checkpoint update

model.py

@@ -103,6 +103,7 @@ Output:'''
             out = self.llm_model.generate(
                 inputs_embeds=combined_embeds,
                 max_new_tokens=max_new_tokens,
+                pad_token_id=self.llm_tokenizer.pad_token_id
             ).cpu().tolist()[0]
 
             output_text = self.llm_tokenizer.decode(out, skip_special_tokens=True)