fix

Author: shihaobai

lightllm/models/qwen3/layer_weights/transformer_layer_weight.py

@@ -19,12 +19,6 @@
 
 class Qwen3TransformerLayerWeight(LlamaTransformerLayerWeight):
     def __init__(self, layer_num, data_type, network_config, mode=[], quant_cfg=None):
-        self.n_routed_experts = network_config["num_experts"]
-        self.is_moe = (
-            network_config["num_experts"] > 0
-            and layer_num not in network_config["mlp_only_layers"]
-            and (layer_num + 1) % network_config["decoder_sparse_step"] == 0
-        )
         super().__init__(layer_num, data_type, network_config, mode, quant_cfg)
         return
 

lightllm/server/api_models.py

@@ -76,7 +76,7 @@ class ChatCompletionRequest(BaseModel):
     ignore_eos: Optional[bool] = False
     role_settings: Optional[Dict[str, str]] = None
     character_settings: Optional[List[Dict[str, str]]] = None
-    chat_template_kwargs: Optional[Dict[str, str]] = None
+    chat_template_kwargs: Optional[Dict[str, bool]] = None
 
 
 class FunctionResponse(BaseModel):