Update README.md
Browse files
README.md
CHANGED
|
@@ -1,10 +1,29 @@
|
|
| 1 |
---
|
| 2 |
title: README
|
| 3 |
-
emoji:
|
| 4 |
-
colorFrom:
|
| 5 |
-
colorTo:
|
| 6 |
sdk: static
|
| 7 |
pinned: false
|
| 8 |
---
|
| 9 |
|
| 10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
title: README
|
| 3 |
+
emoji: ☄️
|
| 4 |
+
colorFrom: purple
|
| 5 |
+
colorTo: blue
|
| 6 |
sdk: static
|
| 7 |
pinned: false
|
| 8 |
---
|
| 9 |
|
| 10 |
+
## 👋 Welcome to met.dev
|
| 11 |
+
We are an independent research team. We craft custom datasets and develop two generations of language models. Our goal is to find the perfect balance between model compactness and real intelligence, making efficient AI accessible to developers. We stand for open source, rapid experimentation, and radical honesty in our results.
|
| 12 |
+
|
| 13 |
+
---
|
| 14 |
+
|
| 15 |
+
### Our Journey: From Mistakes to met2e
|
| 16 |
+
We started with **met(1)**—powerful, but way too resource-heavy. Then we stumbled with **met2**, where we simply lacked the expertise to train it properly. We took all that honest experience, re-engineered our approach, and built our current flagship, **met2e (3B)**—our active experimental generation.
|
| 17 |
+
|
| 18 |
+
---
|
| 19 |
+
|
| 20 |
+
### Model Lineups & Naming
|
| 21 |
+
* **met** — Models built on top of existing third-party architectures.
|
| 22 |
+
* **eos** — Our custom model family, trained completely from scratch.
|
| 23 |
+
|
| 24 |
+
**e** (Experimental) versions are quick preview releases showcasing new features before the main launch. **pro** versions are the smartest, deepest, and heaviest base models in the lineup.
|
| 25 |
+
|
| 26 |
+
---
|
| 27 |
+
|
| 28 |
+
## 🤝 Join Us
|
| 29 |
+
Test our models, grab our datasets, and share your feedback.
|