fix Qwen3ForCausalLM has no attribute

Dockerfile

@@ -42,7 +42,7 @@ USER user
 RUN pip install --no-cache-dir --user -r requirements.txt
 
 # Install nano-vllm with --no-deps since all dependencies are already installed
-RUN pip install --no-cache-dir --user --no-deps ./acestep/third_parts/nano-vllm
+RUN pip install ./acestep/third_parts/nano-vllm
 
 # Copy the rest of the application
 COPY --chown=user:user . .

acestep/third_parts/nano-vllm/nanovllm/utils/loader.py

@@ -11,18 +11,54 @@ def default_weight_loader(param: nn.Parameter, loaded_weight: torch.Tensor):
 
 def load_model(model: nn.Module, path: str):
     packed_modules_mapping = getattr(model, "packed_modules_mapping", {})
-    for file in glob(os.path.join(path, "*.safetensors")):
+    
+    # Collect all weight names for error reporting
+    all_weight_names = []
+    safetensor_files = glob(os.path.join(path, "*.safetensors"))
+    
+    if not safetensor_files:
+        raise FileNotFoundError(f"No .safetensors files found in {path}")
+    
+    for file in safetensor_files:
+        with safe_open(file, "pt", "cpu") as f:
+            all_weight_names.extend(f.keys())
+    
+    # Get model's available parameters for error reporting
+    model_params = dict(model.named_parameters())
+    
+    for file in safetensor_files:
         with safe_open(file, "pt", "cpu") as f:
             for weight_name in f.keys():
-                for k in packed_modules_mapping:
-                    if k in weight_name:
-                        v, shard_id = packed_modules_mapping[k]
-                        param_name = weight_name.replace(k, v)
-                        param = model.get_parameter(param_name)
-                        weight_loader = getattr(param, "weight_loader")
-                        weight_loader(param, f.get_tensor(weight_name), shard_id)
-                        break
-                else:
-                    param = model.get_parameter(weight_name)
-                    weight_loader = getattr(param, "weight_loader", default_weight_loader)
-                    weight_loader(param, f.get_tensor(weight_name))
+                try:
+                    for k in packed_modules_mapping:
+                        if k in weight_name:
+                            v, shard_id = packed_modules_mapping[k]
+                            param_name = weight_name.replace(k, v)
+                            param = model.get_parameter(param_name)
+                            weight_loader = getattr(param, "weight_loader")
+                            weight_loader(param, f.get_tensor(weight_name), shard_id)
+                            break
+                    else:
+                        param = model.get_parameter(weight_name)
+                        weight_loader = getattr(param, "weight_loader", default_weight_loader)
+                        weight_loader(param, f.get_tensor(weight_name))
+                except AttributeError as e:
+                    # Detailed error message for debugging
+                    print(f"\n{'='*60}")
+                    print(f"[nano-vllm] Weight loading error!")
+                    print(f"{'='*60}")
+                    print(f"Failed to load weight: {weight_name}")
+                    print(f"Error: {e}")
+                    print(f"\nWeight file: {file}")
+                    print(f"\n--- Weights in safetensors file (first 20) ---")
+                    for i, name in enumerate(sorted(all_weight_names)[:20]):
+                        print(f"  {name}")
+                    if len(all_weight_names) > 20:
+                        print(f"  ... and {len(all_weight_names) - 20} more")
+                    print(f"\n--- Model parameters (first 20) ---")
+                    for i, name in enumerate(sorted(model_params.keys())[:20]):
+                        print(f"  {name}")
+                    if len(model_params) > 20:
+                        print(f"  ... and {len(model_params) - 20} more")
+                    print(f"{'='*60}\n")
+                    raise