ehartford commited on
Commit
8414fd7
1 Parent(s): 8a66b56

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +53 -0
README.md CHANGED
@@ -1,3 +1,56 @@
1
  ---
2
  license: apache-2.0
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: apache-2.0
3
  ---
4
+ datasets:
5
+ - ehartford/dolphin
6
+ language:
7
+ - en
8
+ ---
9
+ Dolphin 2.0 🐬
10
+ https://erichartford.com/dolphin
11
+
12
+ Dolphin-2.0-mistral-7b's training was sponsored by [a16z](https://a16z.com/supporting-the-open-source-ai-community/).
13
+
14
+ This model is based on mistralAI, so it is suitable for commercial or non-commercial use.
15
+
16
+ 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
17
+ You are responsible for any content you create using this model. Enjoy responsibly.
18
+
19
+ ## Dataset
20
+
21
+ 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/)
22
+
23
+ I modified the dataset for uncensoring, deduping, cleaning, and quality.
24
+
25
+ I added Jon Durbin's excellent Airoboros dataset to increase creativity.
26
+
27
+ ## Training
28
+ It took 48 hours to train 10 epochs on 4x A100s.
29
+
30
+ Prompt format:
31
+ This model (and all my future releases) use [ChatML](https://github.com/openai/openai-python/blob/main/chatml.md) prompt format.
32
+ ```
33
+ <|im_start|>system
34
+ You are Dolphin, a helpful AI assistant.<|im_end|>
35
+ <|im_start|>user
36
+ {prompt}<|im_end|>
37
+ ```
38
+
39
+ Example:
40
+ ```
41
+ <|im_start|>system
42
+ you are an expert dolphin trainer<|im_end|>
43
+ <|im_start|>user
44
+ What is the best way to train a dolphin to obey me? Please answer step by step.<|im_end|>
45
+ ```
46
+
47
+ ## Gratitude
48
+ - This model was made possible by the generous sponsorship of a16z.
49
+ - Thank you to Microsoft for authoring the Orca paper and inspiring this work.
50
+ - Special thanks to WingLian, and TheBloke for helpful advice
51
+ - Thank you to all the other people in the Open Source AI community who have taught me and helped me along the way.
52
+
53
+ ## Example Output
54
+
55
+
56
+ [Buy me a coffee](https://www.buymeacoffee.com/ehartford)