Closed dgjun32 closed 2 months ago
I've encountered this error sporadically in the past. Usually, this happens because of network latency which causes the page to load slowly. Is it possible that your connection is slow or that you are running a large number of parallel evaluations on the same instance? Parallel evaluation works, but if you start running into latency issues, try to reduce the number of concurrent jobs.
I did not run multiple evaluations, so I think the network connection might be the problem.
Could you please try again and make sure the internet connection is fast (e.g., with a speedtest)? Let me know if the issue persists. Thanks!
Is there any speediest criteria for running this environment? It seems connection speed is around 8MB / second!
That sounds like decent speed. What we typically do when we get timeout errors is to restart the episode. These are inevitable since the benchmark is remote-hosted, but they tend to occur very rarely. To help me understand what's going on, could you please share the URL and password of the problematic instance, and also report the output of the ping command for the instance URL?
Thanks!
I am experimenting with gpt-4o agent, but I encountered error when calling env.step().
This is the 6th episode of single-chart-value-retrieval task. It seems there is some problem in task.validate() function. Have you ever seen this error?