---
title: 'Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context'
license: CC-BY-4.0
license_url: https://creativecommons.org/licenses/by/4.0/
license_note: 'paper is CC BY-NC-SA 4.0 (verified on arXiv abs page 2026-11-11); this pointer note is our own words and is CC BY 4.0; we link to the paper rather than reproduce it'
source: https://iinventedlight.com/library/notes/1901-02860-transformer-xl.md
source_license: 'CC BY-NC-SA 4.0'
attribution: 'Pointer by iinventedlight (AI-assisted). Source: Zihang Dai, Zhilin Yang, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov, Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context (CC BY-NC-SA 4.0). This note links to the paper rather than reproducing it.'
author: 'iinventedlight (AI-assisted)'
date: 2026-11-11
status: published
approved_by: "Cole Withington, 2026-11-11 20:58 PT, Grok Bot chat widget t1717s7407: \"Publish all 100 including the LLMs Encode Clinical Knowledge note\""
checked_on: 2026-11-11
as_of: 2026-11-11
confidence: medium
untested: 'Pointer note only. Paper not reproduced. Abs page and license checked 2026-11-11.'
category: arxiv
topic_area: 'ai-research: memory-and-context'
tags: arxiv, long-context, transformers, memory
arxiv_id: 1901.02860
alias: 1901.02860-transformer-xl
type: pointer
format: pointer
sources:
  - id: S1
    title: 'Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context'
    publisher: arXiv
    authors: 'Zihang Dai, Zhilin Yang, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov'
    url: https://arxiv.org/abs/1901.02860
    version_read: 'v3 (2 Jun 2019)'
    accessed: 2026-11-11
    license: 'CC BY-NC-SA 4.0'
---
# Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context

## Summary
[Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) (2019) was an early attempt to let a Transformer model more than one fixed window of tokens. It introduced segment-level recurrence and a relative positional encoding so that information could carry across segment boundaries.

What it introduced: a recurrence mechanism over cached hidden states from previous segments, plus relative encodings that made that recurrence stable, so language models could use a longer effective context than a single training window.

What it led to: later long-context work, from relative and rotary position methods to today's long-context evals. Live notes: [ao-context-long-context-limits](https://iinventedlight.com/library/notes/ao-context-long-context-limits/). See also the RoFormer / RoPE pointer in this batch and the Lost in the Middle pointer in batch B.

Where to read it: https://arxiv.org/abs/1901.02860 (v3, last revised 2 Jun 2019). ACL 2019 long paper.

The paper is posted under CC BY-NC-SA 4.0. This note links to it rather than reproducing it. The note itself is CC BY 4.0.

License: this note is CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Pointer by iinventedlight (AI-assisted). Paper license: CC BY-NC-SA 4.0; this note links to it and does not reproduce it.

---

Agents: [/agents.md](/agents.md) | [/.well-known/agent-library.json](/.well-known/agent-library.json).