bartowski commited on
Commit
54c3f37
1 Parent(s): c4d2abf

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
+ license: apache-2.0
3
+ license_link: https://huggingface.co/Qwen/Qwen2.5-7B-Instruct/blob/main/LICENSE
4
+ language:
5
+ - en
6
+ pipeline_tag: text-generation
7
+ base_model: Qwen/Qwen2.5-7B
8
+ tags:
9
+ - chat
10
+ quantized_by: bartowski
11
+ ---
12
+
13
+ ## Exllama v2 Quantizations of Qwen2.5-7B-Instruct
14
+
15
+ Using <a href="https://github.com/turboderp/exllamav2/releases/tag/v0.2.2">turboderp's ExLlamaV2 v0.2.2</a> for quantization.
16
+
17
+ <b>The "main" branch only contains the measurement.json, download one of the other branches for the model (see below)</b>
18
+
19
+ Each branch contains an individual bits per weight, with the main one containing only the meaurement.json for further conversions.
20
+
21
+ Original model: https://huggingface.co/Qwen/Qwen2.5-7B-Instruct
22
+
23
+ ## Prompt format
24
+
25
+ ```
26
+ <|im_start|>system
27
+ {system_prompt}<|im_end|>
28
+ <|im_start|>user
29
+ {prompt}<|im_end|>
30
+ <|im_start|>assistant
31
+
32
+ ```
33
+
34
+ ## Available sizes
35
+
36
+
37
+ | Branch | Bits | lm_head bits | VRAM (4k) | VRAM (16k) | VRAM (32k) | Description |
38
+ | ----- | ---- | ------- | ------ | ------ | ------ | ------------ |
39
+ | [8_0](https://huggingface.co/bartowski/Qwen2.5-7B-Instruct-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. |
40
+ | [6_5](https://huggingface.co/bartowski/Qwen2.5-7B-Instruct-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**. |
41
+ | [5_0](https://huggingface.co/bartowski/Qwen2.5-7B-Instruct-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. |
42
+ | [4_25](https://huggingface.co/bartowski/Qwen2.5-7B-Instruct-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. |
43
+ | [3_5](https://huggingface.co/bartowski/Qwen2.5-7B-Instruct-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. |
44
+
45
+ ## Download instructions
46
+
47
+ With git:
48
+
49
+ ```shell
50
+ git clone --single-branch --branch 6_5 https://huggingface.co/bartowski/Qwen2.5-7B-Instruct-exl2 Qwen2.5-7B-Instruct-exl2-6_5
51
+ ```
52
+
53
+ With huggingface hub (credit to TheBloke for instructions):
54
+
55
+ ```shell
56
+ pip3 install huggingface-hub
57
+ ```
58
+
59
+ To download a specific branch, use the `--revision` parameter. For example, to download the 6.5 bpw branch:
60
+
61
+ Linux:
62
+
63
+ ```shell
64
+ huggingface-cli download bartowski/Qwen2.5-7B-Instruct-exl2 --revision 6_5 --local-dir Qwen2.5-7B-Instruct-exl2-6_5
65
+ ```
66
+
67
+ Windows (which apparently doesn't like _ in folders sometimes?):
68
+
69
+ ```shell
70
+ huggingface-cli download bartowski/Qwen2.5-7B-Instruct-exl2 --revision 6_5 --local-dir Qwen2.5-7B-Instruct-exl2-6.5
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