Pushing Gaze-Only Models in Reading Comprehension
Enhancing gaze-only models for reading comprehension with lightweight conditioning shows promise. New mechanisms reveal small yet significant AUROC gains.
Read the full articleYou might also wanna read
Study Finds Multimodal Training Provides Selective, Not Global, Benefits for Human-Like Language Processing
Large language models (LLMs) have become increasingly useful computational models of human language processing, but it remains unclear wheth
arxiv.org·1mo agoReading media and multitasking: effects on comprehension and metacomprehension when reading multiple texts
Altamura, Lidia;Stadtler, Marc;Salmerón, Ladislao Metacognition and Learning 01 Dec 2026 : pp.15 Formats: Citation/Abstract , Obtenir à l'Ud
Localizing Factual Recall Circuits in Gemma Models via Activation Patching
Activation patching reveals how facts are stored, routed, and read out across transformer layers, and why the residual stream does most of t
towardsdatascience.com·12d ago
Minimal Transformer Circuits Achieve Perfect Indirect Object Identification with Only Two Attention Heads
Mechanistic interpretability aims to reverse-engineer large language models (LLMs) into human-understandable computational circuits. However
arxiv.org·10d ago
ReverseEOL: A simple method to improve text embeddings by reversing input text in decoder-only LLMs
Recent advances in Large Language Models (LLMs) have opened new avenues for generating training-free text embeddings. However, the causal at
arxiv.org·1mo agoHow attention simplifies mental representations for planning
Human planning is efficient – it frugally deploys limited cognitive resources to accomplish difficult tasks – and flexible – adapting to nov
Comments
Sign in to join the conversation.
No comments yet. Be the first.