microsoft/DeepSpeed

Pull Requests Commits

force upgrade HF dependencies

jeffra committed 2 years ago

1655de0d

added perf test and fix for gpt-j-6B OOM on V100 16GB

mrwyattii committed 2 years ago

2777c23e

increased acceptable ppl diff val

mrwyattii committed 2 years ago

1d21343a

during injection reshape data on cpu to avoid extra GPU memory

jeffra committed 2 years ago

f99fc0d7

fixed device string for lm tests

mrwyattii committed 2 years ago

71a58405

added correctness tests for text-generation gpt models

mrwyattii committed 2 years ago

d3f6dfe9

force inference tests to be seq

jeffra committed 2 years ago

f3e40aea

manual convert gpt-j fp32->fp16, seeing issues with fp16 revision

jeffra committed 2 years ago

50cbfced

remove dep on hf-accelerate

jeffra committed 2 years ago

8b88eeff

Merge branch 'master' into jeffra/fill-mask-tests

jeffra committed 2 years ago

Verified d37f9d9f

DeepSpeed comm backend v1 (#1985)

awan-10 committed 2 years ago

Verified 36ad3119

convert pipe model to half for non-DS case

jeffra committed 2 years ago

3ec3f9f8

use fp16 revision of gpt-j

jeffra committed 2 years ago

2677d20c

add pretty print names for test fixtures

jeffra committed 2 years ago

42841c6e

prune incompatible model+task pairs

jeffra committed 2 years ago

816e746a

ensure torch > 1.10 if cuda graph enabled, also fix typo

jeffra committed 2 years ago

1d08254f

cuda graph requires 1.10+

jeffra committed 2 years ago

0872f9f7

refactored validation of test config, add skip for AMD w/CUDA Graph

mrwyattii committed 2 years ago

2c88ab1a

mrwyattii committed 2 years ago

15ed0a25

revert to not checking GPU memory

mrwyattii committed 2 years ago

d864d2b1

avoid CUDA init when getting GPU memory

mrwyattii committed 2 years ago

3d93c078

fixed bad comparison

mrwyattii committed 2 years ago

9cdd9814

added fp16 tests

mrwyattii committed 2 years ago

31754302

mrwyattii committed 2 years ago

0519a723

Merge branch 'jeffra/fill-mask-tests' of github.com:microsoft/DeepSpeed into jeffra/fill-mask-tests

Reza Yazdani committed 2 years ago

4440aedf

fixing some issue for the Bert-Postln model

Reza Yazdani committed 2 years ago

c10acaf4

added temporary fix to get gpt models passing

mrwyattii committed 2 years ago

9c06f04c

Merge branch 'master' into jeffra/fill-mask-tests

RezaYazdaniAminabadi committed 2 years ago

Verified fbdd6b86

added back gptj model

mrwyattii committed 2 years ago

fc569e9f

keyerror with text-gen assert_fn

mrwyattii committed 2 years ago

fbc05fef

Older