Hello! I have a code build using fastai v0.7 (the symbolic link is towards fastai/old/fastai). The main lines of code related to fastai are these:
from fastai.learner import *
data = ModelData(PATH, trn_dl = train_loader, val_dl = validation_loader)
learn = Learner.from_model_data(SimpleNet(n_variables), data, crit=rmse_loss)
learn.opt_fn = optim.Adam
learn.fit(5e-3, 1, cycle_len=100, use_clr_beta = (20,10,0.95,0.85), wds=wd, use_wd_sched=True)
learn.save(filename)
where train_loader and validation_loader correspond to some numbers (they are not images, or text, just numbers; for example given (1,2,3) it has to predict 7, this is the kind of data). This doesn’t work in fastai v1.0 and I am not sure how can I modify it. Can someone help me? Thank you!