Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Nov 7, 2024

• pnnx write implicit int conversion in python script (#5767)

  nihui

  authored
  e71fdf8
  More actions

Commits on Oct 29, 2024

• disable x86 auto recip optimization for potential precision loss (#5762)

  nihui

  authored
  c32442a
  More actions

Commits on Oct 22, 2024

• pnnx do not fold tensor with dynamic shape, use fp32 module by default (#5755)

  nihui

  authored
  6077adc
  More actions

Commits on Oct 21, 2024

• fix gemm arm int8 scales descales offset (#5750)

  nihui

  authored
  e7602a2
  More actions

• pnnx torch 2.5 (#5748)

  nihui

  authored
  c1f9e95
  More actions

Commits on Oct 18, 2024

• arm neon optimization for layernorm fp32/bf16s/fp16s (#5746)

  nihui

  authored
  8fe6281
  More actions

• Fix #5741 don't crash when vkCreateDevice fails (#5742)

  Upliner

  authored
  cbd17cd
  More actions

• layernorm x86 optimization, re (#5745)

  nihui

  authored
  73d3519
  More actions

Commits on Oct 17, 2024

• blacklist mesa vulkan cooperative matrix feature (#5739)

  nihui

  authored
  bd1f39e
  More actions

• improve compatibility of harmonyos cpu topology abi (#5740)

  nihui

  authored
  8105c75
  More actions

Commits on Oct 15, 2024

• pnnx drop onnx weight-like graph input (#5736)

  nihui

  authored
  f856011
  More actions

• apply code-format changes

  

  nihui

  authored and

  github-actions[bot]

  committed
  121b1fe
  More actions

• multiheadattention int8 quantization (#5733)

  nihui

  authored
  66b54cb
  More actions

Commits on Oct 12, 2024

• gemm int8 quantization (#5706)

  nihui

  authored
  1c7af00
  More actions

Commits on Sep 25, 2024

• fix: typo (#5709)

  ankushgoel27

  authored
  9b5f6a3
  More actions

Commits on Sep 4, 2024

• pnnx fuse t5-layernorm as rmsnorm (#5675)

  nihui

  authored
  80c78a0
  More actions

• update modelwriter for rmsnorm (#5676)

  nihui

  authored
  21e54d8
  More actions

Commits on Sep 3, 2024

• x86 sse2/avx/avx512 optimization for rmsnorm (#5672)

  nihui

  authored
  204583b
  More actions

• arm neon optimzation for rmsnorm (#5668)

  nihui

  authored
  8077d34
  More actions

Commits on Sep 2, 2024

• embed int8 quantization and add embed test (#5667)

  nihui

  authored
  5df5413
  More actions

Commits on Aug 30, 2024

• pnnx fuse mobilevit style selfattention, onnx2pnnx handle more general gemm (#5659)

  nihui

  authored
  5e2d56d
  More actions

Commits on Aug 20, 2024

• update release download

  nihui

  authored
  25a22e0
  More actions

• Fixed bug #5637 (#5640)

  whyb

  authored
  a6d3ef5
  More actions

Commits on Aug 19, 2024

• Update README.md

  nihui

  authored
  27f64a1
  More actions

Commits on Aug 17, 2024

• Add python binding for loading bin from memory (#5164)

  joeyballentine

  authored
  a0c9e77
  More actions

• onnx2pnnx do not fold single constant for gemm weight (#5634)

  nihui

  authored
  4de5369
  More actions

Commits on Aug 16, 2024

• pnnx functionize do not create shadow op for identity consumers (#5632)

  nihui

  authored
  789d868
  More actions

• fix out of range read in convolution im2col aarch64 (#5631)

  nihui

  authored
  70310e9
  More actions

• benchmark: add Kunpeng 920 7260 (#5606)

  violet73

  authored
  07196ee
  More actions

Commits on Aug 15, 2024

• Add yolov8 ncnn example (#5506)

  whyb

  authored
  e550419
  More actions

• RMSNorm (#5630)

  nihui

  authored
  fdf0df3
  More actions

Commits on Aug 14, 2024

• pnnx drop torch.max torch.min indice node if not used (#5629)

  nihui

  authored
  abad90c
  More actions

• pnnx convert nn.RMSNorm F.rms_norm (#5628)

  nihui

  authored
  eb6e084
  More actions

• pnnx convert onnx resize with roi, torch.max torch.min with dim returns tuple (#5627)

  nihui

  authored
  c46278d
  More actions

• ci release ubuntu2404, major release yml refactor (#5624)

  nihui

  authored
  ae17e5e
  More actions

Pagination

PreviousNext