microsoft/DeepSpeed

Pull Requests Commits

Fix CPU Adam JIT compilation (#5780)

lekurile committed 1 year ago

Verified 681be6f5

Find ROCm on Fedora (#5705)

trixirt committed 1 year ago

Verified 550f9c75

Fix op_builder for CUDA 12.5 (#5806)

keshavkowshik committed 1 year ago

Verified 08598dbb

Pin transformers version for MII tests (#5807)

loadams committed 1 year ago

Verified 5e8a27ad

fix: handle exception when loading cache file in test_inference.py (#5802)

HeyangQin committed 1 year ago

Verified 58241b1d

Add doc of compressed backend in Onebit optimizers (#5782)

Liangliang-Ma committed 1 year ago

Verified afe1b9ed

Add fp8-fused gemm kernel (#5764)

sfc-gh-reyazda committed 1 year ago

Verified 4f950672

Update MII tests to pull correct torchvision (#5800)

loadams committed 1 year ago

Verified f8039434

[XPU]Use host time to replace xpu time when IPEX version slower than 2.5. (#5796)

ys950902 committed 1 year ago

Verified 45b36350

Update other workflows to run on Ubuntu 22.04 (#5798)

loadams committed 1 year ago

Verified ffd0a0e3

Unpin transformers version (#5650)

loadams committed 1 year ago

Verified e661ecb3

move is_checkpointable call reducing torch.compile Graph breaks (#5759)

NirSonnenschein committed 1 year ago

Verified 6d0dbf86

Update torch version in cpu-torch-latest and nv-torch-latest-v100 tests to 2.4 (#5797)

loadams committed 1 year ago

Verified bf696ab6

Fix accuracy error of NPUFusedAdam (#5777)

penn513 committed 1 year ago

Verified 5a100f6b

Add new autotp supported model in doc (#5785)

Yejing-Lai committed 1 year ago

Verified acdf1367

Add chatglm2 & chatglm3 autotp (#5540)

Yejing-Lai committed 1 year ago

Verified 0d3bb77b

fix: quantization with DeepSpeed HE (#5624)

Atry committed 1 year ago

Verified 9fa4c424

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

oelayan7 committed 1 year ago

Verified 830d0c0a

Update Ubuntu version for running python tests (#5783)

loadams committed 1 year ago

Verified 85c66fd7

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

taozhiwei committed 1 year ago

Verified f5d6c631

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

loadams committed 1 year ago

Verified 213e2d97

Misplaced global variable `warned` (#5725)

anferico committed 1 year ago

Verified 879c6cd0

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

adk9 committed 1 year ago

Verified 6a163e03

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

beep-bebop committed 1 year ago

Verified 78c6c449

Launcher mode with SSH bypass (#5728)

dogacancolak-kensho committed 1 year ago

Verified acbaca32

[bugfix] promote state in bf16_optimizer (#5767)

billishyahao committed 1 year ago

Verified 98272d14

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

deepcharm committed 1 year ago

Verified 61e07786

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

delock committed 1 year ago

Verified ec6cbb3c

Fix phi3 mini 128k load error (#5765)

Yejing-Lai committed 1 year ago

Verified a07a3c5d

Remove duplicated variable (#5727)

xu-song committed 1 year ago

Verified 0af9ac31

Newer Older