Hugging Face's logo

mradermacher
/
Umbra-v2.1-MoE-4x10.7-GGUF

Transformers
GGUF
English
Mixture of Experts
Merge
mergekit
Solar Moe
Solar
Umbra
Inference Endpoints
Model card Files Community
mradermacher commited on
Commit
7d36555
•
1 Parent(s): 3e70dbb

auto-patch README.md

Browse files
Files changed (1) hide show
  1. README.md +56 -1
README.md CHANGED
@@ -1,3 +1,58 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  static quants of https://huggingface.co/Steelskull/Umbra-v2.1-MoE-4x10.7
2
 
3
- weighted/imatrix quants available at https://huggingface.co/mradermacher/Umbra-v2.1-MoE-4x10.7-i1-GGUF
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - en
4
+ library_name: transformers
5
+ license: apache-2.0
6
+ quantized_by: mradermacher
7
+ tags:
8
+ - moe
9
+ - merge
10
+ - mergekit
11
+ - Solar Moe
12
+ - Solar
13
+ - Umbra
14
+ ---
15
+ ## About
16
+
17
  static quants of https://huggingface.co/Steelskull/Umbra-v2.1-MoE-4x10.7
18
 
19
+ weighted/imatrix quants available at https://huggingface.co/mradermacher/Umbra-v2.1-MoE-4x10.7-i1-GGUF
20
+ <!-- provided-files -->
21
+
22
+ ## Usage
23
+
24
+ If you are unsure how to use GGUF files, refer to one of [TheBloke's
25
+ READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
26
+ more details, including on how to concatenate multi-part files.
27
+
28
+ ## Provided Quants
29
+
30
+ (sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
31
+
32
+ | Link | Type | Size/GB | Notes |
33
+ |:-----|:-----|--------:|:------|
34
+ | [GGUF](https://huggingface.co/mradermacher/Umbra-v2.1-MoE-4x10.7-GGUF/resolve/main/Umbra-v2.1-MoE-4x10.7.Q2_K.gguf) | Q2_K | 13.4 | |
35
+ | [GGUF](https://huggingface.co/mradermacher/Umbra-v2.1-MoE-4x10.7-GGUF/resolve/main/Umbra-v2.1-MoE-4x10.7.IQ3_XS.gguf) | IQ3_XS | 14.9 | |
36
+ | [GGUF](https://huggingface.co/mradermacher/Umbra-v2.1-MoE-4x10.7-GGUF/resolve/main/Umbra-v2.1-MoE-4x10.7.Q3_K_S.gguf) | Q3_K_S | 15.8 | |
37
+ | [GGUF](https://huggingface.co/mradermacher/Umbra-v2.1-MoE-4x10.7-GGUF/resolve/main/Umbra-v2.1-MoE-4x10.7.IQ3_S.gguf) | IQ3_S | 15.8 | fast, beats Q3_K* |
38
+ | [GGUF](https://huggingface.co/mradermacher/Umbra-v2.1-MoE-4x10.7-GGUF/resolve/main/Umbra-v2.1-MoE-4x10.7.IQ3_M.gguf) | IQ3_M | 16.1 | |
39
+ | [GGUF](https://huggingface.co/mradermacher/Umbra-v2.1-MoE-4x10.7-GGUF/resolve/main/Umbra-v2.1-MoE-4x10.7.Q3_K_M.gguf) | Q3_K_M | 17.5 | lower quality |
40
+ | [GGUF](https://huggingface.co/mradermacher/Umbra-v2.1-MoE-4x10.7-GGUF/resolve/main/Umbra-v2.1-MoE-4x10.7.Q3_K_L.gguf) | Q3_K_L | 19.0 | |
41
+ | [GGUF](https://huggingface.co/mradermacher/Umbra-v2.1-MoE-4x10.7-GGUF/resolve/main/Umbra-v2.1-MoE-4x10.7.IQ4_XS.gguf) | IQ4_XS | 19.7 | |
42
+ | [GGUF](https://huggingface.co/mradermacher/Umbra-v2.1-MoE-4x10.7-GGUF/resolve/main/Umbra-v2.1-MoE-4x10.7.Q4_K_S.gguf) | Q4_K_S | 20.8 | fast, medium quality |
43
+ | [GGUF](https://huggingface.co/mradermacher/Umbra-v2.1-MoE-4x10.7-GGUF/resolve/main/Umbra-v2.1-MoE-4x10.7.Q4_K_M.gguf) | Q4_K_M | 22.1 | fast, medium quality |
44
+ | [GGUF](https://huggingface.co/mradermacher/Umbra-v2.1-MoE-4x10.7-GGUF/resolve/main/Umbra-v2.1-MoE-4x10.7.Q5_K_S.gguf) | Q5_K_S | 25.1 | |
45
+ | [GGUF](https://huggingface.co/mradermacher/Umbra-v2.1-MoE-4x10.7-GGUF/resolve/main/Umbra-v2.1-MoE-4x10.7.Q5_K_M.gguf) | Q5_K_M | 25.9 | |
46
+ | [GGUF](https://huggingface.co/mradermacher/Umbra-v2.1-MoE-4x10.7-GGUF/resolve/main/Umbra-v2.1-MoE-4x10.7.Q6_K.gguf) | Q6_K | 29.9 | very good quality |
47
+ | [GGUF](https://huggingface.co/mradermacher/Umbra-v2.1-MoE-4x10.7-GGUF/resolve/main/Umbra-v2.1-MoE-4x10.7.Q8_0.gguf) | Q8_0 | 38.6 | fast, best quality |
48
+
49
+
50
+ Here is a handy graph by ikawrakow comparing some lower-quality quant
51
+ types (lower is better):
52
+
53
+ ![image.png](https://www.nethype.de/huggingface_embed/quantpplgraph.png)
54
+
55
+ And here are Artefact2's thoughts on the matter:
56
+ https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
57
+
58
+ <!-- end -->