Error when running train.py

[gcMs1]

Hello, when I ran train.py following the guide in Readme, I got path error:

When I changed the config_path to match the config file path, I got another error:

How could I solve this problem, Could you give me some suggestion?
Besides, Is the cyberdog_final.ckpt pubilc? I didn't find it.
Looking forward to your reply!
Thank you!

[MasterYip]

cyberdog_final.ckpt can be downloaded from required files, check out the readme.md:

Download [required files](https://osf.io/kxt9w/?view_only=8c4633eaf94e4feaa6a6c92ae37d657e) and place them in DiffuseLoco root folder.

Looks like the training dataset is still pending for publication.

Currently dataset generation is still pending.

[gcMs1]

cyberdog_final.ckpt can be downloaded from required files, check out the readme.md:

Download [required files](https://osf.io/kxt9w/?view_only=8c4633eaf94e4feaa6a6c92ae37d657e) and place them in DiffuseLoco root folder.

Looks like the training dataset is still pending for publication.

Currently dataset generation is still pending.

Thank you for your reply!

[XanderWangXY]

After I download the file, running train.py, it still shows the error in the second picture. How to solve it?

[MasterYip]

After I download the file, running train.py, it still shows the error in the second picture. How to solve it?

The file you downloaded is the final checkpoint of diffusion policy, not the dataset for training.

[MasterYip]

This probably helps:

Dataset Generation

To generate the dataset from source RL checkpoints for each task, run the following command:

source env.sh

python ./scripts/eval.py --checkpoint=./cyberdog_final.ckpt --task=<TASK_NAME>
--online=false --generate_dataset=true

Then, use scripts/combine_dataset.py to combine the generated datasets (skill_filenames should be modified).

After that, you can use scripts/train.py to train the diffusion model.