Sparse autoencoders applied to GPT-2 and Llama models recover semantic features accounting for 94% of peak brain encoding performance and map onto distinct cortical semantic regions across three languages.
and Christianson, Kiel , year =
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
fields
cs.CL 3years
2026 3representative citing papers
Varying the number of simultaneous parses in RNNGs increases predicted garden-path effects but does not fully reconcile LM surprisal with human reading times.
Early layers of language models predict early-pass human reading times better than surprisal, with surprisal superior for late-pass measures and strong variation by language.
citing papers explorer
-
Sparse Autoencoders Map Brain-LLM Alignment onto Cortical Semantic Topography
Sparse autoencoders applied to GPT-2 and Llama models recover semantic features accounting for 94% of peak brain encoding performance and map onto distinct cortical semantic regions across three languages.
-
Why are language models less surprised than humans? Testing the Parse Multiplicity Mismatch Hypothesis
Varying the number of simultaneous parses in RNNGs increases predicted garden-path effects but does not fully reconcile LM surprisal with human reading times.
-
Probing for Reading Times
Early layers of language models predict early-pass human reading times better than surprisal, with surprisal superior for late-pass measures and strong variation by language.