CultriX commited on
Commit
28012db
1 Parent(s): 1ea98a5

Upload folder using huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +77 -0
README.md ADDED
@@ -0,0 +1,77 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - merge
4
+ - mergekit
5
+ - lazymergekit
6
+ - kaitchup/Mayonnaise-4in1-022
7
+ - macadeliccc/WestLake-7B-v2-laser-truthy-dpo
8
+ - vanillaOVO/supermario_v2
9
+ - SC44/Mistral-7B-private-spef
10
+ base_model:
11
+ - kaitchup/Mayonnaise-4in1-022
12
+ - macadeliccc/WestLake-7B-v2-laser-truthy-dpo
13
+ - vanillaOVO/supermario_v2
14
+ - SC44/Mistral-7B-private-spef
15
+ ---
16
+
17
+ # Wernicke-7B-v8
18
+
19
+ Wernicke-7B-v8 is a merge of the following models using [LazyMergekit](https://colab.research.google.com/drive/1obulZ1ROXHjYLn6PPZJwRR6GzgQogxxb?usp=sharing):
20
+ * [kaitchup/Mayonnaise-4in1-022](https://huggingface.co/kaitchup/Mayonnaise-4in1-022)
21
+ * [macadeliccc/WestLake-7B-v2-laser-truthy-dpo](https://huggingface.co/macadeliccc/WestLake-7B-v2-laser-truthy-dpo)
22
+ * [vanillaOVO/supermario_v2](https://huggingface.co/vanillaOVO/supermario_v2)
23
+ * [SC44/Mistral-7B-private-spef](https://huggingface.co/SC44/Mistral-7B-private-spef)
24
+
25
+ ## 🧩 Configuration
26
+
27
+ ```yaml
28
+ models:
29
+ - model: CultriX/Wernicke-7B-v1
30
+ # No parameters necessary for base model
31
+ - model: kaitchup/Mayonnaise-4in1-022
32
+ parameters:
33
+ density: 0.53
34
+ weight: 0.40
35
+ - model: macadeliccc/WestLake-7B-v2-laser-truthy-dpo
36
+ parameters:
37
+ density: 0.53
38
+ weight: 0.25
39
+ - model: vanillaOVO/supermario_v2
40
+ parameters:
41
+ density: 0.53
42
+ weight: 0.25
43
+ - model: SC44/Mistral-7B-private-spef
44
+ parameters:
45
+ density: 0.53
46
+ weight: 0.20
47
+ merge_method: dare_ties
48
+ base_model: CultriX/Wernicke-7B-v1
49
+ parameters:
50
+ int8_mask: true
51
+ dtype: float16
52
+ ```
53
+
54
+ ## 💻 Usage
55
+
56
+ ```python
57
+ !pip install -qU transformers accelerate
58
+
59
+ from transformers import AutoTokenizer
60
+ import transformers
61
+ import torch
62
+
63
+ model = "CultriX/Wernicke-7B-v8"
64
+ messages = [{"role": "user", "content": "What is a large language model?"}]
65
+
66
+ tokenizer = AutoTokenizer.from_pretrained(model)
67
+ prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
68
+ pipeline = transformers.pipeline(
69
+ "text-generation",
70
+ model=model,
71
+ torch_dtype=torch.float16,
72
+ device_map="auto",
73
+ )
74
+
75
+ outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
76
+ print(outputs[0]["generated_text"])
77
+ ```