microsoft/DeepSpeed

Pull Requests Commits

Only run unittest for world size 2

molly-smith committed 2 years ago

9c0942e3

typo replace_method

molly-smith committed 2 years ago

a763b3f1

molly-smith committed 2 years ago

ae473a42

Remove other param instance

molly-smith committed 2 years ago

a14ffbf1

remove extra param

molly-smith committed 2 years ago

d8aedd6e

molly-smith committed 2 years ago

688a0a30

Replace codegen unittest with gptj

molly-smith committed 2 years ago

3edfac8a

Add codegen unittest

molly-smith committed 3 years ago

8e959452

Merge branch 'master' into mosm/module_parser

molly-smith committed 3 years ago

a502ad48

[deepspeed/autotuner] Bug fix for binary search for batch size (#2162)

rahilbathwal5 committed 3 years ago

Verified 384f17b0

Generated text bug fix. Divide bias by mp size

molly-smith committed 3 years ago

158ba8f3

fix some issues

Reza Yazdani committed 3 years ago

ad86f101

change matmul with F.linear (seems to have better performance)

Reza Yazdani committed 3 years ago

a53f298e

Move layer norm to new schedule (#2590)

lokoppakmsft committed 3 years ago

Verified 3a3dfe66

Migrate ops tests to new inference_ops marker (#2599)

cmikeh2 committed 3 years ago

Verified 7425a365

fix blog link (#2600)

conglongli committed 3 years ago

Verified acde873c

DeepSpeed Data Efficiency Library (#2585)

conglongli committed 3 years ago

Verified ef869377

Revert "change matmul with F.linear (seems to have better performance)"

molly-smith committed 3 years ago

818d143a

Revert "fix some issues"

molly-smith committed 3 years ago

d1d3d9b5

jeffra committed 3 years ago

2600db54

get mask token from tokenizer (#2592)

jeffra committed 3 years ago

Verified 2076bf23

Fix issues w. python 3.6 + add py-version checks to CI (#2589)

jeffra committed 3 years ago

Verified 35eabb0a

Updating API docs (#2586)

jomayeri committed 3 years ago

Verified 18713c68

Updating docs README (#2587)

jomayeri committed 3 years ago

Verified 377c770a

Add checkpoint sharding unit tests (#2561)

mrwyattii committed 3 years ago

Verified ccb8eb81

Support N-dimension input in quantization kernel (#2575)

lokoppakmsft committed 3 years ago

Verified 591744eb

Update barrier and reduce_scatter_base to conform to PyTorch signatures (#2570)

Quentin-Anthony committed 3 years ago

Verified 18d55e54

Fix MegatronLayerPolicy to have megatron_v2=True (#2579)

lekurile committed 3 years ago

Verified 731965db

fix some issues

Reza Yazdani committed 3 years ago

6ab70f77

Fix quantized-inference & Add generic support of checkpoint loading (#2547)

RezaYazdaniAminabadi committed 3 years ago

Verified 35b350b2

Older