Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
tags:
|
3 |
+
- moe
|
4 |
+
- llama
|
5 |
+
- '3'
|
6 |
+
- llama 3
|
7 |
+
- 2x8b
|
8 |
+
---
|
9 |
+
# Llama-3-Teal-Instruct-2x8B-MoE
|
10 |
+
This is a experimental MoE created from [meta-llama/Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct) and [nvidia/Llama3-ChatQA-1.5-8B](https://huggingface.co/nvidia/Llama3-ChatQA-1.5-8B) using Mergekit.
|
11 |
+
|
12 |
+
Mergekit yaml file:
|
13 |
+
```
|
14 |
+
base_model: Meta-Llama-3-8B-Instruct
|
15 |
+
experts:
|
16 |
+
- source_model: Meta-Llama-3-8B-Instruct
|
17 |
+
positive_prompts:
|
18 |
+
- "explain"
|
19 |
+
- "chat"
|
20 |
+
- "assistant"
|
21 |
+
- source_model: Llama3-ChatQA-1.5-8B
|
22 |
+
positive_prompts:
|
23 |
+
- "python"
|
24 |
+
- "math"
|
25 |
+
- "solve"
|
26 |
+
- "code"
|
27 |
+
gate_mode: hidden
|
28 |
+
dtype: float16
|
29 |
+
```
|