femiari commited on
Commit
e500664
1 Parent(s): ea144f2

Upload folder using huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +36 -17
README.md CHANGED
@@ -1,7 +1,7 @@
1
  ---
2
  base_model:
3
- - Qwen/Qwen-7B
4
- - TideDra/Qwen-VL-Chat-DPO
5
  license: apache-2.0
6
  tags:
7
  - moe
@@ -9,38 +9,57 @@ tags:
9
  - merge
10
  - mergekit
11
  - lazymergekit
12
- - Qwen/Qwen-7B
13
- - TideDra/Qwen-VL-Chat-DPO
14
  ---
15
 
16
  # QwenMoEAriel
17
 
18
  QwenMoEAriel is a Mixture of Experts (MoE) made with the following models using [LazyMergekit](https://colab.research.google.com/drive/1obulZ1ROXHjYLn6PPZJwRR6GzgQogxxb?usp=sharing):
19
- * [Qwen/Qwen-7B](https://huggingface.co/Qwen/Qwen-7B)
20
- * [TideDra/Qwen-VL-Chat-DPO](https://huggingface.co/TideDra/Qwen-VL-Chat-DPO)
21
 
22
  ## 🧩 Configuration
23
 
24
  ```yaml
25
- base_model: Qwen/Qwen-7B
26
- gate_mode: cheap_embed
27
  experts:
28
- - source_model: Qwen/Qwen-7B
29
  positive_prompts:
 
30
  - "chat"
31
  - "assistant"
32
- - "tell me"
33
- - "explain"
34
- - "I want"
35
- - source_model: TideDra/Qwen-VL-Chat-DPO
 
 
 
 
 
 
 
 
 
 
 
 
36
  positive_prompts:
37
- - "code"
38
  - "python"
39
- - "javascript"
 
 
40
  - "programming"
 
41
  - "algorithm"
42
- shared_experts:
43
- - source_model: Qwen/Qwen-7B
 
 
 
 
 
44
  ```
45
 
46
  ## 💻 Usage
 
1
  ---
2
  base_model:
3
+ - Meta-Llama-3-8B-Instruct
4
+ - Llama-3-8B-Instruct-Coder
5
  license: apache-2.0
6
  tags:
7
  - moe
 
9
  - merge
10
  - mergekit
11
  - lazymergekit
12
+ - Meta-Llama-3-8B-Instruct
13
+ - Llama-3-8B-Instruct-Coder
14
  ---
15
 
16
  # QwenMoEAriel
17
 
18
  QwenMoEAriel is a Mixture of Experts (MoE) made with the following models using [LazyMergekit](https://colab.research.google.com/drive/1obulZ1ROXHjYLn6PPZJwRR6GzgQogxxb?usp=sharing):
19
+ * [Meta-Llama-3-8B-Instruct](https://huggingface.co/Meta-Llama-3-8B-Instruct)
20
+ * [Llama-3-8B-Instruct-Coder](https://huggingface.co/Llama-3-8B-Instruct-Coder)
21
 
22
  ## 🧩 Configuration
23
 
24
  ```yaml
25
+ base_model: Meta-Llama-3-8B-Instruct
 
26
  experts:
27
+ - source_model: Meta-Llama-3-8B-Instruct
28
  positive_prompts:
29
+ - "explain"
30
  - "chat"
31
  - "assistant"
32
+ - "think"
33
+ - "roleplay"
34
+ - "versatile"
35
+ - "helpful"
36
+ - "factual"
37
+ - "integrated"
38
+ - "adaptive"
39
+ - "comprehensive"
40
+ - "balanced"
41
+ negative_prompts:
42
+ - "specialized"
43
+ - "narrow"
44
+ - "focused"
45
+ - "limited"
46
+ - "specific"
47
+ - source_model: Llama-3-8B-Instruct-Coder
48
  positive_prompts:
 
49
  - "python"
50
+ - "math"
51
+ - "solve"
52
+ - "code"
53
  - "programming"
54
+ - "javascript"
55
  - "algorithm"
56
+ - "factual"
57
+ negative_prompts:
58
+ - "sorry"
59
+ - "cannot"
60
+ - "concise"
61
+ - "imaginative"
62
+ - "creative"
63
  ```
64
 
65
  ## 💻 Usage