Home / Curated / Algorithms / NAdam

NAdam

Algorithms AIMech:NAdam DRAFT

Aliases: Nesterov-accelerated adaptive moment estimation

An Adam-family optimizer combining adaptive moment scaling with a Nesterov-style momentum update.

Identity Notes & Concept Scope The Nesterov-style combination distinguishes NAdam from Adam. Optional decoupled weight decay is not required for this identity.

Claims & Grounded Evidence

1 Claim

Mechanism #update-rule

The documented update combines a scheduled, bias-adjusted first-moment term with a current-gradient term and divides the combination by a stabilized, bias-corrected second-moment scale.

supports Source: PyTorch NAdam implementation and algorithm documentation · Locator: NAdam.__doc__, algorithm lines 214–252 and options lines 257–266

⚠️ Curated Limitations & Negative Results

The source specifies a momentum schedule and an optional NAdamW configuration; other implementations need separate exact-equation comparison.
No comparative performance or convergence claim is curated.

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 NAdam implementation and algorithm documentation PyTorch contributors	official documentation documentation	sha256:c23755060c4c…

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/NAdam.yaml