[Model] Replace embedding models with pooling adapter (#10769)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2024-12-01 08:02:54 +08:00
committed by GitHub
parent 7e4bbda573
commit 133707123e
32 changed files with 383 additions and 319 deletions

View File

@@ -4,6 +4,8 @@ Run `pytest tests/models/embedding/language/test_embedding.py`.
"""
import pytest
from vllm.config import PoolerConfig
from ..utils import check_embeddings_close
@@ -33,6 +35,9 @@ def test_models(
dtype: str,
) -> None:
vllm_extra_kwargs = {}
if model == "ssmits/Qwen2-7B-Instruct-embed-base":
vllm_extra_kwargs["override_pooler_config"] = \
PoolerConfig(pooling_type="MEAN")
if model == "Alibaba-NLP/gte-Qwen2-7B-instruct":
vllm_extra_kwargs["hf_overrides"] = {"is_causal": False}