Error while running bash eval rlbench file

[rebecajohn]

Traceback (most recent call last): File "online_evaluation_rlbench/evaluate_policy.py", line 194, in var_success_rates = env.evaluate_task_on_multiple_variations( File "/3d_diffuser_actor/utils/utils_with_rlbench.py", line 464, in evaluate_task_on_multiple_variations sum(var_success_rates.values()) / ZeroDivisionError: division by zero QObject::~QObject: Timers cannot be stopped from another thread

[nickgkan]

From the error you pasted, it seems that you cannot access the test demos. Have you followed the instructions here? https://github.com/nickgkan/3d_diffuser_actor/blob/master/docs/DATA_PREPARATION_RLBENCH.md

[rebecajohn]

Got it , Done. So how much time wiil it take approx. for eval to be completed ?

[rebecajohn]

Its been more than 35 minutes , but its showing the same.

[rebecajohn]

almost 52 minutes still showing the same

[rebecajohn]

almost 52 minutes still showing the same

[nickgkan]

It seems that your program hangs there. This is not the expected behavior, the logs are printed every few seconds. I would suggest that you try to trace where it hangs. Also, I'm suspicious there is some call trying to find the data but cannot fetch it.

[rebecajohn]

why its giving retry for such a long steps after step 9 until last 24?

[nickgkan]

Probably the planner cannot reach the target pose. This happens sometimes. Can you still render the video? I think you're running it in headless mode, but video is helpful for such understanding.

[rebecajohn]

after running the script , i only find .json files in eval_logs/3d_diffuser_actor/seed0/ , but how to get video .mp4 files ?