Hello, and thank you for providing the source code. I encountered a RuntimeError when attempting to run the training script for the rk_diffuser. Here is the traceback:
Getting demos for task open_box
Process Process-2:
Traceback (most recent call last):
File "anaconda3/envs/hdp/lib/python3.10/multiprocessing/process.py", line 314, in _bootstrap
self.run()
File "anaconda3/envs/hdp/lib/python3.10/multiprocessing/process.py", line 108, in run
self._target(*self._args, **self._kwargs)
File "hdp/rk_diffuser/dataset/rl_bench_dataset.py", line 307, in _get_demos
demos = _get_demos_fn()
File "hdp/rk_diffuser/dataset/rl_bench_dataset.py", line 296, in _get_demos_fn
demos = env._task.get_demos(amount=self._num_episodes, live_demos=False)
File "hdp/lib/python3.10/site-packages/rlbench/task_environment.py", line 122, in get_demos
raise RuntimeError(
RuntimeError: Can't ask for a stored demo when no dataset root provided.
I tried to examine the problem, and found that it might be related to the forked yarr repo. In yarr/envs/rlbench_env.py, the dataset_root is modified to not being passed to the inner RLbench environment.
Hello, and thank you for providing the source code. I encountered a RuntimeError when attempting to run the training script for the rk_diffuser. Here is the traceback:
I tried to examine the problem, and found that it might be related to the forked yarr repo. In
yarr/envs/rlbench_env.py
, the dataset_root is modified to not being passed to the inner RLbench environment.https://github.com/Yusufma03/YARR/blob/a431536a0a863d13be7ceb5efd4a465ebf431623/yarr/envs/rlbench_env.py#L161-L166
Thank you for your attention to this matter.