Model Annealing: Data-Efficient Learning through Review and Consolidation
Summary
Model annealing borrows the idea of review and consolidation, letting a small labeled dataset be reused through training instead of merely memorized, so the model learns more from less data.
Why this matters
Covers model and review · Relevant to this topic
Source: arXiv cs.AI · arxiv.org