bartowski commited on
Commit
3980931
·
verified ·
1 Parent(s): 4a0d8de

measurement.json

Browse files
Files changed (2) hide show
  1. README.md +74 -0
  2. measurement.json +0 -0
README.md ADDED
@@ -0,0 +1,74 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model:
3
+ - Qwen/Qwen2.5-Coder-32B-Instruct
4
+ - huihui-ai/Qwen2.5-Coder-32B-Instruct-abliterated
5
+ library_name: transformers
6
+ tags:
7
+ - code
8
+ license: mit
9
+ quantized_by: bartowski
10
+ pipeline_tag: text-generation
11
+ ---
12
+
13
+ ## Exllama v2 Quantizations of tqwendo-36b
14
+
15
+ Using <a href="https://github.com/turboderp/exllamav2/releases/tag/v0.2.4">turboderp's ExLlamaV2 v0.2.4</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
+ Conversion was done using the default calibration dataset.
22
+
23
+ Default arguments used except when the bits per weight is above 6.0, at that point the lm_head layer is quantized at 8 bits per weight instead of the default 6.
24
+
25
+ Original model: https://huggingface.co/nisten/tqwendo-36b
26
+
27
+
28
+ <a href="https://huggingface.co/bartowski/tqwendo-36b-exl2/tree/8_0">8.0 bits per weight</a>
29
+
30
+ <a href="https://huggingface.co/bartowski/tqwendo-36b-exl2/tree/6_5">6.5 bits per weight</a>
31
+
32
+ <a href="https://huggingface.co/bartowski/tqwendo-36b-exl2/tree/5_5">5.5 bits per weight</a>
33
+
34
+ <a href="https://huggingface.co/bartowski/tqwendo-36b-exl2/tree/4_25">4.25 bits per weight</a>
35
+
36
+ <a href="https://huggingface.co/bartowski/tqwendo-36b-exl2/tree/3_5">3.5 bits per weight</a>
37
+
38
+
39
+ ## Download instructions
40
+
41
+ With git:
42
+
43
+ ```shell
44
+ git clone --single-branch --branch 6_5 https://huggingface.co/bartowski/tqwendo-36b-exl2
45
+ ```
46
+
47
+ With huggingface hub (credit to TheBloke for instructions):
48
+
49
+ ```shell
50
+ pip3 install huggingface-hub
51
+ ```
52
+
53
+ To download the `main` (only useful if you only care about measurement.json) branch to a folder called `tqwendo-36b-exl2`:
54
+
55
+ ```shell
56
+ mkdir tqwendo-36b-exl2
57
+ huggingface-cli download bartowski/tqwendo-36b-exl2 --local-dir tqwendo-36b-exl2
58
+ ```
59
+
60
+ To download from a different branch, add the `--revision` parameter:
61
+
62
+ Linux:
63
+
64
+ ```shell
65
+ mkdir tqwendo-36b-exl2-6_5
66
+ huggingface-cli download bartowski/tqwendo-36b-exl2 --revision 6_5 --local-dir tqwendo-36b-exl2-6_5
67
+ ```
68
+
69
+ Windows (which apparently doesn't like _ in folders sometimes?):
70
+
71
+ ```shell
72
+ mkdir tqwendo-36b-exl2-6.5
73
+ huggingface-cli download bartowski/tqwendo-36b-exl2 --revision 6_5 --local-dir tqwendo-36b-exl2-6.5
74
+ ```
measurement.json ADDED
The diff for this file is too large to render. See raw diff