AI Paper Atlas

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

Internal Planning in Language Models: Characterizing Horizon and Branch Awareness

Muhammed Ustaomeroglu, Baris Askin, Gauri Joshi, Carlee Joe-Wong, Guannan Qu

ICLR 2026 · Large language models › Language, safety and interpretability · approach: Reasoning mechanisms and failure modes · 0 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
train with MTP loss instead of NTP lossGPT-3 Small · PF-Longaccuracy: 0.60 -> 0.85Improved
train with MTP loss instead of NTP lossGPT-3 Small · PF-Longalt/decoy MI ratio: 1.45 -> 1.82Improved
train with MTP loss instead of NTP lossGPT-3 Small · PF-Shortaccuracy: 0.92 -> 0.88No clear effect
train with MTP loss instead of NTP lossGPT-3 Small · PF-Shortalt/decoy MI ratio: 7.60 -> 6.29Hurt
train with MTP loss instead of NTP lossGPT-3 Small · CFGnMI decay between prefix and decision stateNo clear effect
increase VQ-VAE codebook size K (ablation)VQ-VAE MI estimator · Synthetic known-MI validationagreement of normalized MI with ground truthImproved
use a task requiring longer paths (PF instead of CFG)GPT-3 Small · CFG vs path findingnMI between prefix and later decision states: nMI at tau=10 about one-fifth of initial on CFGImproved

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