Continuous Self-Improvement of Large Language Models by Test-time Training with Verifier-Driven Sample Selection

VLMLRMRe-assign community