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

[Model] Pooling models default to using chunked prefill & prefix caching if supported. (#20930)

Browse Source 
Signed-off-by: wang.yuqi <noooop@126.com>
This commit is contained in:
wang.yuqi
2025-08-12 00:41:37 +08:00
committed by GitHub
parent 16fb668b61
commit 84cf78acee
31 changed files with 452 additions and 261 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
tests/models/language/pooling/test_bge_reranker_v2_gemma.py
View File

@@ -8,12 +8,12 @@ import torch
from tests.conftest import HfRunner
from .mteb_utils import (RerankModelInfo, VllmMtebEncoder,
mteb_test_rerank_models)
from ...utils import LASTPoolingRerankModelInfo, RerankModelInfo
from .mteb_utils import VllmMtebEncoder, mteb_test_rerank_models
RERANK_MODELS = [
RerankModelInfo("BAAI/bge-reranker-v2-gemma",
architecture="GemmaForSequenceClassification"),
LASTPoolingRerankModelInfo("BAAI/bge-reranker-v2-gemma",
architecture="GemmaForSequenceClassification"),
]
PROMPT = "Given a query A and a passage B, determine whether the passage contains an answer to the query by providing a prediction of either 'Yes' or 'No'." # noqa: E501