Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: apache-2.0
|
| 4 |
+
language:
|
| 5 |
+
- en
|
| 6 |
+
- fr
|
| 7 |
+
- es
|
| 8 |
+
- it
|
| 9 |
+
- pt
|
| 10 |
+
- zh
|
| 11 |
+
- ar
|
| 12 |
+
- ru
|
| 13 |
+
base_model:
|
| 14 |
+
- HuggingFaceTB/SmolLM3-3B-Base
|
| 15 |
+
tags:
|
| 16 |
+
- openvino
|
| 17 |
+
- int4
|
| 18 |
+
- quantization
|
| 19 |
+
- edge-deployment
|
| 20 |
+
- optimization
|
| 21 |
+
- smollm3
|
| 22 |
+
inference: false
|
| 23 |
+
---
|
| 24 |
+
|
| 25 |
+
# SmolLM3 INT4 OpenVINO
|
| 26 |
+
|
| 27 |
+
## 🚀 Optimized for Edge Deployment
|
| 28 |
+
|
| 29 |
+
This is an INT4 quantized version of [SmolLM3-3B](https://huggingface.co/HuggingFaceTB/SmolLM3-3B) using OpenVINO, designed for efficient inference on edge devices and CPUs.
|
| 30 |
+
|
| 31 |
+
## Model Overview
|
| 32 |
+
|
| 33 |
+
- **Base Model:** SmolLM3-3B (3B parameters)
|
| 34 |
+
- **Quantization:** INT4 via OpenVINO
|
| 35 |
+
- **Size Reduction:** ~75% smaller than original
|
| 36 |
+
- **Target Hardware:** CPUs, Intel GPUs, NPUs
|
| 37 |
+
- **Use Cases:** Local inference, edge deployment, resource-constrained environments
|
| 38 |
+
|
| 39 |
+
## 🔧 Technical Details
|
| 40 |
+
|
| 41 |
+
### Quantization Process
|
| 42 |
+
```python
|
| 43 |
+
# Quantized using OpenVINO NNCF
|
| 44 |
+
# INT4 symmetric quantization
|