Update README.md
Browse files
README.md
CHANGED
@@ -25,6 +25,104 @@ tags:
|
|
25 |
This model was converted to GGUF format from [`cognitivecomputations/Dolphin3.0-Mistral-24B`](https://huggingface.co/cognitivecomputations/Dolphin3.0-Mistral-24B) using llama.cpp via the ggml.ai's [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space.
|
26 |
Refer to the [original model card](https://huggingface.co/cognitivecomputations/Dolphin3.0-Mistral-24B) for more details on the model.
|
27 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
28 |
## Use with llama.cpp
|
29 |
Install llama.cpp through brew (works on Mac and Linux)
|
30 |
|
|
|
25 |
This model was converted to GGUF format from [`cognitivecomputations/Dolphin3.0-Mistral-24B`](https://huggingface.co/cognitivecomputations/Dolphin3.0-Mistral-24B) using llama.cpp via the ggml.ai's [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space.
|
26 |
Refer to the [original model card](https://huggingface.co/cognitivecomputations/Dolphin3.0-Mistral-24B) for more details on the model.
|
27 |
|
28 |
+
---
|
29 |
+
Our appreciation for the generous sponsors of Dolphin 3.0:
|
30 |
+
|
31 |
+
|
32 |
+
Dria https://x.com/driaforall - Inference Sponsor
|
33 |
+
Chutes https://x.com/rayon_labs - Compute Sponsor
|
34 |
+
Crusoe Cloud - Compute Sponsor
|
35 |
+
Andreessen Horowitz - provided the grant that originally launched Dolphin
|
36 |
+
|
37 |
+
|
38 |
+
|
39 |
+
|
40 |
+
|
41 |
+
|
42 |
+
|
43 |
+
What is Dolphin?
|
44 |
+
|
45 |
+
|
46 |
+
|
47 |
+
|
48 |
+
Dolphin 3.0 is the next generation of the Dolphin series of
|
49 |
+
instruct-tuned models. Designed to be the ultimate general purpose
|
50 |
+
local model, enabling coding, math, agentic, function calling, and
|
51 |
+
general use cases.
|
52 |
+
|
53 |
+
|
54 |
+
Dolphin aims to be a general purpose instruct model, similar to the
|
55 |
+
models behind ChatGPT, Claude, Gemini. But these models present
|
56 |
+
problems for businesses seeking to include AI in their products.
|
57 |
+
|
58 |
+
|
59 |
+
They maintain control of the system prompt, deprecating and changing things as they wish, often causing software to break.
|
60 |
+
They maintain control of the model versions, sometimes changing
|
61 |
+
things silently, or deprecating older models that your business relies
|
62 |
+
on.
|
63 |
+
They maintain control of the alignment, and in particular the alignment is one-size-fits all, not tailored to the application.
|
64 |
+
They can see all your queries and they can potentially use that data
|
65 |
+
in ways you wouldn't want.
|
66 |
+
Dolphin, in contrast, is steerable and gives control to the system
|
67 |
+
owner. You set the system prompt. You decide the alignment. You have
|
68 |
+
control of your data. Dolphin does not impose its ethics or guidelines
|
69 |
+
on you. You are the one who decides the guidelines.
|
70 |
+
|
71 |
+
|
72 |
+
Dolphin belongs to YOU, it is your tool, an extension of your will.
|
73 |
+
Just as you are personally responsible for what you do with a knife,
|
74 |
+
gun, fire, car, or the internet, you are the creator and originator of
|
75 |
+
any content you generate with Dolphin.
|
76 |
+
|
77 |
+
|
78 |
+
https://erichartford.com/uncensored-models
|
79 |
+
|
80 |
+
Chat Template
|
81 |
+
|
82 |
+
|
83 |
+
|
84 |
+
|
85 |
+
We use ChatML for the chat template.
|
86 |
+
|
87 |
+
|
88 |
+
<|im_start|>system
|
89 |
+
You are Dolphin, a helpful AI assistant.<|im_end|>
|
90 |
+
<|im_start|>user
|
91 |
+
{prompt}<|im_end|>
|
92 |
+
<|im_start|>assistant
|
93 |
+
|
94 |
+
|
95 |
+
|
96 |
+
|
97 |
+
|
98 |
+
|
99 |
+
|
100 |
+
|
101 |
+
System Prompt
|
102 |
+
|
103 |
+
|
104 |
+
|
105 |
+
|
106 |
+
In Dolphin, the system prompt is what you use to set the tone and
|
107 |
+
alignment of the responses. You can set a character, a mood, rules for
|
108 |
+
its behavior, and it will try its best to follow them.
|
109 |
+
|
110 |
+
|
111 |
+
Make sure to set the system prompt in order to set the tone and
|
112 |
+
guidelines for the responses - Otherwise, it will act in a default way
|
113 |
+
that might not be what you want.
|
114 |
+
|
115 |
+
|
116 |
+
Example use of system prompt:
|
117 |
+
|
118 |
+
|
119 |
+
<|im_start|>system
|
120 |
+
You are Dolphin, a golang coding assistant. you only code in golang. If the user requests any other programming language, return the solution in golang instead.<|im_end|>
|
121 |
+
<|im_start|>user
|
122 |
+
Please implement A* using python<|im_end|>
|
123 |
+
<|im_start|>assistant
|
124 |
+
|
125 |
+
---
|
126 |
## Use with llama.cpp
|
127 |
Install llama.cpp through brew (works on Mac and Linux)
|
128 |
|