Optimization and training dynamics · MLD-0232
Cosine decay
A learning-rate schedule following part of a cosine curve.
Original reference page · detailed expansion pending. Original definitions, examples, notes and citations are preserved below. New examples are illustrative, not benchmark results.
Original example
Smoothly reduce the learning rate toward a minimum.
Original caveat
Implementations differ in minimum and restart behavior.
Context and aliases
Level: Intermediate. Stages: Training · Theory.
Tags: ML · DL · SLM · LLM.
Aliases: Cosine annealing.
Related concepts
Concepts linking here
Original references
- Deep Learning: online textbook ↗ · Consulted
- PyTorch optimization algorithms ↗ · Reference link
- PyTorch documentation ↗ · Reference link
- Original reference URL ↗