files with 195 additions and 214 deletions

View File

@@ -4,8 +4,6 @@
import pytest
import torch
from vllm.utils import set_default_torch_num_threads
from ....conftest import VllmRunner
@@ -30,19 +28,17 @@ def _run_test(
} for _ in range(10)
]
with (
set_default_torch_num_threads(1),
vllm_runner(
model,
runner="pooling",
dtype=torch.float16,
enforce_eager=True,
skip_tokenizer_init=True,
# Limit the maximum number of sequences to avoid the
# test going OOM during the warmup run
max_num_seqs=32,
) as vllm_model,
):
with vllm_runner(
model,
runner="pooling",
dtype="half",
enforce_eager=True,
skip_tokenizer_init=True,
# Limit the maximum number of sequences to avoid the
# test going OOM during the warmup run
max_num_seqs=32,
default_torch_num_threads=1,
) as vllm_model:
vllm_model.encode(prompt)