Tag: mechanistic interpretability
Browse our exclusive articles!
Mechanistic Interpretability of Antibody Language Models with SAEs
Explore how Sparse Autoencoders enhance mechanistic interpretability in antibody language models, aiding therapeutic antibody design and optimization.
Weight Patching for Mechanistic Localization in LLMs
Discover Weight Patching, a novel method for source-level mechanistic localization in large language models, enhancing interpretability and model merging.
ConceptTracer: Interactive Tool for Neural Concept Analysis
Explore neural network interpretability with ConceptTracer, analyzing concept saliency and selectivity to enhance AI transparency and understanding.
Boost Sparse Autoencoder Robustness with Masked Regularization
Enhance sparse autoencoder reliability using masked regularization to reduce feature absorption and improve out-of-distribution performance.
Discovering Belief Geometries Using Sparse Autoencoders
Explore how sparse autoencoders reveal belief-state geometries in transformer models, advancing mechanistic interpretability of large language models.