DeepSpeed
Add `pyproject.toml` with legacy build backend to keep most logic in `setup.py`
#7033
Open
Overview
Commits
76
Changes
View On GitHub

Add `pyproject.toml` with legacy build backend to keep most logic in `setup.py` #7033

loadams wants to merge 76 commits into master from loadams/pyproject-toml
loadams
stas00
stas00 commented on 2025-02-14
jeffra
loadams loadams changed the title Add pyproject.toml Add `pyproject.toml` with legacy build backend to keep most logic in `setup.py` 317 days ago
loadams
loadams loadams marked this pull request as ready for review 311 days ago
loadams loadams requested a review from jeffra jeffra 311 days ago
loadams loadams requested a review from sfc-gh-mwyatt sfc-gh-mwyatt 311 days ago
loadams loadams requested a review from tjruwase tjruwase 311 days ago
rraminen [ROCm] Enable fp_quantizer on ROCm (#7027)
230f479e
loadams Add pyproject.toml
130c11c4
loadams Formatting fix
b2e16cbc
GuanhuaWang add gds chinese blog (#7034)
72bfb702
hwchen2017 Add chinese blog for deepspeed windows, and fix format (#7035)
48c02ad4
jomayeri AIO on ROCM (#7023)
4996cae5
loadams Update setuptools min requirement
52709fa2
loadams Switch build to legacy
9eb0618b
loadams Add no-build isolation
36ce3737
tjruwase Control trace cache warnings (#7039)
a0ff11a4
hwchen2017 Update CUDA compute capability to support Blackwell (#7047)
c574b454
loadams Update setup.py handling of ROCm cupy (#7051)
5326873f
loadams nv-ds-chat breaks with latest transformers (#7052)
c22be1ab
loadams Test with non legacy backend
2873a11f
loadams Need legacy backend to execute setup.py, though it executes it differ…
4e884634
loadams Update to actually use legacy backend
6d837ca1
tjruwase Rename aio_thread_count to intra_op_parallelism (#7056)
5b2f7135
inkcherry add autoTP training zero2 tests (#7049)
c0f4235d
wukong1992 Fix, bf16 optimizer remove dup loop (#7054)
01dd3187
loadams Update version.txt after 0.16.4 release (#7063)
41718ada
stas00 fix an outdated doc wrt CUDA_VISIBLE_DEVICES (#7058)
c83ade6f
siqi654321 Tecorigin sdaa accelerator (#6903)
cc79d173
loadams Handle special case of libuv for Windows (#7064)
aa63ba4f
loadams Update README with info on newest accelerator (#7065)
aaf86d68
U-rara Bug Fix for offload_states API (#7050)
630864d3
loadams Add all imports from setup.py to pyproject.toml
5a2eceae
loadams Fix TOCTOU issues, switch to fstat (#7067)
ac52baec
ShellyNR config torch to avoid graph breaks caused by logger (#6999)
fd5273f3
Yejing-Lai Fix meta load tensor imcompatible issue (#7073)
c91d2256
loadams Replace calls to `python setup.py sdist` with `python -m build --sdis…
80ff4201
loadams Update workflow to not do build isolation
355e9c5e
loadams Add more requirements
00addaa7
loadams Revert "Handle special case of libuv for Windows (#7064)" (#7076)
432e71dc
Yejing-Lai Add DeepseekV3 AutoTP. (#7045)
5ab85893
loadams Add pyproject.toml into comments
f325c92a
loadams Improve inference tutorial docs (#7083)
9c4c4447
loadams Pin transformers version on tests that use latest. (#7085)
5ebe9bdc
siddharth9820 Update README.md with ICS '23 MoE paper link (#7087)
9da1b336
loadams Update parallelism for nv-torch-latest/nightly tests due to more GPUs…
f9632484
loadams Remove workflows for very old torch versions (#7090)
8254e1b3
tjruwase Use new dlpack api; Formatting fixes (#7101)
a7ab4ab7
deepcharm Avoid graph breaks by disabling sourceless calls in instrument_w_nvtx…
6eb5baa7
deepcharm Avoid graph breaks in torch.compile caused by inner classes in the ba…
feeaa2e4
hwchen2017 Only run pre-commit on the changes (#7106)
7b7e1122
deepcharm Avoid graph break due to unsupported frozenset (#7105)
674f9bd5
Yejing-Lai Fix fused_qkv print model ValueError (#7109)
240c2a7b
loadams Update references to new X/Twitter handle (#7110)
1c35241b
raza-sikander Update gaudi2 nightly,ci to latest 1.20.0 build (#7093)
ec2e93e3
inkcherry fix keep_module_on_host (#7112)
c7934014
loadams Add sequential pytest mark to TestNVMeCheckpointing to resolve pytest…
62abd88e
tjruwase Training multiple models (#7018)
065084be
loadams Update CONTRIBUTING.md to reflect changes from CLA to DCO (#7135)
e93f0d16
tjruwase Avoid missing attr error (#7133)
eaeae790
A-transformer Add conditional expression (#7119)
8df6245e
loadams Unpin transformers version for most workflows (#7139)
6de20f65
saurabhkoshatwar Conditionally quote env vars (#7071)
e0188ac3
A-transformer Correct the BACKWARD_PREFETCH_SUBMIT mismatch (#7120)
ea61c6ca
raza-sikander Enhance Gaudi2 CI/Nightly Coverage with Model Parallelism and Linear …
9d753116
loadams Update container version that runs on A6000 tests. (#7153)
4ca7ba57
fix leak of z3 buffer
2db922fe
inkcherry hf tp+zero training doc. (#7151)
78ec0254
deepcharm Avoid graph break by removing redundant requires_grad attr change (#7…
4a851f35
tohtana Add destroy to tests to free memory (#7160)
6adcf2e3
c8ef [NFC] Typo fix in SP layer. (#7152)
7b7ac9e3
hwchen2017 Link AutoTP blog in the front page (#7167)
31ec2b77
loadams loadams force pushed from 705edb34 to 31ec2b77 283 days ago
loadams loadams requested a review from tohtana tohtana 283 days ago
loadams loadams requested a review from jomayeri jomayeri 283 days ago
loadams loadams requested a review from hwchen2017 hwchen2017 283 days ago
loadams Merge branch 'master' into loadams/pyproject-toml
e40df22d
loadams Remove unneeded requires in build system declaration
4c32a9d0
sfc-gh-mwyatt
sfc-gh-mwyatt approved these changes on 2025-03-25
loadams Add build to the pyproject
c4b24fd3
loadams Add no torch build triggers
b330b4e5
loadams Remove no build isolation from nv-torch-latest
3601c293
loadams Merge branch 'master' into loadams/pyproject-toml
e0d9ba44
loadams Merge branch 'master' into loadams/pyproject-toml
42e42a08
loadams Merge branch 'master' into loadams/pyproject-toml
3c94f513
agronholm
agronholm commented on 2025-04-09
loadams Merge branch 'master' into loadams/pyproject-toml
221f2be6
loadams Merge branch 'master' into loadams/pyproject-toml
dfcc24d3
loadams Merge branch 'master' into loadams/pyproject-toml
b4ed94a6

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Terms Of Use Privacy Policy Refund Policy Impressum