Masahiro Tanaka 24285d6c73 Add fallback for is_compiling (#6663) ... Importing `torch.compiler.is_compiling` causes an error with an older version of PyTorch. This PR adds a fallback for `is_compiling` to use an equivalent function of older PyTorch versions. This will resolve #6656. Co-authored-by: Logan Adams <114770087+loadams@users.noreply.github.com>		2024-10-25 20:47:22 +00:00
..
__init__.py	Add API for updating ZeRO gradients (#6590)	2024-10-14 17:35:41 +00:00
bwc.py	Support MoE for pipeline models (#5338)	2024-04-08 15:35:53 +00:00
comms_logging.py	fix comm logging for inference (#4043)	2023-07-27 17:52:05 +00:00
config.py	Add throughput timer configuration (#5363)	2024-05-22 20:28:02 +00:00
debug.py	Remove hooks on gradient accumulation on engine/optimizer destroy (#4858)	2023-12-30 02:48:49 +00:00
exceptions.py	Update DeepSpeed copyright license to Apache 2.0 (#3111)	2023-03-30 17:14:38 -07:00
groups.py	sequence parallel for uneven heads (#6392)	2024-10-25 18:26:47 +00:00
init_on_device.py	Update DeepSpeed copyright license to Apache 2.0 (#3111)	2023-03-30 17:14:38 -07:00
logging.py	Add fallback for is_compiling (#6663)	2024-10-25 20:47:22 +00:00
mixed_precision_linkage.py	Add API for updating ZeRO gradients (#6590)	2024-10-14 17:35:41 +00:00
numa.py	Safe usage of popen (#6490)	2024-09-04 21:06:04 +00:00
nvtx.py	Disable nvtx decorator to avoid graph break (#5697)	2024-06-26 22:30:56 +00:00
tensor_fragment.py	Add API for updating ZeRO gradients (#6590)	2024-10-14 17:35:41 +00:00
timer.py	Improve DS logging control (#6602)	2024-10-08 18:38:51 +00:00
torch.py	Cleanup required_torch_version code and references. (#5370)	2024-04-10 15:39:24 +00:00
types.py	Hybrid Engine Refactor and Llama Inference Support (#3425)	2023-05-03 17:20:07 -07:00
z3_leaf_module.py	Delay reduce-scatter for ZeRO3 leaf modules (#5008)	2024-02-02 13:44:08 -08:00
zero_to_fp32.py	Support safetensors export (#6579)	2024-10-15 11:22:31 +00:00