[Model] Deprecate the score task (this will not affect users). (#37537)

Signed-off-by: wang.yuqi <yuqi.wang@daocloud.io>

vllm/tasks.py

@@ -8,7 +8,6 @@ GENERATION_TASKS: tuple[GenerationTask, ...] = get_args(GenerationTask)
 PoolingTask = Literal[
     "embed",
     "classify",
-    "score",
     "token_embed",
     "token_classify",
     "plugin",
@@ -16,10 +15,6 @@ PoolingTask = Literal[
 ]
 POOLING_TASKS: tuple[PoolingTask, ...] = get_args(PoolingTask)
 
-# Score API handles score/rerank for:
-# - "score" task (score_type: cross-encoder models)
-# - "embed" task (score_type: bi-encoder models)
-# - "token_embed" task (score_type: late interaction models)
 ScoreType = Literal["bi-encoder", "cross-encoder", "late-interaction"]
 
 FrontendTask = Literal["render"]

vllm/v1/worker/gpu_model_runner.py

@@ -2834,15 +2834,7 @@ class GPUModelRunner(
         if not is_pooling_model(model):
             return []
 
-        supported_tasks = list(model.pooler.get_supported_tasks())
-
-        if "score" in supported_tasks:
-            num_labels = getattr(self.model_config.hf_config, "num_labels", 0)
-            if num_labels != 1:
-                supported_tasks.remove("score")
-                logger.debug_once("Score API is only enabled for num_labels == 1.")
-
-        return supported_tasks
+        return list(model.pooler.get_supported_tasks())
 
     def get_supported_tasks(self) -> tuple[SupportedTask, ...]:
         tasks = list[SupportedTask]()