A minor issue when training models with maze2d dataset

[zanghyu]

Hi,

Thanks for sharing the nice code! There was a minor issue when I trained the model with the script python scripts/train.py --dataset maze2d-umaze-v1 --horizon 512 --n_diffusion_steps 200:

File ".../diffuser/datasets/buffer.py", line 75, in add_path
    self._dict[key][self._count, :path_length] = array
IndexError: index 10000 is out of bounds for axis 0 with size 10000

and I believe the episode number of the maze2d dataset is more than 10000, which caused this error. So I set max_n_episodes=20000 at the line: https://github.com/jannerm/diffuser/blob/main/diffuser/datasets/sequence.py#L18. And it really fixed this issue. Maybe you can update the corresponding code if I'm correct?

BTW, looking forward to seeing your testing/evaluating code :)

[yilundu]

Hi the maze2d code should be released now -- do you still have issues?

[zanghyu]

Thanks for the reply! But it seems the error still occurs...

[jannerm]

Thanks for reporting. Could you check that you're on the maze2d branch and using the default maze2d config?

git checkout maze2d
python scripts/train.py --config config.maze2d --dataset maze2d-large-v1

If that doesn't solve it, could you send the full stack trace?

[zanghyu]

Here's the stack trace:

Traceback (most recent call last):
  File "scripts/train.py", line 36, in <module>
    dataset = dataset_config()
  File "/data/hyzang/project/rl/representation/diffuser-main/maze2d/diffuser-maze2d/diffuser/utils/config.py", line 65, in __call__
    instance = self._class(*args, **kwargs, **self._dict)
  File "/data/hyzang/project/rl/representation/diffuser-main/maze2d/diffuser-maze2d/diffuser/datasets/sequence.py", line 28, in __init__
    fields.add_path(episode)
  File "/data/hyzang/project/rl/representation/diffuser-main/maze2d/diffuser-maze2d/diffuser/datasets/buffer.py", line 75, in add_path
    self._dict[key][self._count, :path_length] = array
IndexError: index 10000 is out of bounds for axis 0 with size 10000

The default setting of the max_n_episodes being small could still be the issue?

[jannerm]

Could you include everything printed up until that point by the script? I think you're right that the max_n_episodes is what's causing the error locally, but I'm trying to figure out why you're getting the error when I don't, and looking at the printed configs might help me figure that out.

[zanghyu]

Sorry for the late reply, I have re-run the code today, and found it works! Thanks very much!