analiza aiqbench
This research advances mechanistic interpretability in protein language models by demonstrating that sparse autoencoder features capture specific geometric patterns within the protein backbone, rather than just broad sequence labels. By linking latent representations to local 3D structures, the study reveals substructure within known biological functions and improves the understanding of unannotated metagenomic sequences. This work bridges the gap between abstract machine learning representations and concrete structural biology, offering a scalable method for decoding model internals.
Interpreting Latent Protein Language Model Features with Geometric Annotations
arXiv:2608.26419v1 Announce Type: cross Abstract: Protein language models (pLMs) encode information about protein sequences which enable downstream tasks such as structure prediction, but their internal representations are not well understood. Sparse autoencoders (SAEs) provide a promising tool to disentangle latent pLM representations into interpretable features, but existing annotation pipelines largely rely on protein-level annotations derived from database labels and LLM annotations of top activating sequences. Such annotations can overlook the localized residue-level and geometric patterns encoded by sparse features. We introduce an automated and scalable method for interpreting SAE features in ESM-2 by using geometrically inspired features of the protein $\text{C}_\alpha$ backbone. Across ESM-2 8M layers, an FDR-controlled discovery analysis shows that local geometry is significantly associated with many SAE features, with varying levels of predictive strength, expanding coverage beyond database and sequence-based methods. In particular, geometry can distinguish SAE features sharing the same database annotation, revealing substructure within known biological labels. A significant portion of SAE features activate on unannotated metagenomic protein sequences enabling us to use our SAE annotations to better understand these sequences. In addition, ablation experiments at the level of contact prediction show that removing found geometric features shifts ESM-2's predicted contact maps in the direction of the descriptor. This provides a robust method of annotating proteins activated within SAE neurons at a residue level, providing a bridge between mechanistic interpretability and structural biology.
Abstract:Protein language models (pLMs) encode information about protein sequences which enable downstream tasks such as structure prediction, but their internal representations are not well understood. Sparse autoencoders (SAEs) provide a promising tool to disentangle latent pLM representations into interpretable features, but existing annotation pipelines largely rely on protein-level annotations derived from database labels and LLM annotations of top activating sequences. Such annotations can overlook the localized residue-level and geometric patterns encoded by sparse features. We introduce an automated and scalable method for interpreting SAE features in ESM-2 by using geometrically inspired features of the protein $\text{C}_\alpha$ backbone. Across ESM-2 8M layers, an FDR-controlled discovery analysis shows that local geometry is significantly associated with many SAE features, with varying levels of predictive strength, expanding coverage beyond database and sequence-based methods. In particular, geometry can distinguish SAE features sharing the same database annotation, revealing substructure within known biological labels. A significant portion of SAE features activate on unannotated metagenomic protein sequences enabling us to use our SAE annotations to better understand these sequences. In addition, ablation experiments at the level of contact prediction show that removing found geometric features shifts ESM-2's predicted contact maps in the direction of the descriptor. This provides a robust method of annotating proteins activated within SAE neurons at a residue level, providing a bridge between mechanistic interpretability and structural biology.
Quantitative Methods (q-bio.QM); Machine Learning (cs.LG)
arXiv:2608.26419 [q-bio.QM]
(or arXiv:2608.26419v1 [q-bio.QM] for this version)
https://doi.org/10.48550/arXiv.2608.26419 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 320 cuvinte
Subiecte conexe
Circuit Tracing in Autoregressive Protein Language Models
arXiv · 16 iun. 2026
Introducing PIRAMID: Physics-Informed Research for Ambitious Mechanistic Interpretability
LessWrong · 25 iul. 2026
Subspace-Aware Sparse Autoencoders for Effective Mechanistic Interpretability
arXiv · 6 iun. 2026
Mechanistic Analysis of Alignment Algorithms in Language Models
arXiv · 10 iun. 2026