bartowski commited on
Commit
8676a94
1 Parent(s): 0799eb4

measurement.json

Browse files
Files changed (2) hide show
  1. README.md +80 -0
  2. measurement.json +0 -0
README.md ADDED
@@ -0,0 +1,80 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - zh
4
+ - en
5
+ - fr
6
+ - de
7
+ - ja
8
+ - ko
9
+ - it
10
+ - ru
11
+ - fi
12
+
13
+ pipeline_tag: text-generation
14
+ inference: false
15
+ library_name: transformers
16
+ license: other
17
+ license_name: gemma
18
+ license_link: https://ai.google.dev/gemma/terms
19
+ base_model: google/gemma-7b
20
+ quantized_by: bartowski
21
+ ---
22
+
23
+ ## Exllama v2 Quantizations of openbuddy-gemma-7b-v19.1-4k
24
+
25
+ Using <a href="https://github.com/turboderp/exllamav2/releases/tag/v0.0.14">turboderp's ExLlamaV2 v0.0.14</a> for quantization.
26
+
27
+ <b>The "main" branch only contains the measurement.json, download one of the other branches for the model (see below)</b>
28
+
29
+ Each branch contains an individual bits per weight, with the main one containing only the meaurement.json for further conversions.
30
+
31
+ Original model: https://huggingface.co/OpenBuddy/openbuddy-gemma-7b-v19.1-4k
32
+
33
+ No GQA - VRAM requirements will be higher
34
+
35
+ | Branch | Bits | lm_head bits | Size (4k) | Size (16k) | Description |
36
+ | -------------------------------------------------------------- | ---- | ------------ | --------- | ---------- | ----------- |
37
+ | [8_0](https://huggingface.co/bartowski/openbuddy-gemma-7b-v19.1-4k-exl2/tree/8_0) | 8.0 | 8.0 | 14.0 GB | 19.4 GB | Maximum quality that ExLlamaV2 can produce, near unquantized performance. |
38
+ | [6_5](https://huggingface.co/bartowski/openbuddy-gemma-7b-v19.1-4k-exl2/tree/6_5) | 6.5 | 8.0 | 12.5 GB | 17.9 GB | Near unquantized performance at vastly reduced size, **recommended**. |
39
+ | [5_0](https://huggingface.co/bartowski/openbuddy-gemma-7b-v19.1-4k-exl2/tree/5_0) | 5.0 | 6.0 | 10.9 GB | 16.3 GB | Slightly lower quality vs 6.5, but usable on 8GB cards with 4k context. |
40
+ | [4_25](https://huggingface.co/bartowski/openbuddy-gemma-7b-v19.1-4k-exl2/tree/4_25) | 4.25 | 6.0 | 10.2 GB | 15.7 GB | GPTQ equivalent bits per weight. |
41
+ | [3_5](https://huggingface.co/bartowski/openbuddy-gemma-7b-v19.1-4k-exl2/tree/3_5) | 3.5 | 6.0 | 9.5 GB | 14.9 GB | Lower quality, not recommended. |
42
+
43
+ ## Download instructions
44
+
45
+ With git:
46
+
47
+ ```shell
48
+ git clone --single-branch --branch 6_5 https://huggingface.co/bartowski/openbuddy-gemma-7b-v19.1-4k-exl2 openbuddy-gemma-7b-v19.1-4k-exl2-6_5
49
+ ```
50
+
51
+ With huggingface hub (credit to TheBloke for instructions):
52
+
53
+ ```shell
54
+ pip3 install huggingface-hub
55
+ ```
56
+
57
+ To download the `main` (only useful if you only care about measurement.json) branch to a folder called `openbuddy-gemma-7b-v19.1-4k-exl2`:
58
+
59
+ ```shell
60
+ mkdir openbuddy-gemma-7b-v19.1-4k-exl2
61
+ huggingface-cli download bartowski/openbuddy-gemma-7b-v19.1-4k-exl2 --local-dir openbuddy-gemma-7b-v19.1-4k-exl2 --local-dir-use-symlinks False
62
+ ```
63
+
64
+ To download from a different branch, add the `--revision` parameter:
65
+
66
+ Linux:
67
+
68
+ ```shell
69
+ mkdir openbuddy-gemma-7b-v19.1-4k-exl2-6_5
70
+ huggingface-cli download bartowski/openbuddy-gemma-7b-v19.1-4k-exl2 --revision 6_5 --local-dir openbuddy-gemma-7b-v19.1-4k-exl2-6_5 --local-dir-use-symlinks False
71
+ ```
72
+
73
+ Windows (which apparently doesn't like _ in folders sometimes?):
74
+
75
+ ```shell
76
+ mkdir openbuddy-gemma-7b-v19.1-4k-exl2-6.5
77
+ huggingface-cli download bartowski/openbuddy-gemma-7b-v19.1-4k-exl2 --revision 6_5 --local-dir openbuddy-gemma-7b-v19.1-4k-exl2-6.5 --local-dir-use-symlinks False
78
+ ```
79
+
80
+ 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