search

FluxML / FluxTraining.jl

A flexible neural net training library inspired by fast.ai
https://fluxml.ai/FluxTraining.jl
MIT License
118 stars 25 forks source link

Scheduler applies schedules per batch by default #68

Closed DrChainsaw closed 3 years ago

DrChainsaw commented 3 years ago

Documentation implies that first argument given to Schedule is epoch numbers, but the schedule is applied per batch.

Oh, and nice work with the package btw. It seems quite easy to use while still being very flexible.

lorenzoh commented 3 years ago

This kind of got lost, but it will be fixed in #77. Thanks for the feedback 👍