repair-robust-editing-llms

【效率与协作】Perform robust, low-cost model updates via REPAIR: closed-loop feedback, dynamic memory management, and frequent knowledge fusion. Preserve non-target knowledge while achieving 10-30% improvements in editing accuracy. Enable sequential edits without catastrophic forgetting through strong locality guards and stable integration mechanisms.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准