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.

ver217 9506a8beb2 use double buffer to handle grad		3 years ago
..
amp	[fp16] refactored fp16 optimizer (#392)	3 years ago
builder	add pytorch hooks (#179)	3 years ago
communication	fix format (#332)	3 years ago
context	fixed bug in activation checkpointing test (#387)	3 years ago
engine	use double buffer to handle grad	3 years ago
kernel	[formart] format fixed for kernel\cuda_native codes (#335)	3 years ago
logging	[log] better logging display with rich (#426)	3 years ago
nn	removed noisy function during evaluation of MoE router (#419)	3 years ago
registry	add pytorch hooks (#179)	3 years ago
trainer	Added profiler communication operations	3 years ago
utils	[zero] cuda margin space for OS (#418)	3 years ago
zero	use double buffer to handle grad	3 years ago
__init__.py	Develop/experiments (#59)	3 years ago
constants.py	fix format constants.py (#358)	3 years ago
core.py	Develop/experiments (#59)	3 years ago
global_variables.py	Optimized MoE layer and fixed some bugs;	3 years ago
initialize.py	[fp16] refactored fp16 optimizer (#392)	3 years ago