Ayush Shridhar
Ayush Shridhar
Yeah, sounds great. My project mainly focuses on importing ONNX models into the Flux framework. However, my mentor @PhilTomson was interested in making a verilog backend, so that we could...
By verilog, I mean a common backend from which we could export model to different frameworks.
@rened @quaertym Are you still interested in this? Most of my work in ONNX.jl for loading models into Flux is complete, and I'm able to load models with a pretty...
There is an initial repo up [here](https://www.github.com/ayush1999/ONNXKnet.jl). @denizyuret would be great if you could check your mail. Thanks!
It'd be great to try out the models we previously couldn't. There's been a lot of refactoring going on in Flux/NNlib. This includes asymmetric pads/LRN layer and CrossCor for Flux....
@ianshmean There's an error in the onnx model itself: See https://github.com/onnx/models/issues/60.
You could try downloading the pretrained model in some other framework (PyTorch/TensorFlow) and use the newer version of onnx exporters to generate the onnx file, and load it in using...
I'm not sure if this has anything to do with the Julia version. I was just trying to install the OpenCL drivers, following the steps mentioned in the readme.
@aviks Sure, will do. Thanks a lot for notifying me. I wasn't aware of another Keras.jl.
@marco-c I just tried using 2-gram features for the `TfIdfVectorizer`, here is the gist of the results: Before: ``` X: (11007, 21990), y: (11007,) Cross Validation scores: Accuracy: f0.951746383784194 (+/-...