epoch!
function
defined in module
FluxTraining
epoch!(learner, phase[, dataiter])
Train
learner
for one epoch on
dataiter
. Iterates through
dataiter
and
step!
s for each batch/item.
If no data iterator is passed in, use
learner.data[phasedataiter(phase)]
.
The default implementation iterates over every batch in
dataiter
and calls
step!
for each. This behavior can be overloaded by implementing
epoch!(learner, ::MyPhase, dataiter)
.
If you're implementing a custom
epoch!
method, it is recommended you make use of
runepoch
to get begin and end events as well as proper handling of
CancelEpochException
s.
See the default implementation for reference.
There is
1
method for FluxTraining.epoch!
:
The following pages link back here:
training/onecycle.jl , FluxTraining.jl , callbacks/conditional.jl , callbacks/custom.jl , training.jl