About

I made this with data available from the Perseus Project, specifically a dataset I got from GitHub.

I used ChatGPT and Mixtral8x7B to write code to translate the betacode formatted Greek characters in the Perseus data into unicode Greek. The translated characters were then sent, chunk-by-chunk, to a local instance of Mixtral8x7B via an OpenAI-like API for translatation to English. Once the Greek-to-English translation was done, each chunk's translation and Greek text was put into a PostgreSQL database.

To create the webpage and images, I used another script to pull pairs of original Greek and translated text from the database. Every 100th translation was then sent to DALL-E 3 via the OpenAI API for image generation. Each generated image was then downloaded into a structured directory and the database updated to indicate the presence of an image for the relevant translation. The script then generated an HTML file which, along with the directory of images, I rsync'd to my webserver.