You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I encountered an error while running evaluate.py in TaskBench: the data["task_nodes"] key does not exist. Upon inspecting the corresponding data file, I confirmed that the task_nodes field is indeed missing. Based on logical reasoning, it seems that task_nodes might correspond to sampled_nodes and tool_nodes, but I could not find any explanation or distinction between sampled_nodes and tool_nodes in the documentation or the paper. Could you please clarify these differences and confirm whether my assumption is correct?
The text was updated successfully, but these errors were encountered:
I encountered an error while running evaluate.py in TaskBench: the data["task_nodes"] key does not exist. Upon inspecting the corresponding data file, I confirmed that the task_nodes field is indeed missing. Based on logical reasoning, it seems that task_nodes might correspond to sampled_nodes and tool_nodes, but I could not find any explanation or distinction between sampled_nodes and tool_nodes in the documentation or the paper. Could you please clarify these differences and confirm whether my assumption is correct?
The text was updated successfully, but these errors were encountered: