[Bugfix] Fix Qwen3-VL-Reranker model loading for sequence classification (#32089)

Signed-off-by: rickychen-infinirc <ricky.chen@infinirc.com>
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>

vllm/model_executor/models/adapters.py

@@ -401,24 +401,23 @@ def load_weights_using_from_2_way_softmax(
     tokens = cast(list[int], tokens)
     assert len(tokens) == 2
 
-    model.lm_head = ParallelLMHead(
+    language_model = (
+        model.get_language_model() if hasattr(model, "get_language_model") else model
+    )
+    language_model.lm_head = ParallelLMHead(
         text_config.vocab_size, text_config.hidden_size, quant_config=quant_config
     )
     if text_config.tie_word_embeddings:
         # embed_tokens is the assumed name for input embeddings. If the model does not
         # have this attribute, we fall back to get_input_embeddings(), which is used by
         # the Transformers modeling backend.
-        text_backbone = (
-            model.get_language_model().model
-            if hasattr(model, "get_language_model")
-            else model.model
-        )
+        text_backbone = language_model.model
         embed_tokens = (
             text_backbone.embed_tokens
             if hasattr(text_backbone, "embed_tokens")
             else text_backbone.get_input_embeddings()
         )
-        model.lm_head = model.lm_head.tie_weights(embed_tokens)
+        language_model.lm_head = language_model.lm_head.tie_weights(embed_tokens)
 
     # ModelForPooling is dynamically defined inside the _create_pooling_model_cls
     # function, so we need use this hacky method to obtain it.
@@ -438,17 +437,22 @@ 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])
-    score_weight = model.lm_head.weight.data[[true_id]].to(
+    lm_head_weight = language_model.lm_head.weight
+    score_weight = lm_head_weight.data[[true_id]].to(
         torch.float32
-    ) - model.lm_head.weight.data[[false_id]].to(torch.float32)
+    ) - 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, score_weight)
 
-    del model.lm_head
+    del language_model.lm_head
     loaded_weights.add("score.weight")
-    loaded_weights.discard("lm_head.weight")
+
+    lm_head_name = "lm_head.weight"
+    if hf_to_vllm_mapper := getattr(model, "hf_to_vllm_mapper", None):
+        lm_head_name = hf_to_vllm_mapper._map_name(lm_head_name)
+    loaded_weights.discard(lm_head_name)
     return loaded_weights
 
 

vllm/model_executor/models/config.py

@@ -254,7 +254,9 @@ class Qwen3ForSequenceClassificationConfig(VerifyAndUpdateConfig):
             "Try loading the original Qwen3 Reranker?, see: "
             "https://github.com/vllm-project/vllm/tree/main/examples/pooling/score/qwen3_reranker_offline.py"
         )
-        model_config.hf_config.method = "from_2_way_softmax"
+        text_config = config.get_text_config()
+        text_config.method = "from_2_way_softmax"
+        text_config.classifier_from_token = tokens
 
 
 class Qwen3VLForSequenceClassificationConfig(Qwen3ForSequenceClassificationConfig):