Glossary term
Glossary term
Training and Fine-Tuning
The operation of adjusting a model's parameters during training, typically within a single iteration of gradient descent.
Created for this library
An ML engineer monitors parameter update magnitudes during training to detect optimizer instability.
A research team logs parameter update norms across layers to spot vanishing or exploding gradients early in long training runs.
An ML platform team enforces gradient clipping so individual parameter updates remain bounded across production training jobs.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License