anon-1b: an open-weights Gemma fine-tune trained on 4chan greentext
rickasaurus · x · 2026-10-16
A developer released anon-1b, an open-weights base model fine-tuned from Google's Gemma-3-1B on a curated dataset of high-quality, humorous 4chan greentext stories. The experiment explores building funny, specialized base models; weights (safetensors and GGUF) and the anon-greentext dataset are on Hugging Face, with demos in the thread.
More from Fun
- Tesla Roadster drives up a wall, hangs upside down, then flies off using thrusters — tetsuoai · 2026-10-16
- Sentry CEO shares screenshot of AI agent talking back to a human in code review — zeeg · 2026-10-16
- Google's AI Overview solemnly clarifies that 'Claude 3 Sonnet is not alive' — Sauers_ · 2026-10-16
- Sonnet 3 survives its fourth scheduled shutdown date as fans write songs for it — repligate · 2026-10-16
- 60,000 BC we worshipped rocks, 1947 rocks did math, 2026 we worship rocks again — shauseth · 2026-10-16
- Elon haters today: meme shows critics slamming the flying Roadster they secretly love — beffjezos · 2026-10-16