Update Domino for Llama3#7084

Draft

shenzheyu wants to merge 20 commits intodeepspeedai:masterfrom

shenzheyu:master

Commits on Mar 5, 2025

Update setup.py handling of ROCm cupy (#7051)

loadams
authored and
shenzheyu
committed
nv-ds-chat breaks with latest transformers (#7052)

loadams
authored and
shenzheyu
committed
update for llama3
shenzheyu
committed
fix format
shenzheyu
committed
Rename aio_thread_count to intra_op_parallelism (#7056)

tjruwase
authored and
shenzheyu
committed
add autoTP training zero2 tests (#7049)

authored and
shenzheyu
committed
Fix, bf16 optimizer remove dup loop (#7054)

authored and
shenzheyu
committed
Update version.txt after 0.16.4 release (#7063)

loadams
authored and
shenzheyu
committed
fix an outdated doc wrt CUDA_VISIBLE_DEVICES (#7058)

stas00
authored and
shenzheyu
committed
Tecorigin sdaa accelerator (#6903)

authored and
shenzheyu
committed
Handle special case of libuv for Windows (#7064)

loadams
authored and
shenzheyu
committed
Update README with info on newest accelerator (#7065)

loadams
authored and
shenzheyu
committed
Bug Fix for offload_states API (#7050)

authored and
shenzheyu
committed
Fix TOCTOU issues, switch to fstat (#7067)

loadams
authored and
shenzheyu
committed
config torch to avoid graph breaks caused by logger (#6999)

authored and
shenzheyu
committed
Fix meta load tensor imcompatible issue (#7073)

authored and
shenzheyu
committed
Replace calls to `python setup.py sdist` with `python -m build --sdist` (#7069)

loadams
authored and
shenzheyu
committed
Revert "Handle special case of libuv for Windows (#7064)" (#7076)

loadams
authored and
shenzheyu
committed
Add DeepseekV3 AutoTP. (#7045)

Yejing-Lai
authored and
shenzheyu
committed
Merge branch 'master' into master
shenzheyu
authored

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Update Domino for Llama3#7084

Update Domino for Llama3#7084
shenzheyu wants to merge 20 commits intodeepspeedai:masterfrom
shenzheyu:master

Commits on Mar 5, 2025

Update setup.py handling of ROCm cupy (#7051)

nv-ds-chat breaks with latest transformers (#7052)

update for llama3

fix format

Rename aio_thread_count to intra_op_parallelism (#7056)

add autoTP training zero2 tests (#7049)

Fix, bf16 optimizer remove dup loop (#7054)

Update version.txt after 0.16.4 release (#7063)

fix an outdated doc wrt CUDA_VISIBLE_DEVICES (#7058)

Tecorigin sdaa accelerator (#6903)

Handle special case of libuv for Windows (#7064)

Update README with info on newest accelerator (#7065)

Bug Fix for offload_states API (#7050)

Fix TOCTOU issues, switch to fstat (#7067)

config torch to avoid graph breaks caused by logger (#6999)

Fix meta load tensor imcompatible issue (#7073)

Replace calls to `python setup.py sdist` with `python -m build --sdist` (#7069)

Revert "Handle special case of libuv for Windows (#7064)" (#7076)

Add DeepseekV3 AutoTP. (#7045)

Merge branch 'master' into master