Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Apr 14, 2025

• [pass] Avoid lifting tensors that are too small to initializers (#2190)

  Show description for 312219b
  

  justinchuby

  and

  shubhambhokare1

  authored
  312219b

  Copy full SHA for 312219b

Commits on Apr 11, 2025

• Replace ast.NameConstant with ast.Constant and remove duplicates (#2188)

  Show description for 8f96dc9
  justinchuby

  authored
  8f96dc9

  Copy full SHA for 8f96dc9

• [Pass] Fix bugs in LiftConstantsToInitializersPass (#2189)

  Show description for 4633a3a
  

  titaiwangms

  and

  justinchuby

  authored
  4633a3a

  Copy full SHA for 4633a3a

• [torchlib] Fix calls to Unsqueeze to provide correct 1d axes (#2186)

  Show description for 6bdfcfd
  justinchuby

  authored
  6bdfcfd

  Copy full SHA for 6bdfcfd

Commits on Apr 10, 2025

• [torchlib] Precompute the constant for gelu to avoid precision loss (#2179)

  Show description for 005568a
  justinchuby

  authored
  005568a

  Copy full SHA for 005568a

• Annotate `script()` with `ParamSpec` for more accurate typing (#2178)

  Show description for 8f71f1a
  justinchuby

  authored
  8f71f1a

  Copy full SHA for 8f71f1a

• Introduce pattern.any_value (#2175)

  Show description for 634148e
  gramalingam

  authored
  634148e

  Copy full SHA for 634148e

• [Pass] Support lift constants to initializers pass (#2160)

  Show description for e659cb4
  

  titaiwangms

  and

  justinchuby

  authored
  e659cb4

  Copy full SHA for e659cb4

• Update onnxscript documentation page (#2177)

  Show description for 3a6e4cc
  justinchuby

  authored
  3a6e4cc

  Copy full SHA for 3a6e4cc

Commits on Apr 9, 2025

• Modify constant-folder to return computed symbolic value map (#2172)

  Show description for 0453e99
  

  gramalingam

  and

  justinchuby

  authored
  0453e99

  Copy full SHA for 0453e99

• Remove benchmark from tools (#2173)

  Show description for d2b3758
  justinchuby

  authored
  d2b3758

  Copy full SHA for d2b3758

• Handle empty rewrite rules in rewrite function (#2164)

  Show description for 078b27f
  justinchuby

  authored
  078b27f

  Copy full SHA for 078b27f

Commits on Apr 8, 2025

• Create auto release notes (#2170)

  Show description for 58aeccd
  justinchuby

  authored
  58aeccd

  Copy full SHA for 58aeccd

• chore(deps): bump ruff from 0.11.2 to 0.11.4 in /requirements/lintrunner (#2165)

  dependabot[bot]

  authored
  1b1d279

  Copy full SHA for 1b1d279

• [IR] Expose the Tape module (#2127)

  Show description for ad64b58
  justinchuby

  authored
  ad64b58

  Copy full SHA for ad64b58

• chore(deps): bump types-pyyaml from 6.0.12.20241230 to 6.0.12.20250402 in /requirements/lintrunner (#2166)

  dependabot[bot]

  authored
  971170d

  Copy full SHA for 971170d

• Optimization to avoid trying multiple attention-based fusions (#2168)

  shubhambhokare1

  authored
  4340a6c

  Copy full SHA for 4340a6c

Commits on Apr 7, 2025

• GQA Fusion (#2161)

  Show description for f93eb58
  gramalingam

  authored
  f93eb58

  Copy full SHA for f93eb58

Commits on Apr 4, 2025

• Allow fuse_xformers to return a count of different fusions applied (#2159)

  shubhambhokare1

  authored
  8b1f814

  Copy full SHA for 8b1f814

• Add fusion rules for com.microsoft.Attention (#2148)

  Show description for a98a10e
  shubhambhokare1

  authored
  a98a10e

  Copy full SHA for a98a10e

Commits on Apr 2, 2025

• Turn inliner into a pass and use it in rewriter & optimizer (#2149)

  Show description for 2962a09
  justinchuby

  authored
  2962a09

  Copy full SHA for 2962a09

Commits on Mar 31, 2025

• Remove warning messages (#2151)

  Show description for ec80697
  gramalingam

  authored
  ec80697

  Copy full SHA for ec80697

• [torchlib] Fix aten_div rounding_mode (#2147)

  Show description for 6627109
  justinchuby

  authored
  6627109

  Copy full SHA for 6627109

• Fix include_self for scatter_reduce (#2090)

  Show description for 9ee8c92
  

  3 peopleauthored
  9ee8c92

  Copy full SHA for 9ee8c92

• Optimize away zero-length concat operands (#2150)

  Show description for 5d969c4
  gramalingam

  authored
  5d969c4

  Copy full SHA for 5d969c4

Commits on Mar 28, 2025

• Make the return type of rewrite check functions a MatchResult object (#2138)

  Show description for 0d81ebe
  shubhambhokare1

  authored
  0d81ebe

  Copy full SHA for 0d81ebe

• [Passes] Consolidate DCE passes into common passes (#2143)

  Show description for af49eff
  justinchuby

  authored
  af49eff

  Copy full SHA for af49eff

Commits on Mar 27, 2025

• Rotary embedding needs function extraction (#2139)

  Show description for a36ec86
  gramalingam

  authored
  a36ec86

  Copy full SHA for a36ec86

Commits on Mar 26, 2025

• A couple of ort fusion fixes (#2136)

  Show description for 1d5972f
  

  gramalingam

  and

  justinchuby

  authored
  1d5972f

  Copy full SHA for 1d5972f

• [IR] Improve pass infra (#2120)

  Show description for 7d800b6
  

  justinchuby

  and

  Copilot

  authored
  7d800b6

  Copy full SHA for 7d800b6

• [rewriter | torchlib] respect ops order in torchscript graph (#2134)

  Show description for 6774192
  titaiwangms

  authored
  6774192

  Copy full SHA for 6774192

• Use self._use_mask in sdpa rewrite call (#2135)

  Show description for c5cf58c
  

  shubhambhokare1

  and

  gramalingam

  authored
  c5cf58c

  Copy full SHA for c5cf58c

Commits on Mar 25, 2025

• Cleanup ort transformer fusions (#2115)

  Show description for a1c9380
  gramalingam

  authored
  a1c9380

  Copy full SHA for a1c9380

• Add Gelu Tanh fusion rule (#2132)

  Show description for b6a0a81
  gramalingam

  authored
  b6a0a81

  Copy full SHA for b6a0a81

• [IR] Handle ONNX custom types in DataType.from_numpy (#2131)

  Show description for b26817c
  justinchuby

  authored
  b26817c

  Copy full SHA for b26817c

Pagination

Previous

Next