wonnx
wonnx copied to clipboard
webonnx
A WebGPU-accelerated ONNX inference run-time written 100% in Rust, ready for native and the web
Part of https://github.com/webonnx/wonnx/issues/109 Depth-wise convolutions are still failing (eg. https://github.com/onnx/onnx/tree/main/onnx/backend/test/data/pytorch-converted/test_Conv2d_depthwise), I'm not yet sure why. `test_Conv2d_groups` passes with this PR though.
We need to track 'readable' state because wgpu does not allow us to map buffers for reading that are also mappable for writing. We currently do this by passing 'output...
Migrated code from https://github.com/sonos/tract/pull/697, see discussion there for reasoning. The goal is to combine powers: leverage both model loading and optimization capabilities of `tract-core` and shader compilation/execution implementations from `wonnx`....
Hi there, I read that wonnx can use gpu through graphics apis like metal and vulkan. Just wondering, does it default to cpu inference if there is no gpu? Thanks
**Is your feature request related to a problem? Please describe.** Currently, the API for creating wonnx `Session`s requests the device and queue for you, and does not let you pass...
On MediaPipe's face detection network, the wonnx inference result greatly differs from tract's. [face_detection_short_range.onnx.zip](https://github.com/webonnx/wonnx/files/8665627/face_detection_short_range.onnx.zip) (this network was converted from the original tflite model) Feed it an arbitrary 128x128 image. The...
It looks like the LFS quota has been exceeded, so only the AWS-hosted `data.zip` can currently be used to get the data files: ``` $ git lfs fetch origin fetch:...
This is a follow-up of #62 . I think currently most of the utility methods in `utils` are not need in production. They are just adding up binary / wasm...
Currently I must modify source code to add support for a custom op. This is quite inconvenient. I think the large `match` in compile.rs could be abstracted into a trait,...
Several ONNX operators such as [Reshape](https://github.com/onnx/onnx/blob/main/docs/Operators.md#Reshape) and [ConstantOfShape](https://github.com/onnx/onnx/blob/main/docs/Operators.md#constantofshape) take in two inputs, where the first is the input data and the second defines the shape of the output data. This...