mradermacher commited on
Commit
19d72e2
1 Parent(s): a56e810

auto-patch README.md

Browse files
Files changed (1) hide show
  1. README.md +68 -0
README.md CHANGED
@@ -1 +1,69 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  weighted/imatrix quants of https://huggingface.co/CultriX/MoNeuTrix-MoE-4x7B
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model:
3
+ - CultriX/MonaTrix-v4
4
+ - mlabonne/OmniTruthyBeagle-7B-v0
5
+ - CultriX/MoNeuTrix-7B-v1
6
+ - paulml/OmniBeagleSquaredMBX-v3-7B
7
+ exported_from: CultriX/MoNeuTrix-MoE-4x7B
8
+ language:
9
+ - en
10
+ library_name: transformers
11
+ license: apache-2.0
12
+ quantized_by: mradermacher
13
+ tags:
14
+ - moe
15
+ - frankenmoe
16
+ - merge
17
+ - mergekit
18
+ - lazymergekit
19
+ - CultriX/MonaTrix-v4
20
+ - mlabonne/OmniTruthyBeagle-7B-v0
21
+ - CultriX/MoNeuTrix-7B-v1
22
+ - paulml/OmniBeagleSquaredMBX-v3-7B
23
+ ---
24
+ ## About
25
+
26
  weighted/imatrix quants of https://huggingface.co/CultriX/MoNeuTrix-MoE-4x7B
27
+
28
+
29
+ <!-- provided-files -->
30
+ static quants are available at https://huggingface.co/mradermacher/MoNeuTrix-MoE-4x7B-GGUF
31
+ ## Usage
32
+
33
+ If you are unsure how to use GGUF files, refer to one of [TheBloke's
34
+ READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
35
+ more details, including on how to concatenate multi-part files.
36
+
37
+ ## Provided Quants
38
+
39
+ (sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
40
+
41
+ | Link | Type | Size/GB | Notes |
42
+ |:-----|:-----|--------:|:------|
43
+ | [GGUF](https://huggingface.co/mradermacher/MoNeuTrix-MoE-4x7B-i1-GGUF/resolve/main/MoNeuTrix-MoE-4x7B.i1-Q2_K.gguf) | i1-Q2_K | 9.1 | IQ3_XXS probably better |
44
+ | [GGUF](https://huggingface.co/mradermacher/MoNeuTrix-MoE-4x7B-i1-GGUF/resolve/main/MoNeuTrix-MoE-4x7B.i1-Q3_K_S.gguf) | i1-Q3_K_S | 10.7 | IQ3_XS probably better |
45
+ | [GGUF](https://huggingface.co/mradermacher/MoNeuTrix-MoE-4x7B-i1-GGUF/resolve/main/MoNeuTrix-MoE-4x7B.i1-Q3_K_M.gguf) | i1-Q3_K_M | 11.8 | IQ3_S probably better |
46
+ | [GGUF](https://huggingface.co/mradermacher/MoNeuTrix-MoE-4x7B-i1-GGUF/resolve/main/MoNeuTrix-MoE-4x7B.i1-Q3_K_L.gguf) | i1-Q3_K_L | 12.8 | IQ3_M probably better |
47
+ | [GGUF](https://huggingface.co/mradermacher/MoNeuTrix-MoE-4x7B-i1-GGUF/resolve/main/MoNeuTrix-MoE-4x7B.i1-Q4_0.gguf) | i1-Q4_0 | 13.9 | |
48
+ | [GGUF](https://huggingface.co/mradermacher/MoNeuTrix-MoE-4x7B-i1-GGUF/resolve/main/MoNeuTrix-MoE-4x7B.i1-Q4_K_S.gguf) | i1-Q4_K_S | 14.0 | optimal size/speed/quality |
49
+ | [GGUF](https://huggingface.co/mradermacher/MoNeuTrix-MoE-4x7B-i1-GGUF/resolve/main/MoNeuTrix-MoE-4x7B.i1-Q4_K_M.gguf) | i1-Q4_K_M | 14.9 | fast, recommended |
50
+ | [GGUF](https://huggingface.co/mradermacher/MoNeuTrix-MoE-4x7B-i1-GGUF/resolve/main/MoNeuTrix-MoE-4x7B.i1-Q5_K_S.gguf) | i1-Q5_K_S | 16.9 | |
51
+ | [GGUF](https://huggingface.co/mradermacher/MoNeuTrix-MoE-4x7B-i1-GGUF/resolve/main/MoNeuTrix-MoE-4x7B.i1-Q5_K_M.gguf) | i1-Q5_K_M | 17.4 | |
52
+ | [GGUF](https://huggingface.co/mradermacher/MoNeuTrix-MoE-4x7B-i1-GGUF/resolve/main/MoNeuTrix-MoE-4x7B.i1-Q6_K.gguf) | i1-Q6_K | 20.1 | practically like static Q6_K |
53
+
54
+
55
+ Here is a handy graph by ikawrakow comparing some lower-quality quant
56
+ types (lower is better):
57
+
58
+ ![image.png](https://www.nethype.de/huggingface_embed/quantpplgraph.png)
59
+
60
+ And here are Artefact2's thoughts on the matter:
61
+ https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
62
+
63
+ ## Thanks
64
+
65
+ I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
66
+ me use its servers and providing upgrades to my workstation to enable
67
+ this work in my free time.
68
+
69
+ <!-- end -->