microsoft/DeepSpeed

Pull Requests Commits

Merge branch 'master' into loadams/fix-check-valid-version

loadams committed 1 year ago

Verified f960a32a

Update formatting

loadams committed 1 year ago

5fc579a4

Fixup check release version script

loadams committed 1 year ago

7f11e138

add CPU autotp UT (#4263)

Yejing-Lai committed 1 year ago

Verified 388c8483

Add condition when dimension is greater than 2 (#4390)

ajindal1 committed 1 year ago

Verified 28b9d5c2

Update nv-transformers workflow to use cu11.6 (#4412)

loadams committed 1 year ago

Verified 58619402

Add torch no grad condition (#4391)

ajindal1 committed 1 year ago

Verified e3393641

update deepspeed4science blog (#4408)

conglongli committed 1 year ago

Verified aea10eec

Fix incorrect assignment of self.quantized_nontrainable_weights (#4399)

VeryLazyBoy committed 1 year ago

Verified 60bf7845

save/restore step in param groups with zero 1 (#4396)

tohtana committed 1 year ago

Verified f8d3ec7f

Allow multiple inference engines in single script (#4384)

mrwyattii committed 1 year ago

Verified 4c35880b

adds triton flash attention2 kernel (#4337)

stephen-youn committed 1 year ago

Verified 0e0748c5

Fix llama meta tensor loading in AutoTP and kernel injected inference (#3608)

zeyugao committed 1 year ago

Verified 4fc2c8e7

Fix min torch version (#4375)

tjruwase committed 1 year ago

Verified 463dea27

Fix multinode runner to properly append to PDSH_SSH_ARGS_APPEND (#4373)

loadams committed 1 year ago

Verified 17957728

add the missing method (#4363)

cli99 committed 1 year ago

Verified 727609df

Openfold fix (#4368)

cctry committed 1 year ago

Verified c5814647

deepspeed4science japanese blog (#4369)

conglongli committed 1 year ago

Verified a99e5d3f

deepspeed4science chinese blog (#4366)

conglongli committed 1 year ago

Verified 3592a22c

Enable workflow dispatch on Torch 1.10 CI tests (#4361)

loadams committed 1 year ago

Verified dcd3ae19

Update conda env to have max pydantic version (#4362)

loadams committed 1 year ago

Verified dcf649c3

add deepspeed4science blog link (#4364)

conglongli committed 1 year ago

Verified da7a1851

added check to avoid undefined behavior when the input_id length is greater than max_tokens (#4349)

mrwyattii committed 1 year ago

Verified 85334238

Add the policy to run llama model from the official repo (#4313)

RezaYazdaniAminabadi committed 1 year ago

Verified 468882fb

fix deepspeed4science links (#4358)

conglongli committed 1 year ago

Verified 78c3b148

DeepSpeed4Science (#4357)

conglongli committed 1 year ago

Verified f876d81d

Support InternLM (#4137)

wangruohui committed 1 year ago

Verified 367d6f9c

Pass base_dir to model files can be loaded for auto-tp/meta-tensor. (#4348)

awan-10 committed 1 year ago

Verified b9d719a6

Fix autotune to support Triton 2.1 (#4340)

stephen-youn committed 1 year ago

Verified 3f3e9fb1

Fix skipped inference tests (#4336)

mrwyattii committed 1 year ago

Verified 8c7f7fd2

Older