smolvla_task1_5blocks_v3_100ep_b64_50k_v1

Model Card for smolvla_task1_5blocks_v3_100ep_b64_50k_v1

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.

Learn how to train and run it in the LeRobot smolvla guide, or browse the full documentation.

Model Details

• License: apache-2.0
• Fine-tuned from: lerobot/smolvla_base
• Robot type: so_follower (SO-101)
• Cameras: top, wrist
• PEFT Method: LoRA (rank=64, alpha=64)

Inputs & Outputs

The policy consumes these observation features and produces these action features.

Inputs

Outputs

Training Dataset

• Repository: eslab1234/task1_hybrid_5blocks_v3_100ep_merged
• Episodes: 100
• Frames: 138,387
• Frame rate: 30 FPS
• Task: "Pick and place 5 blocks in sequence (red, yellow, wood, green, blue)."

Training Configuration

How to Get Started with the Model

Run the policy on your robot with async policy server or rollout:

# On GPU Policy Server
bash project/scripts/gpu/run_smolvla_red_policy_server.sh

# On SO-101 Robot Client
# MODEL_PATH=eslab1234/smolvla_task1_5blocks_v3_100ep_b64_50k_v1
bash project/scripts/robot/run_smolvla_red_observe_inference.sh

Citation

@misc{cadene2024lerobot,
    author = {Cadene, Remi and Alibert, Simon and Soare, Alexander and Gallouedec, Quentin and Zouitine, Adil and Palma, Steven and Kooijmans, Pepijn and Aractingi, Michel and Shukor, Mustafa and Aubakirova, Dana and Russi, Martino and Capuano, Francesco and Pascal, Caroline and Choghari, Jade and Moss, Jess and Wolf, Thomas},
    title = {LeRobot: State-of-the-art Machine Learning for Real-World Robotics in Pytorch},
    howpublished = {\url{https://github.com/huggingface/lerobot}},
    year = {2024}
}