Update main.py

main.py

@@ -35,8 +35,8 @@ async def chat(request: ChatCompletionRequest, response_mode=None):
     tokens = llm.tokenize(request.prompt)
     async def server_sent_events(chat_chunks, llm):
         for chat_chunk in llm.generate(chat_chunks):
-            yield llm.detokenize(chat_chunk)
-        yield ""
+            yield dict(data=json.dumps(llm.detokenize(chat_chunk))
+        yield dict(data="[DONE]")
 
     return EventSourceResponse(server_sent_events(tokens, llm))