Commit
•
16972a2
1
Parent(s):
fc94ecf
Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
datasets:
|
| 3 |
+
- JeanKaddour/minipile
|
| 4 |
+
language:
|
| 5 |
+
- en
|
| 6 |
+
license: mpl-2.0
|
| 7 |
+
---
|
| 8 |
+
|
| 9 |
+
# RWKV Minipile
|
| 10 |
+
|
| 11 |
+
## Model Specifications
|
| 12 |
+
|
| 13 |
+
- **Architecture**: RWKV
|
| 14 |
+
- **Vocabulary Size**: 65,536
|
| 15 |
+
- **Embedding Size**: 768
|
| 16 |
+
- **Number of Layers**: 12
|
| 17 |
+
- **Context Length**: 512
|
| 18 |
+
- **Data Type**: bfloat16
|
| 19 |
+
- **Dataset**: Minipile
|
| 20 |
+
- **Tokens**: 20,643,840 (20 Million)
|
| 21 |
+
|
| 22 |
+
The model underwent a rigorous training regimen, completing 30 epochs to optimize performance.
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
## Inference
|
| 26 |
+
```bash
|
| 27 |
+
pip install torch numpy
|
| 28 |
+
|
| 29 |
+
python inference.py
|
| 30 |
+
```
|