mradermacher commited on
Commit
c944cec
1 Parent(s): 15b5be8

auto-patch README.md

Browse files
Files changed (1) hide show
  1. README.md +70 -0
README.md CHANGED
@@ -1,6 +1,76 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
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/AALF/gemma-2-27b-it-SimPO-37K-100steps
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: AALF/gemma-2-27b-it-SimPO-37K-100steps
3
+ language:
4
+ - en
5
+ library_name: transformers
6
+ license: gemma
7
+ quantized_by: mradermacher
8
+ tags:
9
+ - alignment-handbook
10
+ - generated_from_trainer
11
+ ---
12
+ ## About
13
+
14
  <!-- ### quantize_version: 2 -->
15
  <!-- ### output_tensor_quantised: 1 -->
16
  <!-- ### convert_type: hf -->
17
  <!-- ### vocab_type: -->
18
  <!-- ### tags: nicoboss -->
19
  weighted/imatrix quants of https://huggingface.co/AALF/gemma-2-27b-it-SimPO-37K-100steps
20
+
21
+ <!-- provided-files -->
22
+ static quants are available at https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-GGUF
23
+ ## Usage
24
+
25
+ If you are unsure how to use GGUF files, refer to one of [TheBloke's
26
+ READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
27
+ more details, including on how to concatenate multi-part files.
28
+
29
+ ## Provided Quants
30
+
31
+ (sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
32
+
33
+ | Link | Type | Size/GB | Notes |
34
+ |:-----|:-----|--------:|:------|
35
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-IQ1_S.gguf) | i1-IQ1_S | 6.2 | for the desperate |
36
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-IQ1_M.gguf) | i1-IQ1_M | 6.8 | mostly desperate |
37
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-IQ2_XXS.gguf) | i1-IQ2_XXS | 7.7 | |
38
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-IQ2_XS.gguf) | i1-IQ2_XS | 8.5 | |
39
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-IQ2_S.gguf) | i1-IQ2_S | 8.8 | |
40
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-IQ2_M.gguf) | i1-IQ2_M | 9.5 | |
41
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-Q2_K.gguf) | i1-Q2_K | 10.5 | IQ3_XXS probably better |
42
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 10.9 | lower quality |
43
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-IQ3_XS.gguf) | i1-IQ3_XS | 11.7 | |
44
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-IQ3_S.gguf) | i1-IQ3_S | 12.3 | beats Q3_K* |
45
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-Q3_K_S.gguf) | i1-Q3_K_S | 12.3 | IQ3_XS probably better |
46
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-IQ3_M.gguf) | i1-IQ3_M | 12.6 | |
47
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-Q3_K_M.gguf) | i1-Q3_K_M | 13.5 | IQ3_S probably better |
48
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-Q3_K_L.gguf) | i1-Q3_K_L | 14.6 | IQ3_M probably better |
49
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-IQ4_XS.gguf) | i1-IQ4_XS | 14.9 | |
50
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-Q4_0.gguf) | i1-Q4_0 | 15.8 | fast, low quality |
51
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-Q4_K_S.gguf) | i1-Q4_K_S | 15.8 | optimal size/speed/quality |
52
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-Q4_K_M.gguf) | i1-Q4_K_M | 16.7 | fast, recommended |
53
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-Q5_K_S.gguf) | i1-Q5_K_S | 19.0 | |
54
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-Q5_K_M.gguf) | i1-Q5_K_M | 19.5 | |
55
+ | [GGUF](https://huggingface.co/mradermacher/gemma-2-27b-it-SimPO-37K-100steps-i1-GGUF/resolve/main/gemma-2-27b-it-SimPO-37K-100steps.i1-Q6_K.gguf) | i1-Q6_K | 22.4 | practically like static Q6_K |
56
+
57
+ Here is a handy graph by ikawrakow comparing some lower-quality quant
58
+ types (lower is better):
59
+
60
+ ![image.png](https://www.nethype.de/huggingface_embed/quantpplgraph.png)
61
+
62
+ And here are Artefact2's thoughts on the matter:
63
+ https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
64
+
65
+ ## FAQ / Model Request
66
+
67
+ See https://huggingface.co/mradermacher/model_requests for some answers to
68
+ questions you might have and/or if you want some other model quantized.
69
+
70
+ ## Thanks
71
+
72
+ I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
73
+ me use its servers and providing upgrades to my workstation to enable
74
+ 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.
75
+
76
+ <!-- end -->