Robotics
LeRobot
Safetensors
English
smolvla
vla
vision-language-action
edge-ai
manipulation
fastvit
jetson
edge-deployment
Eval Results (legacy)
Instructions to use enfuse/edgevla-tiny-fmb with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- LeRobot
How to use enfuse/edgevla-tiny-fmb with LeRobot:
# 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]
# Launch finetuning on your dataset python lerobot/scripts/train.py \ --policy.path=enfuse/edgevla-tiny-fmb \ --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=enfuse/edgevla-tiny-fmb - Notebooks
- Google Colab
- Kaggle
Download edgevla_metadata.json from enfuse/edgevla-tiny-fmb: direct link, hf CLI and curl.
- Browser
- Download file 451 Bytes
-
https://huggingface.co/enfuse/edgevla-tiny-fmb/resolve/main/edgevla_metadata.json
- Command line
-
hf download hf://enfuse/edgevla-tiny-fmb/edgevla_metadata.json
-
curl -L -o edgevla_metadata.json https://huggingface.co/enfuse/edgevla-tiny-fmb/resolve/main/edgevla_metadata.json
451 Bytes
| { | |
| "model_name": "EdgeVLA-Tiny", | |
| "total_params_M": 164, | |
| "trainable_params_M": 30, | |
| "num_vlm_layers": 4, | |
| "num_expert_layers": -1, | |
| "expert_width_multiplier": 0.75, | |
| "fastvit_variant": "fastvit_t8", | |
| "fastvit_input_size": 256, | |
| "dataset": "lerobot/fmb", | |
| "best_round": "r5", | |
| "total_training_steps": 100000, | |
| "best_checkpoint": "outputs/r5_tiny_4L_t8/checkpoint-50000/checkpoint.pt", | |
| "github_repo": "https://github.com/enfuse/edgevla" | |
| } |