microsoft/DeepSpeed

Pull Requests Commits

molly-smith committed 2 years ago

d9edcbb0

AutoTP for llama

molly-smith committed 2 years ago

f87d92f9

Hipify cooperative_groups headers (#3323)

rraminen committed 2 years ago

Verified 6e1cbebe

Fix formatting (#3343)

mrwyattii committed 2 years ago

Verified 4a3ca4e2

jeffra committed 2 years ago

78afab04

Explicitly check for OPT activation function (#3278)

cmikeh2 committed 2 years ago

Verified 793c23e5

Fix missing scale attributes for GPTJ (#3256)

cmikeh2 committed 2 years ago

Verified 145c3a75

Fix for dist not being initialized when constructing main config (#3324)

mrwyattii committed 2 years ago

Verified ad168a69

zero3 checkpoint frozen params (#3205)

tjruwase committed 2 years ago

Verified dd8df20f

Update Dockerfile (#3298)

bobowwb committed 2 years ago

Verified ab1d2f82

Update README.md (#3315)

digger-yu committed 2 years ago

Verified dd902a3c

Add pre-compiling ops test (#3277)

loadams committed 2 years ago

Verified 4de4d2ac

Update MI200 workflow to install apex with changes from pip (#3294)

loadams committed 2 years ago

Verified 61a8b3a0

remove `torch.cuda.is_available()` check when compiling ops (#3085)

jinzhen-lin committed 2 years ago

Verified 036c5d6d

Update README.md (#3282)

digger-yu committed 2 years ago

Verified c381dd16

Fix pydantic and autodoc_pydantic (#3290)

loadams committed 2 years ago

Verified 08905623

[ROCm] temporary workaround till __double2half support enabled in HIP (#3236)

bmedishe committed 2 years ago

Verified f7bfe5e7

Fix cupy install version detection (#3276)

mrwyattii committed 2 years ago

Verified bcccee4d

Update install.sh (#3270)

digger-yu committed 2 years ago

Verified 077e42e6

improving int4 asymmetric quantization accuracy (#3190)

HeyangQin committed 2 years ago

Verified 48297c48

fix typo in autotuner.py (#3269)

eltociear committed 2 years ago

Verified 050aee28

update link to PL docs (#3237)

Borda committed 2 years ago

Verified 6dbff771

[fp16/doc] correct initial_scale_power default value (#3275)

stas00 committed 2 years ago

Verified 72c2d6df

[update] reference in cifar-10.md (#3212)

3xdti committed 2 years ago

Verified cc67f22f

fix for SD afte DeepSpeed-chat merge (#3218)

mrwyattii committed 2 years ago

Verified fef5aa6e

Create deepspeed_chat_bug_report.md

mrwyattii committed 2 years ago

Verified 6fc8e33c

Update torch version check in building sparse_attn (#3152)

loadams committed 2 years ago

Verified 94fbd947

Nested zero.Init() and dynamically defined model class (#2989)

tohtana committed 2 years ago

Verified 717c3020

Update DeepSpeed-Chat docs with latest changes to scripts (#3219)

mrwyattii committed 2 years ago

Verified a8f999e3

change pip install of deepspeed to 0.9.0 (#3216)

mrwyattii committed 2 years ago

Verified ad97ca34

Older