File size: 941 Bytes
401ff25 2d0ee1a 401ff25 2d0ee1a 401ff25 2d0ee1a 401ff25 |
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 |
---
inference: false
language:
- en
tags:
- llama
- llama 2
quantized_by: bartowski
---
# Exllama v2 Quantizations of cat-v1.0-13b at 8.0,6.0,5.0,4.25,4.0,3.75
Using <a href="https://github.com/turboderp/exllamav2/releases/tag/v0.0.7">turboderp's ExLlamaV2 v0.0.7</a> for quantization.
Conversion was done using wikitext-103-raw-v1-test.parquet as calibration dataset.
Original model: https://huggingface.co/Doctor-Shotgun/cat-v1.0-13b
## Download instructions
With git:
```shell
git clone --single-branch --branch 8.0,6.0,5.0,4.25,4.0,3.75 https://huggingface.co/bartowski/cat-v1.0-13b-exl2
```
With huggingface hub (credit to TheBloke for instructions):
```shell
pip3 install huggingface-hub
```
To download from a different branch, add the `--revision` parameter:
```shell
mkdir cat-v1.0-13b-exl2
huggingface-cli download bartowski/cat-v1.0-13b-exl2 --revision --local-dir cat-v1.0-13b-exl2 --local-dir-use-symlinks False
```
|