← 研踪 PaperTrailarXiv:2607.01218 · v2
The State-Prediction Separation Hypothesis
首次提交 2026-07-02 01:55
版本更新 2026-11-06 00:43
雷达观测 2026-11-06 12:09
摘要
Transformers use the same forward computation stream to both predict the next token and store useful state for future token predictions. We formulate the state-prediction separation hypothesis: disentangling the two roles yields better language modeling performance. We design a Transformer variant that uses separate computation streams to separate the two functions, and conduct pretraining experiments across various scales. Our experiments show that state-prediction separation consistently offers better data- and compute-efficiency, improving validation loss and outperforming standard Transformers by 2-3 percentage points on average on downstream tasks. We also conduct extensive empirical analysis that rules out potential confounders and demonstrates the fundamental difference in the gradients our design entails.
作者备注：Published at NeurIPS 2026
观测到的版本
相邻记录可能跳过中间版本；首次发现 v2 及以上也会标为版本更新。
v2 · 2026-11-06 00:43 · 最新观测版本
The State-Prediction Separation Hypothesis
Transformers use the same forward computation stream to both predict the next token and store useful state for future token predictions. We formulate the state-prediction separation hypothesis: disentangling the two roles yields better language modeling performance. We design a Transformer variant that uses separate computation streams to separate the two functions, and conduct pretraining experiments across various scales. Our experiments show that state-prediction separation consistently offers better data- and compute-efficiency, improving validation loss and outperforming standard Transformers by 2-3 percentage points on average on downstream tasks. We also conduct extensive empirical analysis that rules out potential confounders and demonstrates the fundamental difference in the gradients our design entails.
查看此版本解读 →查看此版本原文 ↗