Dynamic Multi-Byte Prediction With Hierarchical Language Models

Hugging Face Daily Papers Papers

Summary

The paper introduces multi-byte prediction to speed up inference in byte-level language models by generating multiple bytes in parallel with minimal performance impact.

Byte-level hierarchical language models (LMs) have recently emerged as a robust alternative to their popular counterparts that use subword tokenization. However, generating one byte at a time remains a bottleneck for inference speed. To address this, we introduce multi-byte prediction (MBP), which generates multiple bytes in parallel, speeding up inference with minimal performance impact and no additional parameters. MBP builds on the popular multi-token prediction (MTP) paradigm with two crucial innovations. First, we introduce a variable-length prediction window that aligns with the latent tokens, or segments, of a hierarchical LM. Second, we implement a novel attention-masking scheme that enables parallel byte prediction without violating causality. We show that multi-byte prediction strikes a Pareto-optimal trade-off across multiple generative tasks, instruction following, question answering, summarization, and machine translation, achieving the best trade-off between performance and inference throughput.
Original Article
View Cached Full Text

Cached at: 08/19/26, 03:59 AM

Paper page - Dynamic Multi-Byte Prediction With Hierarchical Language Models

Source: https://huggingface.co/papers/2608.15454

Abstract

Multi-byte prediction accelerates byte-level hierarchical language models by generating parallel bytes via variable-length windows and causal attention masking, improving inference speed with minimal quality loss.

Byte-level hierarchical language models(LMs) have recently emerged as a robust alternative to their popular counterparts that use subword tokenization. However, generating one byte at a time remains a bottleneck for inference speed. To address this, we introducemulti-byte prediction(MBP), which generates multiple bytes in parallel, speeding up inference with minimal performance impact and no additional parameters. MBP builds on the popularmulti-token prediction(MTP) paradigm with two crucial innovations. First, we introduce avariable-length prediction windowthat aligns with the latent tokens, or segments, of a hierarchical LM. Second, we implement a novelattention-maskingscheme that enables parallel byte prediction without violating causality. We show thatmulti-byte predictionstrikes a Pareto-optimal trade-off across multiple generative tasks, instruction following, question answering, summarization, and machine translation, achieving the best trade-off between performance andinference throughput.

View arXiv pageView PDFGitHub0Add to collection

Get this paper in your agent:

hf papers read 2608\.15454

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2608.15454 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2608.15454 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2608.15454 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

Hierarchical Latent Prediction for Language Models

arXiv cs.CL

This paper introduces HiLP, a hierarchical representation training method that adds multi-scale self-predictive learning to transformer pretraining, aiming to reduce compounding error and improve long-horizon reasoning and speculative decoding efficiency.

Fast Byte Latent Transformer

Hugging Face Daily Papers

This paper introduces BLT Diffusion and speculative decoding techniques for byte-level language models to significantly reduce generation latency and memory bandwidth costs while maintaining quality.

Multi-Block Diffusion Language Models

Hugging Face Daily Papers

This paper proposes Multi-Block Diffusion Language Models (MBD-LMs), extending single-block diffusion to concurrent multi-block decoding with improved training strategies like Multi-block Teacher Forcing and an optimized Block Buffer decoding algorithm. Experiments show increased tokens per forward pass and improved accuracy on benchmarks.