Support embedding models in V1 (#16188)

Signed-off-by: Max de Bayser <mbayser@br.ibm.com>
Signed-off-by: Max de Bayser <maxdebayser@gmail.com>
Signed-off-by: 22quinn <33176974+22quinn@users.noreply.github.com>
Co-authored-by: 22quinn <33176974+22quinn@users.noreply.github.com>
This commit is contained in:
Maximilien de Bayser
2025-06-18 21:36:33 -07:00
committed by GitHub
co-authored by 22quinn
parent 4959915089
commit 799397ee4f
56 changed files with 889 additions and 281 deletions
+3 -2
View File
@@ -21,7 +21,7 @@ from vllm.model_executor.model_loader.weight_utils import default_weight_loader
from vllm.model_executor.pooling_metadata import PoolingMetadata
from vllm.sequence import IntermediateTensors, PoolerOutput
from .interfaces import SupportsCrossEncoding
from .interfaces import SupportsCrossEncoding, SupportsV0Only
from .utils import WeightsMapper, maybe_prefix
@@ -270,7 +270,8 @@ class ModernBertPooler(nn.Module):
return pooled_output
class ModernBertForSequenceClassification(nn.Module, SupportsCrossEncoding):
class ModernBertForSequenceClassification(nn.Module, SupportsV0Only,
SupportsCrossEncoding):
def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
super().__init__()