Correcting Stochastic Update Bias in Preconditioned Language Model Optimizers
The paper discusses a new framework for correcting biases in preconditioned language model optimizers. It identifies two main biases that arise from the stochastic update rules used in these optimizers. The proposed bias-correction methods show improvements in pretraining loss and overall performance.
- ▪Preconditioned optimizers are essential for training language models but often suffer from stochastic update biases.
- ▪The authors propose a bias-correction framework that addresses gradient-preconditioner coupling and nonlinear inversion biases.
- ▪The framework has been shown to reduce pretraining loss in various models, demonstrating its effectiveness.
arXiv cs.AI files mainly under ai research. We currently carry 1,128 of its stories.
Story provenance
Source · retrieval · rights · ranking — open for full record
inspect →
Story provenance
Attribution is not the same as permission. This drawer separates discovery metadata, excerpts, WeSearch-generated summaries, reuse status, and whether the publisher receives the visit. Nothing here claims a legal grant the publisher has not made.
Record
| Original publisher | arXiv cs.AI |
| Canonical URL | https://arxiv.org/abs/2605.20756 |
| Publication time | Fri, 22 May 2026 00:00:00 -0400 |
| Retrieval time | 2026-05-22T04:02:00.009Z |
| Last seen | 2026-05-22T04:02:00.009Z |
| Headline source | Publisher (no WeSearch rewrite) |
| Excerpt source | publisher body |
| Excerpt method | First ~120 words (~800 chars) of extracted publisher body, fair-use limited. |
| Summary | WeSearch · cerebras-chat (WeSearch summarizer) |
| Summary source text | contentText |
| Citation coverage | Summary is a WeSearch-generated derivative; primary citation is the original publisher URL. |
| Cluster | yUgmU2pJvnYp |
| Cluster logic | Grouped by semantic title/content similarity across sources within a rolling window. Same-publisher template collisions are excluded from coverage comparison. |
| Ranking reason | Story pages are not engagement-ranked. Hub feeds use recency, with optional source-diversified chronological ordering (cap consecutive stories per source). No personalized ranking. |
| Publisher visit | Yes — open original |
| Substitutes article? | No — link-out required for full text |
Rights status (four layers)
WeSearch handling by dimension
| Indexing | May the item be indexed (stored, ranked, made findable)? | Allowed |
| Snippet | May a short excerpt of the publisher's text be shown? | Allowed |
| AI summary | May WeSearch generate its own short summary of the article? | Limited |
| Retrieval / RAG | May the content be exposed for third-party retrieval-augmented generation? | Not asserted |
| Model training | May the content be used to train AI models? | Not asserted |
| Commercial reuse | May the content be reused commercially? | Not permitted |
Basis: Derived from the published RSS/Atom feed. Contact: [email protected]. Reviewed: 2026-07-24.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Machine Learning arXiv:2605.20756 (cs) [Submitted on 20 May 2026] Title:Correcting Stochastic Update Bias in Preconditioned Language Model Optimizers Authors:Nikhil Nayak, Julia White, Urchade Zaratiana, Kelton Zhang, Henrijs Princis, Dhruv Atreja, Henry Fawcett, Matthew Thomas, George Hurn-Maloney, Ash Lewis View a PDF of the paper titled Correcting Stochastic Update Bias in Preconditioned Language Model Optimizers, by Nikhil Nayak and 9 other authors View PDF HTML (experimental) Abstract:Preconditioned optimizers are central to language model training, but their stochastic update rules are usually treated as direct approximations to population preconditioned descent. We show that this view misses two finite-sample biases.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
