debug

conversation_public.py

@@ -212,16 +212,14 @@ class Conversation:
                         ret_msg += audio_str
                     else:
                         ret_msg += msg.replace('<image>', '').replace(tts_format, '').strip()
-                    ret.append([ret_msg, None])
+                    ret.append({"role": "user", "content": ret_msg})
                 else:
-                    ret.append([msg, None])
+                    ret.append({"role": "user", "content": msg})
             else:
                 if type(msg) is tuple:
                     audio_b64_str = base64.b64encode(open(msg[1], "rb").read()).decode("utf-8")
                     msg = f'<audio src="data:audio/wav;base64,{audio_b64_str}" controls autoplay></audio>'
-                if ret[-1][-1] is None:
-                    ret[-1][-1] = ()
-                ret[-1][-1] += (msg, )
+                ret.append({"role": "assistant", "content": msg})
         return ret
 
     def copy(self):