Update README.md
Browse files
README.md
CHANGED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
title: SmolLM2-135M From Scratch
|
| 3 |
+
emoji: 🤖
|
| 4 |
+
colorFrom: blue
|
| 5 |
+
colorTo: purple
|
| 6 |
+
sdk: gradio
|
| 7 |
+
sdk_version: 4.44.0
|
| 8 |
+
app_file: app.py
|
| 9 |
+
pinned: false
|
| 10 |
+
license: mit
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
# SmolLM2-135M: Complete From-Scratch Implementation
|
| 14 |
+
|
| 15 |
+
This Space demonstrates a complete reverse-engineered implementation of SmolLM2-135M.
|
| 16 |
+
|
| 17 |
+
## Features
|
| 18 |
+
- 🔍 Reverse-engineered architecture
|
| 19 |
+
- 🏋️ Trained for 5,000+ steps
|
| 20 |
+
- ✅ Checkpoint validation
|
| 21 |
+
- ⚡ Optimized with Flash Attention & Mixed Precision
|
| 22 |
+
|
| 23 |
+
## Links
|
| 24 |
+
- **GitHub Repository**: [abi2024/smollm2-135-implementation](https://github.com/abi2024/smollm2-135-implementation)
|
| 25 |
+
- **Model Details**: See the Model Info tab
|
| 26 |
+
|
| 27 |
+
## Usage
|
| 28 |
+
Enter a prompt and adjust generation parameters to see the model in action!
|