DRATS: Adaptive Task Sampling Boosts Worst-Task Performance in Multi-Task RL
Cohere · youtube · 2026-10-23
In a Cohere Labs talk, UW-Madison PhD Nicholas Edward Corrado presented DRATS, which tackles imbalanced data allocation in multi-task RL. Formalizing MTRL as a feasibility problem yields a minimax objective on worst-case return gap; the algorithm adaptively prioritizes tasks furthest from solved. On MetaWorld-MT10/MT50 it improves data efficiency and worst-task performance vs existing task samplers. Builds on AutoMixAlign (ACL 2025), a DPO-based adaptive data mixing method developed with Amazon's Rufus team.
More from Research
- Dev trains boolean-logic LM with microsecond forward passes on a single CPU thread — cephaloform · 2026-10-23
- AdaDex turns seconds of human hand motion into generalizable dexterous robot skills — mangahomanga · 2026-10-23
- ReAPO paper: conditioning retries on failures beats GRPO in RLVR reasoning training — jasondeanlee · 2026-10-23
- Quantum-JEPA: first JEPA world model deployable on real quantum circuits (QPU) — DrMorganLevine · 2026-10-23
- Omnii Genome: Single Generalist Model Outperforms AlphaGenome Across All Benchmarks — JosephJacks_ · 2026-10-23
- Domingos Cites New Result Claiming RL Is 'Basically Useless' vs Supervised Learning — pmddomingos · 2026-10-23