Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
tags:
|
3 |
+
- merge
|
4 |
+
---
|
5 |
+
# Miquella 120B GGUF
|
6 |
+
GGUF quantized weights for [miquella-120b](https://huggingface.co/alpindale/miquella-120b). Contains *all* quants.
|
7 |
+
|
8 |
+
I used Importance Matrices for the quantization, using random data generated from Q8_0 quant of the model for
|
9 |
+
maximum quality.
|
10 |
+
|
11 |
+
Due to the limitations of HF's file size, the larger files were split into multiple chunks. Instructions below.
|
12 |
+
|
13 |
+
## Linux
|
14 |
+
Example uses Q3_K_L. Replace the names appropriately for your quant of choice.
|
15 |
+
```sh
|
16 |
+
cat miquella-120b.Q3_K_L.gguf_part_* > miquella-120b.Q3_K_L.gguf && rm miquella-120b.Q3_K_L.gguf_part_*
|
17 |
+
```
|
18 |
+
|
19 |
+
## Windows
|
20 |
+
Example uses Q3_K_L. Replace the names appropriately for your quant of choice.
|
21 |
+
```sh
|
22 |
+
COPY /B miquella-120b.Q3_K_L.gguf_part_aa + miquella-120b.Q3_K_L.gguf_part_ab miquella-120b.Q3_K_L.gguf
|
23 |
+
```
|
24 |
+
Then delete the two splits.
|