microsoft/DeepSpeed

Pull Requests Commits

fix if statement for self.mpu (#3497)

mrwyattii committed 2 years ago

Verified 1824cb55

Doc updates (#3520)

goodship1 committed 2 years ago

Verified 82991df0

fix typo with docs/ (#3523)

digger-yu committed 2 years ago

Verified d73f99d8

change actions/checkout@v2 to v3 (#3526)

digger-yu committed 2 years ago

Verified b3956dc9

Changing monitor loss to aggregate loss over gradient accumulation steps (#3428)

jomayeri committed 2 years ago

Verified 4d269c6e

Skip autoTP if tp_size is 1 (#3449)

molly-smith committed 2 years ago

Verified 5979ece8

fix spelling error with deepspeed/runtime/ (#3509)

digger-yu committed 2 years ago

Verified 254663a2

Hybrid Engine Fix Llama (#3505)

lekurile committed 2 years ago

Verified 194053bd

Update README.md (#3504)

goodship1 committed 2 years ago

Verified b303fa8b

Add snip_momentum structured pruning which can support higher sparse ratio with minor accuracy loss (#3300)

ftian1 committed 2 years ago

Verified 6938c449

fix regression in shard checkpoint loading in AutoTP Path caused by qkv_copy() is deleted and add UT case for shard checkpoint loading in AutoTP (#3457)

sywangyi committed 2 years ago

Verified b31b46c0

Update Inference Engine checkpoint loading + meta tensor assertions (#2940)

lekurile committed 2 years ago

Verified db26f8b4

delete the line to keep user_zero_stages (#3473)

MrZhengXin committed 2 years ago

Verified 99dfac9a

fix spelling error with docs/index.md (#3443)

digger-yu committed 2 years ago

Verified 58c4d230

fix spelling error with deepspeed/ (#3494)

digger-yu committed 2 years ago

Verified 19816642

fix a small type error on bf16+pp (#3441)

ys950902 committed 2 years ago

Verified 195563a2

fix spelling error (#3482)

luliyucoordinate committed 2 years ago

Verified f1fab902

Fix spelling errors in comments and documents (#3486)

digger-yu committed 2 years ago

Verified 87edbc8d

Doc corrections (#3435)

goodship1 committed 2 years ago

Verified 41321180

Build: Update license in setup (#3484)

PabloEmidio committed 2 years ago

Verified f3f4c449

[Docs] Update automatic-tensor-parallelism.md (#3198)

sywangyi committed 2 years ago

Verified 9715d802

launcher/multinode_runner.py: mapping env variables in running cmd for mpich runner (#3372)

YizhouZ committed 2 years ago

Verified 4e886f05

add sharded checkpoint loading for AutoTP path to reduce the peak mem… (#3102)

sywangyi committed 2 years ago

Verified d10b8ca0

Hybrid Engine Refactor and Llama Inference Support (#3425)

cmikeh2 committed 2 years ago

Verified 0a61d5d6

add skip for mismatched cuda versions (#3436)

mrwyattii committed 2 years ago

Verified d2e6110e

Enable auto TP policy for llama model (#3170)

jianan-gu committed 2 years ago

Verified 2c63e349

jeffra committed 2 years ago

5314c9a1

Update 2020-09-09-sparse-attention.md (#3432)

goodship1 committed 2 years ago

Verified e0e80850

Adagrad support in ZeRO (#3401)

jomayeri committed 2 years ago

Verified d3550dc8

[zero_to_fp32] fix shared param recovery (#3407)

stas00 committed 2 years ago

Verified 77ebf760

Older