biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Model][2/N] Improve all pooling task | Support multi-vector retrieval (#25370)

Browse Source 
Signed-off-by: wang.yuqi <noooop@126.com>
This commit is contained in:
wang.yuqi
2025-10-15 19:14:41 +08:00
committed by GitHub
parent d4d1a6024f
commit f54f85129e
41 changed files with 786 additions and 399 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
tests/models/language/pooling/test_token_classification.py
View File

@@ -19,7 +19,7 @@ def test_bert_models(
dtype: str,
) -> None:
with vllm_runner(model, max_model_len=None, dtype=dtype) as vllm_model:
vllm_outputs = vllm_model.encode(example_prompts)
vllm_outputs = vllm_model.token_classify(example_prompts)
with hf_runner(
model, dtype=dtype, auto_cls=AutoModelForTokenClassification
@@ -50,7 +50,7 @@ def test_modernbert_models(
dtype: str,
) -> None:
with vllm_runner(model, max_model_len=None, dtype=dtype) as vllm_model:
vllm_outputs = vllm_model.encode(example_prompts)
vllm_outputs = vllm_model.token_classify(example_prompts)
with hf_runner(
model, dtype=dtype, auto_cls=AutoModelForTokenClassification