Instructions to use jaheroth/act_pusht_bs64_dec7_seed1002_160k with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- LeRobot
How to use jaheroth/act_pusht_bs64_dec7_seed1002_160k with LeRobot:
- Notebooks
- Google Colab
- Kaggle
act_pusht_bs64_dec7_seed1002_160k
Best tuned ACT on PushT: 164.4 avg_sum_imputed_reward, 66.7% success (n=5000 seeds). LeRobot ACT, batch 64, lr 2e-5, 7 decoder layers, chunk 100, kl_weight 10, 160k steps.
Part of a 6-week robot-learning training block (https://github.com/JaHeRoth/robot-learning). Evaluated on gym-pusht at n_action_steps=16; avg_sum_imputed_reward imputes 0.95/step to horizon 300 after success.
- Downloads last month
- -