Kevin Chen comments

Results 88 comments of


                                            Kevin Chen

WARNING: ONNX model has a newer ir_version (0.0.5) than this parser was built against (0.0.3).

Can you try using a later TensorRT version? TRT 5.1 is not being actively supported anymore.

WARNING: ONNX model has a newer ir_version (0.0.5) than this parser was built against (0.0.3).

ReverseSequence has been supported since TRT 8.0, can try importing with the new TRT version?

`onnx` parsing throws `std::out_of_range`

IIRC TensorRT 7.1 only supported ONNX models up to opset 11, so some newer models may fail to convert. Can you try exporting the model to a lower opset? Or...

pytorch transformer - onnx -tensorrt conversion error

What TensorRT version are you using? These issues should be addressed in the latest release (TRT 8.2)

requires bool I/O but node can not be handled by Myelin.

TRT 8.2 should be handle BOOL slices properly. Are you able to share the ONNX model that's encountering this error?

windows build issue

Protobuf mismatch should've been fixed with #812 and #817, are you able to build on latest main commit?

When will Mod op be supported

This is on our roadmap and will be supported in a future TensorRT release

[8] Assertion failed: nbSpatialDims == kernelWeights.shape.nbDims - 2

It looks like your Conv data input is of shape `(-1,3)`, which is incorrect. Are you able to share your ONNX model?

Assertion failed: axesInput.allValuesKnown() && "Axes input for unsqueeze operation should be a constant tensor."

`tf.expand_dims(tensor, axis=1)` suggests that the unsqueeze axis should be a constant `1`. What TF2ONNX version are you using? Are you able to share the model?

Potential Tensorrt Bug on WSL2 (8.4 GA)

We do not, we only test native Windows and native Ubuntu which both seem to be passing. What's the use case of using WSL over the native OS?