# Neural Speed Reading with Structural-Jump-LSTM

arXiv: https://arxiv.org/abs/1904.00761 · HTML: https://mlsanity.com/paper/1904.00761/

## Key facts

- Cited by 17 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 20 Mar 2019.

## Built on by (most cited first, 17 of 17)

- [Model Tells You What to Discard: Adaptive KV Cache Compression for LLMs](https://mlsanity.com/paper/2310.01801/) (2023)
- [Dynamic Neural Networks: A Survey](https://mlsanity.com/paper/2102.04906/) (2021)
- [Adaptive Rotated Convolution for Rotated Object Detection](https://mlsanity.com/paper/2303.07820/) (2023)
- [Rank-DETR for High Quality Object Detection](https://mlsanity.com/paper/2310.08854/) (2023)
- [ART: Anonymous Region Transformer for Variable Multi-Layer Transparent Image Generation](https://mlsanity.com/paper/2502.18364/) (2025)
- [A Survey on Dynamic Neural Networks for Natural Language Processing](https://mlsanity.com/paper/2202.07101/) (2022)
- [Transkimmer: Transformer Learns to Layer-wise Skim](https://mlsanity.com/paper/2205.07324/) (2022)
- [Efficient Diffusion Transformer with Step-wise Dynamic Attention Mediators](https://mlsanity.com/paper/2408.05710/) (2024)
- [Block-Skim: Efficient Question Answering for Transformer](https://mlsanity.com/paper/2112.08560/) (2021)
- [Efficient Deep Visual and Inertial Odometry with Adaptive Visual Modality Selection](https://mlsanity.com/paper/2205.06187/) (2022)
- [Task-Customized Self-Supervised Pre-training with Scalable Dynamic Routing](https://mlsanity.com/paper/2205.13267/) (2022)
- [GRA: Detecting Oriented Objects through Group-wise Rotating and Attention](https://mlsanity.com/paper/2403.11127/) (2024)
- [Linear Differential Vision Transformer: Learning Visual Contrasts via Pairwise Differentials](https://mlsanity.com/paper/2511.00833/) (2025)
- [Part & Whole Extraction: Towards A Deep Understanding of Quantitative Facts for Percentages in Text](https://mlsanity.com/paper/2110.13505/) (2021)
- [Switchable Decision: Dynamic Neural Generation Networks](https://mlsanity.com/paper/2405.04513/) (2024)
- [SOI: Scaling Down Computational Complexity by Estimating Partial States of the Model](https://mlsanity.com/paper/2410.03813/) (2024)
- [Rethinking Recurrent Neural Networks for Time Series Forecasting: A Reinforced Recurrent Encoder with Prediction-Oriented Proximal Policy Optimization](https://mlsanity.com/paper/2601.03683/) (2026)

## Abstract (arXiv)

Recurrent neural networks (RNNs) can model natural language by sequentially 'reading' input tokens and outputting a distributed representation of each token. Due to the sequential nature of RNNs, inference time is linearly dependent on the input length, and all inputs are read regardless of their importance. Efforts to speed up this inference, known as 'neural speed reading', either ignore or skim over part of the input. We present Structural-Jump-LSTM: the first neural speed reading model to both skip and jump text during inference. The model consists of a standard LSTM and two agents: one capable of skipping single words when reading, and one capable of exploiting punctuation structure (sub-sentence separators (,:), sentence end symbols (.!?), or end of text markers) to jump ahead after reading a word. A comprehensive experimental evaluation of our model against all five state-of-the-art neural reading models shows that Structural-Jump-LSTM achieves the best overall floating point operations (FLOP) reduction (hence is faster), while keeping the same accuracy or even improving it compared to a vanilla LSTM that reads the whole text.