MTMs replace word counts with SAE features, producing more abstract topics and enabling topic steering, and outperform word-based baselines in LLM and human pairwise evaluations.
InProceedings of the Associa- tion for Computational Linguistics, pages 15504– 15522
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Model Directions, Not Words: Mechanistic Topic Models Using Sparse Autoencoders
MTMs replace word counts with SAE features, producing more abstract topics and enabling topic steering, and outperform word-based baselines in LLM and human pairwise evaluations.