SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.07425 · 2026

Sign-Based Optimizers Are Effective Under Heavy-Tailed Noise

Lijun Zhang, Luo Luo, Dingzhi Yu, Hongyi Tao

arXiv · PDF · Open in the Atlas

Code that ran

6 of 8 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 5 failures of our instrument; 2 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2602.07425.json · badge: /api/ran/2602.07425.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
Dingzhen230/Heavy-tailed-Noise-in-LLMs canonical (classifier label)
classifier basis: census
6 ran · 2 no recorded run
FunctionStatusWhere it lives
entropy_reg Ran
ran; our contract draft was wrong, not the code
Dingzhen230/Heavy-tailed-Noise-in-LLMs/src/models/moe.py
pointer only (licence: NONE) · get_code("1f1b81b26e97e55b")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, Dingzhen230/Heavy-tailed-Noise-in-LLMs when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

load_balancing_loss_ Ran
ran; our contract draft was wrong, not the code
Dingzhen230/Heavy-tailed-Noise-in-LLMs/src/models/moe.py
pointer only (licence: NONE) · get_code("a69212d334eb9bc7")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, Dingzhen230/Heavy-tailed-Noise-in-LLMs when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

log_mean Ran
ran; our contract draft was wrong, not the code
Dingzhen230/Heavy-tailed-Noise-in-LLMs/src/models/moe.py
pointer only (licence: NONE) · get_code("2f520af81f2675ca")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, Dingzhen230/Heavy-tailed-Noise-in-LLMs when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

move_to_cpu Ran
ran; our contract draft was wrong, not the code
Dingzhen230/Heavy-tailed-Noise-in-LLMs/src/logger/logger.py
pointer only (licence: NONE) · get_code("1bd86a911c3f78a3")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, Dingzhen230/Heavy-tailed-Noise-in-LLMs when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

precompute_freqs_cis Ran
ran, honoured the contract we drafted
Dingzhen230/Heavy-tailed-Noise-in-LLMs/src/models/llama.py
pointer only (licence: NONE) · get_code("a1d6f89d43fc42e1")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, Dingzhen230/Heavy-tailed-Noise-in-LLMs when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

self_preserving_overwrite Ran
ran; our contract draft was wrong, not the code
Dingzhen230/Heavy-tailed-Noise-in-LLMs/src/logger/logger.py
pointer only (licence: NONE) · get_code("198aea734522a443")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, Dingzhen230/Heavy-tailed-Noise-in-LLMs when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

apply_rotary_emb Not yet run
unverified (harvested, no recorded run)
Dingzhen230/Heavy-tailed-Noise-in-LLMs/src/models/llama.py
pointer only (licence: NONE) · get_code("9891174b43bd9c9a")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, Dingzhen230/Heavy-tailed-Noise-in-LLMs when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

none_or_str Not yet run
unverified (harvested, no recorded run)
Dingzhen230/Heavy-tailed-Noise-in-LLMs/src/config/base.py
pointer only (licence: NONE) · get_code("2fc6fdb85f13dc41")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, Dingzhen230/Heavy-tailed-Noise-in-LLMs when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 distinct functions with an identity, 8 are also present in other papers whose code we hold (8 with at least one paper dated earlier, 4 the same month, 8 later) and 0 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
none_or_str 16 other papers
14 dated earlier, 2 later
1810.01845 (2018-10)
Task-Oriented Hand Motion Retargeting for Dexterous Manipulation Imitation
1909.10893 (2019-09)
Recurrent Independent Mechanisms
2109.11251 (2021-09)
Trust Region Policy Optimisation in Multi-Agent Reinforcement Learning
and 13 more not named here
precompute_freqs_cis 6 other papers
3 dated earlier, 1 the same month, 2 later
2405.18392 (2024-05)
Scaling Laws and Compute-Optimal Training Beyond Fixed Training Durations
2502.05003 (2025-02)
QuEST: Stable Training of LLMs with 1-Bit Weights and Activations
2507.09846 (2025-07)
Through the River: Understanding the Benefit of Schedule-Free Methods for Language Model Training
and 3 more not named here
apply_rotary_emb 5 other papers
2 dated earlier, 1 the same month, 2 later
2405.18392 (2024-05)
Scaling Laws and Compute-Optimal Training Beyond Fixed Training Durations
2507.09846 (2025-07)
Through the River: Understanding the Benefit of Schedule-Free Methods for Language Model Training
2602.02016 (2026-02)
DASH: Faster Shampoo via Batched Block Preconditioning and Efficient Inverse-Root Solvers
and 2 more not named here
move_to_cpu 5 other papers
3 dated earlier, 1 the same month, 1 later
2405.18392 (2024-05)
Scaling Laws and Compute-Optimal Training Beyond Fixed Training Durations
2502.05003 (2025-02)
QuEST: Stable Training of LLMs with 1-Bit Weights and Activations
2507.09846 (2025-07)
Through the River: Understanding the Benefit of Schedule-Free Methods for Language Model Training
and 2 more not named here
self_preserving_overwrite 5 other papers
3 dated earlier, 1 the same month, 1 later
2405.18392 (2024-05)
Scaling Laws and Compute-Optimal Training Beyond Fixed Training Durations
2502.05003 (2025-02)
QuEST: Stable Training of LLMs with 1-Bit Weights and Activations
2507.09846 (2025-07)
Through the River: Understanding the Benefit of Schedule-Free Methods for Language Model Training
and 2 more not named here
entropy_reg 2 other papers
1 dated earlier, 1 later
2507.09846 (2025-07)
Through the River: Understanding the Benefit of Schedule-Free Methods for Language Model Training
2605.06615 (2026-05)
When and Why SignSGD Outperforms SGD: A Theoretical Study Based on ℓ 1 -norm Lower Bounds
load_balancing_loss_ 2 other papers
1 dated earlier, 1 later
2507.09846 (2025-07)
Through the River: Understanding the Benefit of Schedule-Free Methods for Language Model Training
2605.06615 (2026-05)
When and Why SignSGD Outperforms SGD: A Theoretical Study Based on ℓ 1 -norm Lower Bounds
log_mean 2 other papers
1 dated earlier, 1 later
2507.09846 (2025-07)
Through the River: Understanding the Benefit of Schedule-Free Methods for Language Model Training
2605.06615 (2026-05)
When and Why SignSGD Outperforms SGD: A Theoretical Study Based on ℓ 1 -norm Lower Bounds

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

While adaptive gradient methods are the workhorse of modern machine learning, sign-based optimization algorithms such as Lion and Muon have recently demonstrated superior empirical performance over AdamW in training large language models (LLM). However, a theoretical understanding of why sign-based updates outperform variance-adapted methods remains elusive. In this paper, we aim to bridge the gap between theory and practice through the lens of heavy-tailed gradient noise, a phenomenon frequently observed in language modeling tasks. Theoretically, we introduce a novel generalized heavy-tailed noise condition that captures the behavior of LLMs more accurately than standard finite variance assumptions. Under this noise model, we establish sharp convergence rates of SignSGD and Lion for generalized smooth function classes, matching or surpassing previous best-known bounds. Furthermore, we extend our analysis to Muon and Muonlight, providing what is, to our knowledge, the first rigorous analysis of matrix optimization under heavy-tailed stochasticity. These results offer a strong theoretical justification for the empirical superiority of sign-based optimizers, showcasing that they are naturally suited to handle the noisy gradients associated with heavy tails. Empirically, LLM pretraining experiments validate our theoretical insights and confirm that our proposed noise models are well-aligned with practice.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2602.07425")
get_code_for_paper("2602.07425")
have("2602.07425")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2602.07425.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2602.07425.svg)](https://syntology.ai/paper/2602.07425)

Connect an agent — have() is free.