fastmachinelearning/qonnx

Issues

• RemoveIdentityOps does not correctly handle identity operations following fork-nodes

  #86 opened 4 months ago by iksnagreb
  0

• Add cleanup transformation sorting inputs of commutative operations

  #84 opened 4 months ago by iksnagreb
  0

• Change NodeLocalTransformation Pool to use spawn instead of fork

  #117 opened 7 months ago by bwintermann
  0

• Strage Behavior of FoldConstants Transformation

  #104 opened 10 months ago by iksnagreb
  1

• channels_last transformation support for branching models

  #100 opened a year ago by nghielme
  0

• Multithreshold operator FINN-ONNX execution error after 1D scalar add operator is absorbed into it.

  #98 opened a year ago by shashwat1198
  0

• Missing Shapes after Cleanup transformation

  #95 opened a year ago by Harsh9650
  1

• inference_cost_matmul: Confusion or Bug regarding the MAC-count of Scaled Dot-Product Attention

  #60 opened 2 years ago by iksnagreb
  1

• InferShapes fails after FoldTransposeIntoQuantInit

  #77 opened a year ago by iksnagreb
  1

• Support `QConv2DBatchnorm` and `QDenseBatchnorm` layers

  #10 opened 3 years ago by jmduarte
  1

• Fails to install on Python 3.10 due to pinned onnxruntime dependency

  #57 opened 2 years ago by iksnagreb
  3

• QKeras qconv2d test failure

  #54 opened 2 years ago by maltanar
  0

• Inference Cost calculation bug with pruned quantized model

  #49 opened 2 years ago by sdittmeier
  1

• Inference cost calculation bug with quantized model

  #51 opened 2 years ago by sdittmeier
  5

• Changed behavior after BatchNormToAffine transformation

  #50 opened 2 years ago by auphelia
  0

• GiveReadableTensorNames: Exception("Found multiple get_by_name matches, undefined behavior")

  #33 opened 2 years ago by sducouedic
  1

• Channels last conversion fails when Conv has a bias

  #32 opened 2 years ago by jmitrevs
  0

• Exception: Found multiple get_by_name matches, undefined behavior

  #30 opened 2 years ago by sducouedic
  2

• ConvertToChannelsLastAndClean: New transpose nodes don't have names and intermediate tensors are not kept in place

  #9 opened 3 years ago by heborras
  2

• Transformations should return whether a model was modified or not

  #6 opened 3 years ago by heborras
  1