microsoft/DeepSpeed

Pull Requests Commits

Allowing optimizer params to pass to optimizer callable.

jomayeri committed 2 years ago

bdff8227

Fix Hybrid Engine for BLOOM (#3580)

lekurile committed 2 years ago

Verified 76679884

Fixing bf16 test (#3551)

jomayeri committed 2 years ago

Verified 49d399cd

fix_typo (#3559)

inkcherry committed 2 years ago

Verified d1c3c0df

rraminen committed 2 years ago

Verified 2fc56841

Clone tensors to avoid torch.save bloat (#3348)

tjruwase committed 2 years ago

Verified 5c3ebd7e

[CPU] Support Intel CPU inference (#3041)

delock committed 2 years ago

Verified 1f72082f

[ROCm] Hip headers fix (#3532)

rraminen committed 2 years ago

Verified 5147b90a

fix typo in comments with deepspeed/ (#3537)

digger-yu committed 2 years ago

Verified c8d3f5eb

Syncing FusedAdam with new Apex features (#3434)

jomayeri committed 2 years ago

Verified 9685eb92

share inflight registry between PartitionedParameterCoordinators (#3462)

HeyangQin committed 2 years ago

Verified 4716b0f7

Fix broadcast error on multi-node training with ZeroStage3 and TensorParallel=2 (#2999)

YizhouZ committed 2 years ago

Verified 9f4a8763

Fix passing offload_optimizer_config instead of offload_param_config to (#3420)

mmhab committed 2 years ago

Verified 5c6da1f0

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

Older