masahi

This fixes buidling the galen bin with a recent DD. @frankmcsherry

Currently, `CombinedNonMaxSuppression` converter does not support class-specific boxes, i.e. `q == num_class` case in https://www.tensorflow.org/api_docs/python/tf/image/combined_non_max_suppression. https://github.com/onnx/tensorflow-onnx/blob/01ec092384a8b2096425c2a5de7deb7b3215afdf/tf2onnx/onnx_opset/tensor.py#L1873-L1874 Apparently, "official" TF FasterRCNN / MaskRCNN models uses such variant of combined nms. So...

Hardcoded `UInt8` idtype in FakeQuantWithMinMaxArgs causes error

1

These two lines look odd to me... Why hard code the input dtype to uint8? https://github.com/onnx/tensorflow-onnx/blob/482330f9958eb45c805933f04e2b0a5c7a494f23/tf2onnx/onnx_opset/quantize.py#L57 https://github.com/onnx/tensorflow-onnx/blob/482330f9958eb45c805933f04e2b0a5c7a494f23/tf2onnx/onnx_opset/quantize.py#L63-L68 I got the following error when converting QAT-ed yolo4 model from https://github.com/openvinotoolkit/nncf/tree/develop/examples/tensorflow/object_detection. TF...

`cl_khr_integer_dot_product` support

1

Hi, I'm looking forward to the availability of `cl_khr_integer_dot_product` extension released in OpenCL 3.0.8. Is this being planned?

`pyls` we depend on has not been active https://github.com/palantir/python-language-server Emacs lsp-mode also says `pyls` is deprecated https://emacs-lsp.github.io/lsp-mode/page/lsp-pyls/ Apparently, [pyright](https://github.com/microsoft/pyright) is a popular lang server right now. Probably we should support...

Models I added in https://github.com/tlc-pack/TLCBench/pull/5 were downloadable at least for a few days, but the link is no longer valid, apparently due to the github storage limit (same issue as...

The logic in https://github.com/apache/tvm/blob/52d6b59a39f503fe382b4d7cbac4b02f9e44aae0/python/tvm/topi/generic/conv2d.py#L480-L499 is supposed to legalize the input dtype to be able to apply target-specific intrinsics that only support one of int8 or uint8. For example, the x86...

[Hexagon] Support template-free meta schedule tuning

1

Building on https://github.com/apache/tvm/pull/12845, this PR adds an initial support for template-free auto tuning on Hexagon. Test cases demonstrate: * Auto-scheduler style, template free tuning for fp16 conv2d in NHWC layout....

[Node][Metaschedule] Allow ignoring NDArray raw data in StructuralEqual / Hash

2

Motivation: Currently, `StructuralEqual/Hash` take into account NDArray raw data to determine equality / compute hash. This is problematic for `link-params = True` case, because meta schedule task extraction and database...

[ScatterElements] Clarify the requirement on indices uniqueness

8

[The spec](https://github.com/onnx/onnx/blob/master/docs/Operators.md#ScatterElements) of `ScatterElements` op doesn't tell if elements of `indices` tensor need to be unique or not along the scattered axis. I think there should be an explicit clarification...