Hugging Face's logo

mradermacher
/
bloomz-mt-i1-GGUF

Transformers
GGUF
Inference Endpoints
imatrix
Model card Files Community
mradermacher commited on
Commit
3f87f67
•
1 Parent(s): 347e487

auto-patch README.md

Browse files
Files changed (1) hide show
  1. README.md +114 -0
README.md CHANGED
@@ -1,6 +1,120 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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/bigscience/bloomz-mt
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: bigscience/bloomz-mt
3
+ datasets:
4
+ - bigscience/xP3mt
5
+ language:
6
+ - ak
7
+ - ar
8
+ - as
9
+ - bm
10
+ - bn
11
+ - ca
12
+ - code
13
+ - en
14
+ - es
15
+ - eu
16
+ - fon
17
+ - fr
18
+ - gu
19
+ - hi
20
+ - id
21
+ - ig
22
+ - ki
23
+ - kn
24
+ - lg
25
+ - ln
26
+ - ml
27
+ - mr
28
+ - ne
29
+ - nso
30
+ - ny
31
+ - or
32
+ - pa
33
+ - pt
34
+ - rn
35
+ - rw
36
+ - sn
37
+ - st
38
+ - sw
39
+ - ta
40
+ - te
41
+ - tn
42
+ - ts
43
+ - tum
44
+ - tw
45
+ - ur
46
+ - vi
47
+ - wo
48
+ - xh
49
+ - yo
50
+ - zh
51
+ - zu
52
+ library_name: transformers
53
+ license: bigscience-bloom-rail-1.0
54
+ quantized_by: mradermacher
55
+ ---
56
+ ## About
57
+
58
  <!-- ### quantize_version: 2 -->
59
  <!-- ### output_tensor_quantised: 1 -->
60
  <!-- ### convert_type: hf -->
61
  <!-- ### vocab_type: -->
62
  <!-- ### tags: nicoboss -->
63
  weighted/imatrix quants of https://huggingface.co/bigscience/bloomz-mt
64
+
65
+ <!-- provided-files -->
66
+ static quants are available at https://huggingface.co/mradermacher/bloomz-mt-GGUF
67
+ ## Usage
68
+
69
+ If you are unsure how to use GGUF files, refer to one of [TheBloke's
70
+ READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
71
+ more details, including on how to concatenate multi-part files.
72
+
73
+ ## Provided Quants
74
+
75
+ (sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
76
+
77
+ | Link | Type | Size/GB | Notes |
78
+ |:-----|:-----|--------:|:------|
79
+ | [GGUF](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-IQ1_S.gguf) | i1-IQ1_S | 39.3 | for the desperate |
80
+ | [GGUF](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-IQ1_M.gguf) | i1-IQ1_M | 42.9 | mostly desperate |
81
+ | [GGUF](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-IQ2_XXS.gguf) | i1-IQ2_XXS | 48.8 | |
82
+ | [PART 1](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-IQ2_XS.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-IQ2_XS.gguf.part2of2) | i1-IQ2_XS | 54.0 | |
83
+ | [PART 1](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-IQ2_S.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-IQ2_S.gguf.part2of2) | i1-IQ2_S | 57.0 | |
84
+ | [PART 1](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-IQ2_M.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-IQ2_M.gguf.part2of2) | i1-IQ2_M | 61.8 | |
85
+ | [PART 1](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q2_K.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q2_K.gguf.part2of2) | i1-Q2_K | 68.2 | IQ3_XXS probably better |
86
+ | [PART 1](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-IQ3_XXS.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-IQ3_XXS.gguf.part2of2) | i1-IQ3_XXS | 70.9 | lower quality |
87
+ | [PART 1](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-IQ3_XS.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-IQ3_XS.gguf.part2of2) | i1-IQ3_XS | 76.8 | |
88
+ | [PART 1](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-IQ3_S.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-IQ3_S.gguf.part2of2) | i1-IQ3_S | 78.8 | beats Q3_K* |
89
+ | [PART 1](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q3_K_S.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q3_K_S.gguf.part2of2) | i1-Q3_K_S | 78.8 | IQ3_XS probably better |
90
+ | [PART 1](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-IQ3_M.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-IQ3_M.gguf.part2of2) | i1-IQ3_M | 87.4 | |
91
+ | [PART 1](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q3_K_M.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q3_K_M.gguf.part2of2) | i1-Q3_K_M | 94.5 | IQ3_S probably better |
92
+ | [PART 1](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-IQ4_XS.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-IQ4_XS.gguf.part2of2) | i1-IQ4_XS | 96.7 | |
93
+ | [PART 1](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q4_0.gguf.part1of3) [PART 2](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q4_0.gguf.part2of3) [PART 3](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q4_0.gguf.part3of3) | i1-Q4_0 | 102.7 | fast, low quality |
94
+ | [PART 1](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q4_K_S.gguf.part1of3) [PART 2](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q4_K_S.gguf.part2of3) [PART 3](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q4_K_S.gguf.part3of3) | i1-Q4_K_S | 103.1 | optimal size/speed/quality |
95
+ | [PART 1](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q3_K_L.gguf.part1of3) [PART 2](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q3_K_L.gguf.part2of3) [PART 3](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q3_K_L.gguf.part3of3) | i1-Q3_K_L | 103.1 | IQ3_M probably better |
96
+ | [PART 1](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q4_K_M.gguf.part1of3) [PART 2](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q4_K_M.gguf.part2of3) [PART 3](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q4_K_M.gguf.part3of3) | i1-Q4_K_M | 114.8 | fast, recommended |
97
+ | [PART 1](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q5_K_S.gguf.part1of3) [PART 2](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q5_K_S.gguf.part2of3) [PART 3](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q5_K_S.gguf.part3of3) | i1-Q5_K_S | 124.3 | |
98
+ | [PART 1](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q5_K_M.gguf.part1of3) [PART 2](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q5_K_M.gguf.part2of3) [PART 3](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q5_K_M.gguf.part3of3) | i1-Q5_K_M | 133.7 | |
99
+ | [PART 1](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q6_K.gguf.part1of3) [PART 2](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q6_K.gguf.part2of3) [PART 3](https://huggingface.co/mradermacher/bloomz-mt-i1-GGUF/resolve/main/bloomz-mt.i1-Q6_K.gguf.part3of3) | i1-Q6_K | 147.7 | practically like static Q6_K |
100
+
101
+ Here is a handy graph by ikawrakow comparing some lower-quality quant
102
+ types (lower is better):
103
+
104
+ ![image.png](https://www.nethype.de/huggingface_embed/quantpplgraph.png)
105
+
106
+ And here are Artefact2's thoughts on the matter:
107
+ https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
108
+
109
+ ## FAQ / Model Request
110
+
111
+ See https://huggingface.co/mradermacher/model_requests for some answers to
112
+ questions you might have and/or if you want some other model quantized.
113
+
114
+ ## Thanks
115
+
116
+ I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
117
+ me use its servers and providing upgrades to my workstation to enable
118
+ 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.
119
+
120
+ <!-- end -->