pytorch/ignite

enable automatic mixed precision for xla

Open

#1931 aperta il 12 apr 2021

Vedi su GitHub
 (8 commenti) (1 reazione) (0 assegnatari)Python (602 fork)batch import
enhancementhelp wanted

Metriche repository

Star
 (4313 star)
Metriche merge PR
 (Merge medio 18g 23h) (19 PR mergiate in 30 g)

Descrizione

Feature

Automatic mixed precision for xla has landed in pytorch 1.8.1 and torch/xla nightly. We should enable it in create_supervised_* helper functions.

Suggested solution

Remove xla and amp checks in _check_arg().

  • For create_supervised_trainer, update supervised_training_step_tpu() function to accept scaler argument just like supervised_training_step_amp().
  • For create_supervised_evaluator, just removing xla and amp checks in _check_arg() should work.
  • For tests, we could remove xla checks and only run with pytorch 1.8.1.

Additional context

This feature should not be included in ignite release until the next torch and xla comes out.

Guida contributor