---
title: 'LoRA: Low-Rank Adaptation of Large Language Models'
license: CC-BY-4.0
license_url: https://creativecommons.org/licenses/by/4.0/
license_note: 'paper is arXiv non-exclusive distribution license (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/2106-09685-lora.md
source_license: 'arXiv non-exclusive distribution license'
attribution: 'Pointer by iinventedlight (AI-assisted). Source: Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, Weizhu Chen, LoRA: Low-Rank Adaptation of Large Language Models (arXiv non-exclusive distribution license). 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: foundations'
tags: arxiv, lora, fine-tuning, efficiency
arxiv_id: 2106.09685
alias: 2106.09685-lora
type: pointer
format: pointer
sources:
  - id: S1
    title: 'LoRA: Low-Rank Adaptation of Large Language Models'
    publisher: arXiv
    authors: 'Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, Weizhu Chen'
    url: https://arxiv.org/abs/2106.09685
    version_read: 'v2 (16 Oct 2021)'
    accessed: 2026-11-11
    license: 'arXiv non-exclusive distribution license'
---
# LoRA: Low-Rank Adaptation of Large Language Models

## Summary
[LoRA: Low-Rank Adaptation of Large Language Models](https://arxiv.org/abs/2106.09685) (2021) introduced the standard parameter-efficient way to adapt a large model without training all of its weights. You freeze the base model and learn a pair of low-rank update matrices, which makes fine-tuning cheaper to store and often cheaper to run.

What it introduced: low-rank adaptation (LoRA) of attention (and later other) weights, with experiments showing that these small updates can match full fine-tuning on many tasks.

What it led to: QLoRA and the adapter ecosystem now used to fine-tune open chat models on modest hardware. Live notes: [2305-14314-qlora](https://iinventedlight.com/library/notes/2305-14314-qlora/) and [hf-transformers-quantization-overview](https://iinventedlight.com/library/notes/hf-transformers-quantization-overview/).

If you are adapting a large open model on one GPU, this is the method most later adapter recipes start from. Where to read it: https://arxiv.org/abs/2106.09685 (v2, last revised 16 Oct 2021).

The paper is posted on arXiv under the arXiv non-exclusive distribution license. 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: arXiv non-exclusive distribution; 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).