microsoft/DeepSpeed

Pull Requests Commits

Respond to PR feedback

loadams committed 2 years ago

b34edfea

Merge branch 'master' into loadams/fix-fp16-bf16-logging-issue

loadams committed 2 years ago

Verified 2b985e07

Add links to new azureML examples (#2756)

cassieesvelt committed 2 years ago

Verified 1db4ade3

Merge branch 'master' into loadams/fix-fp16-bf16-logging-issue

loadams committed 2 years ago

Verified fa7c0f2f

Bing/formatting correction (#2764)

xiexbing committed 2 years ago

Verified 8d3b42c2

Merge conflicts messed up pre-commit

loadams committed 2 years ago

3b23ce89

Merge conflicts

loadams committed 2 years ago

e2b28712

Missed once instance

loadams committed 2 years ago

64a61aee

Add model_dtype to print the correct format

loadams committed 2 years ago

503170c8

Add environment variable to make nvcc compilation more verbose (#2759)

cmikeh2 committed 2 years ago

Verified 258d2831

Skip test_bias_gelu unit test if torch < 1.12 (#2754)

lekurile committed 2 years ago

Verified cc3d7cb9

Fix softmax backward (#2709)

RezaYazdaniAminabadi committed 2 years ago

Verified 0b06e0cb

[zero] remove misleading dtype log (#2732)

jeffra committed 2 years ago

Verified a60e31a7

fix a mispelled attribute (#2750)

stas00 committed 2 years ago

Verified 30d3f5df

Fix autotuning so that it records Floating Point Operations per second, not microsecond (#2711)

dashstander committed 2 years ago

Verified d4bfae41

Abstract accelerator (step 3) (#2677)

delock committed 2 years ago

Verified 98cc35b6

[GatheredParameters] fix memory leak (#2665)

stas00 committed 2 years ago

Verified ddd48b36

fixing optimizer sanity check (#2742)

jomayeri committed 2 years ago

Verified 4be8df72

Merge branch 'master' into loadams/fix-fp16-bf16-logging-issue

loadams committed 2 years ago

71402dcc

loadams committed 2 years ago

76316651

Automatic tensor parallelism v2 (#2670)

molly-smith committed 2 years ago

Verified d59b5729

Merge branch 'master' into loadams/fix-fp16-bf16-logging-issue

tjruwase committed 2 years ago

Verified faf565b0

Change zero_grad() argument to match pytorch (#2741)

loadams committed 2 years ago

Verified 34a11688

Revert "Update docutils version to fix errors with sphinx"

loadams committed 2 years ago

293e70e4

Update docutils version to fix errors with sphinx

loadams committed 2 years ago

a8556e00

Remove hardcoded instances to fp16 in log messages.

loadams committed 2 years ago

6b8347ae

Inference Refactor (replace_with_policy, model_implementations) (#2554)

awan-10 committed 2 years ago

Verified 867da307

fix typo (#2718)

mrwyattii committed 2 years ago

Verified 8df50a26

BF16 optimizer for BF16+ZeRO Stage 1 (#2706)

jomayeri committed 2 years ago

Verified 8d87c89e

update for lm-eval==0.3.0 (#2713)

mrwyattii committed 2 years ago

Verified 23e5133c

Older