pytorch-optimizer v2.0.0
Chage Log
- Refactor the package depth
- 4 depths
pytorch_optimizer.lr_scheduler
: lr schedulerspytorch_optimizer.optimizer
: optimizerspytorch_optimizer.base
: base utilspytorch_optimizer.experimental
: any experimental features
pytorch_optimizer.adamp
->pytorch_optimizer.optimizer.adamp
- Still
from pytorch_optimizer import AdamP
is possible
- 4 depths
- Implement lr schedulers
-
CosineAnealingWarmupRestarts
-
- Implement (experimental) lr schedulers
-
DeBERTaV3-large
layer-wise lr scheduler
-
Other changes (bug fixes, small refactors)
- Fix
AGC
(to returning the parameter) - Make a room for
experimental features
(atpytorch_optimizer.experimental
) - base types