mradermacher commited on
Commit
1140f83
1 Parent(s): 49c0d9e

auto-patch README.md

Browse files
Files changed (1) hide show
  1. README.md +68 -0
README.md CHANGED
@@ -1,5 +1,73 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  <!-- ### quantize_version: 2 -->
2
  <!-- ### output_tensor_quantised: 1 -->
3
  <!-- ### convert_type: hf -->
4
  <!-- ### vocab_type: -->
5
  weighted/imatrix quants of https://huggingface.co/HiroseKoichi/Llama-Salad-4x8B
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: HiroseKoichi/Llama-Salad-4x8B
3
+ language:
4
+ - en
5
+ library_name: transformers
6
+ license: llama3
7
+ quantized_by: mradermacher
8
+ tags:
9
+ - nsfw
10
+ - not-for-all-audiences
11
+ - llama-3
12
+ - text-generation-inference
13
+ ---
14
+ ## About
15
+
16
  <!-- ### quantize_version: 2 -->
17
  <!-- ### output_tensor_quantised: 1 -->
18
  <!-- ### convert_type: hf -->
19
  <!-- ### vocab_type: -->
20
  weighted/imatrix quants of https://huggingface.co/HiroseKoichi/Llama-Salad-4x8B
21
+
22
+ <!-- provided-files -->
23
+ static quants are available at https://huggingface.co/mradermacher/Llama-Salad-4x8B-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/Llama-Salad-4x8B-i1-GGUF/resolve/main/Llama-Salad-4x8B.i1-IQ2_XXS.gguf) | i1-IQ2_XXS | 6.9 | |
37
+ | [GGUF](https://huggingface.co/mradermacher/Llama-Salad-4x8B-i1-GGUF/resolve/main/Llama-Salad-4x8B.i1-IQ2_M.gguf) | i1-IQ2_M | 8.5 | |
38
+ | [GGUF](https://huggingface.co/mradermacher/Llama-Salad-4x8B-i1-GGUF/resolve/main/Llama-Salad-4x8B.i1-Q2_K.gguf) | i1-Q2_K | 9.4 | IQ3_XXS probably better |
39
+ | [GGUF](https://huggingface.co/mradermacher/Llama-Salad-4x8B-i1-GGUF/resolve/main/Llama-Salad-4x8B.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 9.9 | lower quality |
40
+ | [GGUF](https://huggingface.co/mradermacher/Llama-Salad-4x8B-i1-GGUF/resolve/main/Llama-Salad-4x8B.i1-IQ3_XS.gguf) | i1-IQ3_XS | 10.5 | |
41
+ | [GGUF](https://huggingface.co/mradermacher/Llama-Salad-4x8B-i1-GGUF/resolve/main/Llama-Salad-4x8B.i1-Q3_K_S.gguf) | i1-Q3_K_S | 11.0 | IQ3_XS probably better |
42
+ | [GGUF](https://huggingface.co/mradermacher/Llama-Salad-4x8B-i1-GGUF/resolve/main/Llama-Salad-4x8B.i1-IQ3_S.gguf) | i1-IQ3_S | 11.1 | beats Q3_K* |
43
+ | [GGUF](https://huggingface.co/mradermacher/Llama-Salad-4x8B-i1-GGUF/resolve/main/Llama-Salad-4x8B.i1-IQ3_M.gguf) | i1-IQ3_M | 11.2 | |
44
+ | [GGUF](https://huggingface.co/mradermacher/Llama-Salad-4x8B-i1-GGUF/resolve/main/Llama-Salad-4x8B.i1-Q3_K_M.gguf) | i1-Q3_K_M | 12.2 | IQ3_S probably better |
45
+ | [GGUF](https://huggingface.co/mradermacher/Llama-Salad-4x8B-i1-GGUF/resolve/main/Llama-Salad-4x8B.i1-Q3_K_L.gguf) | i1-Q3_K_L | 13.1 | IQ3_M probably better |
46
+ | [GGUF](https://huggingface.co/mradermacher/Llama-Salad-4x8B-i1-GGUF/resolve/main/Llama-Salad-4x8B.i1-IQ4_XS.gguf) | i1-IQ4_XS | 13.5 | |
47
+ | [GGUF](https://huggingface.co/mradermacher/Llama-Salad-4x8B-i1-GGUF/resolve/main/Llama-Salad-4x8B.i1-Q4_0.gguf) | i1-Q4_0 | 14.3 | fast, low quality |
48
+ | [GGUF](https://huggingface.co/mradermacher/Llama-Salad-4x8B-i1-GGUF/resolve/main/Llama-Salad-4x8B.i1-Q4_K_S.gguf) | i1-Q4_K_S | 14.4 | optimal size/speed/quality |
49
+ | [GGUF](https://huggingface.co/mradermacher/Llama-Salad-4x8B-i1-GGUF/resolve/main/Llama-Salad-4x8B.i1-Q4_K_M.gguf) | i1-Q4_K_M | 15.3 | fast, recommended |
50
+ | [GGUF](https://huggingface.co/mradermacher/Llama-Salad-4x8B-i1-GGUF/resolve/main/Llama-Salad-4x8B.i1-Q5_K_S.gguf) | i1-Q5_K_S | 17.3 | |
51
+ | [GGUF](https://huggingface.co/mradermacher/Llama-Salad-4x8B-i1-GGUF/resolve/main/Llama-Salad-4x8B.i1-Q5_K_M.gguf) | i1-Q5_K_M | 17.8 | |
52
+ | [GGUF](https://huggingface.co/mradermacher/Llama-Salad-4x8B-i1-GGUF/resolve/main/Llama-Salad-4x8B.i1-Q6_K.gguf) | i1-Q6_K | 20.6 | practically like static Q6_K |
53
+
54
+ Here is a handy graph by ikawrakow comparing some lower-quality quant
55
+ types (lower is better):
56
+
57
+ ![image.png](https://www.nethype.de/huggingface_embed/quantpplgraph.png)
58
+
59
+ And here are Artefact2's thoughts on the matter:
60
+ https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
61
+
62
+ ## FAQ / Model Request
63
+
64
+ See https://huggingface.co/mradermacher/model_requests for some answers to
65
+ questions you might have and/or if you want some other model quantized.
66
+
67
+ ## Thanks
68
+
69
+ I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
70
+ me use its servers and providing upgrades to my workstation to enable
71
+ this work in my free time.
72
+
73
+ <!-- end -->