mradermacher commited on
Commit
ba5c527
1 Parent(s): b50df72

auto-patch README.md

Browse files
Files changed (1) hide show
  1. README.md +58 -0
README.md CHANGED
@@ -1,6 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  <!-- ### quantize_version: 2 -->
2
  <!-- ### output_tensor_quantised: 1 -->
3
  <!-- ### convert_type: hf -->
4
  <!-- ### vocab_type: -->
5
  <!-- ### tags: nicoboss -->
6
  weighted/imatrix quants of https://huggingface.co/llmixer/BigWeave-v25-95b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: llmixer/BigWeave-v25-95b
3
+ language:
4
+ - en
5
+ library_name: transformers
6
+ license: unknown
7
+ quantized_by: mradermacher
8
+ tags:
9
+ - merge
10
+ - frankenmerge
11
+ - 95b
12
+ ---
13
+ ## About
14
+
15
  <!-- ### quantize_version: 2 -->
16
  <!-- ### output_tensor_quantised: 1 -->
17
  <!-- ### convert_type: hf -->
18
  <!-- ### vocab_type: -->
19
  <!-- ### tags: nicoboss -->
20
  weighted/imatrix quants of https://huggingface.co/llmixer/BigWeave-v25-95b
21
+
22
+ <!-- provided-files -->
23
+ static quants are available at https://huggingface.co/mradermacher/BigWeave-v25-95b-GGUF
24
+ ## Usage
25
+
26
+ If you are unsure how to use GGUF files, refer to one of [TheBloke's
27
+ READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
28
+ more details, including on how to concatenate multi-part files.
29
+
30
+ ## Provided Quants
31
+
32
+ (sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
33
+
34
+ | Link | Type | Size/GB | Notes |
35
+ |:-----|:-----|--------:|:------|
36
+ | [GGUF](https://huggingface.co/mradermacher/BigWeave-v25-95b-i1-GGUF/resolve/main/BigWeave-v25-95b.i1-IQ2_M.gguf) | i1-IQ2_M | 31.9 | |
37
+ | [GGUF](https://huggingface.co/mradermacher/BigWeave-v25-95b-i1-GGUF/resolve/main/BigWeave-v25-95b.i1-Q2_K.gguf) | i1-Q2_K | 34.9 | IQ3_XXS probably better |
38
+ | [GGUF](https://huggingface.co/mradermacher/BigWeave-v25-95b-i1-GGUF/resolve/main/BigWeave-v25-95b.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 36.4 | lower quality |
39
+ | [GGUF](https://huggingface.co/mradermacher/BigWeave-v25-95b-i1-GGUF/resolve/main/BigWeave-v25-95b.i1-IQ3_M.gguf) | i1-IQ3_M | 42.4 | |
40
+ | [GGUF](https://huggingface.co/mradermacher/BigWeave-v25-95b-i1-GGUF/resolve/main/BigWeave-v25-95b.i1-Q3_K_M.gguf) | i1-Q3_K_M | 45.6 | IQ3_S probably better |
41
+ | [PART 1](https://huggingface.co/mradermacher/BigWeave-v25-95b-i1-GGUF/resolve/main/BigWeave-v25-95b.i1-Q4_K_S.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/BigWeave-v25-95b-i1-GGUF/resolve/main/BigWeave-v25-95b.i1-Q4_K_S.gguf.part2of2) | i1-Q4_K_S | 53.8 | optimal size/speed/quality |
42
+ | [PART 1](https://huggingface.co/mradermacher/BigWeave-v25-95b-i1-GGUF/resolve/main/BigWeave-v25-95b.i1-Q4_K_M.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/BigWeave-v25-95b-i1-GGUF/resolve/main/BigWeave-v25-95b.i1-Q4_K_M.gguf.part2of2) | i1-Q4_K_M | 56.8 | fast, recommended |
43
+ | [PART 1](https://huggingface.co/mradermacher/BigWeave-v25-95b-i1-GGUF/resolve/main/BigWeave-v25-95b.i1-Q6_K.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/BigWeave-v25-95b-i1-GGUF/resolve/main/BigWeave-v25-95b.i1-Q6_K.gguf.part2of2) | i1-Q6_K | 77.7 | practically like static Q6_K |
44
+
45
+ Here is a handy graph by ikawrakow comparing some lower-quality quant
46
+ types (lower is better):
47
+
48
+ ![image.png](https://www.nethype.de/huggingface_embed/quantpplgraph.png)
49
+
50
+ And here are Artefact2's thoughts on the matter:
51
+ https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
52
+
53
+ ## FAQ / Model Request
54
+
55
+ See https://huggingface.co/mradermacher/model_requests for some answers to
56
+ questions you might have and/or if you want some other model quantized.
57
+
58
+ ## Thanks
59
+
60
+ I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
61
+ me use its servers and providing upgrades to my workstation to enable
62
+ this work in my free time. Additional thanks to [@nicoboss](https://huggingface.co/nicoboss) for giving me access to his private supercomputer, enabling me to provide many more imatrix quants, at much higher quality, than I would otherwise be able to.
63
+
64
+ <!-- end -->