Update modeling_rwkv5.py

modeling_rwkv5.py

@@ -789,6 +789,9 @@ class Rwkv5ForCausalLM(Rwkv5PreTrainedModel):
         # only last token for inputs_ids if the state is passed along.
         if state is not None:
             input_ids = input_ids[:, -1].unsqueeze(-1)
+        else:
+            # add in \n at the beginning
+            input_ids = torch.cat([torch.full([1,1],11,device=input_ids.device,dtype=input_ids.dtype), input_ids])
 
         # if `inputs_embeds` are passed, we only want to use them in the 1st generation step
         if inputs_embeds is not None and state is None: