You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

Zhen Wang 7f7dfccf20 Support pure fp16 training for AMP API. (#29544) ... * add cast ops before and after unsupported fp16 ops. * Keep partial net in FP32 pattern. * Support check_finite_and_unscale and update_loss_scaling for FP16 calculation mode. * Add fp16 support for adam op. * add multi precision attr for adam. * Fix the bug of test_multi_precision_fp16_train UT. * Code format for CI. * Fix the redefine error about MPTypeTrait on windows. * fix bugs of the _create_accumulators func in Momentum. * fix bug when inserting post cast op. * Add the update_loss_scaling op in allow_set of UnusedVarCheck. * Update for ci coverage. * Add some doc for OptimizerWithMixedPrecision. * Fix the code style. * Imporve the doc of `amp_init`. * Change for fp16 testing if users have the infer program defined in separate way.		4 years ago
..
CMakeLists.txt	Compiling operator libraries with Unity build (#29130)	4 years ago
check_finite_and_unscale_op.cc	Update amp_check_finite_and_scale_op and add an updating_loss_scaling op for static graph amp training. (#26240)	4 years ago
check_finite_and_unscale_op.cu	Support pure fp16 training for AMP API. (#29544)	4 years ago
check_finite_and_unscale_op.h	Update amp_check_finite_and_scale_op and add an updating_loss_scaling op for static graph amp training. (#26240)	4 years ago
fp16_type_traits.h	Support pure fp16 training for AMP API. (#29544)	4 years ago
unity_build_rule.cmake	Compiling operator libraries with Unity build (#29130)	4 years ago
update_loss_scaling_op.cc	Support pure fp16 training for AMP API. (#29544)	4 years ago
update_loss_scaling_op.cu	Support pure fp16 training for AMP API. (#29544)	4 years ago
update_loss_scaling_op.h	Support pure fp16 training for AMP API. (#29544)	4 years ago