Fix initializing GGUF weights for ColumnParallelLinear when using tensor parallel > 1 (#13023)

Author: SzymonOzog

vllm/model_executor/layers/linear.py

@@ -335,6 +335,12 @@ def weight_loader(self, param: Parameter, loaded_weight: torch.Tensor):
         tp_rank = get_tensor_model_parallel_rank()
         output_dim = getattr(param, "output_dim", None)
 
+        is_sharded_weight = getattr(param, "is_sharded_weight", False)
+        use_bitsandbytes_4bit = getattr(param, "use_bitsandbytes_4bit", False)
+        # bitsandbytes loads the weights of the specific portion
+        # no need to narrow
+        is_sharded_weight = is_sharded_weight or use_bitsandbytes_4bit
+
         # Special case for GGUF
         is_gguf_weight = getattr(param, "is_gguf_weight", False)
         is_gguf_weight_type = getattr(param, "is_gguf_weight_type", False)
@@ -343,13 +349,12 @@ def weight_loader(self, param: Parameter, loaded_weight: torch.Tensor):
 
         # Materialize GGUF UninitializedParameter
         if is_gguf_weight and isinstance(param, UninitializedParameter):
-            param.materialize(loaded_weight.shape, dtype=loaded_weight.dtype)
-
-        use_bitsandbytes_4bit = getattr(param, "use_bitsandbytes_4bit", False)
-        is_sharded_weight = getattr(param, "is_sharded_weight", False)
-        # bitsandbytes loads the weights of the specific portion
-        # no need to narrow
-        is_sharded_weight = is_sharded_weight or use_bitsandbytes_4bit
+            final_shape = list(loaded_weight.shape)
+            if output_dim is not None:
+                tp_size = get_tensor_model_parallel_world_size()
+                assert final_shape[output_dim] % tp_size == 0
+                final_shape[output_dim] = final_shape[output_dim] // tp_size
+            param.materialize(final_shape, dtype=loaded_weight.dtype)
 
         param_data = param.data
         if output_dim is not None and not is_sharded_weight: