AdamW
Algorithms AIMech:AdamW DRAFT
An Adam optimization variant that applies weight decay directly to parameters separately from gradient moment estimation.
Identity Notes & Concept Scope
Decoupled weight decay distinguishes AdamW from Adam with a penalty included in its gradient moments. It is an optimization algorithm, not a new model architecture.
Claims & Grounded Evidence
1 Claim
Mechanism
#update-rule
Parameters receive a weight-decay shrinkage step, while first and second gradient moments are updated from the objective gradient and used for the adaptive step.
⚠️ Curated Limitations & Negative Results
- The cited implementation also offers an optional AMSGrad variant; that option is not required by the AdamW definition.
- No universal convergence or performance advantage is asserted.
Artificial Intelligence Ontology (AIO) Grounding
Mapped AIO Classes
| AIO Target Class | Mapping Relation | Mapping Rationale |
|---|---|---|
AIO:TrainingStrategy
TrainingStrategy
|
related_match | This algorithm can implement parameter fitting, but the pinned TrainingStrategy definition uses a preprocessing genus. This is a contextual training relationship, not equivalence or a subclass assertion. |
Evidence Sources & Primary Literature
1 Source| Source ID | Publication / Document | Type / Access | Reference & Checksum |
|---|---|---|---|
pytorch-docs
|
PyTorch contributors
|
official documentation
documentation
|
sha256:3112d8799ef9…
|
Curation Provenance
Curator: Codex
Method: ai_assisted
Model: GPT-6
Created: 2026-10-11T03:25:18Z
Updated: 2026-10-11T03:25:18Z
Source File:
data/curated/AdamW.yaml