Methods › General › Large Batch Optimization › Adafactor

Adafactor

733 papers tagged archive 2025-07-28

Introduced by Noam Shazeer et al. in Adafactor: Adaptive Learning Rates with Sublinear Memory Cost

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

Adafactor is a stochastic optimization method based on Adam that reduces memory usage while retaining the empirical benefits of adaptivity. This is achieved through maintaining a factored representation of the squared gradient accumulator across training steps. Specifically, by tracking moving averages of the row and column sums of the squared gradients for matrix-valued variables, we are able to reconstruct a low-rank approximation of the exponentially smoothed accumulator at each training step that is optimal with respect to the generalized Kullback-Leibler divergence. For an n ×m matrix, this reduces the memory requirements from O(n m) to O(n + m).

Instead of defining the optimization algorithm in terms of absolute step sizes {αₜ}ₜ₌₁ᵀ, the authors define the optimization algorithm in terms of relative step sizes {ρₜ}ₜ₌₁ᵀ, which get multiplied by the scale of the parameters. The scale of a parameter vector or matrix is defined as the root-mean-square of its components, lower-bounded by a small constant ϵ₂. The reason for this lower bound is to allow zero-initialized parameters to escape 0.

Proposed hyperparameters are: ϵ₁ = 10⁻³⁰, ϵ₂ = 10⁻³, d=1, pₜ = min(10⁻², 1/(√(t))), β̂_(2ₜ) = 1 - t^(-0.8).

PaperSourceSee Code · DeadAt0m/adafactor-pytorch

In Syntology harvested code stored, below machine verification (unverified_code; harvested), served level V1; retrieve with get_harvested_code_for_paper(arxiv_id='1804.04235'). Repositories: 5 listed of 5 in the graph, strongest link: mentioned on GitHub. The lookup matched a paper by its title, arXiv:1804.04235, not a method entry. This answer is about arXiv:1804.04235, the paper the archive credits with this method.

Origin paper of this answer, arXiv:1804.04235 (its page on this site · its page on Syntology): Syntology's run record for its harvested code, 4 ran (of which 0 constructed an object rather than computing a result; 4 with no instrument failure: 0 honoured, 0 violated, 4 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified (of 4 harvested samples). The instrument figure counts failures of Syntology's instrument, not of the code. Ran means executed on a synthesized input, not that the code is correct.

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.

What each level means, quoted from Syntology's server card: V0 “imports”; V1 “executes against a synthesized fixture and returns a value of the recorded type”; V2 “two implementations GENERATED from the paper's signature and spec agree numerically (1e-6) on fixture inputs” “neither of the two ever read the authors' own code, so V2 measures how unambiguously the paper specifies the method, not fidelity to the authors' implementation (measured 2026-09-11, commit 75eaa77: cross_checked_by is a second LLM on 2,628 of 2,831 served samples)”; V3 “paper-derived property tests pass on both of those same two generations.” The card adds: “No rung on this ladder compares the served code against the authors' implementation, so treat every level as evidence about runnability and internal agreement, not about matching the paper's results.” A level quoted in two parts is one definition the card writes as two clauses joined by a double hyphen, each clause quoted on its own (how these levels are quoted).

Papers archive 2025-07-28

archive papers tagged: 733 · with a code link: 363 · where Syntology ran a sample: 103 (89 with a run with no instrument failure, 14 where every run was a failure of Syntology's instrument) Syntology

Show: all tagged papersonly where code ran (103 of 733 tagged: 89 with a run with no instrument failure, 14 where every run was a failure of Syntology's instrument)

Page 1 of 8: papers 1 to 100 of 733, 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 492 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
Language Modelling128
Language Modeling100
Question Answering86
Decoder79
Text Generation65
Sentence58
Translation47
Machine Translation41
Retrieval40
Transfer Learning40
Natural Language Understanding29
Abstractive Text Summarization23
Semantic Parsing23
Sentiment Analysis22
Natural Language Inference21
Code Generation20
Data Augmentation20
Text Summarization19
Diversity18
Knowledge Graphs17

Usage over time archive 2025-07-28

Papers per year tagged with Adafactor: 2018 to 2025, peak 202 202 0 2018: 1 paper 2018 2019: 2 papers 2019 2020: 37 papers 2020 2021: 112 papers 2021 2022: 168 papers 2022 2023: 202 papers 2023 2024: 158 papers 2024 2025: 53 papers 2025
Papers per year the archive tags with this method, by the paper's archive date (733 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 OptimizationStochastic 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