Resources for AutoMaster, an autonomous post-training system that designs skills to help a fixed teacher model generate better demonstrations.
AI & ML interests
None defined yet.
Recent Activity
View all activity
Papers
AutoMaster: Autonomous Post-Training by Making the Same Model a Better Teacher
Spectral-Driven Adaptive-Rank Optimizer for Large Language Model Reinforcement Learning with Verifiable Rewards
Resources for AutoMaster, an autonomous post-training system that designs skills to help a fixed teacher model generate better demonstrations.
Resources for Agent-Editing World Model (AEWM), a world model for improving long-horizon LLM agents through decision-effect modeling and state rev
Models trained by SWE-Master and SWE-World, including both policy models and verifiers.