Interpretable sparse fine-tuning with TopKLoRA enables faithful circuit discovery

Paper recorded by Signals 4 on 2026-10-23 in cs.AI. Abstract reproduced from arXiv; link to the original below.

Published 2026-10-23 on arXiv · recorded by Signals 4 on 2026-10-26

Category: cs.AI · 人工智能 · first seen 2026-10-26

Abstract

Fine-tuning on datasets of unknown origin can create a backdoor in the trained model when the data includes poisoned examples. Difficult to localise and remove, backdoors can persist through safety training. While parameter-efficient fine-tuning methods capture the full parameter changes introduced during training, they are not easily interpretable. We present the full method of TopKLoRA, extendin

Read on arXiv →