File size: 917 Bytes
81541e1
 
 
 
0115474
81541e1
 
 
 
 
 
 
 
 
 
 
 
0115474
81541e1
 
 
 
 
 
 
 
 
 
 
 
0115474
81541e1
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
---
quantized_by: bartowski
---

# Exllama v2 Quantizations of dolphin-2.7-mixtral-8x7b at 3.0 bits per weight

Using <a href="https://github.com/turboderp/exllamav2/releases/tag/v0.0.11">turboderp's ExLlamaV2 v0.0.11</a> for quantization.

Conversion was done using the default calibration dataset.

Original model: https://huggingface.co/cognitivecomputations/dolphin-2.7-mixtral-8x7b

## Download instructions

With git:

```shell
git clone --single-branch --branch 3.0 https://huggingface.co/bartowski/dolphin-2.7-mixtral-8x7b-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 dolphin-2.7-mixtral-8x7b-exl2
huggingface-cli download bartowski/dolphin-2.7-mixtral-8x7b-exl2 --revision 3_0 --local-dir dolphin-2.7-mixtral-8x7b-exl2 --local-dir-use-symlinks False
```