Metadata-Version: 2.4
Name: lerobot_policy_act_ttt
Version: 0.1.2
Summary: ACT policy with Test-Time Tuning (TTT) for LeRobot
Project-URL: Homepage, https://github.com/sacovo/lerobot_policy_act_ttt
Project-URL: Repository, https://github.com/sacovo/lerobot_policy_act_ttt
Author: Sandro Covo
License: MIT
Keywords: act,imitation-learning,lerobot,robotics,test-time-tuning
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Classifier: Programming Language :: Python :: 3
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Requires-Python: >=3.12
Requires-Dist: lerobot>=0.4.2
Requires-Dist: torch>=2.7.1
Requires-Dist: torchvision>=0.22.1
Description-Content-Type: text/markdown

# Test-time Training for Action Chunking Transformer

This repository uses [Test-time Training](https://yueatsprograms.github.io/ttt/home.html) to improve the performance of [Action Chunking Transfomers](https://huggingface.co/docs/lerobot/act).

The policy can be installed with `pip install lerobot_policy_act_ttt`. 

You need the latest git version of `lerobot` that has support for custom policies: https://huggingface.co/docs/lerobot/bring_your_own_policies.
