openpi
openpi copied to clipboard
Physical-Intelligence
I try to train pi0_fast on my own dataset. However, I find that when the token_str is '\uddcb', it raised the error: *** TypeError: TextEncodeInput must be Union[TextInputSequence, Tuple[InputSequence, InputSequence]]....
with the code you shared, I fine-tuned and tested it, and found that the arm follows the intended trajectory well, but the gripper repeatedly opens and closes. we have modified...
First, thank you for making this incredible work open source! I'm currently working on bimanual ARX R5, and I'm building my inference environment following the `aloha_real` example. I noticed that...
Hello, To facilitate including openpi into various simulation frameworks as a supported model for training / inference, it would be very helpful if you could include the `train.py` and `compute_norm_stats.py`...
The paper mentions that the number of heads is set to **18**. However, in the code it seems to be **8**, not 18.
When I run "python scripts/train_pytorch.py pi05_full_droid_finetune", I got the error of "PyTorch RLDS data loader is not supported yet" TrainConfig( # This config is for fine-tuning pi05 on the *full*...
Hi! Awesome respository, very neatly organized. I have a question though, in the paper it is mentioned you did experiments training the model with the action expert trained from scratch....
Hi all, When testing with the following script, I found that the output of pi05’s weight before and after conversion is not the same given identical input. May I please...
Currently, by setting a custom dataset storage path in the repo_id of TrainConfig, what if multiple sets of dataset paths are configured for pi05 training? Look forward to your reply！
TrainConfig( name="pi0_aloha_pen_uncap", model=pi0_config.Pi0Config(), data=LeRobotAlohaDataConfig( repo_id="physical-intelligence/aloha_pen_uncap_diverse", assets=AssetsConfig( assets_dir="gs://openpi-assets/checkpoints/pi0_base/assets", asset_id="trossen", ), default_prompt="uncap the pen", repack_transforms=_transforms.Group( inputs=[ _transforms.RepackTransform( { "images": { "cam_high": "observation.images.cam_high", "cam_left_wrist": "observation.images.cam_left_wrist", "cam_right_wrist": "observation.images.cam_right_wrist", }, "state": "observation.state", "actions": "action", }...
