Methods › General › Large Batch Optimization › LARS

LARS

77 papers tagged archive 2025-07-28

Introduced by Yang You et al. in Large Batch Training of Convolutional Networks

archive 2025-07-28 Description, source and code snippet are the archive's method entry.

Layer-wise Adaptive Rate Scaling, or LARS, is a large batch optimization technique. There are two notable differences between LARS and other adaptive algorithms such as Adam or RMSProp: first, LARS uses a separate learning rate for each layer and not for each weight. And second, the magnitude of the update is controlled with respect to the weight norm for better control of training speed.

mₜ = β₁mₜ₋₁ + (1-β₁)(gₜ + λxₜ) xₜ₊₁⁽ⁱ⁾ = xₜ⁽ⁱ⁾ - ηₜ(ϕ(|| xₜ⁽ⁱ⁾ ||))/(|| mₜ⁽ⁱ⁾ ||)mₜ⁽ⁱ⁾

PaperSourceSee Code · kakaobrain/torchlars

In Syntology about another paper, arXiv:2308.09618: repository known, no verified implementation (declared_code; repositories: 2 listed of 2 in the graph, strongest link: official). The lookup matched a paper by its title, arXiv:2308.09618, not a method entry. About another paper: this answer is about arXiv:2308.09618, while the archive credits arXiv:1708.03888.

From the lookup's note on this name: The same name also belongs to “Layer-wise Adaptive Rate Scaling (LARS)”, a method in the paper “Large Batch Training of Convolutional Networks” (arXiv:1708.03888, 977 citations), which the lookup finds cited far more than the paper this answer is about (citations not recorded), so this answer may be about a different sense of the name. For arXiv:1708.03888: its page on this site · ask the lookup about it.

Origin paper of this answer, arXiv:2308.09618 (its page on this site · its page on Syntology).

Syntology's have check for this name (free, it answers coverage only), read 2026-10-05; ask it again live. A level is the served verification level of that one answer; nothing here is summed across methods.

Papers archive 2025-07-28

archive papers tagged: 77 · with a code link: 34 · where Syntology ran a sample: 16 (13 with a run with no instrument failure, 3 where every run was a failure of Syntology's instrument) Syntology

Show: all tagged papersonly where code ran (16 of 77 tagged: 13 with a run with no instrument failure, 3 where every run was a failure of Syntology's instrument)

Page 1 of 1: papers 1 to 77 of 77, newest first by the archive's date (ties by slug), in archive order.

Repository counts are the archive's code-links table. A Syntology line states what Syntology ran from that paper's harvested code, as “N ran (of which C constructed an object rather than computing a result; K with no instrument failure: H honoured, V violated, P with no contract checked; I where Syntology's instrument failed) · U unverified”; the instrument figure counts failures of Syntology's instrument, not of the code. It is per sample and not a correctness claim. When the archive marks a repository official for the paper, the line starts with that repository's state (the archive's flag, not a verdict on who wrote the code; “community repositories only” when every sample that ran came from a community repository, “official: no sample here; runs from other or unrecorded repositories” when some came from a repository the paper names or has in its text, or from none recorded); hover it for the repositories the samples that ran came from.

Tasks archive 2025-07-28

20 shown of 106 tasks the archive attaches to papers tagged with this method, by distinct papers. A task without a page in the catalog is plain text.

TaskPapers
Self-Supervised Learning20
Contrastive Learning10
Image Classification10
Representation Learning7
Clustering6
Question Answering5
Semantic Segmentation5
Survival Analysis5
Transfer Learning5
Data Augmentation4
Object Detection4
Prognosis4
Semi-Supervised Image Classification4
image-classification4
object-detection4
Diversity3
Reinforcement Learning3
Self-Supervised Image Classification3
reinforcement-learning3
16k2

Usage over time archive 2025-07-28

Papers per year tagged with LARS: 2017 to 2025, peak 18 18 0 2017: 2 papers 2017 2018: 3 papers 2018 2019: 1 paper 2019 2020: 6 papers 2020 2021: 16 papers 2021 2022: 16 papers 2022 2023: 18 papers 2023 2024: 9 papers 2024 2025: 6 papers 2025
Papers per year the archive tags with this method, by the paper's archive date (77 dated). Bars are counts, not a trend claim.

Components: the archive holds no method-to-method composition, so PwC's Components table cannot be rebuilt; the Papers list carries no Results column for the same reason (the archive does not join its leaderboard rows to method tags).

Categories archive 2025-07-28

Large Batch Optimization

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections