Upload 3 files
Browse files- .gitattributes +1 -0
- README.md +233 -0
- sovythos.png +3 -0
- sovythos_tokenizer.json +0 -0
.gitattributes
CHANGED
|
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
+
sovythos.png filter=lfs diff=lfs merge=lfs -text
|
README.md
CHANGED
|
@@ -1,3 +1,236 @@
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
+
- ar
|
| 6 |
+
tags:
|
| 7 |
+
- sovereign
|
| 8 |
+
- sovythos
|
| 9 |
+
- llm
|
| 10 |
+
- transformer
|
| 11 |
+
- pytorch
|
| 12 |
+
- arabic
|
| 13 |
+
- english
|
| 14 |
+
- code
|
| 15 |
+
- open-source
|
| 16 |
+
pipeline_tag: text-generation
|
| 17 |
---
|
| 18 |
+
|
| 19 |
+
<p align="center">
|
| 20 |
+
<img src="https://huggingface.co/buckets/my0919175/Sovythos-66M-Base/resolve/sovythos.png?download=true" width="320" alt="SOVYTHOS">
|
| 21 |
+
</p>
|
| 22 |
+
|
| 23 |
+
<h1 align="center">π¦ SOVYTHOS-66M-Base</h1>
|
| 24 |
+
|
| 25 |
+
<p align="center">
|
| 26 |
+
<b>The first open-source pretrained language model developed completely from scratch under the Sovereign AI Project.</b>
|
| 27 |
+
</p>
|
| 28 |
+
|
| 29 |
+
<p align="center">
|
| 30 |
+
|
| 31 |
+

|
| 32 |
+

|
| 33 |
+

|
| 34 |
+

|
| 35 |
+

|
| 36 |
+
|
| 37 |
+
</p>
|
| 38 |
+
|
| 39 |
+
---
|
| 40 |
+
|
| 41 |
+
# π¦ Overview
|
| 42 |
+
|
| 43 |
+
**SOVYTHOS-66M-Base** is a decoder-only Transformer language model trained **entirely from scratch** as part of the **Sovereign AI Project**.
|
| 44 |
+
|
| 45 |
+
Unlike models that continue training from existing checkpoints, SOVYTHOS begins from randomly initialized weights and learns language directly from its training corpus.
|
| 46 |
+
|
| 47 |
+
This repository contains the **Base (Pretrained)** release intended for:
|
| 48 |
+
|
| 49 |
+
- Continued Pretraining
|
| 50 |
+
- Fine-tuning
|
| 51 |
+
- Instruction Tuning
|
| 52 |
+
- Research
|
| 53 |
+
- Educational Purposes
|
| 54 |
+
- Building downstream AI applications
|
| 55 |
+
|
| 56 |
+
> **This model is NOT instruction tuned and is not intended to behave like a production chatbot.**
|
| 57 |
+
|
| 58 |
+
---
|
| 59 |
+
|
| 60 |
+
# π Experience the Sovereign Platform
|
| 61 |
+
|
| 62 |
+
SOVYTHOS is the foundation of the **Sovereign AI Platform**.
|
| 63 |
+
|
| 64 |
+
If you'd like to experience the latest AI capabilities through a complete interface, you can try the live platform below.
|
| 65 |
+
|
| 66 |
+
## π Live Demo
|
| 67 |
+
|
| 68 |
+
**https://awaken-creation-guidance.ngrok-free.dev/**
|
| 69 |
+
|
| 70 |
+
### Available Features
|
| 71 |
+
|
| 72 |
+
- π¬ Modern AI Chat Interface
|
| 73 |
+
- β‘ Flash AI Models
|
| 74 |
+
- π§ Advanced Reasoning Models
|
| 75 |
+
- π» Coding Assistant
|
| 76 |
+
- π Arabic & English Support
|
| 77 |
+
- π₯ Continuous Platform Updates
|
| 78 |
+
|
| 79 |
+
> **Note:** The Sovereign platform may use newer models, APIs, and experimental capabilities beyond this Base release.
|
| 80 |
+
|
| 81 |
+
---
|
| 82 |
+
|
| 83 |
+
# π Model Specifications
|
| 84 |
+
|
| 85 |
+
| Property | Value |
|
| 86 |
+
|----------|--------|
|
| 87 |
+
| Model | SOVYTHOS-66M-Base |
|
| 88 |
+
| Parameters | ~66.7 Million |
|
| 89 |
+
| Architecture | Decoder-only Transformer |
|
| 90 |
+
| Hidden Size | 512 |
|
| 91 |
+
| Layers | 12 |
|
| 92 |
+
| Attention Heads | 16 |
|
| 93 |
+
| Framework | PyTorch |
|
| 94 |
+
| Tokenizer | Custom |
|
| 95 |
+
| Context Length | Configured during training |
|
| 96 |
+
| Training | From Scratch |
|
| 97 |
+
| Release | Base |
|
| 98 |
+
|
| 99 |
+
---
|
| 100 |
+
|
| 101 |
+
# π Languages
|
| 102 |
+
|
| 103 |
+
SOVYTHOS has been trained on multilingual data including:
|
| 104 |
+
|
| 105 |
+
- πͺπ¬ Arabic
|
| 106 |
+
- πͺπ¬ Egyptian Arabic
|
| 107 |
+
- πΊπΈ English
|
| 108 |
+
- Programming Languages
|
| 109 |
+
- Technical Documentation
|
| 110 |
+
- Mathematical Content
|
| 111 |
+
- General Web Text
|
| 112 |
+
|
| 113 |
+
---
|
| 114 |
+
|
| 115 |
+
# π― Intended Uses
|
| 116 |
+
|
| 117 |
+
This release is designed for:
|
| 118 |
+
|
| 119 |
+
- Research
|
| 120 |
+
- Fine-tuning
|
| 121 |
+
- Continued Pretraining
|
| 122 |
+
- Instruction Tuning
|
| 123 |
+
- Educational Experiments
|
| 124 |
+
- Language Model Development
|
| 125 |
+
- Custom AI Applications
|
| 126 |
+
|
| 127 |
+
---
|
| 128 |
+
|
| 129 |
+
# β οΈ Current Limitations
|
| 130 |
+
|
| 131 |
+
This repository contains the **Base Language Model** only.
|
| 132 |
+
|
| 133 |
+
It has **NOT** been instruction tuned.
|
| 134 |
+
|
| 135 |
+
Current limitations include:
|
| 136 |
+
|
| 137 |
+
- Hallucination of factual information
|
| 138 |
+
- Incorrect reasoning in some cases
|
| 139 |
+
- Repetition
|
| 140 |
+
- Incomplete generations
|
| 141 |
+
- Limited instruction following
|
| 142 |
+
|
| 143 |
+
These limitations are expected for a pretrained base model and will be addressed in future releases.
|
| 144 |
+
|
| 145 |
+
---
|
| 146 |
+
|
| 147 |
+
# π» Example
|
| 148 |
+
|
| 149 |
+
```python
|
| 150 |
+
prompt = "The capital of Egypt is"
|
| 151 |
+
|
| 152 |
+
output = model.generate(prompt)
|
| 153 |
+
```
|
| 154 |
+
|
| 155 |
+
---
|
| 156 |
+
|
| 157 |
+
# ποΈ Training
|
| 158 |
+
|
| 159 |
+
SOVYTHOS was trained completely from scratch using:
|
| 160 |
+
|
| 161 |
+
- Custom Tokenizer
|
| 162 |
+
- PyTorch
|
| 163 |
+
- CUDA
|
| 164 |
+
- Mixed Precision Training
|
| 165 |
+
- Automatic Best Checkpoint Saving
|
| 166 |
+
|
| 167 |
+
**No pretrained language model weights were used.**
|
| 168 |
+
|
| 169 |
+
---
|
| 170 |
+
|
| 171 |
+
# π£οΈ Roadmap
|
| 172 |
+
|
| 173 |
+
β
SOVYTHOS-66M-Base
|
| 174 |
+
|
| 175 |
+
π‘ SOVYTHOS-66M-Instruct *(In Development)*
|
| 176 |
+
|
| 177 |
+
β¬ SOVYTHOS Chat
|
| 178 |
+
|
| 179 |
+
β¬ Larger Parameter Models
|
| 180 |
+
|
| 181 |
+
β¬ Advanced Reasoning
|
| 182 |
+
|
| 183 |
+
β¬ Improved Coding
|
| 184 |
+
|
| 185 |
+
β¬ Better Arabic Understanding
|
| 186 |
+
|
| 187 |
+
β¬ Long Context Support
|
| 188 |
+
|
| 189 |
+
---
|
| 190 |
+
|
| 191 |
+
# π Sovereign Project
|
| 192 |
+
|
| 193 |
+
SOVYTHOS is the first public language model released under the **Sovereign AI Project**.
|
| 194 |
+
|
| 195 |
+
The long-term vision of Sovereign is to build an open ecosystem of multilingual language models with strong Arabic support, modern reasoning capabilities, and accessible open-source AI technologies.
|
| 196 |
+
|
| 197 |
+
---
|
| 198 |
+
|
| 199 |
+
# π Citation
|
| 200 |
+
|
| 201 |
+
```bibtex
|
| 202 |
+
@misc{sovythos66m,
|
| 203 |
+
title={SOVYTHOS-66M-Base},
|
| 204 |
+
author={Mahmoud Yasser},
|
| 205 |
+
year={2026},
|
| 206 |
+
publisher={Hugging Face},
|
| 207 |
+
}
|
| 208 |
+
```
|
| 209 |
+
|
| 210 |
+
---
|
| 211 |
+
|
| 212 |
+
# π¨βπ» Creator
|
| 213 |
+
|
| 214 |
+
**Mahmoud Yasser**
|
| 215 |
+
|
| 216 |
+
Founder of the **Sovereign AI Project**
|
| 217 |
+
|
| 218 |
+
---
|
| 219 |
+
|
| 220 |
+
# π License
|
| 221 |
+
|
| 222 |
+
**Apache License 2.0**
|
| 223 |
+
|
| 224 |
+
---
|
| 225 |
+
|
| 226 |
+
<p align="center">
|
| 227 |
+
|
| 228 |
+
### π¦ Welcome to the beginning of the Sovereign ecosystem.
|
| 229 |
+
|
| 230 |
+
The current release represents the first public pretrained foundation model.
|
| 231 |
+
|
| 232 |
+
Future releases will introduce instruction tuning, stronger reasoning, improved coding capabilities, and more advanced multilingual intelligence.
|
| 233 |
+
|
| 234 |
+
β If you find this project interesting, consider following the repository and trying the Sovereign AI Platform.
|
| 235 |
+
|
| 236 |
+
</p>
|
sovythos.png
ADDED
|
Git LFS Details
|
sovythos_tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|