A neural network based file sorter. Trains an autoencoder to sort images or audio based on the similarity of their encodings, or uses the OpenAI CLIP model.
MIT License
27
stars
1
forks
source link
Some stride values not allowed due to deconvolution (ConvTranspose2d) output_padding #1
Need to ensure that de-convolution parameters work in such a way that out image will always be the right size -- ideally no cropping.