[FIX] Fix the case when input_is_parallel=False for `ScaledActivati…

…on` (vllm-project#1737)

vllm/model_executor/layers/activation.py

@@ -61,6 +61,7 @@ def __init__(
     ):
         super().__init__()
         self.act = act_module
+        self.input_is_parallel = input_is_parallel
         if input_is_parallel:
             tp_size = get_tensor_model_parallel_world_size()
             intermediate_size_per_partition = divide(intermediate_size,
@@ -79,11 +80,12 @@ def forward(self, x: torch.Tensor) -> torch.Tensor:
         return self.act(x) / self.scales
 
     def weight_loader(self, param: nn.Parameter, loaded_weight: torch.Tensor):
-        tp_rank = get_tensor_model_parallel_rank()
         param_data = param.data
-        shard_size = param_data.shape[0]
-        start_idx = tp_rank * shard_size
-        loaded_weight = loaded_weight.narrow(0, start_idx, shard_size)
+        if self.input_is_parallel:
+            tp_rank = get_tensor_model_parallel_rank()
+            shard_size = param_data.shape[0]
+            start_idx = tp_rank * shard_size
+            loaded_weight = loaded_weight.narrow(0, start_idx, shard_size)
         assert param_data.shape == loaded_weight.shape
         param_data.copy_(loaded_weight)