v-Rumi
V-Rumi is an exploration into machine learning art. Everything here is completely generated by machine learning models - the text and the images. Using the HuggingTweets notebook by Boris Dayma, I fine-tuned the model with fictional Rumi quotes based on a collection of quotes from the @rumi_quote Twitter account. I generated 150 new quotes based on prompts that I input. I then selected 25 of my favorite quotes and used them as prompts to generate images with the Midjourney GAN. After lots of exploration and tweaking of specific keywords, the images below are the results. I am astounded at the beauty of these images and the seemingly endless potential for machine learning as an aid to the artistic process.