Paper Detail
Lu Dai, Ziyang Rao, Yili Wang, Hanqing Wang, Hao Liu, Hui Xiong
Fine-tuning LLMs to inject new knowledge faces a critical challenge: LLMs can quickly memorize new facts, yet fail to use them for downstream reasoning tasks. We formalize this failure as the \textbf{Knowing--Using Gap}, characterized by an accuracy gap and a temporal lag between memorization and generalization. To understand this phenomenon, we fine-tune LLMs with unseen knowledge and monitor the spatial permeation dynamics of the knowledge internally using a novel intervention technique called self-patching. Self-patching identifies activation locations where relocating representations substantially improves failed generalization cases. These results are consistent with a knowledge-circuit misalignment hypothesis: memorized representations can exist internally but may not be routed to computation-effective layers. To demonstrate the practicality of this diagnostic finding, we design a simple heuristic strategy which recovers 58--75\% of the oracle headroom in generalization failure. Experiments are done cross-domain for the robustness of this finding.
No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.
No ranking explanation is available yet.
No tags.
@misc{dai2026mechanistically,
title = {Towards Mechanistically Understanding Why Memorized Knowledge Fails to Generalize in Large Language Model Finetuning},
author = {Lu Dai and Ziyang Rao and Yili Wang and Hanqing Wang and Hao Liu and Hui Xiong},
year = {2026},
abstract = {Fine-tuning LLMs to inject new knowledge faces a critical challenge: LLMs can quickly memorize new facts, yet fail to use them for downstream reasoning tasks. We formalize this failure as the \textbackslash{}textbf\{Knowing--Using Gap\}, characterized by an accuracy gap and a temporal lag between memorization and generalization. To understand this phenomenon, we fine-tune LLMs with unseen knowledge and monitor the spatial permeation dynamics of the knowledge internally using a novel intervention technique called},
url = {https://huggingface.co/papers/2607.08393},
keywords = {huggingface daily},
eprint = {2607.08393},
archiveprefix = {arXiv},
}
{}