报错Network Error: Request Failed

[DonkeFly]

PS E:\X-PLUG\PCAgent\MyPcAgent> python run.py --instruction="打开 “此电脑”" --mac 0
2026-03-02 17:30:50,933 - modelscope - INFO - PyTorch version 2.10.0 Found.
2026-03-02 17:30:50,934 - modelscope - INFO - Loading ast index from C:\Users\Administrator.cache\modelscope\ast_indexer
2026-03-02 17:30:51,125 - modelscope - INFO - Loading done! Current index file version is 1.15.0, with md5 c4022a2bbc7c06b7df540266ce6f1e75 and a total number of 980 components indexed
打开 此电脑
task_1/screenshot_1.png
Network Error:
Request Failed
Network Error:
Request Failed
Network Error:
Request Failed
Network Error:
Request Failed
Network Error:
Request Failed
Traceback (most recent call last):
File "E:\X-PLUG\PCAgent\MyPcAgent\run.py", line 661, in
output_action = inference_chat(chat_action, vl_model_version, API_url, token)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "E:\X-PLUG\PCAgent\MyPcAgent\PCAgent\api.py", line 62, in inference_chat
print("completion"+completion)
^^^^^^^^^^
UnboundLocalError: cannot access local variable 'completion' where it is not associated with a value

纯小白，跪求作者大大回复

[JiwaniZakir]

The root cause is the repeated Network Error: Request Failed — the API calls inside inference_chat (PCAgent/api.py) are failing before completion is ever assigned a value, so the print("completion"+completion) at line 62 crashes with UnboundLocalError because the variable only gets set on a successful response. Check that API_url and token in your configuration are correct and reachable from your machine; the five consecutive failures suggest the endpoint is either misconfigured, requires a proxy, or the token is invalid/expired. You can verify by making a raw curl or requests.post call to API_url with your token outside of the agent to confirm connectivity.