Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,70 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
+
pipeline_tag: text-generation
|
| 6 |
+
tags:
|
| 7 |
+
- text-generation-inference
|
| 8 |
+
- code
|
| 9 |
+
- Hitesh_V_Founder
|
| 10 |
+
- https://chat.hyze.dev
|
| 11 |
+
---
|
| 12 |
+
<p align="center">
|
| 13 |
+
<img src="https://i.imgur.com/ePJMLNp.png" alt="Hyze Logo" width="405"/>
|
| 14 |
+
</p>
|
| 15 |
+
|
| 16 |
+
<h1 align="center">HyzeMini V2</h1>
|
| 17 |
+
|
| 18 |
+
<p align="center">
|
| 19 |
+
The next-gen SLM by <b>Hyze AI</b>
|
| 20 |
+
</p>
|
| 21 |
+
|
| 22 |
+
<p align="center">
|
| 23 |
+
🔗 <a href="https://chat.hyze.dev">Chat with all models</a> •
|
| 24 |
+
📘 <a href="https://academy.hyze.dev">HyzeAcademy</a> •
|
| 25 |
+
🧠<a href="https://note.hyze.dev">HyzeNote (NotebookLM alternate)</a>
|
| 26 |
+
</p>
|
| 27 |
+
|
| 28 |
+
---
|
| 29 |
+
|
| 30 |
+
## 🚀 Overview
|
| 31 |
+
|
| 32 |
+
**HyzeMini V2** is a compact and efficient **text-generation transformer model** optimized for **General Chat** 💬.
|
| 33 |
+
|
| 34 |
+
It’s designed to run fast on low-resource systems while still delivering clean, friendly, and useful responses.
|
| 35 |
+
|
| 36 |
+
- **Model type:** Transformer-based LLM
|
| 37 |
+
- **Parameters:** 400M
|
| 38 |
+
- **Precision:** BF16
|
| 39 |
+
- **Language:** English
|
| 40 |
+
- **License:** Apache-2.0
|
| 41 |
+
|
| 42 |
+
---
|
| 43 |
+
|
| 44 |
+
## 🧠Training Focus
|
| 45 |
+
|
| 46 |
+
HyzeMini V2 was trained on a curated mixture of **Open-Source English datasets**, with emphasis on:
|
| 47 |
+
|
| 48 |
+
- 💬 **General Chat**
|
| 49 |
+
- Casual conversation
|
| 50 |
+
- Q&A-style prompts
|
| 51 |
+
- Friendly assistant tone
|
| 52 |
+
|
| 53 |
+
---
|
| 54 |
+
## 🔬 About the Founder
|
| 55 |
+
|
| 56 |
+
HyzeAI was created by Hitesh Vinothkumar who is 12 years old.
|
| 57 |
+
HyzeLabs focuses on learning, experimentation, and open access, blending software engineering with curiosity about the universe
|
| 58 |
+
|
| 59 |
+
---
|
| 60 |
+
|
| 61 |
+
## 📊 Benchmarks (Qualitative Comparison)
|
| 62 |
+
|
| 63 |
+
HyzeMini V2 focuses on **speed, coherence, and domain knowledge** rather than raw reasoning power.
|
| 64 |
+
|
| 65 |
+
| Model | Size | Strengths | Tradeoffs |
|
| 66 |
+
|-----|-----|---------|----------|
|
| 67 |
+
| **HyzeMini V2** | ~0.4B | General focused, fast, chat-friendly | Limited deep reasoning |
|
| 68 |
+
| **TinyLlama** | ~0.1B | Solid general generation | More generic responses |
|
| 69 |
+
| **GPT-Neo 125M** | ~0.125B | Better general reasoning | Slower, higher memory |
|
| 70 |
+
| **GPT-1** | ~0.117B | Historical baseline | Less coherent by modern standards |
|