diff --git a/vllm/entrypoints/pooling/base/serving.py b/vllm/entrypoints/pooling/base/serving.py index cfda3d6563..e94d2ce2e3 100644 --- a/vllm/entrypoints/pooling/base/serving.py +++ b/vllm/entrypoints/pooling/base/serving.py @@ -108,7 +108,6 @@ class PoolingBaseServing(ABC, BaseServing): request: AnyPoolingRequest, raw_request: Request | None = None, ): - model_name = self.models.model_name() base_request_id = self._base_request_id( raw_request, getattr(request, "request_id", None) ) @@ -117,6 +116,7 @@ class PoolingBaseServing(ABC, BaseServing): pooling_params = io_processor.create_pooling_params(request) lora_request = self._maybe_get_adapters(request) + model_name = self.models.model_name(lora_request) priorities = getattr(request, "priority", 0) prompt_extras = { k: v diff --git a/vllm/entrypoints/pooling/scoring/serving.py b/vllm/entrypoints/pooling/scoring/serving.py index 3b1e138115..b32a7a5730 100644 --- a/vllm/entrypoints/pooling/scoring/serving.py +++ b/vllm/entrypoints/pooling/scoring/serving.py @@ -79,7 +79,7 @@ class ServingScores(PoolingServing): final_res_batch = ctx.final_res_batch request_id = ctx.request_id created_time = ctx.created_time - model_name = self.models.model_name() + model_name = ctx.model_name if isinstance(ctx.request, ScoreRequest): return self._request_output_to_score_response(