LeRobot Adds Zero-Shot Reward Models
RemiCadene · x · 2026-07-14
LeRobot v0.6 introduces a unified reward models API along with two zero-shot reward functions:
- Robometer: A general-purpose reward model trained on over 1 million trajectories across 21 robot morphologies. It can directly process video and language instructions to assess task progress and success without requiring task-specific training.
- TOPReward: A lighter alternative that directly uses the output probability of an off-the-shelf VLM to determine "True," eliminating the need for dedicated reward weights.
Both components come with annotation scripts that can write frame-by-frame progress curves back into the dataset, facilitating subsequent reward-aware behavioral training.
Related event: LeRobot v0.6 Adds Zero-Shot Reward Model API(2 posts)→
More from coding & agent
- agents-best-practices: a provider-neutral Agent Skill for designing and auditing agentic harnesses — tom_doerr · 2026-09-11
- Cognition's SWE-2 uses a KKT duality argument in RL to shift the effort Pareto curve — YouJiacheng · 2026-09-11
- First-ever Three.js Conference lands in Paris, with a panel on AI-shortened design workflows — OdinLovis · 2026-09-11
- Data engineering, not agent frameworks, is the real bottleneck for enterprise AI agents — dhruv2038 · 2026-09-11
- RTK Terminal Compression Cuts Tokens but Leaves Your AI Coding Bill Unchanged — Bartaseth · 2026-09-11
- GPT-6 Astra beats Factorio with enemies in 44 in-game hours at ~$4,500 API cost — liminal_bardo · 2026-09-11