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

Improve the output precision of embedding models (#19092)

Browse Source
This commit is contained in:
wang.yuqi
2025-06-04 19:48:57 +08:00
committed by GitHub
parent 8711bc5e68
commit 35cf32df30
8 changed files with 69 additions and 28 deletions
Download Patch File Download Diff File Expand all files Collapse all files

46
tests/models/language/pooling/test_intfloat.py Normal file
View File

@@ -0,0 +1,46 @@
# SPDX-License-Identifier: Apache-2.0
import pytest
from ...utils import EmbedModelInfo
from .embed_utils import correctness_test_embed_models
from .mteb_utils import mteb_test_embed_models
MODELS = [
########## BertModel
EmbedModelInfo("intfloat/e5-small",
architecture="BertModel",
enable_test=True),
EmbedModelInfo("intfloat/e5-base",
architecture="BertModel",
enable_test=False),
EmbedModelInfo("intfloat/e5-large",
architecture="BertModel",
enable_test=False),
EmbedModelInfo("intfloat/multilingual-e5-small",
architecture="BertModel",
enable_test=False),
########## XLMRobertaModel
EmbedModelInfo("intfloat/multilingual-e5-base",
architecture="XLMRobertaModel",
enable_test=True),
EmbedModelInfo("intfloat/multilingual-e5-large",
architecture="XLMRobertaModel",
enable_test=False),
EmbedModelInfo("intfloat/multilingual-e5-large-instruct",
architecture="XLMRobertaModel",
enable_test=False),
]
@pytest.mark.parametrize("model_info", MODELS)
def test_embed_models_mteb(hf_runner, vllm_runner,
model_info: EmbedModelInfo) -> None:
mteb_test_embed_models(hf_runner, vllm_runner, model_info)
@pytest.mark.parametrize("model_info", MODELS)
def test_embed_models_correctness(hf_runner, vllm_runner,
model_info: EmbedModelInfo,
example_prompts) -> None:
correctness_test_embed_models(hf_runner, vllm_runner, model_info,
example_prompts)