mradermacher commited on
Commit
56b6e3a
1 Parent(s): cc27e7e

auto-patch README.md

Browse files
Files changed (1) hide show
  1. README.md +157 -0
README.md CHANGED
@@ -1,6 +1,163 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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/LeroyDyer/LCARS_AI_001
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: LeroyDyer/LCARS_AI_001
3
+ datasets:
4
+ - gretelai/synthetic_text_to_sql
5
+ - HuggingFaceTB/cosmopedia
6
+ - teknium/OpenHermes-2.5
7
+ - Open-Orca/SlimOrca
8
+ - Open-Orca/OpenOrca
9
+ - cognitivecomputations/dolphin-coder
10
+ - databricks/databricks-dolly-15k
11
+ - yahma/alpaca-cleaned
12
+ - uonlp/CulturaX
13
+ - mwitiderrick/SwahiliPlatypus
14
+ - swahili
15
+ - Rogendo/English-Swahili-Sentence-Pairs
16
+ - ise-uiuc/Magicoder-Evol-Instruct-110K
17
+ - meta-math/MetaMathQA
18
+ - abacusai/ARC_DPO_FewShot
19
+ - abacusai/MetaMath_DPO_FewShot
20
+ - abacusai/HellaSwag_DPO_FewShot
21
+ - HaltiaAI/Her-The-Movie-Samantha-and-Theodore-Dataset
22
+ - HuggingFaceFW/fineweb
23
+ - occiglot/occiglot-fineweb-v0.5
24
+ - omi-health/medical-dialogue-to-soap-summary
25
+ - keivalya/MedQuad-MedicalQnADataset
26
+ - ruslanmv/ai-medical-dataset
27
+ - Shekswess/medical_llama3_instruct_dataset_short
28
+ - ShenRuililin/MedicalQnA
29
+ - virattt/financial-qa-10K
30
+ - PatronusAI/financebench
31
+ - takala/financial_phrasebank
32
+ - Replete-AI/code_bagel
33
+ - athirdpath/DPO_Pairs-Roleplay-Alpaca-NSFW
34
+ - IlyaGusev/gpt_roleplay_realm
35
+ - rickRossie/bluemoon_roleplay_chat_data_300k_messages
36
+ - jtatman/hypnosis_dataset
37
+ - Hypersniper/philosophy_dialogue
38
+ - Locutusque/function-calling-chatml
39
+ - bible-nlp/biblenlp-corpus
40
+ - DatadudeDev/Bible
41
+ - Helsinki-NLP/bible_para
42
+ - HausaNLP/AfriSenti-Twitter
43
+ - aixsatoshi/Chat-with-cosmopedia
44
+ - HuggingFaceTB/cosmopedia-100k
45
+ - HuggingFaceFW/fineweb-edu
46
+ - m-a-p/CodeFeedback-Filtered-Instruction
47
+ - heliosbrahma/mental_health_chatbot_dataset
48
+ language:
49
+ - en
50
+ - sw
51
+ - ig
52
+ - tw
53
+ - es
54
+ library_name: transformers
55
+ license: apache-2.0
56
+ quantized_by: mradermacher
57
+ tags:
58
+ - text-generation-inference
59
+ - transformers
60
+ - unsloth
61
+ - mistral
62
+ - trl
63
+ - chemistry
64
+ - biology
65
+ - legal
66
+ - art
67
+ - music
68
+ - finance
69
+ - code
70
+ - medical
71
+ - not-for-all-audiences
72
+ - merge
73
+ - climate
74
+ - chain-of-thought
75
+ - tree-of-knowledge
76
+ - forest-of-thoughts
77
+ - visual-spacial-sketchpad
78
+ - alpha-mind
79
+ - knowledge-graph
80
+ - entity-detection
81
+ - encyclopedia
82
+ - wikipedia
83
+ - stack-exchange
84
+ - Reddit
85
+ - Cyber-series
86
+ - MegaMind
87
+ - Cybertron
88
+ - SpydazWeb
89
+ - Spydaz
90
+ - LCARS
91
+ - star-trek
92
+ - mega-transformers
93
+ - Mulit-Mega-Merge
94
+ - Multi-Lingual
95
+ - Afro-Centric
96
+ - African-Model
97
+ - Ancient-One
98
+ ---
99
+ ## About
100
+
101
  <!-- ### quantize_version: 2 -->
102
  <!-- ### output_tensor_quantised: 1 -->
103
  <!-- ### convert_type: hf -->
104
  <!-- ### vocab_type: -->
105
  <!-- ### tags: nicoboss -->
106
  weighted/imatrix quants of https://huggingface.co/LeroyDyer/LCARS_AI_001
107
+
108
+ <!-- provided-files -->
109
+ static quants are available at https://huggingface.co/mradermacher/LCARS_AI_001-GGUF
110
+ ## Usage
111
+
112
+ If you are unsure how to use GGUF files, refer to one of [TheBloke's
113
+ READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
114
+ more details, including on how to concatenate multi-part files.
115
+
116
+ ## Provided Quants
117
+
118
+ (sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
119
+
120
+ | Link | Type | Size/GB | Notes |
121
+ |:-----|:-----|--------:|:------|
122
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-IQ1_S.gguf) | i1-IQ1_S | 1.7 | for the desperate |
123
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-IQ1_M.gguf) | i1-IQ1_M | 1.9 | mostly desperate |
124
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-IQ2_XXS.gguf) | i1-IQ2_XXS | 2.1 | |
125
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-IQ2_XS.gguf) | i1-IQ2_XS | 2.3 | |
126
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-IQ2_S.gguf) | i1-IQ2_S | 2.4 | |
127
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-IQ2_M.gguf) | i1-IQ2_M | 2.6 | |
128
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-Q2_K.gguf) | i1-Q2_K | 2.8 | IQ3_XXS probably better |
129
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 2.9 | lower quality |
130
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-IQ3_XS.gguf) | i1-IQ3_XS | 3.1 | |
131
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-Q3_K_S.gguf) | i1-Q3_K_S | 3.3 | IQ3_XS probably better |
132
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-IQ3_S.gguf) | i1-IQ3_S | 3.3 | beats Q3_K* |
133
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-IQ3_M.gguf) | i1-IQ3_M | 3.4 | |
134
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-Q3_K_M.gguf) | i1-Q3_K_M | 3.6 | IQ3_S probably better |
135
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-Q3_K_L.gguf) | i1-Q3_K_L | 3.9 | IQ3_M probably better |
136
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-IQ4_XS.gguf) | i1-IQ4_XS | 4.0 | |
137
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-Q4_0.gguf) | i1-Q4_0 | 4.2 | fast, low quality |
138
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-Q4_K_S.gguf) | i1-Q4_K_S | 4.2 | optimal size/speed/quality |
139
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-Q4_K_M.gguf) | i1-Q4_K_M | 4.5 | fast, recommended |
140
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-Q5_K_S.gguf) | i1-Q5_K_S | 5.1 | |
141
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-Q5_K_M.gguf) | i1-Q5_K_M | 5.2 | |
142
+ | [GGUF](https://huggingface.co/mradermacher/LCARS_AI_001-i1-GGUF/resolve/main/LCARS_AI_001.i1-Q6_K.gguf) | i1-Q6_K | 6.0 | practically like static Q6_K |
143
+
144
+ Here is a handy graph by ikawrakow comparing some lower-quality quant
145
+ types (lower is better):
146
+
147
+ ![image.png](https://www.nethype.de/huggingface_embed/quantpplgraph.png)
148
+
149
+ And here are Artefact2's thoughts on the matter:
150
+ https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
151
+
152
+ ## FAQ / Model Request
153
+
154
+ See https://huggingface.co/mradermacher/model_requests for some answers to
155
+ questions you might have and/or if you want some other model quantized.
156
+
157
+ ## Thanks
158
+
159
+ I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
160
+ me use its servers and providing upgrades to my workstation to enable
161
+ this work in my free time. Additional thanks to [@nicoboss](https://huggingface.co/nicoboss) for giving me access to his hardware for calculating the imatrix for these quants.
162
+
163
+ <!-- end -->