projects
Kiran Sanjeevan Cabeza
pytorch Spectrogram + CNN + LSTM networks for audio classification on the UrbanSound8K dataset
contributions to pytorch/audio
acknowledgement in the torchaudio paper; implementations such as phase vocoder on GPU, or spectrogram time and frequency masking:
keras / tensorflow implementation of the state-of-the-art object detection system You only look once
fauxgrad
a toy library and tutorial for learning the idea behind autodiff
contribution to mlflow
changes to allow mlflow docker projects to use GPU:
ml-image-compression
learned image compression based on CNNs in tensorflow
simple-receptive-field
numerically compute the receptive field of a conv block in pytorch
density estimation using random forests and KDE
implementation of Mapper (Topological Data Analysis technique) for extracting insights from high dimensional data
torchparse
simple pytorch model parser from a .cfg file
