Lost in the Prompt Order: Revealing the Limitations of Causal Attention in Language Models

Hyunjong Ok, Jaeho Lee

Lost in the Prompt Order: Revealing the Limitations of Causal Attention in Language Models: 6 upvotes on Hugging Face Daily Papers, #18 of 26 papers on 2026-01-22. Day-by-day upvote history.

Large language models exhibit surprising sensitivity to the structure of the prompt, but the mechanisms underlying this sensitivity remain poorly understood. In this work, we conduct an in-depth investigation on a striking case: in multiple-choice question answering, placing context before the questions and options (CQO) outperforms the reverse order (QOC) by over 14%p, consistently over a wide range of models and datasets. Through systematic architectural analysis, we identify causal attention as the core mechanism: in QOC prompts, the causal mask prevents option tokens from attending to context, creating an information bottleneck where context becomes invisible to options.

Paper page on Hugging Face · arXiv

Data: hysts-bot-data/daily-papers-stats and the Daily Papers API. Open data: tardellirs/paper-pulse-data. Sister project: Model Pulse, the download history of every model on the Hub.