Update main.py

main.py

@@ -277,14 +277,15 @@ async def _stream_sse(text: str, req: ChatCompletionRequest) -> AsyncGenerator[s
         yield f"data: {json.dumps(chunk)}\n\n"
         await asyncio.sleep(0.01)
 
-    yield f"data: {json.dumps({
-        'id': cid,
-        'object': 'chat.completion.chunk',
-        'created': created,
-        'model': req.model,
-        'choices': [{'index': 0, 'delta': {}, 'finish_reason': 'stop'}],
-    })}\n\n"
-
+    # final chunk
+    final_chunk = {
+        "id": cid,
+        "object": "chat.completion.chunk",
+        "created": created,
+        "model": req.model,
+        "choices": [{"index": 0, "delta": {}, "finish_reason": "stop"}],
+    }
+    yield f"data: {json.dumps(final_chunk)}\n\n"
     yield "data: [DONE]\n\n"