hello.mmseg.lr module

LRFinder(https://docs.fast.ai/callback.schedule.html#lrfinder)

\[\begin{split}\begin{align} & lr_i = lr_{min} \times q^i , \quad i \in [0, N-1]\\ & lr_{min} = 1e-07 , \quad lr_{max} = 10\\ & q = \lgroup \frac{lr_{max}}{lr_{min}} \rgroup ^ {\frac{1}{N-1}} \end{align}\end{split}\]

hello.mmseg.lr.func(text_logs, out_dir, lr_field='lr', loss_field='loss', reduction='mean', format='.png')

hello.mmseg.lr.load_text_log(text_log)

hello.mmseg.lr.main(args=None)

hello.mmseg.lr.parse_args(args=None)

hello.mmseg.lr.plot_lr_loss(log_dict, lr_field='lr', loss_field='loss', reduction='mean')

hello.mmseg.lr.to_dict(text)

hello.mmseg.lr.valley(lrs: list, losses: list)

Suggests a learning rate from the longest valley and returns its index.

https://github.com/fastai/fastai/blob/master/fastai/callback/schedule.py

Parameters:

• lrs (list) – _description_

• losses (list) – _description_