Wholesome Greentext LLM

Building and fine‑tuning a small LLM that generates coherent, wholesome greentext stories, analyzing model size impact, data collection, and synthetic data generation.

Overview

I built/finetuned an small-LLM (lol) to reproduce coherent stories using idiosyncratic language. The project is inspired by TinyStories by MSFT research.
It’s the aim to reproduce small little wholesome stories in the style of 4chan greentexts and serve that on a website.