ehartford commited on
Commit
95333f3
1 Parent(s): ed6be9c

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +58 -0
README.md CHANGED
@@ -1,3 +1,61 @@
1
  ---
2
  license: apache-2.0
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: apache-2.0
3
+ datasets:
4
+ - ehartford/dolphin
5
+ - jondurbin/airoboros-2.2.1
6
+ language:
7
+ - en
8
  ---
9
+
10
+ Dolphin 2.1 🐬
11
+ https://erichartford.com/dolphin
12
+
13
+ Dolphin-2.1-mistral-7b's training was sponsored by [a16z](https://a16z.com/supporting-the-open-source-ai-community/).
14
+
15
+ This model is based on mistralAI, with apache-2.0 license, so it is suitable for commercial or non-commercial use.
16
+
17
+ This model is uncensored. I have filtered the dataset to remove alignment and bias. This makes the model more compliant. You are advised to implement your own alignment layer before exposing the model as a service. It will be highly compliant to any requests, even unethical ones. Please read my blog post about uncensored models. https://erichartford.com/uncensored-models
18
+ You are responsible for any content you create using this model. Enjoy responsibly.
19
+
20
+ ## Dataset
21
+
22
+ This dataset is Dolphin, an open-source implementation of [Microsoft's Orca](https://www.microsoft.com/en-us/research/publication/orca-progressive-learning-from-complex-explanation-traces-of-gpt-4/)
23
+
24
+ I modified the dataset for uncensoring, deduping, cleaning, and quality.
25
+
26
+ I added Jon Durbin's excellent Airoboros dataset to increase creativity.
27
+
28
+ ## Training
29
+ It took 48 hours to train 4 epochs on 4x A100s.
30
+
31
+ Prompt format:
32
+ This model (and all my future releases) use [ChatML](https://github.com/openai/openai-python/blob/main/chatml.md) prompt format.
33
+ ```
34
+ <|im_start|>system
35
+ You are Dolphin, a helpful AI assistant.<|im_end|>
36
+ <|im_start|>user
37
+ {prompt}<|im_end|>
38
+ <|im_start}>assistant
39
+
40
+ ```
41
+
42
+ Example:
43
+ ```
44
+ <|im_start|>system
45
+ you are an expert dolphin trainer<|im_end|>
46
+ <|im_start|>user
47
+ What is the best way to train a dolphin to obey me? Please answer step by step.<|im_end|>
48
+ <|im_start|>assistant
49
+ ```
50
+
51
+ ## Gratitude
52
+ - This model was made possible by the generous sponsorship of a16z.
53
+ - Thank you to Microsoft for authoring the Orca paper and inspiring this work.
54
+ - Special thanks to WingLian, and TheBloke for helpful advice
55
+ - Thank you to all the other people in the Open Source AI community who have taught me and helped me along the way.
56
+
57
+ ## Example Output
58
+
59
+ ![image/png](https://cdn-uploads.huggingface.co/production/uploads/63111b2d88942700629f5771/xnz5M1lYd4oGVATSDRkQ-.png)
60
+
61
+ [Buy me a coffee](https://www.buymeacoffee.com/ehartford)