microsoft/DeepSpeed

Pull Requests Commits

Update torch version

loadams committed 1 year ago

fac6a0c1

add check for cuda available

mrwyattii committed 1 year ago

9e10370f

update inference pages to point to FastGen (#5029)

mrwyattii committed 1 year ago

Verified 24f20ef0

Fix nv-accelerate and nv-torch-latest-v100. (#5035)

loadams committed 1 year ago

Verified 546d08ee

Fix for nightly torch CI (#5039)

mrwyattii committed 1 year ago

Verified f0e5bece

Fix P40 CI failures (#5037)

mrwyattii committed 1 year ago

Verified 6857061a

Remove Megatron-DeepSpeed CI workflow (#5038)

mrwyattii committed 1 year ago

Verified db22e7a7

Remove coverage reports from workflows and fix for inference CI (#5028)

loadams committed 1 year ago

Verified f5920457

Remove MI100 badge from landing page (#5036)

mrwyattii committed 1 year ago

Verified 8c30e7c4

Graph capture support on HPU accelerators (#5013)

deepcharm committed 1 year ago

Verified 8ec1cc3b

launcher/launcher_helper.py: fix PMI name and add EnvironmentError (#5025)

YizhouZ committed 1 year ago

Verified 66fb9241

deepspeed/launcher: add launcher_helper as each rank's start portal (#4699)

YizhouZ committed 2 years ago

Verified 6e1a6801

[minor] Improve logging for multiprocesses (#5004)

ByronHsu committed 2 years ago

Verified 9500ab7d

DS #4993 #662 : autotune single node hostfile bugfix (#4996)

oushu1zhangxiangxuan1 committed 2 years ago

Verified c0525451

Fix error report of DSElasticAgent._set_master_addr_port() (#4985)

RobinDong committed 2 years ago

Verified d2e9adce

Update falcon fused type order (#5007)

Yejing-Lai committed 2 years ago

Verified 62afafe8

[minor] improve code quality and readablilty (#5011)

ByronHsu committed 2 years ago

Verified e8136931

Enable workflow dispatch on all workflows (#5016)

loadams committed 2 years ago

Verified 0268d551

Enable hpz based on secondary tensor presence (#4906)

HeyangQin committed 2 years ago

Verified 75ed63c9

Allow nightly tests dispatch (#5014)

mrwyattii committed 2 years ago

Verified bc8b1ae3

Support `exclude_frozen_parameters` for `save_16bit_model` (#4999)

LZHgrla committed 2 years ago

Verified 0aecdcb5

Update version.txt after 0.13.1 release (#5002)

mrwyattii committed 2 years ago

Verified 0dd0c615

Refactor the Qwen positional emebdding config code (#4955)

ZonePG committed 2 years ago

Verified 1d35db76

Implement some APIs of HPU accelerator (#4935)

mmhab committed 2 years ago

Verified 5a8bf3f8

Make installable without torch (#5001)

mrwyattii committed 2 years ago

Verified 7d511398

Delete unused --deepspeed_mpi command line argument (#4981)

ShukantPal committed 2 years ago

Verified 538ffb4b

fix for DS_ENV issue (#4992)

jeffra committed 2 years ago

Verified a59455fa

Fix placeholder value in FastGen Blog (#5000)

mrwyattii committed 2 years ago

Verified 4738a5e6

fix some typo under blogs/ (#4988)

digger-yu committed 2 years ago

Verified b81bed69

Fix the MoE-params gradient-scaling (#4957)

RezaYazdaniAminabadi committed 2 years ago

Verified 9d2660d2

Older