Saved in:
Bibliographic Details
Main Author: Wang, Tongxi
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2601.21708
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913014747234304
author Wang, Tongxi
author_facet Wang, Tongxi
contents Large language models (LLMs) excel across many tasks, yet inference is still dominated by strictly token-by-token autoregression. Existing acceleration methods largely patch this pipeline and miss core human-reading ingredients: content-adaptive foresight, chunk-structure-aware compute allocation, and train-test consistency for preview/skimming. We propose the Fovea-Block-Skip Transformer (FBS), which injects a causal, trainable loop into Transformers via Parafovea-Attention Window (PAW), Chunk-Head (CH), and Skip-Gate (SG). Across diverse benchmarks, FBS improves the quality-efficiency trade-off without increasing parameters, and ablations show the three modules are complementary.
format Preprint
id arxiv_https___arxiv_org_abs_2601_21708
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle FBS: Modeling Native Parallel Reading inside a Transformer
Wang, Tongxi
Artificial Intelligence
Computation and Language
Large language models (LLMs) excel across many tasks, yet inference is still dominated by strictly token-by-token autoregression. Existing acceleration methods largely patch this pipeline and miss core human-reading ingredients: content-adaptive foresight, chunk-structure-aware compute allocation, and train-test consistency for preview/skimming. We propose the Fovea-Block-Skip Transformer (FBS), which injects a causal, trainable loop into Transformers via Parafovea-Attention Window (PAW), Chunk-Head (CH), and Skip-Gate (SG). Across diverse benchmarks, FBS improves the quality-efficiency trade-off without increasing parameters, and ablations show the three modules are complementary.
title FBS: Modeling Native Parallel Reading inside a Transformer
topic Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2601.21708