deepspeedai/DeepSpeed

Pull Requests Commits

Use new dlpack api; Formatting fixes (#7101)

tjruwase committed 1 year ago

a7ab4ab7

Remove workflows for very old torch versions (#7090)

loadams committed 1 year ago

8254e1b3

Update parallelism for nv-torch-latest/nightly tests due to more GPUs/runner (#7086)

loadams committed 1 year ago

f9632484

Update README.md with ICS '23 MoE paper link (#7087)

siddharth9820 committed 1 year ago

9da1b336

Pin transformers version on tests that use latest. (#7085)

loadams committed 1 year ago

5ebe9bdc

Improve inference tutorial docs (#7083)

loadams committed 1 year ago

9c4c4447

Add pyproject.toml into comments

loadams committed 1 year ago

f325c92a

Add DeepseekV3 AutoTP. (#7045)

Yejing-Lai committed 1 year ago

5ab85893

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

loadams committed 1 year ago

432e71dc

Add more requirements

loadams committed 1 year ago

00addaa7

Update workflow to not do build isolation

loadams committed 1 year ago

355e9c5e

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

loadams committed 1 year ago

80ff4201

Fix meta load tensor imcompatible issue (#7073)

Yejing-Lai committed 1 year ago

c91d2256

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

ShellyNR committed 1 year ago

fd5273f3

Fix TOCTOU issues, switch to fstat (#7067)

loadams committed 1 year ago

ac52baec

Add all imports from setup.py to pyproject.toml

loadams committed 1 year ago

5a2eceae

Bug Fix for offload_states API (#7050)

U-rara committed 1 year ago

630864d3

Update README with info on newest accelerator (#7065)

loadams committed 1 year ago

aaf86d68

Handle special case of libuv for Windows (#7064)

loadams committed 1 year ago

aa63ba4f

Tecorigin sdaa accelerator (#6903)

siqi654321 committed 1 year ago

cc79d173

fix an outdated doc wrt CUDA_VISIBLE_DEVICES (#7058)

stas00 committed 1 year ago

c83ade6f

Update version.txt after 0.16.4 release (#7063)

loadams committed 1 year ago

41718ada

Fix, bf16 optimizer remove dup loop (#7054)

wukong1992 committed 1 year ago

01dd3187

add autoTP training zero2 tests (#7049)

inkcherry committed 1 year ago

c0f4235d

Rename aio_thread_count to intra_op_parallelism (#7056)

tjruwase committed 1 year ago

5b2f7135

Update to actually use legacy backend

loadams committed 1 year ago

6d837ca1

Need legacy backend to execute setup.py, though it executes it differently, so we aren't seeing cupy installed.

loadams committed 1 year ago

4e884634

Test with non legacy backend

loadams committed 1 year ago

2873a11f

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

loadams committed 1 year ago

c22be1ab

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

loadams committed 1 year ago

5326873f

Newer Older