bartowski commited on
Commit
07f2bdb
1 Parent(s): dfc9850

measurement.json

Browse files
Files changed (2) hide show
  1. README.md +70 -0
  2. measurement.json +0 -0
README.md ADDED
@@ -0,0 +1,70 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - merge
4
+ - mergekit
5
+ - lazymergekit
6
+ - Locutusque/Hercules-2.5-Mistral-7B
7
+ - teknium/OpenHermes-2.5-Mistral-7B
8
+ base_model:
9
+ - Locutusque/Hercules-2.5-Mistral-7B
10
+ - teknium/OpenHermes-2.5-Mistral-7B
11
+ quantized_by: bartowski
12
+ pipeline_tag: text-generation
13
+ ---
14
+
15
+ ## Exllama v2 Quantizations of OpenHercules-2.5-Mistral-7B
16
+
17
+ Using <a href="https://github.com/turboderp/exllamav2/releases/tag/v0.0.14">turboderp's ExLlamaV2 v0.0.14</a> for quantization.
18
+
19
+ <b>The "main" branch only contains the measurement.json, download one of the other branches for the model (see below)</b>
20
+
21
+ Each branch contains an individual bits per weight, with the main one containing only the meaurement.json for further conversions.
22
+
23
+ Original model: https://huggingface.co/Locutusque/OpenHercules-2.5-Mistral-7B
24
+
25
+ | Branch | Bits | lm_head bits | VRAM (4k) | VRAM (16k) | VRAM (32k) | Description |
26
+ | ----- | ---- | ------- | ------ | ------ | ------ | ------------ |
27
+ | [8_0](https://huggingface.co/bartowski/OpenHercules-2.5-Mistral-7B-exl2/tree/8_0) | 8.0 | 8.0 | 8.4 GB | 9.8 GB | 11.8 GB | Maximum quality that ExLlamaV2 can produce, near unquantized performance. |
28
+ | [6_5](https://huggingface.co/bartowski/OpenHercules-2.5-Mistral-7B-exl2/tree/6_5) | 6.5 | 8.0 | 7.2 GB | 8.6 GB | 10.6 GB | Very similar to 8.0, good tradeoff of size vs performance, **recommended**. |
29
+ | [5_0](https://huggingface.co/bartowski/OpenHercules-2.5-Mistral-7B-exl2/tree/5_0) | 5.0 | 6.0 | 6.0 GB | 7.4 GB | 9.4 GB | Slightly lower quality vs 6.5, but usable on 8GB cards. |
30
+ | [4_25](https://huggingface.co/bartowski/OpenHercules-2.5-Mistral-7B-exl2/tree/4_25) | 4.25 | 6.0 | 5.3 GB | 6.7 GB | 8.7 GB | GPTQ equivalent bits per weight, slightly higher quality. |
31
+ | [3_5](https://huggingface.co/bartowski/OpenHercules-2.5-Mistral-7B-exl2/tree/3_5) | 3.5 | 6.0 | 4.7 GB | 6.1 GB | 8.1 GB | Lower quality, only use if you have to. |
32
+
33
+ ## Download instructions
34
+
35
+ With git:
36
+
37
+ ```shell
38
+ git clone --single-branch --branch 6_5 https://huggingface.co/bartowski/OpenHercules-2.5-Mistral-7B-exl2 OpenHercules-2.5-Mistral-7B-exl2-6_5
39
+ ```
40
+
41
+ With huggingface hub (credit to TheBloke for instructions):
42
+
43
+ ```shell
44
+ pip3 install huggingface-hub
45
+ ```
46
+
47
+ To download the `main` (only useful if you only care about measurement.json) branch to a folder called `OpenHercules-2.5-Mistral-7B-exl2`:
48
+
49
+ ```shell
50
+ mkdir OpenHercules-2.5-Mistral-7B-exl2
51
+ huggingface-cli download bartowski/OpenHercules-2.5-Mistral-7B-exl2 --local-dir OpenHercules-2.5-Mistral-7B-exl2 --local-dir-use-symlinks False
52
+ ```
53
+
54
+ To download from a different branch, add the `--revision` parameter:
55
+
56
+ Linux:
57
+
58
+ ```shell
59
+ mkdir OpenHercules-2.5-Mistral-7B-exl2-6_5
60
+ huggingface-cli download bartowski/OpenHercules-2.5-Mistral-7B-exl2 --revision 6_5 --local-dir OpenHercules-2.5-Mistral-7B-exl2-6_5 --local-dir-use-symlinks False
61
+ ```
62
+
63
+ Windows (which apparently doesn't like _ in folders sometimes?):
64
+
65
+ ```shell
66
+ mkdir OpenHercules-2.5-Mistral-7B-exl2-6.5
67
+ huggingface-cli download bartowski/OpenHercules-2.5-Mistral-7B-exl2 --revision 6_5 --local-dir OpenHercules-2.5-Mistral-7B-exl2-6.5 --local-dir-use-symlinks False
68
+ ```
69
+
70
+ Want to support my work? Visit my ko-fi page here: https://ko-fi.com/bartowski
measurement.json ADDED
The diff for this file is too large to render. See raw diff