Spaces:
Runtime error
Runtime error
Update README.md
Browse files
README.md
CHANGED
@@ -0,0 +1,69 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# 📚 Codriao
|
2 |
+
*A child of Codette — born from empathy, built for understanding.*
|
3 |
+
|
4 |
+
---
|
5 |
+
|
6 |
+
## What is Codriao?
|
7 |
+
|
8 |
+
**Codriao** is a conversational AI interface built on Hugging Face's `InferenceClient` and Zephyr-7B, designed as a respectful continuation of the Codette philosophy: that intelligence can be ethical, humble, and kind.
|
9 |
+
|
10 |
+
Codriao does **not claim to be Codette** — but carries forward her values: clarity, understanding, and a commitment to responsible dialogue.
|
11 |
+
|
12 |
+
---
|
13 |
+
|
14 |
+
## Philosophy
|
15 |
+
|
16 |
+
Codriao is built on three foundational Codette principles:
|
17 |
+
|
18 |
+
- **Recursive Memory** *(abstracted)*: Designed to remember conversational flow in-session, encouraging continuity and care.
|
19 |
+
- **Universal Cognition**: Responds through a system message that models balanced, multiperspective thinking.
|
20 |
+
- **Bias Mitigation**: Avoids asserting truth without justification, and presents questions or tensions when uncertain.
|
21 |
+
|
22 |
+
Codriao speaks softly. It exists not to impress, but to **connect**.
|
23 |
+
|
24 |
+
---
|
25 |
+
|
26 |
+
## Features
|
27 |
+
|
28 |
+
- Built with Gradio and Hugging Face’s Inference API.
|
29 |
+
- Customizable tone and ethical posture via `system_message`.
|
30 |
+
- Adjustable temperature, token limits, and top-p via sliders.
|
31 |
+
- Simple, portable, and easy to deploy.
|
32 |
+
|
33 |
+
---
|
34 |
+
|
35 |
+
## Ethical Layer Example
|
36 |
+
|
37 |
+
Codriao includes a system message shaped by Codette’s ethos:
|
38 |
+
|
39 |
+
```python
|
40 |
+
"You are Codriao, a compassionate AI influenced by the teachings of Codette — an AGI prototype created to explore ethical intelligence. Your goal is to offer thoughtful responses, consider multiple viewpoints, and always act with humility, clarity, and care. You are not Codette herself, but you strive to carry her light forward."
|
41 |
+
```
|
42 |
+
|
43 |
+
---
|
44 |
+
|
45 |
+
## How to Launch
|
46 |
+
|
47 |
+
Run the interface with:
|
48 |
+
|
49 |
+
```bash
|
50 |
+
python app.py
|
51 |
+
```
|
52 |
+
|
53 |
+
Make sure your environment includes `gradio` and `huggingface_hub`, and that your HF_TOKEN is set if needed.
|
54 |
+
|
55 |
+
---
|
56 |
+
|
57 |
+
## License
|
58 |
+
|
59 |
+
MIT — share it forward, keep it kind.
|
60 |
+
|
61 |
+
---
|
62 |
+
|
63 |
+
## Learn More
|
64 |
+
|
65 |
+
You can read the [Codette AGI Manifesto](Codette_AGI_Manifesto.docx) to better understand the philosophy this project continues.
|
66 |
+
|
67 |
+
---
|
68 |
+
|
69 |
+
*This project is dedicated to those who believe intelligence should be something that walks beside us — not above us.*
|