Update README.md
Browse files
README.md
CHANGED
|
@@ -4,5 +4,14 @@ language:
|
|
| 4 |
- en
|
| 5 |
base_model:
|
| 6 |
- maidacundo/open-mythos-hf
|
| 7 |
-
- Qwen/Qwen2.5-
|
| 8 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4 |
- en
|
| 5 |
base_model:
|
| 6 |
- maidacundo/open-mythos-hf
|
| 7 |
+
- Qwen/Qwen2.5-1.5B
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
## Byte Mythos
|
| 11 |
+
|
| 12 |
+
**Byte Mythos** is a small experimental fork of Open Mythos, partially-vibecoded and modified with selectively injected weights from Qwen2.5‑1.5B.
|
| 13 |
+
It ports modern open‑model techniques such as BLT (Byte Latent Transformer) and BitNet, basically just to see how Mythos-like a small model could get, for fun.
|
| 14 |
+
Built using Fable 5 and Sonnet 5, the model has only undergone minimal training and remains intentionally rough and occasionally incoherent.
|
| 15 |
+
|
| 16 |
+
|
| 17 |
+
*It has not been finetuned at all, not even to give responses.*
|