bartowski commited on
Commit
3453e7a
1 Parent(s): 625c7e4

measurement.json

Browse files
Files changed (2) hide show
  1. README.md +63 -0
  2. measurement.json +0 -0
README.md ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ library_name: transformers
4
+ pipeline_tag: text-generation
5
+ quantized_by: bartowski
6
+ ---
7
+
8
+ ## Exllama v2 Quantizations of Starling-LM-10.7B-beta
9
+
10
+ Using <a href="https://github.com/turboderp/exllamav2/releases/tag/v0.0.17">turboderp's ExLlamaV2 v0.0.17</a> for quantization.
11
+
12
+ <b>The "main" branch only contains the measurement.json, download one of the other branches for the model (see below)</b>
13
+
14
+ Each branch contains an individual bits per weight, with the main one containing only the meaurement.json for further conversions.
15
+
16
+ Original model: https://huggingface.co/ddh0/Starling-LM-10.7B-beta
17
+
18
+ | Branch | Bits | lm_head bits | VRAM (4k) | VRAM (16k) | VRAM (32k) | Description |
19
+ | ----- | ---- | ------- | ------ | ------ | ------ | ------------ |
20
+ | [8_0](https://huggingface.co/bartowski/Starling-LM-10.7B-beta-exl2/tree/8_0) | 8.0 | 8.0 | 11.9 GB | 13.3 GB | 15.3 GB | Maximum quality that ExLlamaV2 can produce, near unquantized performance. |
21
+ | [6_5](https://huggingface.co/bartowski/Starling-LM-10.7B-beta-exl2/tree/6_5) | 6.5 | 8.0 | 10.3 GB | 11.7 GB | 13.7 GB | Very similar to 8.0, good tradeoff of size vs performance, **recommended**. |
22
+ | [5_0](https://huggingface.co/bartowski/Starling-LM-10.7B-beta-exl2/tree/5_0) | 5.0 | 6.0 | 8.3 GB | 9.7 GB | 11.7 GB | Slightly lower quality vs 6.5, but usable on 8GB cards. |
23
+ | [4_25](https://huggingface.co/bartowski/Starling-LM-10.7B-beta-exl2/tree/4_25) | 4.25 | 6.0 | 7.4 GB | 8.6 GB | 10.6 GB | GPTQ equivalent bits per weight, slightly higher quality. |
24
+ | [3_5](https://huggingface.co/bartowski/Starling-LM-10.7B-beta-exl2/tree/3_5) | 3.5 | 6.0 | 6.4 GB | 7.8 GB | 9.8 GB | Lower quality, only use if you have to. |
25
+
26
+ ## Download instructions
27
+
28
+ With git:
29
+
30
+ ```shell
31
+ git clone --single-branch --branch 6_5 https://huggingface.co/bartowski/Starling-LM-10.7B-beta-exl2 Starling-LM-10.7B-beta-exl2-6_5
32
+ ```
33
+
34
+ With huggingface hub (credit to TheBloke for instructions):
35
+
36
+ ```shell
37
+ pip3 install huggingface-hub
38
+ ```
39
+
40
+ To download the `main` (only useful if you only care about measurement.json) branch to a folder called `Starling-LM-10.7B-beta-exl2`:
41
+
42
+ ```shell
43
+ mkdir Starling-LM-10.7B-beta-exl2
44
+ huggingface-cli download bartowski/Starling-LM-10.7B-beta-exl2 --local-dir Starling-LM-10.7B-beta-exl2 --local-dir-use-symlinks False
45
+ ```
46
+
47
+ To download from a different branch, add the `--revision` parameter:
48
+
49
+ Linux:
50
+
51
+ ```shell
52
+ mkdir Starling-LM-10.7B-beta-exl2-6_5
53
+ huggingface-cli download bartowski/Starling-LM-10.7B-beta-exl2 --revision 6_5 --local-dir Starling-LM-10.7B-beta-exl2-6_5 --local-dir-use-symlinks False
54
+ ```
55
+
56
+ Windows (which apparently doesn't like _ in folders sometimes?):
57
+
58
+ ```shell
59
+ mkdir Starling-LM-10.7B-beta-exl2-6.5
60
+ huggingface-cli download bartowski/Starling-LM-10.7B-beta-exl2 --revision 6_5 --local-dir Starling-LM-10.7B-beta-exl2-6.5 --local-dir-use-symlinks False
61
+ ```
62
+
63
+ 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