File size: 2,329 Bytes
75e6414
 
 
 
 
 
 
 
 
2df6127
75e6414
2df6127
bdeb729
 
2df6127
 
 
 
 
75e6414
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
64
65
66
67
68
69
70
71
72
73
74
---
license: apache-2.0
tags:
- merge
- mergekit
- lazymergekit
- udkai/Turdus
- flemmingmiguel/DareBeagle-7B
---
## Exl2 version of [Undi95/OpenDolphinMaid-4x7b](https://huggingface.co/Undi95/OpenDolphinMaid-4x7b)  

## branch  
main : 8bpw h8  
b8h8 : 8bpw h8  

Using ThePile [0007.parquet](https://huggingface.co/datasets/EleutherAI/the_pile_deduplicated/resolve/refs%2Fconvert%2Fparquet/default/train/0007.parquet) as dataset  
 
Quantization settings : ```python convert.py -i models/flemmingmiguel_TurdusDareBeagle-7B -o TurdusDareBeagle-7B-temp -cf TurdusDareBeagle-7B-8bpw-h8-exl2 -c 0007.parquet -l 8192 -b 8 -hb 8 -ml 8192```  
### below this line is original readme 
# TurdusDareBeagle-7B

TurdusDareBeagle-7B is a merge of the following models using [LazyMergekit](https://colab.research.google.com/drive/1obulZ1ROXHjYLn6PPZJwRR6GzgQogxxb?usp=sharing):
* [udkai/Turdus](https://huggingface.co/udkai/Turdus)
* [flemmingmiguel/DareBeagle-7B](https://huggingface.co/flemmingmiguel/DareBeagle-7B)

As an experiment to find the best base merge to further fine-tuning, expect a lot of experiments named using parts of the component models until a clear winner emerges in the benchmarks

In this case .

## 🧩 Configuration

```yaml
slices:
    - sources:
      - model: udkai/Turdus
        layer_range: [0, 32]
      - model: flemmingmiguel/DareBeagle-7B
        layer_range: [0, 32]
merge_method: slerp
base_model: flemmingmiguel/DareBeagle-7B
parameters:
  t:
    - filter: self_attn
      value: [0, 0.5, 0.3, 0.7, 1]
    - filter: mlp
      value: [1, 0.5, 0.7, 0.3, 0]
    - value: 0.45 # fallback for rest of tensors
dtype: float16
```

## 💻 Usage

```python
!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "flemmingmiguel/TurdusDareBeagle-7B"
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"])
```