Update README.md
Browse files
README.md
CHANGED
|
@@ -17,15 +17,6 @@ Pre-quantized version of Moondream 3 Preview for MLX inference.
|
|
| 17 |
- **Other weights**: bf16 (unchanged)
|
| 18 |
- **Memory savings**: ~60% reduction in MoE weight memory
|
| 19 |
|
| 20 |
-
## Usage
|
| 21 |
-
|
| 22 |
-
This model is designed for use with the moondream-station MLX backend.
|
| 23 |
-
|
| 24 |
-
```python
|
| 25 |
-
# In moondream-station, use with:
|
| 26 |
-
moondream-station serve --backend mlx
|
| 27 |
-
```
|
| 28 |
-
|
| 29 |
## Source
|
| 30 |
|
| 31 |
Quantized from [moondream/moondream3-preview](https://huggingface.co/moondream/moondream3-preview)
|
|
|
|
| 17 |
- **Other weights**: bf16 (unchanged)
|
| 18 |
- **Memory savings**: ~60% reduction in MoE weight memory
|
| 19 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
## Source
|
| 21 |
|
| 22 |
Quantized from [moondream/moondream3-preview](https://huggingface.co/moondream/moondream3-preview)
|