https://github.com/audiodude/endless-love-letters

Scripts for generating love letters for my wife via AI
https://github.com/audiodude/endless-love-letters

Last synced: about 2 months ago
JSON representation

Scripts for generating love letters for my wife via AI

• Host: GitHub
• URL: https://github.com/audiodude/endless-love-letters
• Owner: audiodude
• License: mit
• Created: 2023-09-14T16:53:38.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2024-09-09T05:56:57.000Z (4 months ago)
• Last Synced: 2024-10-15T00:47:08.723Z (3 months ago)
• Language: Python
• Size: 333 KB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Endless Love Letters

This is the source code for the site at https://endless-love-letters.com

Since 2019, I have written my wife Abby a love letter every day, for a total of 1458 captured

at the time of the Google Takeout export.

These scripts take an `.mbox` file with a bunch of love letter emails, parse them out, feed them to

the OpenAI GPT 3.5 finetuning API to finetune a model, then generate new love letters.

The web interface includes both v1 (static love letter list generated with some pre-GPT charNN-like

solution whose name has been lost) and v2 (love letters generated on the fly with ChatGPT).

## Populating the SECRETS file

Look at `SECRETS.example` for the tokens which should go into `SECRETS`, which is a JSON file.

Neither the site nor the command line tools will function without the `SECRETS` being set.

## Generating a love letter offline

Offline meaning without the web app, this still contacts the OpenAI API.

Just run:

```bash

python generate.py

```

Note that the prompts and model id are hardcoded into the script.

## Generating the database

In MySQL, run:

```sql

CREATE DATABASE endless_love_leeters;

```

Then load the schema:

```bash

mysql -u  -p endless_love_letters < schema.sql

```

## Running the web app locally

```bash

FLASK_DEBUG=1 FLASK_APP=app pipenv run flask run

```

## Deploying to fly,io

1. Install the Fly.io command line tool, `flyctl`

```bash

fly deploy

```