Code

Update Domino for Llama3#7084

Code

Open

shenzheyu wants to merge 20 commits intodeepspeedai:masterdeepspeedai/DeepSpeed:masterfrom shenzheyu:mastershenzheyu/DeepSpeed:masterCopy head branch name to clipboard

+62-37

Commits

Commits on Mar 5, 2025

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

  Show description for 963f11b
  

  loadams

  authored and

  shenzheyu

  committed
  963f11b

  Copy full SHA for 963f11b

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

  Show description for f538f55
  

  loadams

  authored and

  shenzheyu

  committed
  f538f55

  Copy full SHA for f538f55

• update for llama3

  Show description for ef6c29b
  shenzheyu

  committed
  ef6c29b

  Copy full SHA for ef6c29b

• fix format

  Show description for 54a1421
  shenzheyu

  committed
  54a1421

  Copy full SHA for 54a1421

• Rename aio_thread_count to intra_op_parallelism (deepspeedai#7056)

  Show description for 4239526
  

  tjruwase

  authored and

  shenzheyu

  committed
  4239526

  Copy full SHA for 4239526

• add autoTP training zero2 tests (deepspeedai#7049)

  Show description for f3ce29f
  

  2 peopleauthored and

  shenzheyu

  committed
  f3ce29f

  Copy full SHA for f3ce29f

• Fix, bf16 optimizer remove dup loop (deepspeedai#7054)

  Show description for 5a72572
  

  3 peopleauthored and

  shenzheyu

  committed
  5a72572

  Copy full SHA for 5a72572

• Update version.txt after 0.16.4 release (deepspeedai#7063)

  Show description for adb4e08
  

  loadams

  authored and

  shenzheyu

  committed
  adb4e08

  Copy full SHA for adb4e08

• fix an outdated doc wrt CUDA_VISIBLE_DEVICES (deepspeedai#7058)

  Show description for aeaf0ce
  

  stas00

  authored and

  shenzheyu

  committed
  aeaf0ce

  Copy full SHA for aeaf0ce

• Tecorigin sdaa accelerator (deepspeedai#6903)

  Show description for ef1cbd0
  

  4 peopleauthored and

  shenzheyu

  committed
  ef1cbd0

  Copy full SHA for ef1cbd0

• Handle special case of libuv for Windows (deepspeedai#7064)

  Show description for 9df70c2
  

  loadams

  authored and

  shenzheyu

  committed
  9df70c2

  Copy full SHA for 9df70c2

• Update README with info on newest accelerator (deepspeedai#7065)

  Show description for 1faaf1e
  

  loadams

  authored and

  shenzheyu

  committed
  1faaf1e

  Copy full SHA for 1faaf1e

• Bug Fix for offload_states API (deepspeedai#7050)

  Show description for fb6d9a8
  

  2 peopleauthored and

  shenzheyu

  committed
  fb6d9a8

  Copy full SHA for fb6d9a8

• Fix TOCTOU issues, switch to fstat (deepspeedai#7067)

  Show description for 3638f9c
  

  loadams

  authored and

  shenzheyu

  committed
  3638f9c

  Copy full SHA for 3638f9c

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

  Show description for f0db010
  

  2 peopleauthored and

  shenzheyu

  committed
  f0db010

  Copy full SHA for f0db010

• Fix meta load tensor imcompatible issue (deepspeedai#7073)

  Show description for b9a77e2
  

  2 peopleauthored and

  shenzheyu

  committed
  b9a77e2

  Copy full SHA for b9a77e2

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

  Show description for 6830975
  

  loadams

  authored and

  shenzheyu

  committed
  6830975

  Copy full SHA for 6830975

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

  Show description for dddc7cf
  

  loadams

  authored and

  shenzheyu

  committed
  dddc7cf

  Copy full SHA for dddc7cf

• Add DeepseekV3 AutoTP. (deepspeedai#7045)

  Show description for 91d05e2
  

  Yejing-Lai

  authored and

  shenzheyu

  committed
  91d05e2

  Copy full SHA for 91d05e2

• Merge branch 'master' into master

  shenzheyu

  authored
  be199d2

  Copy full SHA for be199d2