microsoft/DeepSpeed

Pull Requests Commits

fix: quantization with DeepSpeed HE (#5624)

Atry committed 2 years ago

Verified 9fa4c424

[INF] Add Qwen2RMSNorm to loaded layers in auto_tp (#5786)

oelayan7 committed 2 years ago

Verified 830d0c0a

Update Ubuntu version for running python tests (#5783)

loadams committed 2 years ago

Verified 85c66fd7

reduce all-to-all communication volume when both expert and non-expert are tensor-parallel (#5626)

taozhiwei committed 2 years ago

Verified f5d6c631

Fixes for latest Huggingface_hub changes on modelId -> id (#5789)

loadams committed 2 years ago

Verified 213e2d97

Misplaced global variable `warned` (#5725)

anferico committed 2 years ago

Verified 879c6cd0

Add support for Microsoft Phi-3 model to DeepSpeed-FastGen (#5559)

adk9 committed 2 years ago

Verified 6a163e03

Update the list of supported models in the Chinese README of fastgen (#5773)

beep-bebop committed 2 years ago

Verified 78c6c449

Launcher mode with SSH bypass (#5728)

dogacancolak-kensho committed 2 years ago

Verified acbaca32

[bugfix] promote state in bf16_optimizer (#5767)

billishyahao committed 2 years ago

Verified 98272d14

Added wrappers for hpu tensors based on dtype (#5771)

deepcharm committed 2 years ago

Verified 61e07786

[CPU] Allow deepspeed.comm.inference_all_reduce in torch.compile graph (#5604)

delock committed 2 years ago

Verified ec6cbb3c

Fix phi3 mini 128k load error (#5765)

Yejing-Lai committed 2 years ago

Verified a07a3c5d

Remove duplicated variable (#5727)

xu-song committed 2 years ago

Verified 0af9ac31

Fix memory leak for pipelined optimizer swapper (#5700)

mauryaavinash95 committed 2 years ago

Verified db5a875b

Unit Test: Add error handling for rate limit exceeded in model list (#5715)

HeyangQin committed 2 years ago

Verified 83aa1843

Add Windows scripts (deepspeed, ds_report). (#5699)

costin-eseanu committed 2 years ago

Verified 74f3dcab

[INF] Enable torch compile for inference (#5612)

oelayan7 committed 2 years ago

Verified 7b1ea225

Update checkout action for nv-human-eval workflow (#5757)

loadams committed 2 years ago

Verified 2105976e

fix the missing argument in test and typo (#5730)

xylian86 committed 2 years ago

Verified 774b8977

Update node16 check on self-hosted runners and remove python 3.6 (#5756)

loadams committed 2 years ago

Verified 84118165

Fix tutorial links (#5714)

samadejacobs committed 2 years ago

Verified 3d347276

UCP Chinese Blog (#5713)

HeyangQin committed 2 years ago

Verified dd7a5be5

DeepSpeed Universal Checkpointing: Blog and Tutorial (#5711)

samadejacobs committed 2 years ago

Verified 121efdbd

update xpu fusedadam opbuilder for pytorch 2.3 (#5702)

baodii committed 2 years ago

Verified e3922967

Update XPU docker version (#5712)

loadams committed 2 years ago

Verified df58a784

Add additional paths to trigger xpu tests (#5707)

loadams committed 2 years ago

Verified aecfec7f

Change source of CPUAdam for xpu accelerator (#5703)

Liangliang-Ma committed 2 years ago

Verified 4b8a4a07

Add an argument to enable the injection of missing state during the conversion of universal checkpoints (#5608)

xylian86 committed 2 years ago

Verified f0e3f01d

Disable nvtx decorator to avoid graph break (#5697)

tohtana committed 2 years ago

Verified b421e8c8

Newer Older