README.md

---
title: Glassbox
emoji: 🔍
colorFrom: gray
colorTo: red
sdk: static
pinned: false
license: mit
short_description: Small models, narrow tasks
---

# Glassbox

Small models, narrow tasks. Trained from scratch, under 100M parameters, each doing one thing.

The 91M Bulgarian model against INSAIT's BgGPT-Gemma-2-2.6B, 28× larger:

| | ours | theirs |
|---|---|---|
| fluency judging, 2000 pairs | **0.9305** | 0.9175 |
| bits per character, same text | 1.3041 | **1.1600** |

One win, one loss, both on the [card](https://huggingface.co/glassbox/gpt-alpha-bg-91m).
[bg-eval](https://github.com/iksnerd/bg-eval) re-runs them without trusting us.

[restorer-interp-lab.iksnerd.workers.dev](https://restorer-interp-lab.iksnerd.workers.dev)