Saved in:
Bibliographic Details
Main Author: Chopra, Ruhaan
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.11426
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909035165384704
author Chopra, Ruhaan
author_facet Chopra, Ruhaan
contents The cosine similarity between a large language model's hidden activations before and after Supervised Fine-Tuning (SFT) remains very high. This, at first glance, suggests that SFT leaves the model's activation geometry largely undisturbed. However, projecting both sets of activations through a Sparse Autoencoder (SAE) pretrained on the base model reveals that the underlying sparse latents diverge significantly. We introduce a novel investigative pipeline which utilizes these pretrained SAEs as a high-resolution diagnostic tool to mechanistically investigate the drivers of this representational divergence. Through our analytical pipeline, we discover task-specific and layer-specific distributions of the precise semantic features that are systematically altered during supervised fine-tuning. We additionally identify a layer-wise update profile specific to safety alignment. All code, experimental scripts, and analysis files associated with this work are publicly available at: https://github.com/ruhzi/sae-investigation.
format Preprint
id arxiv_https___arxiv_org_abs_2605_11426
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle A Mechanistic Investigation of Supervised Fine Tuning
Chopra, Ruhaan
Artificial Intelligence
The cosine similarity between a large language model's hidden activations before and after Supervised Fine-Tuning (SFT) remains very high. This, at first glance, suggests that SFT leaves the model's activation geometry largely undisturbed. However, projecting both sets of activations through a Sparse Autoencoder (SAE) pretrained on the base model reveals that the underlying sparse latents diverge significantly. We introduce a novel investigative pipeline which utilizes these pretrained SAEs as a high-resolution diagnostic tool to mechanistically investigate the drivers of this representational divergence. Through our analytical pipeline, we discover task-specific and layer-specific distributions of the precise semantic features that are systematically altered during supervised fine-tuning. We additionally identify a layer-wise update profile specific to safety alignment. All code, experimental scripts, and analysis files associated with this work are publicly available at: https://github.com/ruhzi/sae-investigation.
title A Mechanistic Investigation of Supervised Fine Tuning
topic Artificial Intelligence
url https://arxiv.org/abs/2605.11426