act_so101_grasp_50_v2 is an open-source robotics model implementing Action Chunking with Transformers (ACT) for imitation learning. It predicts short action chunks from teleoperated data, helping robotics researchers and developers train and evaluate robotic policies efficiently. The model is available for use via Hugging Face and integrates with tools like LeRobot.
In the Other AI space, Act So101 Grasp 50 takes a focused approach. It enables robotics researchers to train and evaluate imitation learning policies for action chunking without building models from scratch. Act So101 Grasp 50 is an open-source project aimed at robotics researchers. The project is open source (Apache-2.0). Act So101 Grasp 50 is available on the web, API, and the command line.
Behind Act So101 Grasp 50 is iptihar, and the product first shipped in 2024. The project is developed in the open on GitHub with 25.8k stars and 160 commits in the last 90 days. Across PulseGate's embedding index, Act So101 Grasp 50 has few near neighbours, marking it as relatively distinct. Among its 5 catalogued features are imitation learning, action chunking, and robotics policy.
Latest indexed changes and source events
iptihar/act_so101_grasp_50_v2 discovered by the PulseGate indexer
