Your LR Schedule May Not Be Principled: Effective Learning Rate Scheduling in Weight-Normalized Training
READ THE ORIGINAL SOURCEOpen original sourcehuggingface.co · Opens in a new tabThe paper argues that gains from weight-normalized training over conventional weight decay are driven mainly by the effective learning rate trajectory rather than normalization alone. Matching effective learning rate trajectories largely collapses the training loss dynamics of the two methods. The authors propose AIR, an affine inverse-square-root decay schedule with prescribed peak and terminal learning rates. In experiments on 128B-token Mixture-of-Experts pretraining, AIR outperforms cosine and other candidate schedules in training loss, downstream performance, and hyperparameter robustness.
One primary source; other links are context, not independent corroboration.
- Primary sourceHugging Face Daily Papers
- Discovery linkHugging Face
- Linked in sourceView arXiv page
- Linked in sourceView PDF
The paper page is linked. Peer review and findings have not been independently checked by AINewsIQ.
Open full article view →- Original paper+45
- Fetched source page+30
- No editor link review+0
- Original non-discussion link+5
This measures the stored source trail and editorial steps. It does not certify accuracy, peer review, or independent corroboration. Popularity adds no points.
View full scoring →Read the scoring rubric →What the source says
The paper argues that gains from weight-normalized training over conventional weight decay are driven mainly by the effective learning rate trajectory rather than normalization alone. Matching effective learning rate trajectories largely collapses the training loss dynamics of the two methods. The authors propose AIR, an affine inverse-square-root decay schedule with prescribed peak and terminal learning rates. In experiments on 128B-token Mixture-of-Experts pretraining, AIR outperforms cosine and other candidate schedules in training loss, downstream performance, and hyperparameter robustness.
A READER'S EVIDENCE CHECKHow to read this signalSix questions about claims, sources, risks, and open checks
This is a signal-level guide until claim-level evidence is available. “Not recorded” means the evidence has not been assessed here; it does not mean a result is false. How we check AI news →
- 01Who is making each claim?
The claimant and whether each claim was measured or asserted are not recorded in this public brief.
- 02What was actually tested?
Not recorded per claim. Check the source for sample size, tasks, model and benchmark versions, and reproducibility.
- 03Where are the sources?
Original link: huggingface.co. Brief basis: fetched source page. Primary page date and quote locations are not established in this panel.
- 04What could mislead?
Source claims not independently checked; results come from a single large-scale MoE pretraining setup and the reported improvement is a modest loss gap.
- 05What is at risk if wrong?
Not assessed per claim. Consider safety, privacy, cost, and adoption decisions before relying on the result.
- 06What would change the view?
No claim-specific check is recorded. An inspectable primary method or independent replication could change the assessment.
Key findings from the stored brief
- Matching effective learning rate trajectories largely collapses the training loss dynamics of weight-normalized training and weight decay.
- AIR combines a theoretically derived 1/sqrt(t) decay with an affine adjustment to hit prescribed peak and terminal learning rates.
- On 128B-token MoE pretraining, AIR reduces terminal training loss by about 0.01 relative to cosine at matched cumulative learning rate.
Source claims not independently checked; results come from a single large-scale MoE pretraining setup and the reported improvement is a modest loss gap.
This is a source-based abstract, not independent validation of the source's claims.
Evidence & sources
These links show where the signal came from and what the source page linked to. A linked page is context, not independent corroboration.
The paper page is linked. Peer review and findings have not been independently checked by AINewsIQ.
Timeline & updates
- Signal dated by source or discovery feed
This timestamp places the signal in the radar. It may be the discovery post date rather than the primary page's first publication date.
- Shared brief stored
AINewsIQ prepared one abstract from fetched source page.
A publication revision history is not stored for this signal. New engagement snapshots do not mean the source itself changed.
Why this may matter
If effective learning rate, rather than nominal learning rate, is the right object for schedule design, practitioners may tune schedules differently. The claim that the benefits extend to weight decay suggests the finding is not limited to fixed-norm training.
Limits and open questions
Source claims not independently checked; results come from a single large-scale MoE pretraining setup and the reported improvement is a modest loss gap.
Organization-specific financial, security, and operational effects require the organization's own context and review.
Discussion
Comments are reviewed before publication. Discuss the evidence, not the person.
No published comments yet.
Reader discussion opens with accounts. You can read the original source and published comments now.