Commit
•
4f883c3
1
Parent(s):
088443f
Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
Official [AQLM](https://arxiv.org/abs/2401.06118) quantization of `mistralai/Mixtral-8x7B-v0.1`.
|
2 |
+
|
3 |
+
For this quantization, we used 1 codebook of 16 bits.
|
4 |
+
|
5 |
+
Selected evaluation results for this and other models:
|
6 |
+
|
7 |
+
| Model | AQLM scheme | WikiText 2 PPL | Model size, Gb | Hub link |
|
8 |
+
|------------|-------------|----------------|----------------|--------------------------------------------------------------------------|
|
9 |
+
| Llama-2-7b | 1x16 | 6.31 | 2.4 | [Link](https://huggingface.co/BlackSamorez/Llama-2-7b-AQLM-2Bit-1x16-hf) |
|
10 |
+
| Llama-2-7b | 2x8 | 7.98 | 2.2 | [Link](https://huggingface.co/BlackSamorez/Llama-2-7b-AQLM-2Bit-2x8-hf) |
|
11 |
+
| Llama-2-7b | 8x8 | 7.83 | 2.2 | [Link](https://huggingface.co/BlackSamorez/Llama-2-7b-AQLM-2Bit-8x8-hf) |
|
12 |
+
| Llama-2-13b| 1x16 | 5.41 | 4.1 | [Link](https://huggingface.co/BlackSamorez/Llama-2-13b-AQLM-2Bit-1x16-hf)|
|
13 |
+
| Llama-2-70b| 1x16 | 3.96 | 18.8 | [Link](https://huggingface.co/BlackSamorez/Llama-2-70b-AQLM-2Bit-1x16-hf)|
|
14 |
+
| Mixtral-8x7b (THIS)| 1x16 | 4.37 | 12.6 | [Link](https://huggingface.co/BlackSamorez/Mixtral-8x7b-AQLM-2Bit-1x16-hf)|
|
15 |
+
|
16 |
+
To learn more about the inference and how to quantize models yourself, please refer to the [official GitHub repo](https://github.com/Vahe1994/AQLM).
|