Update app.py

app.py

@@ -10,7 +10,7 @@ from starlette.responses import StreamingResponse, JSONResponse
 """
 For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
 """
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
+client = InferenceClient("microsoft/Phi-3-mini-4k-instruct")
 
 
 def respond(
@@ -76,7 +76,7 @@ def chat_get(message: str = "", system_message: str = "You are a friendly Chatbo
     predict_response = requests.post('http://localhost:7860/call/chat', json={'data': [message, system_message, max_tokens, temperature, top_p]}).json()
     if "event_id" not in predict_response:
         return predict_response
-    return JSONResponse(json.loads(requests.get(f'http://localhost:7860/call/chat/{predict_response["event_id"]}').text.split("data: ", 1)[-1]))
+    return JSONResponse(json.loads(requests.get(f'http://localhost:7860/call/chat/{predict_response["event_id"]}').text.rsplit("\n\nevent: error", 1)[0].split("data: ", 1)[-1]))
 
 if __name__ == "__main__":
     app = gr.mount_gradio_app(app, demo, path="/")