mradermacher commited on
Commit
6e86cb0
1 Parent(s): 600fb2a

auto-patch README.md

Browse files
Files changed (1) hide show
  1. README.md +64 -0
README.md CHANGED
@@ -1,3 +1,67 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  <!-- ### convert_type: -->
2
  <!-- ### vocab_type: -->
3
  static quants of https://huggingface.co/son-of-man/Twizzler-7B
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model:
3
+ - Severian/Nexus-IKM-Mistral-Instruct-v0.2-7B
4
+ - son-of-man/HoloViolet-7B-test3
5
+ - alpindale/Mistral-7B-v0.2-hf
6
+ - localfultonextractor/Erosumika-7B-v3
7
+ exported_from: son-of-man/Twizzler-7B
8
+ language:
9
+ - en
10
+ library_name: transformers
11
+ quantized_by: mradermacher
12
+ tags:
13
+ - mergekit
14
+ - merge
15
+ ---
16
+ ## About
17
+
18
  <!-- ### convert_type: -->
19
  <!-- ### vocab_type: -->
20
  static quants of https://huggingface.co/son-of-man/Twizzler-7B
21
+
22
+
23
+ <!-- provided-files -->
24
+ weighted/imatrix quants seem not to be available (by me) at this time. If they do not show up a week or so after the static ones, I have probably not planned for them. Feel free to request them by opening a Community Discussion.
25
+ ## Usage
26
+
27
+ If you are unsure how to use GGUF files, refer to one of [TheBloke's
28
+ READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
29
+ more details, including on how to concatenate multi-part files.
30
+
31
+ ## Provided Quants
32
+
33
+ (sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
34
+
35
+ | Link | Type | Size/GB | Notes |
36
+ |:-----|:-----|--------:|:------|
37
+ | [GGUF](https://huggingface.co/mradermacher/Twizzler-7B-GGUF/resolve/main/Twizzler-7B.Q2_K.gguf) | Q2_K | 3.0 | |
38
+ | [GGUF](https://huggingface.co/mradermacher/Twizzler-7B-GGUF/resolve/main/Twizzler-7B.IQ3_XS.gguf) | IQ3_XS | 3.3 | |
39
+ | [GGUF](https://huggingface.co/mradermacher/Twizzler-7B-GGUF/resolve/main/Twizzler-7B.Q3_K_S.gguf) | Q3_K_S | 3.4 | |
40
+ | [GGUF](https://huggingface.co/mradermacher/Twizzler-7B-GGUF/resolve/main/Twizzler-7B.IQ3_S.gguf) | IQ3_S | 3.4 | beats Q3_K* |
41
+ | [GGUF](https://huggingface.co/mradermacher/Twizzler-7B-GGUF/resolve/main/Twizzler-7B.IQ3_M.gguf) | IQ3_M | 3.5 | |
42
+ | [GGUF](https://huggingface.co/mradermacher/Twizzler-7B-GGUF/resolve/main/Twizzler-7B.Q3_K_M.gguf) | Q3_K_M | 3.8 | lower quality |
43
+ | [GGUF](https://huggingface.co/mradermacher/Twizzler-7B-GGUF/resolve/main/Twizzler-7B.Q3_K_L.gguf) | Q3_K_L | 4.1 | |
44
+ | [GGUF](https://huggingface.co/mradermacher/Twizzler-7B-GGUF/resolve/main/Twizzler-7B.IQ4_XS.gguf) | IQ4_XS | 4.2 | |
45
+ | [GGUF](https://huggingface.co/mradermacher/Twizzler-7B-GGUF/resolve/main/Twizzler-7B.Q4_K_S.gguf) | Q4_K_S | 4.4 | fast, recommended |
46
+ | [GGUF](https://huggingface.co/mradermacher/Twizzler-7B-GGUF/resolve/main/Twizzler-7B.Q4_K_M.gguf) | Q4_K_M | 4.6 | fast, recommended |
47
+ | [GGUF](https://huggingface.co/mradermacher/Twizzler-7B-GGUF/resolve/main/Twizzler-7B.Q5_K_S.gguf) | Q5_K_S | 5.3 | |
48
+ | [GGUF](https://huggingface.co/mradermacher/Twizzler-7B-GGUF/resolve/main/Twizzler-7B.Q5_K_M.gguf) | Q5_K_M | 5.4 | |
49
+ | [GGUF](https://huggingface.co/mradermacher/Twizzler-7B-GGUF/resolve/main/Twizzler-7B.Q6_K.gguf) | Q6_K | 6.2 | very good quality |
50
+ | [GGUF](https://huggingface.co/mradermacher/Twizzler-7B-GGUF/resolve/main/Twizzler-7B.Q8_0.gguf) | Q8_0 | 7.9 | fast, best quality |
51
+
52
+
53
+ Here is a handy graph by ikawrakow comparing some lower-quality quant
54
+ types (lower is better):
55
+
56
+ ![image.png](https://www.nethype.de/huggingface_embed/quantpplgraph.png)
57
+
58
+ And here are Artefact2's thoughts on the matter:
59
+ https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
60
+
61
+ ## Thanks
62
+
63
+ I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
64
+ me use its servers and providing upgrades to my workstation to enable
65
+ this work in my free time.
66
+
67
+ <!-- end -->