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

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

docs/models/pooling_models/README.md

@@ -31,28 +31,29 @@ Of course, we also have "plugin" tasks that allow users to customize input and o
 
 ### Pooling Tasks
 
-| Pooling Tasks      | Granularity   | Outputs                                         |
-|--------------------|---------------|-------------------------------------------------|
-| `classify`         | Sequence-wise | probability vector of classes for each sequence |
-| `score` (see note) | Sequence-wise | reranker score for each sequence                |
-| `embed`            | Sequence-wise | vector representations for each sequence        |
-| `token_classify`   | Token-wise    | probability vector of classes for each token    |
-| `token_embed`      | Token-wise    | vector representations for each token           |
+| Pooling Tasks         | Granularity   | Outputs                                         |
+|-----------------------|---------------|-------------------------------------------------|
+| `classify` (see note) | Sequence-wise | probability vector of classes for each sequence |
+| `embed`               | Sequence-wise | vector representations for each sequence        |
+| `token_classify`      | Token-wise    | probability vector of classes for each token    |
+| `token_embed`         | Token-wise    | vector representations for each token           |
 
 !!! note
     Within classification tasks, there is a specialized subcategory: Cross-encoder (aka reranker) models. These models are a subset of classification models that accept two prompts as input and output num_labels equal to 1.
 
 ### Score Types
 
-| Pooling Tasks      | Granularity   | Outputs                                         | Score Types        | scoring function         |
-|--------------------|---------------|-------------------------------------------------|--------------------|--------------------------|
-| `classify`         | Sequence-wise | probability vector of classes for each sequence | nan                | nan                      |
-| `score` (see note) | Sequence-wise | reranker score for each sequence                | `cross-encoder`    | linear classifier        |
-| `embed`            | Sequence-wise | vector representations for each sequence        | `bi-encoder`       | cosine similarity        |
-| `token_classify`   | Token-wise    | probability vector of classes for each token    | nan                | nan                      |
-| `token_embed`      | Token-wise    | vector representations for each token           | `late-interaction` | late interaction(MaxSim) |
+The scoring models is designed to compute similarity scores between two input prompts. It supports three model types (aka `score_type`): `cross-encoder`, `late-interaction`, and `bi-encoder`.
 
-The score models is designed to compute similarity scores between two input prompts. It supports three model types (aka `score_type`): `cross-encoder`, `late-interaction`, and `bi-encoder`.
+| Pooling Tasks         | Granularity   | Outputs                                      | Score Types        | scoring function         |
+|-----------------------|---------------|----------------------------------------------|--------------------|--------------------------|
+| `classify` (see note) | Sequence-wise | reranker score for each sequence             | `cross-encoder`    | linear classifier        |
+| `embed`               | Sequence-wise | vector representations for each sequence     | `bi-encoder`       | cosine similarity        |
+| `token_classify`      | Token-wise    | probability vector of classes for each token | nan                | nan                      |
+| `token_embed`         | Token-wise    | vector representations for each token        | `late-interaction` | late interaction(MaxSim) |
+
+!!! note
+    Only when a classification model outputs num_labels equal to 1 can it be used as a scoring model and have its scoring API enabled.
 
 ### Pooling Usages
 
@@ -85,14 +86,16 @@ enabling the corresponding APIs.
 
 ### Offline APIs corresponding to pooling tasks
 
-| Task             | APIs                                                                       |
-|------------------|----------------------------------------------------------------------------|
-| `embed`          | `LLM.embed(...)`,`LLM.encode(..., pooling_task="embed")`, `LLM.score(...)` |
-| `classify`       | `LLM.classify(...)`, `LLM.encode(..., pooling_task="classify")`            |
-| `score`          | `LLM.score(...)`                                                           |
-| `token_classify` | `LLM.reward(...)`, `LLM.encode(..., pooling_task="token_classify")`        |
-| `token_embed`    | `LLM.encode(..., pooling_task="token_embed")`, `LLM.score(...)`            |
-| `plugin`         | `LLM.encode(..., pooling_task="plugin")`                                   |
+| Task             | APIs                                                                                  |
+|------------------|---------------------------------------------------------------------------------------|
+| `embed`          | `LLM.embed(...)`, `LLM.encode(..., pooling_task="embed")`, `LLM.score(...)`(see note) |
+| `classify`       | `LLM.classify(...)`, `LLM.encode(..., pooling_task="classify")`, `LLM.score(...)`     |
+| `token_classify` | `LLM.reward(...)`, `LLM.encode(..., pooling_task="token_classify")`                   |
+| `token_embed`    | `LLM.encode(..., pooling_task="token_embed")`, `LLM.score(...)`                       |
+| `plugin`         | `LLM.encode(..., pooling_task="plugin")`                                              |
+
+!!! note
+    Only when a classification model outputs num_labels equal to 1 can it be used as a scoring model and have its scoring API enabled.
 
 ### `LLM.classify`
 
@@ -206,11 +209,11 @@ If `--runner pooling` has been set (manually or automatically) but the model doe
 vLLM will attempt to automatically convert the model according to the architecture names
 shown in the table below.
 
-| Architecture                                    | `--convert` | Supported pooling tasks               |
-| ----------------------------------------------- | ----------- | ------------------------------------- |
-| `*ForTextEncoding`, `*EmbeddingModel`, `*Model` | `embed`     | `token_embed`, `embed`                |
-| `*ForRewardModeling`, `*RewardModel`            | `embed`     | `token_embed`, `embed`                |
-| `*For*Classification`, `*ClassificationModel`   | `classify`  | `token_classify`, `classify`, `score` |
+| Architecture                                    | `--convert` | Supported pooling tasks      |
+|-------------------------------------------------|-------------|------------------------------|
+| `*ForTextEncoding`, `*EmbeddingModel`, `*Model` | `embed`     | `token_embed`, `embed`       |
+| `*ForRewardModeling`, `*RewardModel`            | `embed`     | `token_embed`, `embed`       |
+| `*For*Classification`, `*ClassificationModel`   | `classify`  | `token_classify`, `classify` |
 
 !!! tip
     You can explicitly set `--convert <type>` to specify how to convert the model.
@@ -251,3 +254,7 @@ Pooling models now default support all pooling, you can use it without any setti
 
 - Extracting hidden states prefers using `token_embed` task.
 - Named Entity Recognition (NER) and reward models prefers using `token_classify` task.
+
+### Score task
+
+`score` task is deprecated and will be removed in v0.20. Please use `classify` instead. Only when a classification model outputs num_labels equal to 1 can it be used as a scoring model and have its scoring API enabled.

docs/models/pooling_models/classify.md

@@ -17,6 +17,8 @@ The key distinction between (sequence) classification and token classification l
 
 Many classification models support both (sequence) classification and token classification. For further details on token classification, please refer to [this page](token_classify.md).
 
+Only when a classification model outputs num_labels equal to 1 can it be used as a scoring model and have its scoring API enabled, please refer to [this page](scoring.md).
+
 ## Typical Use Cases
 
 ### Classification
@@ -54,7 +56,7 @@ If your model is not in the above list, we will try to automatically convert the
 
 Cross-encoder (aka reranker) models are a subset of classification models that accept two prompts as input and output num_labels equal to 1. Most classification models can also be used as [cross-encoder models](scoring.md#cross-encoder-models). For more information on cross-encoder models, please refer to [this page](scoring.md).
 
---8<-- "docs/models/pooling_models/scoring.md:supported-score-models"
+--8<-- "docs/models/pooling_models/scoring.md:supported-cross-encoder-models"
 
 ### Reward Models
 

docs/models/pooling_models/scoring.md

@@ -10,11 +10,11 @@ The score models is designed to compute similarity scores between two input prom
 - Model Usage: Scoring
 - Pooling Task:
 
-| Score Types        | Pooling Tasks | scoring function         |
-|--------------------|---------------|--------------------------|
-| `cross-encoder`    | `score`       | linear classifier        |
-| `late-interaction` | `token_embed` | late interaction(MaxSim) |
-| `bi-encoder`       | `embed`       | cosine similarity        |
+| Score Types        | Pooling Tasks         | scoring function         |
+|--------------------|-----------------------|--------------------------|
+| `cross-encoder`    | `classify` (see note) | linear classifier        |
+| `late-interaction` | `token_embed`         | late interaction(MaxSim) |
+| `bi-encoder`       | `embed`               | cosine similarity        |
 
 - Offline APIs:
     - `LLM.score`
@@ -22,13 +22,16 @@ The score models is designed to compute similarity scores between two input prom
     - [Score API](scoring.md#score-api) (`/score`)
     - [Rerank API](scoring.md#rerank-api) (`/rerank`, `/v1/rerank`, `/v2/rerank`)
 
+!!! note
+    Only when a classification model outputs num_labels equal to 1 can it be used as a scoring model and have its scoring API enabled.
+
 ## Supported Models
 
 ### Cross-encoder models
 
 [Cross-encoder](https://www.sbert.net/examples/applications/cross-encoder/README.html) (aka reranker) models are a subset of classification models that accept two prompts as input and output num_labels equal to 1.
 
---8<-- [start:supported-score-models]
+--8<-- [start:supported-cross-encoder-models]
 
 #### Text-only Models
 
@@ -99,7 +102,7 @@ The score models is designed to compute similarity scores between two input prom
     vllm serve Qwen/Qwen3-VL-Reranker-2B --hf_overrides '{"architectures": ["Qwen3VLForSequenceClassification"],"classifier_from_token": ["no", "yes"],"is_original_qwen3_reranker": true}'
     ```
 
---8<-- [end:supported-score-models]
+--8<-- [end:supported-cross-encoder-models]
 
 ### Late-interaction models
 

tests/test_pooling_params.py

@@ -74,7 +74,7 @@ def test_embed_dimensions(model_info: EmbedModelInfo):
         pooling_params.verify(model_config)
 
 
-@pytest.mark.parametrize("task", ["score", "classify"])
+@pytest.mark.parametrize("task", ["classify"])
 def test_classify(task):
     model_config = MockModelConfig(pooler_config=PoolerConfig(seq_pooling_type="CLS"))
 

vllm/config/model.py

@@ -1435,10 +1435,10 @@ class ModelConfig:
     @property
     def score_type(self) -> ScoreType:
         """
-        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)
+        Scoring API handles score/rerank for:\n
+        - "classify" task (score_type: cross-encoder models)\n
+        - "embed" task (score_type: bi-encoder models)\n
+        - "token_embed" task (score_type: late interaction models)\n
         """
         # fixme: self._model_info.score_type is the score type before
         #  as_seq_cls_model, which is "bi-encoder", rather than the

vllm/entrypoints/llm.py

@@ -1477,9 +1477,9 @@ class LLM:
             data_1 = data_1 * len(data_2)
 
         if pooling_params is None:
-            pooling_params = PoolingParams(task="score")
+            pooling_params = PoolingParams(task="classify")
         elif pooling_params.task is None:
-            pooling_params.task = "score"
+            pooling_params.task = "classify"
 
         pooling_params_list = list[PoolingParams]()
 

vllm/entrypoints/openai/api_server.py

@@ -22,7 +22,7 @@ from fastapi.middleware.cors import CORSMiddleware
 from starlette.datastructures import State
 
 import vllm.envs as envs
-from vllm.config import VllmConfig
+from vllm.config import ModelConfig, VllmConfig
 from vllm.engine.arg_utils import AsyncEngineArgs
 from vllm.engine.protocol import EngineClient
 from vllm.entrypoints.chat_utils import load_chat_template
@@ -155,7 +155,9 @@ async def build_async_engine_client_from_engine_args(
 
 
 def build_app(
-    args: Namespace, supported_tasks: tuple["SupportedTask", ...] | None = None
+    args: Namespace,
+    supported_tasks: tuple["SupportedTask", ...] | None = None,
+    model_config: ModelConfig | None = None,
 ) -> FastAPI:
     if supported_tasks is None:
         warnings.warn(
@@ -191,7 +193,7 @@ def build_app(
         attach_router as register_sagemaker_api_router,
     )
 
-    register_sagemaker_api_router(app, supported_tasks)
+    register_sagemaker_api_router(app, supported_tasks, model_config)
 
     if "generate" in supported_tasks:
         from vllm.entrypoints.openai.generate.api_router import (
@@ -242,7 +244,7 @@ def build_app(
     if any(task in POOLING_TASKS for task in supported_tasks):
         from vllm.entrypoints.pooling import register_pooling_api_routers
 
-        register_pooling_api_routers(app, supported_tasks)
+        register_pooling_api_routers(app, supported_tasks, model_config)
 
     app.root_path = args.root_path
     app.add_middleware(
@@ -583,8 +585,10 @@ async def build_and_serve(
         uvicorn_kwargs["log_config"] = log_config
 
     supported_tasks = await engine_client.get_supported_tasks()
+    model_config = engine_client.model_config
+
     logger.info("Supported tasks: %s", supported_tasks)
-    app = build_app(args, supported_tasks)
+    app = build_app(args, supported_tasks, model_config)
     await init_app_state(engine_client, app.state, args, supported_tasks)
 
     logger.info("Starting vLLM server on %s", listen_address)

vllm/entrypoints/pooling/__init__.py

@@ -5,6 +5,9 @@ from typing import TYPE_CHECKING
 
 from fastapi import FastAPI
 
+from vllm.config import ModelConfig
+from vllm.logger import init_logger
+
 if TYPE_CHECKING:
     from argparse import Namespace
 
@@ -17,9 +20,30 @@ else:
     RequestLogger = object
     SupportedTask = object
 
+logger = init_logger(__name__)
+
+
+def enable_scoring_api(
+    supported_tasks: tuple["SupportedTask", ...],
+    model_config: ModelConfig | None = None,
+) -> bool:
+    if any(t in supported_tasks for t in ("embed", "token_embed")):
+        return True
+
+    if model_config is not None and "classify" in supported_tasks:
+        num_labels = getattr(model_config.hf_config, "num_labels", 0)
+        if num_labels != 1:
+            logger.debug_once("Score API is only enabled for num_labels == 1.")
+            return False
+        return True
+
+    return False
+
 
 def register_pooling_api_routers(
-    app: FastAPI, supported_tasks: tuple["SupportedTask", ...]
+    app: FastAPI,
+    supported_tasks: tuple["SupportedTask", ...],
+    model_config: ModelConfig | None = None,
 ):
     from vllm.entrypoints.pooling.pooling.api_router import router as pooling_router
 
@@ -37,11 +61,7 @@ def register_pooling_api_routers(
 
         app.include_router(embed_router)
 
-    # 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)
-    if any(t in supported_tasks for t in ("score", "embed", "token_embed")):
+    if enable_scoring_api(supported_tasks, model_config):
         from vllm.entrypoints.pooling.score.api_router import router as score_router
 
         app.include_router(score_router)
@@ -61,6 +81,8 @@ def init_pooling_state(
     from vllm.entrypoints.pooling.score.serving import ServingScores
     from vllm.tasks import POOLING_TASKS
 
+    model_config = engine_client.model_config
+
     resolved_chat_template = load_chat_template(args.chat_template)
 
     state.serving_pooling = (
@@ -102,10 +124,6 @@ def init_pooling_state(
         if "classify" in supported_tasks
         else None
     )
-    # 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)
     state.serving_scores = (
         ServingScores(
             engine_client,
@@ -114,6 +132,6 @@ def init_pooling_state(
             score_template=resolved_chat_template,
             log_error_stack=args.log_error_stack,
         )
-        if any(t in supported_tasks for t in ("embed", "score", "token_embed"))
+        if enable_scoring_api(supported_tasks, model_config)
         else None
     )

vllm/entrypoints/pooling/score/protocol.py

@@ -35,7 +35,7 @@ class ScoreRequestMixin(PoolingBasicRequestMixin, ClassifyRequestMixin):
             max_total_tokens_param="max_model_len",
         )
 
-    def to_pooling_params(self, task: PoolingTask = "score"):
+    def to_pooling_params(self, task: PoolingTask = "classify"):
         return PoolingParams(
             task=task,
             use_activation=self.use_activation,
@@ -111,7 +111,7 @@ class RerankRequest(PoolingBasicRequestMixin, ClassifyRequestMixin):
             max_total_tokens_param="max_model_len",
         )
 
-    def to_pooling_params(self, task: PoolingTask = "score"):
+    def to_pooling_params(self, task: PoolingTask = "classify"):
         return PoolingParams(
             task=task,
             use_activation=self.use_activation,

vllm/entrypoints/pooling/score/serving.py

@@ -413,7 +413,7 @@ class ServingScores(OpenAIServing):
         # Schedule the request and get the result generator.
         generators: list[AsyncGenerator[PoolingRequestOutput, None]] = []
 
-        default_pooling_params = request.to_pooling_params("score")
+        default_pooling_params = request.to_pooling_params("classify")
 
         for i, engine_prompt in enumerate(engine_prompts):
             request_id_item = f"{request_id}-{i}"

vllm/entrypoints/sagemaker/api_router.py

@@ -10,9 +10,11 @@ import pydantic
 from fastapi import APIRouter, Depends, FastAPI, HTTPException, Request
 from fastapi.responses import JSONResponse, Response
 
+from vllm.config import ModelConfig
 from vllm.entrypoints.openai.engine.protocol import ErrorResponse
 from vllm.entrypoints.openai.engine.serving import OpenAIServing
 from vllm.entrypoints.openai.utils import validate_json_request
+from vllm.entrypoints.pooling import enable_scoring_api
 from vllm.entrypoints.pooling.base.serving import PoolingServing
 from vllm.entrypoints.serve.instrumentator.basic import base
 from vllm.entrypoints.serve.instrumentator.health import health
@@ -25,7 +27,10 @@ GetHandlerFn = Callable[[Request], OpenAIServing | PoolingServing | None]
 EndpointFn = Callable[[RequestType, Request], Awaitable[Any]]
 
 
-def get_invocation_types(supported_tasks: tuple["SupportedTask", ...]):
+def get_invocation_types(
+    supported_tasks: tuple["SupportedTask", ...],
+    model_config: ModelConfig | None = None,
+):
     # NOTE: Items defined earlier take higher priority
     INVOCATION_TYPES: list[tuple[RequestType, tuple[GetHandlerFn, EndpointFn]]] = []
 
@@ -70,7 +75,7 @@ def get_invocation_types(supported_tasks: tuple["SupportedTask", ...]):
             (ClassificationRequest, (classify, create_classify)),
         ]
 
-    if "score" in supported_tasks:
+    if enable_scoring_api(supported_tasks, model_config):
         from vllm.entrypoints.pooling.score.api_router import do_rerank, rerank
         from vllm.entrypoints.pooling.score.protocol import RerankRequest
 
@@ -78,7 +83,6 @@ def get_invocation_types(supported_tasks: tuple["SupportedTask", ...]):
             (RerankRequest, (rerank, do_rerank)),
         ]
 
-    if "score" in supported_tasks or "embed" in supported_tasks:
         from vllm.entrypoints.pooling.score.api_router import create_score, score
         from vllm.entrypoints.pooling.score.protocol import ScoreRequest
 
@@ -97,11 +101,15 @@ def get_invocation_types(supported_tasks: tuple["SupportedTask", ...]):
     return INVOCATION_TYPES
 
 
-def attach_router(app: FastAPI, supported_tasks: tuple["SupportedTask", ...]):
+def attach_router(
+    app: FastAPI,
+    supported_tasks: tuple["SupportedTask", ...],
+    model_config: ModelConfig | None = None,
+):
     router = APIRouter()
 
     # NOTE: Construct the TypeAdapters only once
-    INVOCATION_TYPES = get_invocation_types(supported_tasks)
+    INVOCATION_TYPES = get_invocation_types(supported_tasks, model_config)
     INVOCATION_VALIDATORS = [
         (pydantic.TypeAdapter(request_type), (get_handler, endpoint))
         for request_type, (get_handler, endpoint) in INVOCATION_TYPES

vllm/model_executor/layers/pooler/activations.py

@@ -16,25 +16,22 @@ from vllm.utils.import_utils import resolve_obj_by_qualname
 logger = init_logger(__name__)
 
 
-def get_classification_act_fn(
+def get_act_fn(
     config: PretrainedConfig,
+    static_num_labels: bool = True,
 ) -> "PoolerActivation":
+    # get classification act_fn
     # Implement alignment with transformers ForSequenceClassificationLoss
     # https://github.com/huggingface/transformers/blob/57bb6db6ee4cfaccc45b8d474dfad5a17811ca60/src/transformers/loss/loss_utils.py#L92
     problem_type = getattr(config, "problem_type", "")
     if problem_type == "regression":
         return PoolerIdentity()
     if problem_type == "single_label_classification":
-        return PoolerClassify()
+        return PoolerClassify(static_num_labels=static_num_labels)
     if problem_type == "multi_label_classification":
         return PoolerMultiLabelClassify()
 
-    return PoolerClassify()
-
-
-def get_cross_encoder_act_fn(
-    config: PretrainedConfig,
-) -> "PoolerActivation":
+    # get cross_encoder act_fn
     function_name: str | None = None
     if (
         hasattr(config, "sentence_transformers")
@@ -55,24 +52,16 @@ def get_cross_encoder_act_fn(
         fn = resolve_obj_by_qualname(function_name)()
         return PoolerActivation.wraps(fn)
 
-    return PoolerClassify()
+    return PoolerClassify(static_num_labels=static_num_labels)
 
 
 def resolve_classifier_act_fn(
     model_config: ModelConfig,
     static_num_labels: bool = True,
-    act_fn: "PoolerActivation | str | None" = None,
+    act_fn: "PoolerActivation | None" = None,
 ):
-    if isinstance(act_fn, str):
-        if act_fn == "classify":
-            return get_classification_act_fn(model_config.hf_config)
-        if act_fn == "score":
-            return get_cross_encoder_act_fn(model_config.hf_config)
-
-        raise ValueError(f"act_fn [{act_fn=}] not supported.")
-
     if act_fn is None:
-        return PoolerClassify(static_num_labels=static_num_labels)
+        return get_act_fn(model_config.hf_config, static_num_labels)
 
     assert callable(act_fn)
     return act_fn
@@ -97,9 +86,8 @@ class PoolerActivation(nn.Module, ABC):
 
     def forward(self, pooled_data: _T) -> _T:
         # shape:
-        # classify (& score) -> (batch_size, num_classes)
-        # embed -> (batch_size, embedding_dim) or list(embedding_dim)
-        #          (batch_size, dimensions) or list(dimensions) if using MRL
+        # classify -> (batch_size, num_classes)
+        # embed -> (batch_size, embedding_size) or list(embedding_size)
         if isinstance(pooled_data, list):
             return [self.forward_chunk(data) for data in pooled_data]
 

vllm/model_executor/layers/pooler/seqwise/heads.py

@@ -56,29 +56,31 @@ class EmbeddingPoolerHead(SequencePoolerHead):
 
         if isinstance(pooled_data, list):
             pooled_data = torch.stack(pooled_data)
-        # pooled_data shape: [batchsize, hidden_dimension]
+        # pooled_data shape: [batchsize, hidden_size]
 
         if self.head_dtype is not None:
             pooled_data = pooled_data.to(self.head_dtype)
 
         # Apply ST projector
         if self.projector is not None:
-            pooled_data = self.projector(pooled_data)
-        # pooled_data shape: [batchsize, embedding_dimension]
+            embeddings = self.projector(pooled_data)
+        else:
+            embeddings = pooled_data
+        # embeddings shape: [batchsize, embedding_size]
 
         # for matryoshka representation
         dimensions_list = [pooling_param.dimensions for pooling_param in pooling_params]
         if any(d is not None for d in dimensions_list):
             # change the output dimension
-            assert len(pooled_data) == len(dimensions_list)
-            if len(set(dimensions_list)) == 1 and not isinstance(pooled_data, list):
+            assert len(embeddings) == len(dimensions_list)
+            if len(set(dimensions_list)) == 1 and not isinstance(embeddings, list):
                 # if all dimensions are the same
                 d = dimensions_list[0]
-                pooled_data = pooled_data[..., :d]
+                embeddings = embeddings[..., :d]
             else:
-                pooled_data = [
+                embeddings = [
                     vecs if d is None else vecs[..., :d]
-                    for vecs, d in zip(pooled_data, dimensions_list)
+                    for vecs, d in zip(embeddings, dimensions_list)
                 ]
 
         # for normalize
@@ -86,15 +88,15 @@ class EmbeddingPoolerHead(SequencePoolerHead):
             flags = [p.use_activation for p in pooling_params]
             if len(set(flags)) == 1:
                 if flags[0]:
-                    pooled_data = self.activation(pooled_data)
+                    embeddings = self.activation(embeddings)
             else:
-                pooled_data = [
+                embeddings = [
                     self.activation(vecs) if f else vecs
-                    for vecs, f in zip(pooled_data, flags)
+                    for vecs, f in zip(embeddings, flags)
                 ]
 
-        # pooled_data shape: [batchsize, embedding_dimension]
-        return pooled_data
+        # embeddings shape: [batchsize, embedding_size]
+        return embeddings
 
 
 class ClassifierPoolerHead(SequencePoolerHead):
@@ -113,7 +115,7 @@ class ClassifierPoolerHead(SequencePoolerHead):
         self.activation = activation
 
     def get_supported_tasks(self) -> Set[PoolingTask]:
-        return {"classify", "score"}
+        return {"classify"}
 
     def forward(
         self,
@@ -131,21 +133,23 @@ class ClassifierPoolerHead(SequencePoolerHead):
             pooled_data = pooled_data.to(self.head_dtype)
 
         if self.classifier is not None:
-            pooled_data = self.classifier(pooled_data)
-        # pooled_data shape: [batchsize, num_labels]
+            logits = self.classifier(pooled_data)
+        else:
+            logits = pooled_data
 
+        # logits shape: [batchsize, num_labels]
         if self.logit_bias is not None:
-            pooled_data -= self.logit_bias
+            logits -= self.logit_bias
 
         if self.activation is not None:
             flags = [p.use_activation for p in pooling_params]
             if len(set(flags)) == 1:
-                pooled_data = self.activation(pooled_data) if flags[0] else pooled_data
+                logits = self.activation(logits) if flags[0] else logits
             else:
-                pooled_data = [
+                logits = [
                     self.activation(vecs) if f else vecs
-                    for vecs, f in zip(pooled_data, flags)
+                    for vecs, f in zip(logits, flags)
                 ]
 
-        # pooled_data shape: [batchsize, num_labels]
-        return pooled_data
+        # logits shape: [batchsize, num_labels]
+        return logits

vllm/model_executor/layers/pooler/seqwise/methods.py

@@ -17,7 +17,7 @@ SequencePoolingMethodOutput: TypeAlias = torch.Tensor | list[torch.Tensor]
 
 class SequencePoolingMethod(nn.Module, ABC):
     def get_supported_tasks(self) -> Set[PoolingTask]:
-        return {"token_embed", "token_classify", "embed", "classify", "score"}
+        return {"token_embed", "token_classify", "embed", "classify"}
 
     def get_pooling_updates(self, task: PoolingTask) -> PoolingParamsUpdate:
         return PoolingParamsUpdate()

vllm/model_executor/layers/pooler/seqwise/poolers.py

@@ -108,7 +108,7 @@ def pooler_for_classify(
     *,
     pooling: SequencePoolingMethod | SequencePoolingFn | None = None,
     classifier: ClassifierFn | None = None,
-    act_fn: PoolerActivation | str | None = None,
+    act_fn: PoolerActivation | None = None,
 ):
     if pooling is None:
         pooling = get_seq_pooling_method(pooler_config.get_seq_pooling_type())

vllm/model_executor/layers/pooler/special.py

@@ -52,13 +52,6 @@ class DispatchPooler(Pooler):
                     pooler_config,
                     pooling=pooling,
                     classifier=classifier,
-                    act_fn="classify",
-                ),
-                "score": pooler_for_classify(
-                    pooler_config,
-                    pooling=pooling,
-                    classifier=classifier,
-                    act_fn="score",
                 ),
             }
         )
@@ -115,7 +108,7 @@ class DispatchPooler(Pooler):
 
 class IdentityPooler(Pooler):
     def get_supported_tasks(self) -> Set[PoolingTask]:
-        return {"plugin", "score"}
+        return {"plugin"}
 
     def forward(
         self,

vllm/model_executor/layers/pooler/tokwise/heads.py

@@ -68,22 +68,24 @@ class TokenEmbeddingPoolerHead(TokenPoolerHead):
 
         if self.head_dtype is not None:
             pooled_data = pooled_data.to(self.head_dtype)
-        # pooled_data shape: [n_tokens, hidden_dimension]
+        # pooled_data shape: [n_tokens, hidden_size]
 
         # Apply ST projector
         if self.projector is not None:
-            pooled_data = self.projector(pooled_data)
-        # pooled_data shape: [n_tokens, embedding_dimension]
+            embeddings = self.projector(pooled_data)
+        else:
+            embeddings = pooled_data
+        # embeddings shape: [n_tokens, embedding_size]
 
         # for matryoshka representation
-        pooled_data = pooled_data[..., : pooling_param.dimensions]
+        embeddings = embeddings[..., : pooling_param.dimensions]
 
         # for normalize
         if self.activation is not None and pooling_param.use_activation:
-            pooled_data = self.activation(pooled_data)
+            embeddings = self.activation(embeddings)
 
-        # pooled_data shape: [n_tokens, embedding_dimension]
-        return pooled_data
+        # embeddings shape: [n_tokens, embedding_size]
+        return embeddings
 
 
 class TokenClassifierPoolerHead(TokenPoolerHead):
@@ -118,16 +120,16 @@ class TokenClassifierPoolerHead(TokenPoolerHead):
         # hidden_states shape: [n_token, hidden_size]
 
         if self.classifier is not None:
-            scores = self.classifier(pooled_data)
+            logits = self.classifier(pooled_data)
         else:
-            scores = pooled_data
-        # scores shape: [n_token, num_labels]
+            logits = pooled_data
+        # logits shape: [n_token, num_labels]
 
         if self.logit_bias is not None:
-            scores -= self.logit_bias
+            logits -= self.logit_bias
 
         if self.activation is not None and pooling_param.use_activation:
-            scores = self.activation(scores)
+            logits = self.activation(logits)
 
-        # scores shape: [n_token, num_labels]
-        return scores
+        # logits shape: [n_token, num_labels]
+        return logits

vllm/model_executor/layers/pooler/tokwise/poolers.py

@@ -116,7 +116,7 @@ def pooler_for_token_classify(
     *,
     pooling: TokenPoolingMethod | TokenPoolingFn | None = None,
     classifier: ClassifierFn | None = None,
-    act_fn: PoolerActivation | str | None = None,
+    act_fn: PoolerActivation | None = None,
 ):
     if pooling is None:
         pooling = get_tok_pooling_method(pooler_config.get_tok_pooling_type())

vllm/model_executor/models/interfaces_base.py

@@ -194,18 +194,18 @@ class VllmModelForPooling(VllmModel[T_co], Protocol[T_co]):
     [vllm.config.model.ModelConfig.score_type][]
     to use by default.
     
-    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)
+    Scoring API handles score/rerank for:\n
+    - "classify" task (score_type: cross-encoder models)\n
+    - "embed" task (score_type: bi-encoder models)\n
+    - "token_embed" task (score_type: late interaction models)\n
     
-    score_type defaults to bi-encoder, then the Score API uses the "embed" task.
+    score_type defaults to bi-encoder, then the Score API uses the "embed" task.\n
     If you set score_type to cross-encoder via 
     [vllm.model_executor.models.interfaces.SupportsCrossEncoding][], 
-    then the Score API uses the "score" task.
+    then the Score API uses the "score" task.\n
     If you set score_type to late-interaction via 
     [vllm.model_executor.models.interfaces.SupportsLateInteraction][], 
-    then the Score API uses the "token_embed" task.    
+    then the Score API uses the "token_embed" task.\n
     """
 
     pooler: Pooler

vllm/pooling_params.py

@@ -7,9 +7,12 @@ from typing import Any
 import msgspec
 
 from vllm.config import ModelConfig, PoolerConfig
+from vllm.logger import init_logger
 from vllm.sampling_params import RequestOutputKind
 from vllm.tasks import PoolingTask
 
+logger = init_logger(__name__)
+
 
 class LateInteractionParams(
     msgspec.Struct,
@@ -54,10 +57,6 @@ class PoolingParams(
     dimensions: int | None = None
     # --8<-- [end:embed-pooling-params]
 
-    ## for classification, scoring and rerank
-    # --8<-- [start:classify-pooling-params]
-    # --8<-- [end:classify-pooling-params]
-
     ## for step pooling models
     step_tag_id: int | None = None
     returned_token_ids: list[int] | None = None
@@ -79,7 +78,6 @@ class PoolingParams(
         return {
             "embed": ["dimensions", "use_activation"],
             "classify": ["use_activation"],
-            "score": ["use_activation"],
             "token_embed": ["dimensions", "use_activation"],
             "token_classify": ["use_activation"],
         }
@@ -89,6 +87,13 @@ class PoolingParams(
         return deepcopy(self)
 
     def verify(self, model_config: ModelConfig) -> None:
+        if self.task == "score":
+            logger.warning_once(
+                "`score` task is deprecated and will be removed in v0.20. "
+                "Please use `classify` instead."
+            )
+            self.task = "classify"
+
         # plugin task uses io_processor.parse_request to verify inputs,
         # skipping PoolingParams verify
         if self.task == "plugin":
@@ -184,7 +189,7 @@ class PoolingParams(
                 elif self.dimensions < 1:
                     raise ValueError("Dimensions must be greater than 0")
 
-        elif self.task in ["classify", "score", "token_classify"]:
+        elif self.task in ["classify", "token_classify"]:
             if self.use_activation is None:
                 self.use_activation = True
         else:

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]()