Python package

python-package.yml

1066 workflow runs

Migrate all kbit kernels to uint8 E4M4 absmax, add fp16 absmax path
failed:
Python package#2248:Commit7e0063cpushedbyTimDettmers
feature/kbit-gemv-v8
Feb 17, 20267m 39s
Remove dead warpspec/dqonce kernels, add deployment analysis docs
failed:
Python package#2247:Commitb02b657pushedbyTimDettmers
feature/kbit-gemv-v8
Feb 17, 20267m 1s
Simplest MPS backend: adding kernels using kernel builder and kernels-community
completed successfully:
Python package#2246:Pull request#1875synchronizebyMekkCyber
MekkCyber:add-bitsandbytes
Feb 17, 202610m 50s
Simplest MPS backend: adding kernels using kernel builder and kernels-community
cancelled:
Python package#2245:Pull request#1875synchronizebyMekkCyber
MekkCyber:add-bitsandbytes
Feb 17, 20265m 59s
[CI] Test against PyTorch 2.10
completed successfully:
Python package#2244:Commita334ed0pushedbymatthewdouglas
ci-torch-210
Feb 17, 20269m 59s
Simplest MPS backend: adding kernels using kernel builder and kernels-community
cancelled:
Python package#2243:Pull request#1875synchronizebyMekkCyber
MekkCyber:add-bitsandbytes
Feb 17, 20265m 27s
CI: add NVIDIA A10 runner for tests
completed successfully:
Python package#2241:Commitabdc17bpushedbymatthewdouglas
ci-add-a10
Feb 17, 20269m 40s
Bump dev version
completed successfully:
Python package#2237:Commit9dd8b70pushedbymatthewdouglas
main
Feb 16, 20269m 27s
0.49.2
completed successfully:
Python package#2236:Release0.49.2publishedbymatthewdouglas
Feb 16, 20269m 46s
Release 0.49.2
completed successfully:
Python package#2235:Commitf0e6ca3pushedbymatthewdouglas
0.49.2
Feb 16, 202610m 7s
Release 0.49.2
completed successfully:
Python package#2234:Commitf0e6ca3pushedbymatthewdouglas
main
Feb 16, 20269m 30s
Fix AdEMAMix scheduler guard and add state_dict round-trip test (#1861)
completed successfully:
Python package#2233:Commitcace65cpushedbymatthewdouglas
main
Feb 16, 20269m 31s
Handle non-contiguous tensors in quantize/dequantize ops (#1859)
cancelled:
Python package#2232:Commit505a00apushedbymatthewdouglas
main
Feb 16, 20269m 14s
[ROCm] Make blocksize=64 default for 4bit (#1873)
cancelled:
Python package#2231:Commitc2ae381pushedbymatthewdouglas
main
Feb 16, 20262m 6s
[ROCm] Make blocksize=64 default for 4bit
completed successfully:
Python package#2230:Pull request#1873openedbymatthewdouglas
rocm-blocksize-64-default
Feb 16, 20269m 48s
Update test
completed successfully:
Python package#2229:Commita96cd65pushedbymatthewdouglas
rocm-blocksize-64-default
Feb 16, 20269m 43s
[ROCm] Make blocksize=64 default for 4bit
cancelled:
Python package#2228:Commit73b5e7apushedbymatthewdouglas
rocm-blocksize-64-default
Feb 16, 20264m 12s
Skip typo check on agent markdown files
completed successfully:
Python package#2227:Commit943e42dpushedbymatthewdouglas
main
Feb 16, 20269m 34s
Migrate all kbit kernels to uint8 E4M4 absmax, add fp16 absmax path
failed:
Python package#2225:Commit193317fpushedbyTimDettmers
experiment/scalar-gemv-int8-absmax
Feb 16, 20267m 48s
update docs (#1857)
completed successfully:
Python package#2224:Commitb075afcpushedbymatthewdouglas
main
Feb 16, 202617m 47s
Remove deprecated APIs: research module, non-blockwise optimizers, and legacy quantization functions
completed successfully:
Python package#2223:Pull request#1871synchronizebyTimDettmers
deprecation
Feb 16, 20269m 18s
Remove deprecated APIs: research module, non-blockwise optimizers, an…
completed successfully:
Python package#2222:Commit62d6963pushedbyTimDettmers
deprecation
Feb 16, 202610m 45s
docs: Add PR review posting guide, agent reference docs, and lint fixes
completed successfully:
Python package#2221:Commitde89ff7pushedbyTimDettmers
main
Feb 16, 202612m 17s
Grouped MMA: TILE_N=64 + k_splits for small-N MoE shapes
failed:
Python package#2220:Commit23f92e5pushedbyTimDettmers
feature/kbit-gemv-v8
Feb 16, 20268m 17s
Add workload-weighted kernel analysis and vLLM deployment model
failed:
Python package#2219:Commitdf55dafpushedbyTimDettmers
feature/kbit-gemv-v8
Feb 16, 20267m 11s

Navigation Menu

Python package

1066 workflow runs

Migrate all kbit kernels to uint8 E4M4 absmax, add fp16 absmax path

Remove dead warpspec/dqonce kernels, add deployment analysis docs

Simplest MPS backend: adding kernels using kernel builder and kernels-community

Simplest MPS backend: adding kernels using kernel builder and kernels-community

[CI] Test against PyTorch 2.10

Simplest MPS backend: adding kernels using kernel builder and kernels-community

CI: add NVIDIA A10 runner for tests

Bump dev version

0.49.2

Release 0.49.2

Release 0.49.2

Fix AdEMAMix scheduler guard and add state_dict round-trip test (#1861)

Handle non-contiguous tensors in quantize/dequantize ops (#1859)

[ROCm] Make blocksize=64 default for 4bit (#1873)

[ROCm] Make blocksize=64 default for 4bit

Update test

[ROCm] Make blocksize=64 default for 4bit

Skip typo check on agent markdown files

Migrate all kbit kernels to uint8 E4M4 absmax, add fp16 absmax path

update docs (#1857)

Remove deprecated APIs: research module, non-blockwise optimizers, and legacy quantization functions

Remove deprecated APIs: research module, non-blockwise optimizers, an…

docs: Add PR review posting guide, agent reference docs, and lint fixes

Grouped MMA: TILE_N=64 + k_splits for small-N MoE shapes

Add workload-weighted kernel analysis and vLLM deployment model

ProTip! You can narrow down the results and go further in time using the filter or the other filters available.

Workflow runs for Python package

1066 workflow runs

ProTip! You can narrow down the results and go further in time using the created: filter or the other filters available.

ProTip! You can narrow down the results and go further in time using the filter or the other filters available.