microsoft/DeepSpeed

Pull Requests Commits

fix the case where kwargs is None

cli99 committed 2 years ago

7aeb1c48

update pytorch pool operator function signiture

cli99 committed 2 years ago

670e4d64

Fix Bug #2319 (#2438)

jomayeri committed 2 years ago

Verified 7d113633

jeffra committed 2 years ago

a5248643

Fix broken link to DeepSpeed Megatron fork (#2440)

lekurile committed 2 years ago

Verified 877a8818

parallelize writing of layer checkpoint files across data parallel instances (#1419)

adammoody committed 2 years ago

Verified b8fb9c3f

[memory estimators] new config args sync (#2431)

stas00 committed 2 years ago

Verified 99fde3b7

Add TestInjectionPolicy inference unittest class for testing custom injection policies (#2426)

lekurile committed 2 years ago

Verified b2a724e2

only add deps if extra is explictly called (#2432)

jeffra committed 2 years ago

Verified 1b7c6791

Universal checkpoint for zero stage 1 (#2284)

tjruwase committed 2 years ago

Verified 799120e7

Fixing bug 2361 (#2410)

jomayeri committed 2 years ago

Verified 906b4a02

Fix for inference gpt-j test (#2430)

mrwyattii committed 2 years ago

Verified 34fb6d19

fixes #2389 (#2411)

Alexander Jipa committed 2 years ago

Verified cfead551

Make error regex more generic in collect_results.py (#2415)

molly-smith committed 2 years ago

Verified 5eafb8c7

fix checkpoint loading when it is a dictionary (#2425)

RezaYazdaniAminabadi committed 2 years ago

Verified 537e8581

add SD injection policy (#2381)

jeffra committed 2 years ago

Verified ec13da6b

[docs] update mii blog title (#2423)

jeffra committed 2 years ago

Verified d7165374

CI fixes related to triton (#2422)

jeffra committed 2 years ago

Verified 1c39fe13

MII-Public and MII-Azure subheading in mii post

samyam committed 2 years ago

Verified 1ffcfebf

Fix GPT Neo-X multi-gpu inference (#2401)

andrewchernyh committed 2 years ago

Verified cd3a7095

DeepSpeed-MII title change in website

samyam committed 2 years ago

Verified 843fd4c1

MII blog title update on Readme

samyam committed 2 years ago

Verified 3789dc5c

Fix issue with corrupted output on long generation for GPT (#2359)

andrewchernyh committed 2 years ago

Verified d5d10b0c

Add SLURM Multinode Runner (#2404)

dashstander committed 2 years ago

Verified 3db0b5e2

[docs] add mii repo link

jeffra committed 2 years ago

5a162132

[docs] update news items

jeffra committed 2 years ago

ab77677d

Fix figure reference (#2419)

awan-10 committed 2 years ago

Verified d0c6b295

MII blog post (#2418)

jeffra committed 2 years ago

Verified b8b98c60

Add predicated global load (#2373)

cmikeh2 committed 2 years ago

Verified c3001324

Checkpoint backwards-compatbility workaround (#2384)

tjruwase committed 2 years ago

Verified f4a92a19

Older