microsoft/DeepSpeed

Pull Requests Commits

Test updating workflows to 2.0

loadams committed 2 years ago

538895ba

Merge branch 'loadams/fix-nccl-comm-torch-check' into loadams/skip-on-torch-rocm

loadams committed 2 years ago

3a5855e7

Fix torch requirements on nccl

loadams committed 2 years ago

a10edc01

Fix formatting and make it more readable

loadams committed 2 years ago

6ea179aa

Updating torch requirements to better handle torch 2+

loadams committed 2 years ago

d927abe6

Merge branch 'master' into loadams/skip-on-torch-rocm

loadams committed 2 years ago

2199d7cc

Update assert with more info

loadams committed 2 years ago

e48e54da

fix CI badges (#3138)

mrwyattii committed 2 years ago

Verified f2c9a827

Merge branch 'master' into loadams/skip-on-torch-rocm

loadams committed 2 years ago

Verified f7334e88

Fix launch issue (#3137)

tjruwase committed 2 years ago

Verified fcb868e2

Merge branch 'master' into loadams/skip-on-torch-rocm

loadams committed 2 years ago

Verified 6fc73d61

Clean up formatting

loadams committed 2 years ago

e90b761f

Replace old torch version checks with existing function

loadams committed 2 years ago

7b4eb61b

Update SD triton version in requirements-sd.txt (#3135)

lekurile committed 2 years ago

Verified fcc0d9c0

[docs] fix blog links

jeffra committed 2 years ago

62db3c0d

VL MoE Blog (#3120)

yaozhewei committed 2 years ago

Verified 1242d8b4

added full apache license (#3119)

mrwyattii committed 2 years ago

Verified 6c4782b0

Update DeepSpeed copyright license to Apache 2.0 (#3111)

mrwyattii committed 2 years ago

Verified b361c727

Make fp32 default communication data type (#2970)

tjruwase committed 2 years ago

Verified 261d6370

♻️ replace deprecated functions for communication (#2995)

mayank31398 committed 2 years ago

Verified a6317eb5

Disable Stage 1&2 CPUAdam pathways (#3097)

mrwyattii committed 2 years ago

Verified 4b6d7c15

Fix comms benchmark import issues and support MPI/slurm launching (#2932)

Quentin-Anthony committed 2 years ago

Verified 9726bd46

update formatter version and style settings (#3098)

jeffra committed 2 years ago

Verified 91d63e02

Move cuda check into utils (#3074)

loadams committed 2 years ago

Verified b3ec1c97

pre-commit check for torch.cuda in code (#2981)

delock committed 2 years ago

Verified 090d49e7

Empty ZeRO3 partition cache (#3060)

tjruwase committed 2 years ago

Verified e80ae088

Goodbye Torch 1.8 (#3082)

mrwyattii committed 2 years ago

Verified 5cdf3593

allow list (#3042)

satpalsr committed 2 years ago

Verified 5c2a81c2

Fix nebula in save_16bit_model issue (#3023)

FreyaRao committed 2 years ago

Verified a78d6b89

Softmax Scheduling Cleanup (#3046)

cmikeh2 committed 2 years ago

Verified 1286e374

Older