ConceptTest-Time Reinforcement Learning
Test-Time Reinforcement Learning
A machine learning paradigm that, at inference time and without labeled data, constructs reward signals from the model's own predictions to update parameters or policies online. It is distinct from supervised fine-tuning and from test-time adaptation that only adjusts prompts.
Timeline (last 90 days)
Oct 30
Label-free test-time reinforcement learning uses rewards generated from the model's own predictions for self-adjustment, without requiring labeled data.
Unverified50%