SmolVLA is a compact, efficient vision-language-action model that achieves competitive performance at reduced computational costs and can be deployed on consumer-grade hardware.
This policy has been trained and pushed to the Hub using
LeRobot.
See the full documentation at
LeRobot Docs.
For a complete walkthrough, see the
training guide.
Below is the short version on how to train and run inference/eval:
1lerobot-train \
2 --dataset.repo_id=${HF_USER}/<dataset> \
3 --policy.type=act \
4 --output_dir=outputs/train/<desired_policy_repo_id> \
5 --job_name=lerobot_training \
6 --policy.device=cuda \
7 --policy.repo_id=${HF_USER}/<desired_policy_repo_id>
8 --wandb.enable=true
1lerobot-record \
2 --robot.type=so100_follower \
3 --dataset.repo_id=<hf_user>/eval_<dataset> \
4 --policy.path=<hf_user>/<desired_policy_repo_id> \
5 --episodes=10
