# Convergence (machine learning)

> A state reached when loss values change very little or not at all with each iteration. A model **converges** when additional training won't improve the model. In deep model, loss values sometimes stay constant or nearly so for many iterations before finally descending. During a long period of constant loss values, you may temporarily get a false sense of convergence.[^1]

A state reached when [loss](https://wiki.g15e.com/pages/Loss%20(machine%20learning.txt)) values change very little or not at all with each [iteration](https://wiki.g15e.com/pages/Iteration%20(machine%20learning.txt)). A model **converges** when additional [training](https://wiki.g15e.com/pages/Training%20(machine%20learning.txt)) won't improve the model. In [deep model](https://wiki.g15e.com/pages/Deep%20neural%20network.txt), loss values sometimes stay constant or nearly so for many iterations before finally descending. During a long period of constant loss values, you may temporarily get a false sense of convergence.[^1]

See also [early stopping](https://wiki.g15e.com/pages/Early%20stopping.txt).

## Footnotes

[^1]: https://developers.google.com/machine-learning/glossary#convergence