bartowski commited on
Commit
1e54feb
1 Parent(s): 9ffd4b4

measurement.json

Browse files
Files changed (2) hide show
  1. README.md +73 -0
  2. measurement.json +0 -0
README.md ADDED
@@ -0,0 +1,73 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - en
4
+ license: other
5
+ datasets:
6
+ - teknium/OpenHermes-2.5
7
+ - m-a-p/Code-Feedback
8
+ - m-a-p/CodeFeedback-Filtered-Instruction
9
+ - abacusai/SystemChat
10
+ license_name: tongyi-qianwen
11
+ license_link: https://huggingface.co/Qwen/Qwen1.5-72B/blob/main/LICENSE
12
+ quantized_by: bartowski
13
+ pipeline_tag: text-generation
14
+ ---
15
+
16
+ ## Exllama v2 Quantizations of Liberated-Qwen1.5-7B
17
+
18
+ Using <a href="https://github.com/turboderp/exllamav2/releases/tag/v0.0.15">turboderp's ExLlamaV2 v0.0.15</a> for quantization.
19
+
20
+ <b>The "main" branch only contains the measurement.json, download one of the other branches for the model (see below)</b>
21
+
22
+ Each branch contains an individual bits per weight, with the main one containing only the meaurement.json for further conversions.
23
+
24
+ Original model: https://huggingface.co/abacusai/Liberated-Qwen1.5-7B
25
+
26
+ No GQA - VRAM requirements will be higher
27
+
28
+ | Branch | Bits | lm_head bits | Size (4k) | Size (16k) | Description |
29
+ | -------------------------------------------------------------- | ---- | ------------ | --------- | ---------- | ----------- |
30
+ | [8_0](https://huggingface.co/bartowski/Liberated-Qwen1.5-7B-exl2/tree/8_0) | 8.0 | 8.0 | 9.4 GB | 15.6 GB | Maximum quality that ExLlamaV2 can produce, near unquantized performance. |
31
+ | [6_5](https://huggingface.co/bartowski/Liberated-Qwen1.5-7B-exl2/tree/6_5) | 6.5 | 8.0 | 8.6 GB | 14.8 GB | Near unquantized performance at vastly reduced size, **recommended**. |
32
+ | [5_0](https://huggingface.co/bartowski/Liberated-Qwen1.5-7B-exl2/tree/5_0) | 5.0 | 6.0 | 7.2 GB | 13.4 GB | Slightly lower quality vs 6.5, but usable on 8GB cards with 4k context. |
33
+ | [4_25](https://huggingface.co/bartowski/Liberated-Qwen1.5-7B-exl2/tree/4_25) | 4.25 | 6.0 | 6.5 GB | 12.7 GB | GPTQ equivalent bits per weight. |
34
+ | [3_5](https://huggingface.co/bartowski/Liberated-Qwen1.5-7B-exl2/tree/3_5) | 3.5 | 6.0 | 5.9 GB | 12.1 GB | Lower quality, not recommended. |
35
+
36
+ ## Download instructions
37
+
38
+ With git:
39
+
40
+ ```shell
41
+ git clone --single-branch --branch 6_5 https://huggingface.co/bartowski/Liberated-Qwen1.5-7B-exl2 Liberated-Qwen1.5-7B-exl2-6_5
42
+ ```
43
+
44
+ With huggingface hub (credit to TheBloke for instructions):
45
+
46
+ ```shell
47
+ pip3 install huggingface-hub
48
+ ```
49
+
50
+ To download the `main` (only useful if you only care about measurement.json) branch to a folder called `Liberated-Qwen1.5-7B-exl2`:
51
+
52
+ ```shell
53
+ mkdir Liberated-Qwen1.5-7B-exl2
54
+ huggingface-cli download bartowski/Liberated-Qwen1.5-7B-exl2 --local-dir Liberated-Qwen1.5-7B-exl2 --local-dir-use-symlinks False
55
+ ```
56
+
57
+ To download from a different branch, add the `--revision` parameter:
58
+
59
+ Linux:
60
+
61
+ ```shell
62
+ mkdir Liberated-Qwen1.5-7B-exl2-6_5
63
+ huggingface-cli download bartowski/Liberated-Qwen1.5-7B-exl2 --revision 6_5 --local-dir Liberated-Qwen1.5-7B-exl2-6_5 --local-dir-use-symlinks False
64
+ ```
65
+
66
+ Windows (which apparently doesn't like _ in folders sometimes?):
67
+
68
+ ```shell
69
+ mkdir Liberated-Qwen1.5-7B-exl2-6.5
70
+ huggingface-cli download bartowski/Liberated-Qwen1.5-7B-exl2 --revision 6_5 --local-dir Liberated-Qwen1.5-7B-exl2-6.5 --local-dir-use-symlinks False
71
+ ```
72
+
73
+ 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