Bug: fix lm head weights in Qwen models (#3777)

python/sglang/srt/models/qwen2.py

@@ -379,8 +379,6 @@ class Qwen2ForCausalLM(nn.Module):
                 continue
             if name.startswith("model.vision_tower") and name not in params_dict:
                 continue
-            if name.startswith("lm_head"):
-                continue
 
             for param_name, weight_name, shard_id in stacked_params_mapping:
                 if weight_name not in name:

python/sglang/srt/models/qwen2_rm.py

@@ -62,7 +62,11 @@ class Qwen2ForRewardModel(nn.Module):
         return EmbeddingPoolerOutput(pooled_logits)
 
     def load_weights(self, weights: Iterable[Tuple[str, torch.Tensor]]):
-        return Qwen2ForCausalLM.load_weights(self, weights)
+        # Filter out lm_head weights of Qwen2ForCausalLM
+        filtered_weights = [
+            (name, w) for name, w in weights if not name.startswith("lm_head")
+        ]
+        return Qwen2ForCausalLM.load_weights(self, filtered_weights)
 
 
 EntryClass = [