arXiv preprint ·
Sparse Autoencoder Attribution Graphs Are Decided at the Margin
- Published
- Source
- arXiv preprint, posted before peer review
- Original
- Read it at arXiv
- Topics
Abstract
Attribution graphs over the features of sparse autoencoders (SAEs) are used to predict what an ablation does and are compared across training seeds. A TopK SAE keeps the largest preactivations at each token, and a feature's margin is its preactivation's distance to the cutoff. On five public TopK SAE pairs from three Pythia models, zeroing one feature moves a next layer feature across the cutoff in 0.74 of cases or more, and the crossings carry most of the response's energy. Over writes that scale a feature's code by three factors, the graph built on the features active beforehand has a median relative error of 0.247 or more. The margin graph adds the first order change to every preactivation and reranks, at the same cost, and errs by at most 0.033. At the median, the crossing features' decoded change is 1.24 times the model's own change in norm or more. Between seeds of one testbed, sparse maps between features explain 0.85 of a graph's energy on its active features, and 0.47 once the other seed's graph keeps only its own. The largest model studied is pythia-160m.
The authors’ abstract, as arXiv lists it. arXiv’s listings, abstracts included, are free for anyone to reuse.
Read it on ReadSearch
Each opens after you create an account, and brings you back to this paper.
More in Machine Learning
Spotting the Artificial: Dialogic Humor Improvisation in Humans and LLMs
XDecideBench: Evaluating Typed Decision Models and Confidence Policies Across Languages
Swap, Replicate, Read the Mass: A Protocol for Auditing Demographic Bias in System One Decision Models
Information-Theoretic Design of Sparse Autoencoder-Based Steering
Robustness Starts with the Objective: Designing and Certifying Robust LLM-Generated Heuristics
Get the few papers worth your time in Machine Learning every morning.
ReadSearch reads the new papers in your field each day and brings you the few worth reading, each with its key finding in a sentence.