fix

modeling_internlm2.py

@@ -1218,7 +1218,7 @@ class InternLM2ForCausalLM(InternLM2PreTrainedModel):
                     return
 
                 token = self.tokenizer.decode([value[-1]], skip_special_tokens=True)
-                if token.strip() != "<eoa>":
+                if token.strip() != "[UNUSED_TOKEN_145]":
                     self.response = self.response + token
                     history = self.history + [(self.query, self.response)]
                     self.queue.put((self.response, history))