maywell commited on
Commit
5ad3d86
β€’
1 Parent(s): 660b823

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +83 -0
README.md CHANGED
@@ -1,3 +1,86 @@
1
  ---
2
  license: cc-by-sa-4.0
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: cc-by-sa-4.0
3
+ language:
4
+ - ko
5
+ - en
6
+ tags:
7
+ - moe
8
  ---
9
+
10
+ # **Synatra-Mixtral-8x7B**
11
+ <img src="./Synatra-Mixtral.png" alt="Synatra-Mixtral-8x7B" width="512"/>
12
+
13
+
14
+ **Synatra-Mixtral-8x7B** is a fine-tuned version of the Mixtral-8x7B-Instruct-v0.1 model using **Korean** datasets.
15
+
16
+ This model features overwhelmingly superior comprehension and inference capabilities and is licensed under CC-BY-SA.
17
+
18
+ # **EXL2 Info**
19
+
20
+ [measurement.json](./measurement.json)
21
+
22
+ 8.0bpw, 6.0bpw, 4.0bpw, 3.5bpw, 3.0bpw, 2.5bpw
23
+
24
+ <img src="./measurement.png" alt="Measurement" width="512"/>
25
+
26
+
27
+
28
+ # **License**
29
+
30
+ The "Model" is completely free (ie. base model, derivates, merges/mixes) to use for non-commercial purposes as long as the the included **cc-by-sa-4.0** license in any parent repository, and the non-commercial use statute remains, regardless of other models' licences.
31
+
32
+ # **Model Details**
33
+
34
+ **Base Model**
35
+ [mistralai/Mixtral-8x7B-Instruct-v0.1](https://huggingface.co/mistralai/Mixtral-8x7B-Instruct-v0.1)
36
+
37
+ **Trained On**
38
+ A100 80GB * 6
39
+
40
+ **Instruction format**
41
+
42
+ It follows **Alpaca** format.
43
+ ```
44
+ Below is an instruction that describes a task. Write a response that appropriately completes the request.
45
+
46
+ ### Instruction:
47
+ {input}
48
+
49
+ ### Response:
50
+ {output}
51
+ ```
52
+
53
+ # **Model Benchmark**
54
+ TBD
55
+
56
+ # **Implementation Code**
57
+
58
+ ```python
59
+ from transformers import AutoModelForCausalLM, AutoTokenizer
60
+
61
+ device = "cuda" # the device to load the model onto
62
+
63
+ model = AutoModelForCausalLM.from_pretrained("maywell/Synatra-Mixtral-8x7B")
64
+ tokenizer = AutoTokenizer.from_pretrained("maywell/Synatra-Mixtral-8x7B")
65
+
66
+ messages = [
67
+ {"role": "user", "content": "μ•„μΈμŠˆνƒ€μΈμ˜ μƒλŒ€μ„±μ΄λ‘ μ— λŒ€ν•΄μ„œ μžμ„Ένžˆ μ„€λͺ…ν•΄μ€˜."},
68
+ ]
69
+
70
+ encodeds = tokenizer.apply_chat_template(messages, return_tensors="pt")
71
+
72
+ model_inputs = encodeds.to(device)
73
+ model.to(device)
74
+
75
+ generated_ids = model.generate(model_inputs, max_new_tokens=1000, do_sample=True)
76
+ decoded = tokenizer.batch_decode(generated_ids)
77
+ print(decoded[0])
78
+ ```
79
+
80
+ # **Author's Message**
81
+
82
+ This model's training got sponsered by no one but support from people around Earth.
83
+
84
+ [Support Me](https://www.buymeacoffee.com/mwell)
85
+
86
+ Follow me on twitter: https://twitter.com/stablefluffy