Python package

Workflow runs for Python package

• fix(cpu): correctly handle 1D output shape in dequantize_4bit

  failed:
  Python package#2969:Pull request#2055openedbySparshM8
  SparshM8:fix/cpu-dequantize-1d-shape

  Aug 23, 202618d 19h 34m 30s
  More actions for workflow run 2969

• Fuse nested 4-bit scale reconstruction on SM103

  failed:
  Python package#2968:Pull request#2051synchronizebyheiheiha798
  heiheiha798:b300/sm103-nested-4bit-dequant

  Aug 22, 202630d 0h 1m 7s
  More actions for workflow run 2968

• Fuse nested 4-bit scale reconstruction on SM103

  failed:
  Python package#2967:Pull request#2051synchronizebyheiheiha798
  heiheiha798:b300/sm103-nested-4bit-dequant

  Aug 22, 202630d 0h 4m 26s
  More actions for workflow run 2967

• Tune SM103 fused 4-bit GEMM SIMT dispatch

  failed:
  Python package#2966:Pull request#2054openedbyheiheiha798
  heiheiha798:b300/sm103-4bit-simt-mma-dispatch

  Aug 22, 202630d 0h 0m 25s
  More actions for workflow run 2966

• Remove redundant CUB barriers from blockwise dequantization

  failed:
  Python package#2965:Pull request#2053openedbyheiheiha798
  heiheiha798:b300/dequantize-blockwise-cub-barriers

  Aug 22, 202630d 0h 1m 38s
  More actions for workflow run 2965

• Respect the current CUDA stream in blockwise quantization

  failed:
  Python package#2964:Pull request#2052openedbyheiheiha798
  heiheiha798:b300/quantize-blockwise-current-stream

  Aug 22, 202630d 0h 0m 18s
  More actions for workflow run 2964

• Fuse nested 4-bit scale reconstruction on SM103

  failed:
  Python package#2963:Pull request#2051openedbyheiheiha798
  heiheiha798:b300/sm103-nested-4bit-dequant

  Aug 22, 202630d 0h 4m 16s
  More actions for workflow run 2963

• Use the current CUDA stream for non-paged optimizer updates

  failed:
  Python package#2962:Pull request#2050openedbyheiheiha798
  heiheiha798:b300/nonpaged-optimizer-current-stream

  Aug 22, 202630d 0h 4m 23s
  More actions for workflow run 2962

• ci: expand ROCm architecture coverage

  failed at startup:
  Python package#2961:Pull request#2046synchronizebysstamenk
  sstamenk:feature/rocm-7.14-rdna1

  Aug 21, 20263m 12s
  More actions for workflow run 2961

• Fix CPU dequantize_4bit output shape for 1-D inputs

  failed:
  Python package#2959:Pull request#2048openedby2sumtech
  2sumtech:fix/dequantize-4bit-1d-shape

  Aug 20, 202630d 0h 5m 6s
  More actions for workflow run 2959

• ci: expand ROCm architecture coverage

  completed successfully:
  Python package#2958:Pull request#2046synchronizebysstamenk
  sstamenk:feature/rocm-7.14-rdna1

  Aug 20, 20269m 34s
  More actions for workflow run 2958

• [ROCm] Harden HIP version for native library naming and selection

  completed successfully:
  Python package#2955:Pull request#1888synchronizebysstamenk
  sstamenk:fix/rocm-build-rework

  Aug 20, 202615m 43s
  More actions for workflow run 2955

• chore(deps): bump the actions group across 1 directory with 2 updates…

  completed successfully:
  Python package#2954:Commit95f9af3pushedbymatthewdouglas
  main

  Aug 19, 202612m 41s
  More actions for workflow run 2954

• Fix mixed INT8 FakeTensor output metadata

  failed:
  Python package#2953:Pull request#2044openedbytandede
  tandede:fix/int8-mixed-fake-metadata

  Aug 19, 202630d 0h 1m 15s
  More actions for workflow run 2953

• Add sm121 (GB10) to the wide_n_simt 4-bit GEMM dispatch rule (#2039)

  completed successfully:
  Python package#2952:Commitd2d08f6pushedbymatthewdouglas
  main

  Aug 18, 20269m 18s
  More actions for workflow run 2952

• Mask the absmax store in the Triton 4-bit quantize kernels (#2043)

  cancelled:
  Python package#2951:Commitd8ccdd9pushedbymatthewdouglas
  main

  Aug 18, 20265m 26s
  More actions for workflow run 2951

• Mask the absmax store in the Triton 4-bit quantize kernels

  completed successfully:
  Python package#2950:Pull request#2043synchronizebytruong-v
  truong-v:fix/triton-4bit-absmax-store-mask

  Aug 17, 20268m 11s
  More actions for workflow run 2950

• chore(deps): bump the actions group across 1 directory with 2 updates

  completed successfully:
  Python package#2949:Pull request#2036synchronizebydependabot[bot]Bot
  dependabot/github_actions/actions-a4d04eb940

  Aug 17, 202647s
  More actions for workflow run 2949

• chore(deps): bump the actions group across 1 directory with 2 updates

  completed successfully:
  Python package#2948:Commitf7ec8e2pushedbydependabot[bot]Bot
  dependabot/github_actions/actions-a4d04eb940

  Aug 17, 20267m 19s
  More actions for workflow run 2948

• Mask the absmax store in the Triton 4-bit quantize kernels

  completed successfully:
  Python package#2947:Pull request#2043openedbytruong-v
  truong-v:fix/triton-4bit-absmax-store-mask

  Aug 17, 202622m 27s
  More actions for workflow run 2947

• [ROCm] Harden HIP version for native library naming and selection

  failed:
  Python package#2946:Pull request#1888synchronizebysstamenk
  sstamenk:fix/rocm-build-rework

  Aug 14, 202636s
  More actions for workflow run 2946

• Fix decoupled weight decay ordering in the CUDA Adam/AdEMAMix kernels

  completed successfully:
  Python package#2941:Pull request#2040openedbyyentur
  yentur:fix/issue-2010-weight-decay-order

  Aug 14, 202613m 49s
  More actions for workflow run 2941

• Add sm121 (GB10) to the wide_n_simt 4-bit GEMM dispatch rule

  completed successfully:
  Python package#2940:Pull request#2039openedbyyashb98
  yashb98:wide-n-simt-sm121

  Aug 14, 20268m 6s
  More actions for workflow run 2940

• Bump dev version

  completed successfully:
  Python package#2939:Commit2b6cfb7pushedbymatthewdouglas
  main

  Aug 13, 202629m 27s
  More actions for workflow run 2939

• 0.50.1: RTX Spark Support

  completed successfully:
  Python package#2938:Release0.50.1publishedbymatthewdouglas
  Aug 13, 202617m 12s
  More actions for workflow run 2938