Search papers, labs, and topics across Lattice.
This paper introduces a forward-looking localization framework that predicts post-Supervised Fine-Tuning (SFT) interpretability by leveraging pre-SFT parameters and target datasets, addressing the limitations of traditional mechanistic interpretability. By modeling SFT as a continuous parameter evolution and employing Taylor expansion, the authors effectively connect pre-SFT dynamics with post-tuning objectives. Experimental results reveal that this approach not only enhances SFT guidance but also maintains robust performance and scalability across various model sizes, marking a significant advancement in the intersection of interpretability and optimization.
Traditional interpretability fails to predict task-critical mechanisms before training, but this new framework bridges that gap, enabling more effective fine-tuning strategies.
Mechanistic Localization bridges mechanistic interpretability and post-training optimization by isolating critical parameters via interpretative approaches and then guiding parameter-efficient Supervised Fine-Tuning (SFT) in a ``locating-then-tuning'' paradigm. However, due to the retrospective nature of mechanistic interpretability, directly interpreting pre-SFT models introduces misleading conclusions. Specifically for novel tasks, initially identified neurons differ drastically from those governing the final model, introducing biases that actively disrupt SFT. To address this, we propose a forward-looking localization framework that accurately estimates the post-SFT interpretability state using only pre-SFT parameters and the target dataset. Theoretically, we model SFT as a continuous parameter evolution, leveraging Taylor expansion to rigorously bridge the post-tuning mechanistic objective with the pre-SFT model's dynamic gradients. Practically, we design dual-granularity (neuron- and component-level) localization pipelines. Extensive experiments demonstrate that our approach not only provides superior SFT guidance but also exhibits robust performance and temporal scalability across increasing model sizes. This work transcends the fundamental limitation of traditional interpretability-its inability to identify task-critical mechanisms before they are trained-pioneering a predictive frontier that unites mechanistic interpretability with targeted optimization.
