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

[Model] Reorganize pooling layers (#31973)

Browse Source 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2026-01-09 19:02:14 +08:00
committed by GitHub
parent 020732800c
commit c8ed39b9dd
34 changed files with 1290 additions and 1143 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
vllm/model_executor/models/siglip.py
View File

@@ -27,7 +27,7 @@ from vllm.model_executor.layers.linear import (
QKVParallelLinear,
RowParallelLinear,
)
from vllm.model_executor.layers.pooler import DispatchPooler, Pooler
from vllm.model_executor.layers.pooler import DispatchPooler
from vllm.model_executor.layers.quantization import QuantizationConfig
from vllm.model_executor.layers.vocab_parallel_embedding import VocabParallelEmbedding
from vllm.model_executor.model_loader.weight_utils import (
@@ -1050,12 +1050,7 @@ class SiglipEmbeddingModel(nn.Module, SupportsMultiModal, SupportsQuant):
assert pooler_config is not None
self.pooler_config = pooler_config
self.pooler = DispatchPooler(
{
"token_embed": Pooler.for_token_embed(pooler_config),
"embed": Pooler.for_embed(pooler_config),
}
)
self.pooler = DispatchPooler.for_embedding(pooler_config)
self._is_text_input = True