Model release

This release concludes the project with release of the 15 models we trained over the course of 2 weeks. Each model has it's own strengths and weaknesses. Flower 15 seems to be producing the most coherent outputs. Through, feel free to experiment and find out which model produces the outputs you like.

Some models are plain broken, so don't bother trying to make sense of it all, refer to the training docs to find out what might've led to such behaviour.

The release has been split into 2 parts because of GitHub's 2GB/file size limit. pretrained_models_1.zip contains models 1-8, and pretrained_models_2.zip contains models 9-15

NOTE: Whenever you start draw2pix, it loads all models available in the pretrained_models directory. So, all these combined, you might need about 3GB RAM to have all these 15 models loaded at the same time

NOTE 2 : The powershell scripts used to download/update these models might take some time because of abysmal download speeds. Atleast from what we've seen. We're not sure if it's github throttling speed, or just powershell using a single thread to download. Nonetheless, just something to keep an eye out for. Feel free to go the manual route if that isn't sufficient