ExBurn.Training.ReduceLROnPlateauCallback (ex_burn v0.5.0)

Copy Markdown View Source

Reduces learning rate when validation loss stops improving.

Monitors validation loss and multiplies the learning rate by factor after patience epochs without improvement.

Usage

callbacks: [
  ExBurn.Training.ReduceLROnPlateauCallback.new(patience: 5, factor: 0.5, min_lr: 1.0e-6)
]

Summary

Functions

new(opts \\ [])

@spec new(keyword()) :: (map() -> map())