[Model] Re-add the implicit conversion feature for as_seq_cls_model (vllm-project#21103)

Signed-off-by: wang.yuqi <noooop@126.com>
Signed-off-by: Himanshu Jaju <hj@mistral.ai>

Author: noooop

tests/models/registry.py

@@ -265,7 +265,6 @@ def check_available_online(
     "Qwen2MoeForCausalLM": _HfExamplesInfo("Qwen/Qwen1.5-MoE-A2.7B-Chat"),
     "Qwen3ForCausalLM": _HfExamplesInfo("Qwen/Qwen3-8B"),
     "Qwen3MoeForCausalLM": _HfExamplesInfo("Qwen/Qwen3-30B-A3B"),
-    "Qwen3ForSequenceClassification": _HfExamplesInfo("tomaarsen/Qwen3-Reranker-0.6B-seq-cls"),  # noqa: E501
     "RWForCausalLM": _HfExamplesInfo("tiiuae/falcon-40b"),
     "StableLMEpochForCausalLM": _HfExamplesInfo("stabilityai/stablelm-zephyr-3b"),  # noqa: E501
     "StableLmForCausalLM": _HfExamplesInfo("stabilityai/stablelm-3b-4e1t"),
@@ -292,7 +291,6 @@ def check_available_online(
     # [Text-only]
     "BertModel": _HfExamplesInfo("BAAI/bge-base-en-v1.5", v0_only=True),
     "Gemma2Model": _HfExamplesInfo("BAAI/bge-multilingual-gemma2", v0_only=True),  # noqa: E501
-    "GPT2ForSequenceClassification": _HfExamplesInfo("nie3e/sentiment-polish-gpt2-small"),  # noqa: E501
     "GritLM": _HfExamplesInfo("parasail-ai/GritLM-7B-vllm"),
     "GteModel": _HfExamplesInfo("Snowflake/snowflake-arctic-embed-m-v2.0",
                                                trust_remote_code=True),
@@ -311,7 +309,6 @@ def check_available_online(
     "Qwen2Model": _HfExamplesInfo("ssmits/Qwen2-7B-Instruct-embed-base"),
     "Qwen2ForRewardModel": _HfExamplesInfo("Qwen/Qwen2.5-Math-RM-72B"),
     "Qwen2ForProcessRewardModel": _HfExamplesInfo("Qwen/Qwen2.5-Math-PRM-7B"),
-    "Qwen2ForSequenceClassification": _HfExamplesInfo("jason9693/Qwen2.5-1.5B-apeach"),  # noqa: E501
     "RobertaModel": _HfExamplesInfo("sentence-transformers/stsb-roberta-base-v2", v0_only=True),  # noqa: E501
     "RobertaForMaskedLM": _HfExamplesInfo("sentence-transformers/all-roberta-large-v1", v0_only=True),  # noqa: E501
     "XLMRobertaModel": _HfExamplesInfo("intfloat/multilingual-e5-small", v0_only=True),  # noqa: E501
@@ -324,20 +321,29 @@ def check_available_online(
                                             is_available_online=False),  # noqa: E501
 }
 
-_CROSS_ENCODER_EXAMPLE_MODELS = {
-    # [Text-only]
+_SEQUENCE_CLASSIFICATION_EXAMPLE_MODELS = {
+    # [Decoder-only]
+    "GPT2ForSequenceClassification": _HfExamplesInfo("nie3e/sentiment-polish-gpt2-small"),  # noqa: E501
+
+    # [Cross-encoder]
     "BertForSequenceClassification": _HfExamplesInfo("cross-encoder/ms-marco-MiniLM-L-6-v2", v0_only=True),  # noqa: E501
-    "GemmaForSequenceClassification": _HfExamplesInfo("BAAI/bge-reranker-v2-gemma", # noqa: E501
-                                                      v0_only=True,
-                                                      hf_overrides={"architectures": ["GemmaForSequenceClassification"], # noqa: E501
-                                                                    "classifier_from_token": ["Yes"], # noqa: E501
-                                                                    "method": "no_post_processing"}), # noqa: E501
-    "LlamaForSequenceClassification": _HfExamplesInfo("Skywork/Skywork-Reward-V2-Llama-3.2-1B"), # noqa: E501
     "ModernBertForSequenceClassification": _HfExamplesInfo("Alibaba-NLP/gte-reranker-modernbert-base", v0_only=True), # noqa: E501
     "RobertaForSequenceClassification": _HfExamplesInfo("cross-encoder/quora-roberta-base", v0_only=True),  # noqa: E501
     "XLMRobertaForSequenceClassification": _HfExamplesInfo("BAAI/bge-reranker-v2-m3", v0_only=True),  # noqa: E501
 }
 
+_AUTOMATIC_CONVERTED_MODELS = {
+    # Use as_seq_cls_model for automatic conversion
+    "GemmaForSequenceClassification": _HfExamplesInfo("BAAI/bge-reranker-v2-gemma",  # noqa: E501
+                                                      v0_only=True,
+                                                      hf_overrides={"architectures": ["GemmaForSequenceClassification"], # noqa: E501
+                                                                    "classifier_from_token": ["Yes"],  # noqa: E501
+                                                                    "method": "no_post_processing"}),  # noqa: E501
+    "LlamaForSequenceClassification": _HfExamplesInfo("Skywork/Skywork-Reward-V2-Llama-3.2-1B"),  # noqa: E501
+    "Qwen2ForSequenceClassification": _HfExamplesInfo("jason9693/Qwen2.5-1.5B-apeach"),  # noqa: E501
+    "Qwen3ForSequenceClassification": _HfExamplesInfo("tomaarsen/Qwen3-Reranker-0.6B-seq-cls"),  # noqa: E501
+}
+
 _MULTIMODAL_EXAMPLE_MODELS = {
     # [Decoder-only]
     "AriaForConditionalGeneration": _HfExamplesInfo("rhymes-ai/Aria"),
@@ -449,6 +455,7 @@ def check_available_online(
     "JinaVLForRanking": _HfExamplesInfo("jinaai/jina-reranker-m0"),   # noqa: E501
 }
 
+
 _SPECULATIVE_DECODING_EXAMPLE_MODELS = {
     "EAGLEModel": _HfExamplesInfo("JackFram/llama-68m",
                                   speculative_model="abhigoyal/vllm-eagle-llama-68m-random"),  # noqa: E501
@@ -489,7 +496,7 @@ def check_available_online(
 _EXAMPLE_MODELS = {
     **_TEXT_GENERATION_EXAMPLE_MODELS,
     **_EMBEDDING_EXAMPLE_MODELS,
-    **_CROSS_ENCODER_EXAMPLE_MODELS,
+    **_SEQUENCE_CLASSIFICATION_EXAMPLE_MODELS,
     **_MULTIMODAL_EXAMPLE_MODELS,
     **_SPECULATIVE_DECODING_EXAMPLE_MODELS,
     **_TRANSFORMERS_MODELS,
@@ -522,3 +529,4 @@ def find_hf_info(self, model_id: str) -> _HfExamplesInfo:
 
 
 HF_EXAMPLE_MODELS = HfExampleModels(_EXAMPLE_MODELS)
+AUTO_EXAMPLE_MODELS = HfExampleModels(_AUTOMATIC_CONVERTED_MODELS)

tests/models/test_initialization.py

@@ -13,20 +13,21 @@
 from vllm.v1.engine.core import EngineCore as V1EngineCore
 
 from ..utils import create_new_process_for_each_test
-from .registry import HF_EXAMPLE_MODELS
+from .registry import AUTO_EXAMPLE_MODELS, HF_EXAMPLE_MODELS, HfExampleModels
 
 
-@pytest.mark.parametrize("model_arch", HF_EXAMPLE_MODELS.get_supported_archs())
 @create_new_process_for_each_test()
-def test_can_initialize(model_arch: str, monkeypatch: pytest.MonkeyPatch):
-    """The reason for using create_new_process_for_each_test is to avoid 
-    the WARNING: 
-        "We must use the 'spawn' multiprocessing start method. Overriding 
+def can_initialize(model_arch: str, monkeypatch: pytest.MonkeyPatch,
+                   EXAMPLE_MODELS: HfExampleModels):
+    """The reason for using create_new_process_for_each_test is to avoid
+    the WARNING:
+        "We must use the 'spawn' multiprocessing start method. Overriding
         VLLM_WORKER_MULTIPROC_METHOD to 'spawn'."
-    The spawn process causes the _initialize_kv_caches_v1 function below to 
+    The spawn process causes the _initialize_kv_caches_v1 function below to
     become ineffective.
     """
-    model_info = HF_EXAMPLE_MODELS.get_hf_info(model_arch)
+
+    model_info = EXAMPLE_MODELS.get_hf_info(model_arch)
     model_info.check_available_online(on_fail="skip")
     model_info.check_transformers_version(on_fail="skip")
 
@@ -127,3 +128,15 @@ def _initialize_kv_caches_v1(self, vllm_config):
             load_format="dummy",
             hf_overrides=hf_overrides,
         )
+
+
+@pytest.mark.parametrize("model_arch", HF_EXAMPLE_MODELS.get_supported_archs())
+def test_can_initialize(model_arch: str, monkeypatch: pytest.MonkeyPatch):
+    can_initialize(model_arch, monkeypatch, HF_EXAMPLE_MODELS)
+
+
+@pytest.mark.parametrize("model_arch",
+                         AUTO_EXAMPLE_MODELS.get_supported_archs())
+def test_implicit_converted_models(model_arch: str,
+                                   monkeypatch: pytest.MonkeyPatch):
+    can_initialize(model_arch, monkeypatch, AUTO_EXAMPLE_MODELS)

tests/models/test_transformers.py

@@ -138,3 +138,38 @@ def test_quantization(
         name_0="transformers",
         name_1="vllm",
     )
+
+
+@pytest.mark.parametrize(
+    "model",
+    ["jason9693/Qwen2.5-1.5B-apeach"],
+)
+@pytest.mark.parametrize("dtype", ["half"])
+def test_classify(
+    hf_runner,
+    vllm_runner,
+    example_prompts,
+    model: str,
+    dtype: str,
+    monkeypatch,
+) -> None:
+    import torch
+    from transformers import AutoModelForSequenceClassification
+
+    with vllm_runner(model,
+                     max_model_len=512,
+                     dtype=dtype,
+                     model_impl="transformers") as vllm_model:
+        vllm_outputs = vllm_model.classify(example_prompts)
+
+    with hf_runner(model,
+                   dtype=dtype,
+                   auto_cls=AutoModelForSequenceClassification) as hf_model:
+        hf_outputs = hf_model.classify(example_prompts)
+
+    for hf_output, vllm_output in zip(hf_outputs, vllm_outputs):
+        hf_output = torch.tensor(hf_output)
+        vllm_output = torch.tensor(vllm_output)
+
+        assert torch.allclose(hf_output, vllm_output,
+                              1e-3 if dtype == "float" else 1e-2)

vllm/config.py

@@ -551,7 +551,7 @@ def __post_init__(self) -> None:
         # For pooling models, self.task is used to indicate the
         # user-selected task
         if self.task == "score":
-            if self.registry.is_cross_encoder_model(self.architectures):
+            if self._is_classify_task(self.architectures):
                 self.task = "classify"
             else:
                 self.task = "embed"
@@ -806,21 +806,24 @@ def _verify_tokenizer_mode(self) -> None:
                 f"one of {get_args(TokenizerMode)}.")
         self.tokenizer_mode = tokenizer_mode
 
+    def _is_classify_task(self, architectures: list[str]):
+        for arch in architectures:
+            if arch.endswith("ForSequenceClassification"):
+                return True
+        return self.registry.is_cross_encoder_model(architectures)
+
     def _get_preferred_pooling_task(
         self,
         architectures: list[str],
     ) -> _ResolvedTask:
         model_id = self.model
         if get_pooling_config(model_id, self.revision):
             return "embed"
-        if self.registry.is_cross_encoder_model(architectures):
-            return "classify"
         if self.registry.is_transcription_model(architectures):
             return "transcription"
 
         suffix_to_preferred_task: list[tuple[str, _ResolvedTask]] = [
             # Other models follow this pattern
-            ("ForSequenceClassification", "classify"),
             ("EmbeddingModel", "embed"),
             ("RewardModel", "reward"),
         ]
@@ -878,11 +881,14 @@ def _get_supported_tasks(
         self,
         task_option: TaskOption,
     ) -> dict[RunnerType, list[_ResolvedTask]]:
-        return {
-            "generate": self._get_supported_generation_tasks(task_option),
-            "pooling": self._get_supported_pooling_tasks(task_option),
-            "draft": ["draft"]
-        }
+        if self._is_classify_task(self.architectures):
+            return {"generate": [], "pooling": ["classify"], "draft": []}
+        else:
+            return {
+                "generate": self._get_supported_generation_tasks(task_option),
+                "pooling": self._get_supported_pooling_tasks(task_option),
+                "draft": ["draft"]
+            }
 
     def _get_supported_runner_types(
         self,
@@ -925,12 +931,16 @@ def _resolve_runner(
                     f"Available tasks for runner={task_runner!r}: "
                     f"{supported_tasks[task_runner]}")
 
+        if "classify" in supported_tasks.get("pooling", []):
+            # When multiple pooling tasks are present, default to
+            # pooling (eg cross-encoder) for non-standard architectures.
+            return "pooling"
+
         suffix_to_preferred_runner: list[tuple[str, RunnerType]] = [
             ("ForCausalLM", "generate"),
             ("ForConditionalGeneration", "generate"),
             ("ChatModel", "generate"),
             ("LMHeadModel", "generate"),
-            ("ForSequenceClassification", "pooling"),
             ("EmbeddingModel", "pooling"),
             ("RewardModel", "pooling"),
         ]
@@ -940,10 +950,6 @@ def _resolve_runner(
             if arch.endswith(suffix) and pref_runner in supported_runner_types:
                 return pref_runner
 
-        if "classify" in supported_tasks.get("pooling", []):
-            # When multiple pooling tasks are present, default to
-            # pooling (eg cross-encoder) for non-standard architectures.
-            return "pooling"
         if "generate" in supported_runner_types:
             return "generate"
         if "pooling" in supported_runner_types:
@@ -1525,7 +1531,7 @@ def is_v1_compatible(self) -> bool:
 
     @property
     def is_matryoshka(self) -> bool:
-        return (hasattr(self.hf_config, "matryoshka_dimensions")
+        return (bool(getattr(self.hf_config, "matryoshka_dimensions", None))
                 or getattr(self.hf_config, "is_matryoshka", False))
 
     @property
@@ -1539,13 +1545,11 @@ def use_pad_token(self) -> bool:
         return getattr(self.hf_config, "use_pad_token", True)
 
     def get_and_verify_max_len(self, max_model_len: int):
-        # For pooling models, the tokenizer's `model_max_length` is often a
-        # reliable source for the maximum sequence length. However, for
-        # generative models, this can be incorrect and unduly limit the
-        # context window (e.g., DeepSeek-R1). Therefore, we only consider
-        # tokenizer_config for pooling models.
+        # Consider max_model_len in tokenizer_config only when
+        # pooling models use absolute position_embedding.
         tokenizer_config = None
-        if self.runner_type == "pooling":
+        if (self.runner_type == "pooling" and getattr(
+                self.hf_config, "position_embedding_type", "") == "absolute"):
             tokenizer_config = try_get_tokenizer_config(
                 self.tokenizer,
                 trust_remote_code=self.trust_remote_code,

vllm/model_executor/model_loader/utils.py

@@ -22,7 +22,8 @@
     QuantizationConfig, QuantizeMethodBase)
 from vllm.model_executor.models import ModelRegistry
 from vllm.model_executor.models.adapters import (as_embedding_model,
-                                                 as_reward_model)
+                                                 as_reward_model,
+                                                 as_seq_cls_model)
 from vllm.model_executor.models.interfaces import SupportsQuant
 from vllm.utils import is_pin_memory_available
 
@@ -238,22 +239,43 @@ def get_model_architecture(
     vllm_supported_archs = ModelRegistry.get_supported_archs()
     vllm_not_supported = not any(arch in vllm_supported_archs
                                  for arch in architectures)
+
+    if vllm_not_supported:
+        # try automatic conversion in adapters.py
+        for arch in architectures:
+            if not arch.endswith("ForSequenceClassification"):
+                continue
+
+            assert model_config.task == "classify"
+            causal_lm_arch = arch.replace("ForSequenceClassification",
+                                          "ForCausalLM")
+            causal_lm_arch_vllm_supported = (causal_lm_arch
+                                             in vllm_supported_archs)
+            if not causal_lm_arch_vllm_supported:
+                continue
+
+            architectures = [causal_lm_arch]
+            vllm_not_supported = False
+            break
+
     if (model_config.model_impl == ModelImpl.TRANSFORMERS or
             model_config.model_impl != ModelImpl.VLLM and vllm_not_supported):
         architectures = resolve_transformers_arch(model_config, architectures)
+        logger.debug_once("Resolve transformers arch %s", str(architectures))
     elif (model_config.quantization is not None
           and model_config.quantization not in mixtral_supported
           and "MixtralForCausalLM" in architectures):
         architectures = ["QuantMixtralForCausalLM"]
 
     model_cls, arch = ModelRegistry.resolve_model_cls(architectures)
     if model_config.task == "embed":
+        logger.debug_once("Automatic conversion using `as_embedding_model`.")
         model_cls = as_embedding_model(model_cls)
     elif model_config.task == "classify":
-        # Cannot automatically run as_seq_cls_model,
-        # otherwise it will cause a circular reference on is_cross_encoder_model
-        pass
+        logger.debug_once("Automatic conversion using `as_seq_cls_model`.")
+        model_cls = as_seq_cls_model(model_cls)
     elif model_config.task == "reward":
+        logger.debug_once("Automatic conversion using `as_reward_model`.")
         model_cls = as_reward_model(model_cls)
 
     return model_cls, arch

vllm/model_executor/models/adapters.py

@@ -331,13 +331,13 @@ def load_weights_using_from_2_way_softmax(
 
     false_id = tokenizer.convert_tokens_to_ids(tokens[0])
     true_id = tokenizer.convert_tokens_to_ids(tokens[1])
-    weight = model.lm_head.weight.data[[true_id]].to(
+    score_weight = model.lm_head.weight.data[[true_id]].to(
         torch.float32) - model.lm_head.weight.data[[false_id]].to(
             torch.float32)
 
     param = model.score.weight
     weight_loader = getattr(param, "weight_loader", default_weight_loader)
-    weight_loader(param, weight)
+    weight_loader(param, score_weight)
 
     del model.lm_head
     loaded_weights.add("score.weight")
@@ -350,15 +350,15 @@ def load_weights_no_post_processing(model,
                                                             torch.Tensor]]):
     from vllm.model_executor.layers.vocab_parallel_embedding import (
         ParallelLMHead)
+    from vllm.model_executor.model_loader.weight_utils import (
+        default_weight_loader)
     from vllm.model_executor.models.utils import AutoWeightsLoader
 
     model_config = model.vllm_config.model_config
     tokens = getattr(model.config, "classifier_from_token", [])
     tokens = cast(list[int], tokens)
     assert len(tokens) > 0
 
-    device = model.score.weight.device
-
     if model.config.tie_word_embeddings:
         model.lm_head = model.model.embed_tokens
     else:
@@ -376,8 +376,11 @@ def load_weights_no_post_processing(model,
                               trust_remote_code=model_config.trust_remote_code)
 
     token_ids = [tokenizer.convert_tokens_to_ids(t) for t in tokens]
-    score_weight = model.lm_head.weight.data[token_ids].to(device)
-    model.score.weight.data.copy_(score_weight)
+    score_weight = model.lm_head.weight.data[token_ids]
+
+    param = model.score.weight
+    weight_loader = getattr(param, "weight_loader", default_weight_loader)
+    weight_loader(param, score_weight)
 
     del model.lm_head
     loaded_weights.add("score.weight")