# Launch finetuning on your dataset
python lerobot/scripts/train.py \
--policy.path=makepluscode/ch09-02-train-eval-smolvla-20000-step \
--dataset.repo_id=lerobot/svla_so101_pickplace \
--batch_size=64 \
--steps=20000 \
--output_dir=outputs/train/my_smolvla \
--job_name=my_smolvla_training \
--policy.device=cuda \
--wandb.enable=true# Run the policy using the record function
python -m lerobot.record \
--robot.type=so101_follower \
--robot.port=/dev/ttyACM0 \ # <- Use your port
--robot.id=my_blue_follower_arm \ # <- Use your robot id
--robot.cameras="{ front: {type: opencv, index_or_path: 8, width: 640, height: 480, fps: 30}}" \ # <- Use your cameras
--dataset.single_task="Grasp a lego block and put it in the bin." \ # <- Use the same task description you used in your dataset recording
--dataset.repo_id=HF_USER/dataset_name \ # <- This will be the dataset name on HF Hub
--dataset.episode_time_s=50 \
--dataset.num_episodes=10 \
--policy.path=makepluscode/ch09-02-train-eval-smolvla-20000-stepsmolvla-20000-step
VLA λ‘보ν±μ€ μ λ¬Έμ μμ μ© SmolVLA νμΈνλ 체ν¬ν¬μΈνΈμ λλ€.
κ²½λ‘ (GitHub code/ μ λμ)
| νλͺ© | κ° |
|---|---|
| λ Όλ¦¬ κ²½λ‘ | vla-robotics/ch09/02-train-eval/smolvla-20000-step |
| GitHub | vla-robotics-examples |
| μμ ν΄λ | code/ch09/02-train-eval/ |
| Hub λͺ¨λΈ ID | makepluscode/ch09-02-train-eval-smolvla-20000-step |
| νμ΅ step | 20000 |
| λ°μ΄ν°μ | local/so_arm101_block_picking_aug200 |
| λ² μ΄μ€ | lerobot/smolvla_base |
λ‘λ
from lerobot.policies.smolvla.modeling_smolvla import SmolVLAPolicy
policy = SmolVLAPolicy.from_pretrained("makepluscode/ch09-02-train-eval-smolvla-20000-step")
μ Β·νμ²λ¦¬λ λμΌ ν΄λμ policy_preprocessor.json /
policy_postprocessor.json μ from_pretrained λ‘ ν¨κ» λ‘λν©λλ€.
- Downloads last month
- 6
# See https://github.com/huggingface/lerobot?tab=readme-ov-file#installation for more details git clone https://github.com/huggingface/lerobot.git cd lerobot pip install -e .[smolvla]