Harvard Paper: Simple Parameter Averaging Beats Replay in Continual Learning
A Harvard-led paper proposes SFA, which mitigates catastrophic forgetting by simply averaging sequentially fine-tuned models, outperforming replay methods while cutting training time by up to 36%.
2026-11-20 ~ 2026-11-20 · 2 related posts
- SFA: model averaging beats replay in continual learning, no replay or auxiliary loss needed — jefrankle · 2026-11-20
- SFA paper: simple parameter averaging beats replay in continual learning with 36% less training — jefrankle · 2026-11-20