File size: 711 Bytes
09c90f3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
00d2f33
09c90f3
 
ddb56ce
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
---
library_name: transformers
tags:
- llama
- facebook
- meta
- llama-3
- conversational
- text-generation-inference
---
Official [AQLM](https://arxiv.org/abs/2401.06118) quantization of [meta-llama/Meta-Llama-3-70B
](https://huggingface.co/meta-llama/Meta-Llama-3-70B).

For this quantization, we used 1 codebook of 16 bits.

Results (measured with `lm_eval==4.0`): 
| Model      | Quantization | MMLU (5-shot) | ArcC| ArcE| Hellaswag | Winogrande | PiQA | Model size, Gb |
|------|------|-------|------|------|------|------|------|------|
|meta-llama/Meta-Llama-3-70B | - | 0.7868 | 0.6041 | 0.8670 | 0.6634 | 0.8248 | 0.8090 | 141.2 |
|  | 1x16 | 0.7486 | 0.5009	| 0.7896 | 0.6346 | 0.7954 | 0.7814 | 21.9 |