deepspeedai/DeepSpeed

Pull Requests Commits

Merge branch 'master' into loadams/transformers-torch-update

loadams committed 2 years ago

Verified e855ef30

Merge branch 'master' into loadams/transformers-torch-update

loadams committed 2 years ago

Verified 7a63eede

[NPU] Fix npu offload bug (#4883)

CurryRice233 committed 2 years ago

Verified d873ce61

Fix f-string messages (#4865)

li-plus committed 2 years ago

Verified 9ec55bd9

Merge branch 'master' into loadams/transformers-torch-update

loadams committed 2 years ago

Verified 3f48690f

optimize grad_norm calculation in stage3.py (#4436)

mmhab committed 2 years ago

Verified ea0d8114

Remove hooks on gradient accumulation on engine/optimizer destroy (#4858)

chiragjn committed 2 years ago

Verified 40342055

deepspeed engine: fp16 support validation on init (#4843)

nelyahu committed 2 years ago

Verified 3e94f8c7

Fix exception handling in get_all_ranks_from_group() function (#4862)

HeyangQin committed 2 years ago

Verified c37fe9cb

doc corrections (#4861)

goodship1 committed 2 years ago

Verified 75c77202

Update version.txt after 0.12.6 release (#4850)

mrwyattii committed 2 years ago

Verified 827e1ca8

Mixtral FastGen Support (#4828)

cmikeh2 committed 2 years ago

Verified c00388a2

engine.py: remove unused _curr_save_path (#4844)

nelyahu committed 2 years ago

Verified 18643914

Checkpointing: Avoid assigning tensor storage with different device (#4836)

deepcharm committed 2 years ago

Verified 9e455d76

Capture short kernel sequences to graph (#4318)

inkcherry committed 2 years ago

Verified d5a7c1e0

Add cuda_accelerator.py to triggers for A6000 test (#4848)

mrwyattii committed 2 years ago

Verified 060a8e18

[NPU] load EXPORT_ENV based on different accelerators to support multi-node training on other devices (#4830)

minchao-sun committed 2 years ago

Verified 6d7b44a8

Update zeropp.md (#4835)

goodship1 committed 2 years ago

Verified a00bdde8

inference engine: check if accelerator supports FP16 (#4832)

nelyahu committed 2 years ago

Verified 3f1c3c2a

support baichuan model: (#4721)

baodii committed 2 years ago

Verified c20f6fa4

Update CODEOWNERS (#4838)

mrwyattii committed 2 years ago

Verified 83fa673a

Inference changes for incorporating meta loading checkpoint (#4692)

oelayan7 committed 2 years ago

Verified 4c2cac03

Cache metadata for TP activations and grads (#4360)

BacharL committed 2 years ago

Verified 4559dadd

Merge branch 'master' into loadams/transformers-torch-update

loadams committed 2 years ago

Verified 0fd2dbbc

Update version.txt after 0.12.5 release (#4826)

mrwyattii committed 2 years ago

Verified 4d866bd5

Merge branch 'master' into loadams/transformers-torch-update

loadams committed 2 years ago

Verified a31af1fe

Fix 4649 (#4650)

Alienfeel committed 2 years ago

Verified 65b77277

Refactor launcher user arg parsing (#4824)

mrwyattii committed 2 years ago

Verified d37fc25d

mv DeepSpeedEngine param_names dict init post _configure_distributed_model (#4803)

nelyahu committed 2 years ago

Verified 449e454f

fix falcon model load from_config meta_data error (#4783)

baodii committed 2 years ago

Verified faa00b13

Older