avoid duplicate generate args

modeling_minicpmo.py

@@ -636,6 +636,7 @@ class MiniCPMO(MiniCPMOPreTrainedModel):
         return self.llm(input_ids=None, position_ids=position_ids, inputs_embeds=vllm_embedding, **kwargs)
 
     def _decode(self, inputs_embeds, tokenizer, attention_mask, **kwargs):
+        kwargs.pop("output_hidden_states", None)
         terminators = [tokenizer.convert_tokens_to_ids(i) for i in self.terminators]
         outputs = self.llm.generate(
             inputs_embeds=inputs_embeds,
@@ -649,7 +650,6 @@ class MiniCPMO(MiniCPMOPreTrainedModel):
         return outputs
 
     def _decode_stream(self, inputs_embeds, tokenizer, **kwargs):
-        kwargs.pop("output_hidden_states", None)
         terminators = [tokenizer.convert_tokens_to_ids(i) for i in self.terminators]
         streamer = TextIteratorStreamer(tokenizer=tokenizer)
         generation_kwargs = {