Update README.md
Browse files
README.md
CHANGED
|
@@ -1,7 +1,22 @@
|
|
| 1 |
---
|
| 2 |
-
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
- mlx
|
| 6 |
library_name: mlx
|
| 7 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
base_model:
|
| 3 |
+
- CohereLabs/BLS-Mini-Code-1.0
|
| 4 |
+
base_model_relation: quantized
|
|
|
|
| 5 |
library_name: mlx
|
| 6 |
---
|
| 7 |
+
|
| 8 |
+
Generated via:
|
| 9 |
+
|
| 10 |
+
```
|
| 11 |
+
git clone https://github.com/eauchs/mlx-lm --single-branch --depth=1 --branch add-cohere2-moe
|
| 12 |
+
uv venv venv
|
| 13 |
+
source venv/bin/activate.fish
|
| 14 |
+
uv pip install mlx[cpu] ./mlx-lm
|
| 15 |
+
python -m mlx_lm.convert \
|
| 16 |
+
--hf-path "./BLS-Mini-Code-1.0" \
|
| 17 |
+
-q \
|
| 18 |
+
--q-bits 4 --q-group-size 64 \
|
| 19 |
+
--mlx-path ./cohere-mlx-q4
|
| 20 |
+
```
|
| 21 |
+
|
| 22 |
+
Requires this fork of mlx_lm to run (main doesn't include support for cohere2_moe yet):
|