Olatunji Ruwase 470dd6dceb Precisely track nvme optimizer offload (#6963) ... Fix #4998		2025-01-23 16:42:06 +00:00
..
autotuning	docs: fix HF links (#6780)	2024-11-25 10:10:08 -08:00
checkpoint	Removes unnecessary cloning (#6761)	2024-11-21 17:37:29 +00:00
comm	Allow to compile collective for PT>2.3 (#6899)	2024-12-19 09:26:50 -08:00
compression	…
elasticity	…
inference	[inf] Add config var to enable keeping module on host (#6846)	2025-01-15 19:25:29 +00:00
launcher	Fix --enable_each_rank_log when used with PDSH multi-node runner (#6863)	2024-12-17 09:33:09 -08:00
linear	…
model_implementations	…
module_inject	Add extra_repr to Linear classes for debugging purpose (#6954)	2025-01-16 18:11:07 +00:00
moe	…
monitor	…
nebula	…
nvme	…
ops	Fix building on Windows with presence of Triton (#6749)	2025-01-08 18:59:41 +00:00
pipe	…
profiling	Flops profiler support einops.einsum (#6755)	2024-12-09 09:56:54 -08:00
runtime	Precisely track nvme optimizer offload (#6963)	2025-01-23 16:42:06 +00:00
sequence	Add the missing view operations from sequence parallel(async). (#6750)	2025-01-21 16:49:06 +00:00
utils	Stage3: Use new torch grad accumulation hooks API (#6773)	2025-01-03 07:48:24 -08:00
__init__.py	…
accelerator	…
constants.py	…
env_report.py	…
git_version_info.py	…