File size: 1,695 Bytes
dcea1df
 
 
 
 
 
 
 
 
 
 
 
7823ba1
dcea1df
7823ba1
dcea1df
 
 
 
 
 
 
7823ba1
dcea1df
 
 
 
 
 
 
 
 
 
7823ba1
dcea1df
 
 
 
 
 
 
 
 
 
 
 
 
 
7823ba1
dcea1df
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
---
tags:
- merge
- mergekit
- lazymergekit
- pankajmathur/orca_mini_v5_8b_dpo
- mlabonne/Daredevil-8B-abliterated
base_model:
- pankajmathur/orca_mini_v5_8b_dpo
- mlabonne/Daredevil-8B-abliterated
---

# Llama-3-8B-combo-merge

Llama-3-8B-combo-merge is a merge of the following models using [LazyMergekit](https://colab.research.google.com/drive/1obulZ1ROXHjYLn6PPZJwRR6GzgQogxxb?usp=sharing):
* [pankajmathur/orca_mini_v5_8b_dpo](https://huggingface.co/pankajmathur/orca_mini_v5_8b_dpo)
* [mlabonne/Daredevil-8B-abliterated](https://huggingface.co/mlabonne/Daredevil-8B-abliterated)

## 🧩 Configuration

```yaml
models:
  - model: DavidAhn/Llama-3-8B-combo-merge
    # no parameters necessary for base model
  - model: pankajmathur/orca_mini_v5_8b_dpo
    parameters:
      density: 0.5
      weight: 0.5
  - model: mlabonne/Daredevil-8B-abliterated
    parameters:
      density: 0.5
      weight: 0.3
merge_method: ties
base_model: DavidAhn/Llama-3-8B-combo-merge
parameters:
  normalize: true
dtype: float16
```

## 💻 Usage

```python
!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "DavidAhn/Llama-3-8B-combo-merge"
messages = [{"role": "user", "content": "What is a large language model?"}]

tokenizer = AutoTokenizer.from_pretrained(model)
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    torch_dtype=torch.float16,
    device_map="auto",
)

outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])
```