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

[Refactor] Separate sequence and token pooling types (#32026)

Browse Source 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2026-01-10 12:53:24 +08:00
committed by GitHub
parent 52d428295d
commit 583a90e005
42 changed files with 324 additions and 204 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
tests/models/language/pooling_mteb_test/test_gte.py
View File

@@ -18,7 +18,7 @@ MODELS = [
"thenlper/gte-large",
mteb_score=0.76807651,
architecture="BertModel",
pooling_type="MEAN",
seq_pooling_type="MEAN",
attn_type="encoder_only",
is_prefix_caching_supported=False,
is_chunked_prefill_supported=False,
@@ -44,7 +44,7 @@ MODELS = [
architecture="GteNewModel",
mteb_score=0.775074696,
hf_overrides={"architectures": ["GteNewModel"]},
pooling_type="CLS",
seq_pooling_type="CLS",
attn_type="encoder_only",
is_prefix_caching_supported=False,
is_chunked_prefill_supported=False,
@@ -67,7 +67,7 @@ MODELS = [
"Alibaba-NLP/gte-Qwen2-1.5B-instruct",
mteb_score=0.758473459018872,
architecture="Qwen2ForCausalLM",
pooling_type="LAST",
seq_pooling_type="LAST",
attn_type="encoder_only",
is_prefix_caching_supported=False,
is_chunked_prefill_supported=False,
@@ -78,7 +78,7 @@ MODELS = [
"Alibaba-NLP/gte-modernbert-base",
mteb_score=0.748193353,
architecture="ModernBertModel",
pooling_type="CLS",
seq_pooling_type="CLS",
attn_type="encoder_only",
is_prefix_caching_supported=False,
is_chunked_prefill_supported=False,
@@ -89,7 +89,7 @@ MODELS = [
"Qwen/Qwen3-Embedding-0.6B",
mteb_score=0.771163695,
architecture="Qwen3ForCausalLM",
pooling_type="LAST",
seq_pooling_type="LAST",
attn_type="decoder",
is_prefix_caching_supported=True,
is_chunked_prefill_supported=True,
@@ -108,7 +108,7 @@ RERANK_MODELS = [
"Alibaba-NLP/gte-reranker-modernbert-base",
mteb_score=0.33386,
architecture="ModernBertForSequenceClassification",
pooling_type="CLS",
seq_pooling_type="CLS",
attn_type="encoder_only",
is_prefix_caching_supported=False,
is_chunked_prefill_supported=False,
@@ -119,7 +119,7 @@ RERANK_MODELS = [
mteb_score=0.33062,
architecture="GteNewForSequenceClassification",
hf_overrides={"architectures": ["GteNewForSequenceClassification"]},
pooling_type="CLS",
seq_pooling_type="CLS",
attn_type="encoder_only",
is_prefix_caching_supported=False,
is_chunked_prefill_supported=False,