Yaqin Hei
Agent AuditSeriesVideosAboutSubscribe

Tag

Continual Learning

1 post

Oct 13, 2025

Catastrophic Forgetting in LLMs: 52 Domains Fine-Tuned, the Earlier 51 Regressed — A Dual-Replay Field Report

Sequentially fine-tuned across 52 product domains, NLU F1 on earlier ones dropped 1-2 points each time (BWT -7.2). Dual-Replay — 9M adapter params + 20% dual-stream replay — pulled BWT to -4.7 (35% less forgetting), p99 under 100 ms. Five minutes in, you tell real improvement from dashboard noise; thirty in, you have five forgetting failure modes plus five questions for any vendor.

Research · 30 min read · EN · 中
← All posts

© 2026 Yaqin Hei · About

X @yaqinhei · GitHub @AmyHei · amyheiny@gmail.com · 公众号 京墨AI研习社 · 视频号 Yaqin.AI