Migrating from Lightning

Intro

Lightning disentangles PyTorch code to decouple the science from the engineering.

Lightning with fastai

library(fastai)
library(magrittr)

model = LitModel()

data = Data_Loaders(model$train_dataloader(), model$val_dataloader())$cuda()

nn = nn()
learn = Learner(data, model, loss_func = nn$functional$cross_entropy, opt_func = Adam,
                metrics = accuracy)
learn %>% fit_one_cycle(1, 0.001)

epoch     train_loss  valid_loss  accuracy  time    
0         0.354146    0.334655    0.911300  00:09