MT5, Adafactor optimizer, additional schedulers
·
683 commits
to master
since this release
MT5, Adafactor optimizer, additional schedulers
Breaking change
T5Modelnow has a requiredmodel_typeparameter ("t5"or"mt5")
Added
- Added support for MT5
- Added support for Adafactor optimizer
- Added support for various schedulers:
- get_constant_schedule
- get_constant_schedule_with_warmup
- get_linear_schedule_with_warmup
- get_cosine_schedule_with_warmup
- get_cosine_with_hard_restarts_schedule_with_warmup
- get_polynomial_decay_schedule_with_warmup
Changed
T5Modelnow has a requiredmodel_typeparameter ("t5"or"mt5")
Fixed
- Fixed issue with class weights not working in
ClassificationModelwhen using mult-GPU training