-
I'm not too familiar with the model loading process but is it possible to use this library to load in different model formats (like h5) other than the provided .model extension?
P.S. amazing work o…
-
Hi, I intend to use `tract` for inference with `AWS Lambda`. I've observed that the initialization and optimization of ONNX models (from &[u8]) can be 2-3 times slower than the actual model execution.…
-
Is it on the roadmap to allow `tract`'s crates to be used without the standard library?
The use case we have in mind is using `tract` on embedded devices because a pure Rust library is a lot easier…
-
What was the rationale to give `sum_reduce` a `normalize` attribute while also providing a `mean_reduce` operation, instead of just having `sum_reduce` and `mean_reduce` operations without the `normal…
zoeoz updated
5 years ago
-
I believe this issue has been seen before, I forget what the fix is or if there had been updates made that should fix this. When trying to run 'python setup.py install' within the parser folder, the f…
-
Supported layers are limited, for a good reason, to those for which a code generator exists. It would be interesting to at least provide a Python implementation and specification for additional layer …
-
Hi there.
I see no discussions about [ONNX](https://github.com/onnx/onnx), an open source framework to share models among different deep learning libraries. I don't know if this is better implement…
-
It would be great to take advantage of the cost_model setup for arbitrary ARM CPUs (like the a57) and generate better cost models for then, but I’m not really sure on the procedure.
Digging in a li…
-
Hi, really excited about this project. Have been looking into using tract for object detection inference on CPU and can get around 73-75ms inference speed with onnxruntime but only around 300ms with t…
-
Hello again!
As usual, this is something I'm still investigating but want to open up for input. I've been debugging a text model over the last few days, trying to figure out why Tract is so much sl…