onnxruntime
MatMulNBits + Add fusion
#20587
Merged
Overview
Commits
30
Changes
View On GitHub
Commits
  • save work
    edgchen1 committed 2 years ago
  • handle bias in fallback impl
    edgchen1 committed 2 years ago
  • reorder includes
    edgchen1 committed 2 years ago
  • save work - initial impl of optimizer
    edgchen1 committed 2 years ago
  • Merge remote-tracking branch 'origin/main' into edgchen1/matmul_nbits_bias
    edgchen1 committed 2 years ago
  • Add test, refine impl.
    edgchen1 committed 2 years ago
  • Merge remote-tracking branch 'origin/main' into edgchen1/matmul_nbits_bias
    edgchen1 committed 2 years ago
  • fix fusion, update test
    edgchen1 committed 2 years ago
  • Enable bias in SQNBitGemm benchmark.
    edgchen1 committed 2 years ago
  • clean up
    edgchen1 committed 2 years ago
  • move test functions around to avoid unused function warnings
    edgchen1 committed 2 years ago
  • add /bigobj for graph_tarnsform_test.cc
    edgchen1 committed 2 years ago
  • lint
    edgchen1 committed 2 years ago
  • Add Neural Speed build to post-merge-jobs.yml.
    edgchen1 committed 2 years ago
  • move graph_utils::IsSupportedProvider() in header so it is available in minimal extended build
    edgchen1 committed 2 years ago
  • put MatMulNBitsBiasFusion test in contrib ops ifdef
    edgchen1 committed 2 years ago
  • update ContribOperators.md
    edgchen1 committed 2 years ago
  • call RunWithConfig instead of Run
    edgchen1 committed 2 years ago
  • update OperatorKernels.md
    edgchen1 committed 2 years ago
  • add matmul_nbits_fusion files to extended minimal build source
    edgchen1 committed 2 years ago
  • Merge remote-tracking branch 'origin/main' into edgchen1/matmul_nbits_bias
    edgchen1 committed 2 years ago
  • Make DML MatMulNBits input check accomodate new inputs.
    edgchen1 committed 2 years ago
  • expect input arg counts of 0 for missing optional inputs
    edgchen1 committed 2 years ago
  • Merge branch 'edgchen1/matmul_nbits_bias' of github.com:microsoft/onnxruntime into edgchen1/matmul_nbits_bias
    edgchen1 committed 2 years ago
  • add runtime optimization test for matmulnbits and add fusion
    edgchen1 committed 2 years ago
  • add test onnx file
    edgchen1 committed 2 years ago
  • lint
    edgchen1 committed 2 years ago
  • Add !ORT_NEURAL_SPEED ifdefs around adding MatMulNBitsFusion transformer.
    edgchen1 committed 2 years ago
  • add ifdef around test
    edgchen1 committed 2 years ago
  • Merge branch 'main' into edgchen1/matmul_nbits_bias
    edgchen1 committed 2 years ago
FAQ Terms Privacy Refunds Impressum
Loading