// ⚙ Fine-tuning & PEFT

DoRA (Weight-Decomposed LoRA)

Splits weights into magnitude and direction, using LoRA for direction.

DoRA 2024 Intermediate extended

IN PLAIN WORDS

Full fine-tuning and LoRA update differently; DoRA learns per-column magnitude separately and uses LoRA for direction, closing the gap at equal parameters.

Used in: Option in the PEFT library

TIMELINE · prerequisites → this → what builds on it

2021 LoRA (Low-Ran…2024 DoRA (Weight-Deco…

01Original paper

DoRA: Weight-Decomposed Low-Rank Adaptation

Shih-Yang Liu, Chien-Yi Wang, Hongxu Yin, et al. · 2024

arXiv ↗ ✓ Verified · arXiv API

02Where it sits

Learn the left column first; the right column builds on this one.

Learn first

LoRA (Low-Rank Adaptation)

DoRA (Weight-Decomposed LoRA)

Then you can learn

Nothing in the atlas builds on it yet

Related:QLoRA

Left: prerequisites · right: builds on this · dashed: related. Full map →

03Learning resources

Official docs, explainers, courses and reference code. All links checked.

Official docs

huggingface.co ↗EN

04Where the course teaches it

U50 Instruction Tuning & PEFT SFT, synthetic data, LoRA/QLoRA/DoRA, adapters and model merging. 10 h8 algorithms3 paperslesson ready

For AI / data

Verification: verified, checked 2026-10-21.
JSON: /api/learn/algorithms.json (id: dora)