bartowski commited on
Commit
50b8617
1 Parent(s): 871f2e4

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
+ license: apache-2.0
3
+ datasets:
4
+ - Open-Orca/SlimOrca
5
+ - ajibawa-2023/SlimOrca-ShareGPT
6
+ language:
7
+ - en
8
+ tags:
9
+ - Q&A
10
+ - mcq
11
+ - Summary Generation
12
+ - Reviews
13
+ - Hypothesis Generation
14
+ - Story
15
+ - Articles
16
+ quantized_by: bartowski
17
+ pipeline_tag: text-generation
18
+ ---
19
+
20
+ ## Exllama v2 Quantizations of SlimOrca-Llama-3-8B
21
+
22
+ Using <a href="https://github.com/turboderp/exllamav2/releases/tag/v0.1.0">turboderp's ExLlamaV2 v0.1.0</a> for quantization.
23
+
24
+ <b>The "main" branch only contains the measurement.json, download one of the other branches for the model (see below)</b>
25
+
26
+ Each branch contains an individual bits per weight, with the main one containing only the meaurement.json for further conversions.
27
+
28
+ Original model: https://huggingface.co/ajibawa-2023/SlimOrca-Llama-3-8B
29
+
30
+ ## Prompt format
31
+
32
+ ```
33
+ <|im_start|>system
34
+ {system_prompt}<|im_end|>
35
+ <|im_start|>user
36
+ {prompt}<|im_end|>
37
+ <|im_start|>assistant
38
+
39
+ ```
40
+
41
+ ## Available sizes
42
+
43
+
44
+ | Branch | Bits | lm_head bits | VRAM (4k) | VRAM (8K) | VRAM (16k) | VRAM (32k) | Description |
45
+ | ----- | ---- | ------- | ------ | ------ | ------ | ------ | ------------ |
46
+ | [8_0](https://huggingface.co/bartowski/SlimOrca-Llama-3-8B-exl2/tree/8_0) | 8.0 | 8.0 | 10.1 GB | 10.5 GB | 11.5 GB | 13.6 GB | Maximum quality that ExLlamaV2 can produce, near unquantized performance. |
47
+ | [6_5](https://huggingface.co/bartowski/SlimOrca-Llama-3-8B-exl2/tree/6_5) | 6.5 | 8.0 | 8.9 GB | 9.3 GB | 10.3 GB | 12.4 GB | Very similar to 8.0, good tradeoff of size vs performance, **recommended**. |
48
+ | [5_0](https://huggingface.co/bartowski/SlimOrca-Llama-3-8B-exl2/tree/5_0) | 5.0 | 6.0 | 7.7 GB | 8.1 GB | 9.1 GB | 11.2 GB | Slightly lower quality vs 6.5, but usable on 8GB cards. |
49
+ | [4_25](https://huggingface.co/bartowski/SlimOrca-Llama-3-8B-exl2/tree/4_25) | 4.25 | 6.0 | 7.0 GB | 7.4 GB | 8.4 GB | 10.5 GB | GPTQ equivalent bits per weight, slightly higher quality. |
50
+ | [3_5](https://huggingface.co/bartowski/SlimOrca-Llama-3-8B-exl2/tree/3_5) | 3.5 | 6.0 | 6.4 GB | 6.8 GB | 7.8 GB | 9.9 GB | Lower quality, only use if you have to. |
51
+
52
+ ## Download instructions
53
+
54
+ With git:
55
+
56
+ ```shell
57
+ git clone --single-branch --branch 6_5 https://huggingface.co/bartowski/SlimOrca-Llama-3-8B-exl2 SlimOrca-Llama-3-8B-exl2-6_5
58
+ ```
59
+
60
+ With huggingface hub (credit to TheBloke for instructions):
61
+
62
+ ```shell
63
+ pip3 install huggingface-hub
64
+ ```
65
+
66
+ To download a specific branch, use the `--revision` parameter. For example, to download the 6.5 bpw branch:
67
+
68
+ Linux:
69
+
70
+ ```shell
71
+ huggingface-cli download bartowski/SlimOrca-Llama-3-8B-exl2 --revision 6_5 --local-dir SlimOrca-Llama-3-8B-exl2-6_5
72
+ ```
73
+
74
+ Windows (which apparently doesn't like _ in folders sometimes?):
75
+
76
+ ```shell
77
+ huggingface-cli download bartowski/SlimOrca-Llama-3-8B-exl2 --revision 6_5 --local-dir SlimOrca-Llama-3-8B-exl2-6.5
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