AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Incorporating Context into Language Encoding Models for fMRI

Shailee Jain, Alexander Huth

NeurIPS 2018 · Large language models › Language, safety and interpretability · approach: Linguistic encoding in LLM representations · 163 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use LSTM LM contextual representations instead of word embeddings985-d word embedding encoding model · Moth Radio Hour fMRI (6 subjects)summed voxel r^2 (encoding performance)Improved
increase context length (0 to 19 words) (ablation)LSTM contextual encoding model · Moth Radio Hour fMRI (6 subjects)summed voxel r^2 (encoding performance)Improved
replace true context with random words (ablation)LSTM contextual encoding model · Moth Radio Hour fMRI (6 subjects)summed voxel r^2 (encoding performance)Hurt
swap context with another occurrence of the same word (ablation)LSTM contextual encoding model · Moth Radio Hour fMRI (6 subjects)summed voxel r^2 (encoding performance)Hurt
use LSTM layer 2 instead of layers 1 or 3 (ablation)LSTM contextual encoding model · Moth Radio Hour fMRI (6 subjects)summed voxel r^2 (encoding performance)Improved
use layer 1 with zero context instead of input embedding (ablation)embedding encoding model · Moth Radio Hour fMRI (6 subjects)summed voxel r^2 (encoding performance)Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy