File size: 4,812 Bytes
cf05e07
 
 
 
 
 
 
 
 
 
 
6781bb1
cf05e07
 
 
 
 
 
 
 
 
 
 
6781bb1
cf05e07
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c9da6ad
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
<!-- markdownlint-disable first-line-h1 -->
<!-- markdownlint-disable html -->
<!-- markdownlint-disable no-duplicate-header -->

![image/png](https://cdn-uploads.huggingface.co/production/uploads/677064c8117cc54ff842798e/0JrIXszV26Eruk2zcsxLu.png)

<div align="center">
  <img src="https://github.com/souschef-ai/SousChef/blob/main/assets/logo.svg?raw=true" width="60%" alt="SousChef AI" />
</div>
<hr>
<div align="center" style="line-height: 1;">
  <a href="https://www.cookprotocol.xyz/" target="_blank" style="margin: 2px;">
    <img alt="Homepage" src="https://github.com/souschef-ai/SousChef/blob/main/assets/badge.svg?raw=true" style="display: inline-block; vertical-align: middle;"/>
  </a>
  <a href="https://chat.souschef.ai/" target="_blank" style="margin: 2px;">
    <img alt="Chat" src="https://img.shields.io/badge/Chat-SousChef%20AI-6c5ce7?color=6c5ce7&logoColor=white" style="display: inline-block; vertical-align: middle;"/>
  </a>
  <a href="https://huggingface.co/souschef-ai" target="_blank" style="margin: 2px;">
    <img alt="Hugging Face" src="https://img.shields.io/badge/Hugging%20Face-SousChef%20AI-fbc531?color=fbc531&logoColor=white" style="display: inline-block; vertical-align: middle;"/>
  </a>
</div>

<div align="center" style="line-height: 1;">
  <a href="https://x.com/cookprotocol" target="_blank" style="margin: 2px;">
    <img alt="Twitter Follow" src="https://img.shields.io/badge/Twitter-souschef_ai-1da1f2?logo=x&logoColor=white" style="display: inline-block; vertical-align: middle;"/>
  </a>
</div>

<div align="center" style="line-height: 1;">
  <a href="https://github.com/souschef-ai/SousChef/blob/main/LICENSE-CODE" style="margin: 2px;">
    <img alt="Code License" src="https://img.shields.io/badge/Code_License-MIT-f5c542?color=f5c542" style="display: inline-block; vertical-align: middle;"/>
  </a>
  <a href="https://github.com/souschef-ai/SousChef/blob/main/LICENSE-MODEL" style="margin: 2px;">
    <img alt="Model License" src="https://img.shields.io/badge/Model_License-Open%20RAIL-M-4caf50?color=4caf50" style="display: inline-block; vertical-align: middle;"/>
  </a>
</div>

<p align="center">
  <a href="https://github.com/souschef-ai/SousChef/blob/main/SousChef_Paper.pdf"><b>Whitepaper</b> 📄</a>
</p>

## 1. Introduction

SousChef v1 is an open-source AI model designed to assist developers and power users in building, managing, and innovating on the Hyperliquid ecosystem. With state-of-the-art performance across a wide range of benchmarks, SousChef AI is tailored to deliver actionable insights, streamline portfolio management, and generate optimized strategies. 

### Key Features:
- **Precision**: Advanced reasoning and decision-making capabilities using cutting-edge training objectives.
- **Scalability**: Efficient inference on both consumer-grade GPUs and high-performance clusters.
- **Versatility**: Pretrained for multi-domain tasks, including mathematics, code generation, and portfolio optimization.

SousChef v1 is a step towards making Hyperliquid accessible, transparent, and efficient for everyone.

---

## 2. Model Overview

### **Architecture**
SousChef v1 employs a Hybrid Transformer architecture with:
- **256 Expert Modules** for specific task optimization.
- **Multi-Context Analysis** for longer attention spans (up to 128K tokens).
- **Dynamic Knowledge Fusion**, enabling real-time adaptation to new data.

### **Training Methodology**
1. **Pretraining**: 10 trillion high-quality tokens, covering diverse domains (finance, mathematics, development).
2. **Fine-tuning**: Reinforcement learning on real-world user scenarios.
3. **Ethical Constraints**: Integrated Responsible AI License (RAIL-M) to prevent misuse.

### **Performance**
<p align="center">
  <img width="80%" src="assets/benchmark_souschef_v1.png">
</p>

SousChef v1 consistently outperforms comparable models on key benchmarks, demonstrating exceptional performance in code generation, math reasoning, and portfolio strategies.

---

## 3. Model Downloads

<div align="center">

| **Model**     | **# Parameters** | **Context Length** | **Download**                           |
| :-----------: | :--------------: | :----------------: | :------------------------------------: |
| SousChef-v1   | 250B             | 128K               | [🤗 HuggingFace](https://huggingface.co/souschef-ai/SousChef-v1) |

</div>

### Installation
Clone the repository and install dependencies:
```bash
pip install souschef-ai
```

---

## 4. License

The code and pre-trained weights for SousChef v1 are licensed under:
- **Code**: MIT License
- **Model**: Open RAIL-M License

---

## 5. Contact

Questions or feedback? Reach out via:
- Email: [[email protected]](mailto:[email protected])
- Discord: [Join the SousChef AI community](https://discord.gg/souschef-ai)