---
license: apache-2.0
datasets:
- lerobot/pusht
pipeline_tag: robotics
---
# Model Card for Transformer based Diffusion Policy / PushT

Transformer based Diffusion Policy (as per [Diffusion Policy: Visuomotor Policy
Learning via Action Diffusion](https://arxiv.org/abs/2303.04137)) trained for the `PushT` environment from [gym-pusht](https://github.com/huggingface/gym-pusht).

## How to Get Started with the Model

See the [LeRobot library](https://github.com/huggingface/lerobot) (particularly the [evaluation script](https://github.com/huggingface/lerobot/blob/main/lerobot/scripts/eval.py)) for instructions on how to load and evaluate this model.

## Training Details

The model was trained using [LeRobot's training script](https://github.com/huggingface/lerobot/blob/d747195c5733c4f68d4bfbe62632d6fc1b605712/lerobot/scripts/train.py) and with the [pusht](https://huggingface.co/datasets/lerobot/pusht/tree/v1.3) dataset, using this command:

```bash
python lerobot/scripts/train.py \
  hydra.run.dir=outputs/train/diffusion_pusht \
  hydra.job.name=diffusion_pusht \
  policy=diffusion training.save_model=true \
  env=pusht \
  env.task=PushT-v0 \
  dataset_repo_id=lerobot/pusht \
  training.save_freq=25000 \
  training.eval_freq=10000 \
  wandb.enable=true \
  device=cuda
```


The training and eval curves may be found at https://api.wandb.ai/links/none7/rd2trav7

This took about 6 hours to train on an Nvida Tesla P100 GPU.

## Evaluation

The model was evaluated on the `PushT` environment from [gym-pusht](https://github.com/huggingface/gym-pusht) and compared to a similar model trained with the original [Diffusion Policy code](https://github.com/real-stanford/diffusion_policy). There are two evaluation metrics on a per-episode basis:

- Maximum overlap with target (seen as `eval/avg_max_reward` in the charts above). This ranges in [0, 1].
- Success: whether or not the maximum overlap is at least 95%.

Here are the metrics for 500 episodes worth of evaluation. 

<blank>|Ours|Theirs
-|-|-
Average max. overlap ratio | 0.000 | 0.000
Success rate for 500 episodes (%) | 0.00 | 0.00

The results of each of the individual rollouts may be found in [eval_info.json](eval_info.json).