| license: mit | |
| tags: | |
| - worldkit | |
| - world-model | |
| - jepa | |
| - robotics | |
| - push-t | |
| library_name: worldkit | |
| pipeline_tag: reinforcement-learning | |
| # WorldKit / Push-T (Default) | |
| The default Push-T world model for [WorldKit](https://github.com/DilpreetBansi/worldkit). | |
| This is the `base` config (13M params, 192-D latent space). | |
| ```python | |
| from worldkit import WorldModel | |
| model = WorldModel.from_hub("DilpreetBansi/pusht") | |
| plan = model.plan(current_frame, goal_frame) | |
| ``` | |
| See [DilpreetBansi/pusht-base](https://huggingface.co/DilpreetBansi/pusht-base) for full details. | |
| Built by [Dilpreet Bansi](https://github.com/DilpreetBansi). | |