All pull requests

Pull requests

• docs: fix 'offical' typo in README install section

  #10690·25f1001183-Saurabh opened Mar 29, 2026·

  ·
  2

• Dev npu support ssd

  #10667·Flowingsun007 opened Jul 15, 2025Contributor·

  ·

• Dev fused nms

  apifeatureop
  #10643·ShawnXuan opened Jun 1, 2025Collaborator·

  ·
  6

• Npu use cpu argwhere

  featureop
  #10640·ShawnXuan opened May 21, 2025Collaborator·

  ·
  4

• Dev nll loss

  #10633·ShawnXuan opened May 2, 2025Collaborator·

  ·

• Modify dropout mask tensor shape and bits format for NPU

  #10620·MagicalMogu opened Mar 5, 2025Contributor·

  ·
  1

• fix autocast to support global tensor

  enhancementglobalop
  #10605·fpzh2011 opened Dec 18, 2024Contributor·

  ·
  6

• Dev support graph 2d parallel on npu

  #10566·Flowingsun007 opened Nov 18, 2024Contributor·

  ·
  1

• Remove CUDA-only restriction for multi-tensor model updates in optimizer

  eagerenhancementgraphop
  #10564·ShawnXuan opened Nov 14, 2024Collaborator·

  ·
  5

• layer_norm_grad for npu

  #10560·fpzh2011 opened Nov 7, 2024Contributor·

  ·

• Update README.md, adding python 3.12 and cuda version cu121, cu122, and things related to PEP 668

  #10557·Neutron3529 opened Sep 12, 2024·

  ·
  2

• add Tensor method: new and new_ones

  enhancementop
  #10553·crazy-JiangDongHua opened Aug 21, 2024Contributor·

  ·
  2

• Add softmax parameter stacklevel

  #10552·Dmovic opened Aug 20, 2024·

  ·

• autocast support new devices

  enhancementfeaturesystem
  #10539·fpzh2011 opened Jun 26, 2024Contributor·

  ·
  9

• profiler adapt to ascend npu

  #10536·Flowingsun007 opened Jun 18, 2024Contributor·

  ·

• Support input check for pool operator

  enhancementsystem
  #10532·Dmovic opened May 24, 2024·

  ·
  18

  18

• flash attention use mirror

  automergebuildenhancement
  #10530·mosout opened May 22, 2024Contributor·

  ·
  3

• Use cpm and move flash_attention to external

  #10529·mosout opened May 22, 2024Contributor·

  ·

• Use Opentelemetry

  #10526·mosout opened May 15, 2024Contributor·

  ·

• [feature]add tensor folder map to graph

  #10525·crazy-JiangDongHua opened May 13, 2024Contributor·

  ·

• add FusedApplyRotaryEmbGradKernel

  enhancementop
  #10517·cccddd77 opened May 11, 2024Contributor·

  ·
  6

• Update dependency constraints on cudnn

  buildenhancement
  #10515·jackalcooper opened May 6, 2024Contributor·

  ·
  2

• modify clip_grad with no to_global

  enhancementsystem
  #10443·hanwen-sun opened Mar 11, 2024·

  ·
  24

• Update cudnn convolution kernel

  enhancementfeatureop
  #10440·linzs148 opened Mar 6, 2024Contributor·

  ·
  10

• [bugfix] enable faster rcnn and sd model with oneflow backend

  automergebugsystem
  #10439·crazy-JiangDongHua opened Mar 5, 2024Contributor·

  ·
  4