File size: 878 Bytes
32bf49b
 
bad1d2d
 
3f81155
bad1d2d
 
 
 
32bf49b
bad1d2d
32bf49b
 
554b8a5
32bf49b
 
 
 
 
 
 
 
 
 
 
 
 
1bfb29f
 
 
 
 
 
 
 
 
32bf49b
1bfb29f
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
---
license: llama3.1
base_model:
- meta-llama/Llama-3.1-8B-Instruct
base_model_relation: quantized
tags:
- VPTQ
- Quantized
- Quantization
---

**Disclaimer**: 

The model is reproduced based on the paper *VPTQ: Extreme Low-bit Vector Post-Training Quantization for Large Language Models* [github](https://github.com/microsoft/vptq) and [arXiv](https://arxiv.org/abs/2409.17066)

The model itself is sourced from a community release. 

It is intended only for experimental purposes. 

Users are responsible for any consequences arising from the use of this model.

**Note**: 

The PPL test results are for reference only and were collected using GPTQ testing script.

```json
{
    "ctx_2048": {
        "wikitext2": 8.166712760925293
    },
    "ctx_4096": {
        "wikitext2": 7.6312713623046875
    },
    "ctx_8192": {
        "wikitext2": 7.3152079582214355
    }
}
```