Update README.md
Browse files
README.md
CHANGED
|
@@ -7,4 +7,39 @@ base_model:
|
|
| 7 |
pipeline_tag: text-generation
|
| 8 |
---
|
| 9 |
|
| 10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
pipeline_tag: text-generation
|
| 8 |
---
|
| 9 |
|
| 10 |
+
# Zeta 1 - First Step of Zeta Project, This is Epoch.
|
| 11 |
+
Zeta 1 is the first step that I, an ordinary person and consumer, have taken in pursuit of my dream as an individual.
|
| 12 |
+
|
| 13 |
+
## About Zeta 1
|
| 14 |
+
Zeta 1 is an LLM with about 400 million parameters.
|
| 15 |
+
|
| 16 |
+
It might be better to call it an SLM.
|
| 17 |
+
|
| 18 |
+
Zeta 1 is an SLM that was painstakingly created on a consumer computer.
|
| 19 |
+
|
| 20 |
+
### Computer Spec
|
| 21 |
+
|
| 22 |
+
**Machine:** *Mac Mini (M2 Pro, 32GB RAM, 2023)*
|
| 23 |
+
|
| 24 |
+
This is a technical matter, but on a Mac with Apple Silicon you can only choose between CPU or MPS (Apple's proprietary API).
|
| 25 |
+
|
| 26 |
+
CPU has some compatibility but is too slow.
|
| 27 |
+
|
| 28 |
+
MPS is somewhat faster but doesn't allow the use of optimization systems such as fp16.
|
| 29 |
+
|
| 30 |
+
Zeta 1 was built using MPS.
|
| 31 |
+
|
| 32 |
+
### Arguments of Trainer
|
| 33 |
+
- **train epochs:** 3
|
| 34 |
+
- **warmup steps:** 100
|
| 35 |
+
|
| 36 |
+
### Datasets
|
| 37 |
+
Details of the dataset used can be found [here](https://github.com/DiamondGotCat/Zeta-Dataset/releases/tag/zeta-1)
|
| 38 |
+
|
| 39 |
+
## Links
|
| 40 |
+
GitHub: [Zeta](https://github.com/DiamondGotCat/Zeta)
|
| 41 |
+
|
| 42 |
+
---
|
| 43 |
+
|
| 44 |
+
Zeta is just a small SLM.
|
| 45 |
+
But don't forget that it has big dreams inside.
|