forked from lmstudio-ai/model-catalog
-
Notifications
You must be signed in to change notification settings - Fork 0
/
catalog.json
1095 lines (1095 loc) · 63.8 KB
/
catalog.json
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
[
{
"_descriptorVersion": "0.0.1",
"datePublished": "2023-06-03T05:34:04.000Z",
"name": "Nous-Hermes-13b",
"description": "Nous-Hermes-13b is a state-of-the-art language model fine-tuned on over 300,000 instructions. This model was fine-tuned by Nous Research, with Teknium and Karan4D leading the fine tuning process and dataset curation, Redmond AI sponsoring the compute, and several other contributors. The result is an enhanced Llama 13b model that rivals GPT-3.5-turbo in performance across a variety of tasks. This model stands out for its long responses, low hallucination rate, and absence of OpenAI censorship mechanisms. The fine-tuning process was performed with a 2000 sequence length on an 8x a100 80GB DGX machine for over 50 hours.",
"author": {
"name": "Nous Research",
"url": "https://nousresearch.com",
"blurb": "Nous Research is dedicated to advancing the field of natural language processing, in collaboration with the open-source community, through bleeding-edge research and a commitment to symbiotic development."
},
"numParameters": "13B",
"resources": {
"canonicalUrl": "https://huggingface.co/NousResearch/Nous-Hermes-13b",
"downloadUrl": "https://huggingface.co/TheBloke/Nous-Hermes-13B-GGML"
},
"trainedFor": "chat",
"arch": "llama",
"files": {
"highlighted": {
"economical": {
"name": "nous-hermes-13b.ggmlv3.q3_K_S.bin"
},
"most_capable": {
"name": "nous-hermes-13b.ggmlv3.q6_K.bin"
}
},
"all": [
{
"name": "nous-hermes-13b.ggmlv3.q3_K_S.bin",
"url": "https://huggingface.co/TheBloke/Nous-Hermes-13B-GGML/resolve/main/nous-hermes-13b.ggmlv3.q3_K_S.bin",
"sizeBytes": 5594695104,
"quantization": "Q4_K_S",
"format": "ggml",
"sha256checksum": "591a49f1ef4dbc2cf43943c5ec9bd617e6086264e30f66718bc764bc55286b5e",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/Nous-Hermes-13B-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/Nous-Hermes-13B-GGML"
},
{
"name": "nous-hermes-13b.ggmlv3.q6_K.bin",
"url": "https://huggingface.co/TheBloke/Nous-Hermes-13B-GGML/resolve/main/nous-hermes-13b.ggmlv3.q6_K.bin",
"sizeBytes": 10678859104,
"quantization": "Q6_K",
"format": "ggml",
"sha256checksum": "efe8ffe14aa97c3c5f45f2bc8e80a02933c5e907813deb685c93341bf671f44e",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/Nous-Hermes-13B-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/Nous-Hermes-13B-GGML"
}
]
}
},
{
"_descriptorVersion": "0.0.1",
"datePublished": "2023-08-24T21:39:59",
"name": "CodeLlama 7B Instruct",
"description": "MetaAI has released Code Llama, a comprehensive family of large language models for code. These models are based on Llama 2 and exhibit state-of-the-art performance among openly available models. They offer advanced infilling capabilities, can accommodate large input contexts, and have the ability to follow instructions for programming tasks without prior training. There are various versions available to cater to a wide array of applications: foundation models (Code Llama), Python-specific models (Code Llama - Python), and models for following instructions (Code Llama - Instruct). These versions come with 7B, 13B, and 34B parameters respectively. All models are trained on 16k token sequences and show improvements even on inputs with up to 100k tokens. The 7B and 13B models of Code Llama and Code Llama - Instruct have the ability to infill based on surrounding content. In terms of performance, Code Llama has set new standards among open models on several code benchmarks, achieving scores of up to 53% on HumanEval and 55% on MBPP. Notably, the Python version of Code Llama 7B surpasses the performance of Llama 2 70B on HumanEval and MBPP. All of MetaAI's models outperform every other publicly available model on MultiPL-E. Code Llama has been released under a permissive license that enables both research and commercial use.",
"author": {
"name": "Meta AI",
"url": "https://ai.meta.com",
"blurb": "Pushing the boundaries of AI through research, infrastructure and product innovation."
},
"numParameters": "7B",
"resources": {
"canonicalUrl": "https://ai.meta.com/blog/code-llama-large-language-model-coding/",
"paperUrl": "https://ai.meta.com/research/publications/code-llama-open-foundation-models-for-code/",
"downloadUrl": "https://huggingface.co/TheBloke/CodeLlama-7B-Instruct-GGUF"
},
"trainedFor": "chat",
"arch": "llama",
"files": {
"highlighted": {
"economical": {
"name": "codellama-7b-instruct.Q4_K_S.gguf"
},
"most_capable": {
"name": "codellama-7b-instruct.Q6_K.gguf"
}
},
"all": [
{
"name": "codellama-7b-instruct.Q4_K_S.gguf",
"url": "https://huggingface.co/TheBloke/CodeLlama-7B-Instruct-GGUF/resolve/main/codellama-7b-instruct.Q4_K_S.gguf",
"sizeBytes": 3856831168,
"quantization": "Q4_K_S",
"format": "gguf",
"sha256checksum": "2e44d2b7ae28bbe3a2ed698e259cbd3a6bf7fe8f9d351e14b2be17fb690d7f95",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/CodeLlama-7B-Instruct-GGUF",
"repositoryUrl": "https://huggingface.co/TheBloke/CodeLlama-7B-Instruct-GGUF"
},
{
"name": "codellama-7b-instruct.Q6_K.gguf",
"url": "https://huggingface.co/TheBloke/CodeLlama-7B-Instruct-GGUF/resolve/main/codellama-7b-instruct.Q6_K.gguf",
"sizeBytes": 5529302208,
"quantization": "Q6_K",
"format": "gguf",
"sha256checksum": "2f516cd9c16181832ffceaf94b13e8600d88c9bc8d7f75717d25d8c9cf9aa973",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/CodeLlama-7B-Instruct-GGUF",
"repositoryUrl": "https://huggingface.co/TheBloke/CodeLlama-7B-Instruct-GGUF"
}
]
}
},
{
"_descriptorVersion": "0.0.1",
"datePublished": "2023-07-18T21:08:14.000Z",
"name": "Llama-2-7B-Chat-GGML",
"description": "This is the 7B model from the Llama 2 family of large language models (LLMs), a collection of pretrained and fine-tuned generative text models ranging in scale from 7 billion to 70 billion parameters. Meta's fine-tuned LLMs, called Llama-2-Chat, are optimized for dialogue use cases. Llama-2-Chat models outperform open-source chat models on most benchmarks we tested, and in Meta's human evaluations for helpfulness and safety, are on par with some popular closed-source models like ChatGPT and PaLM.",
"author": {
"name": "Meta AI",
"url": "https://ai.meta.com",
"blurb": "Pushing the boundaries of AI through research, infrastructure and product innovation."
},
"numParameters": "7B",
"resources": {
"canonicalUrl": "https://huggingface.co/meta-llama/Llama-2-7b-chat-hf",
"paperUrl": "https://ai.meta.com/research/publications/llama-2-open-foundation-and-fine-tuned-chat-models/",
"downloadUrl": "https://huggingface.co/TheBloke/Llama-2-7B-Chat-GGML"
},
"trainedFor": "chat",
"arch": "llama",
"files": {
"highlighted": {
"economical": {
"name": "llama-2-7b-chat.ggmlv3.q4_K_S.bin"
},
"most_capable": {
"name": "llama-2-7b-chat.ggmlv3.q6_K.bin"
}
},
"all": [
{
"name": "llama-2-7b-chat.ggmlv3.q4_K_S.bin",
"url": "https://huggingface.co/TheBloke/Llama-2-7B-Chat-GGML/resolve/main/llama-2-7b-chat.ggmlv3.q4_K_S.bin",
"sizeBytes": 3825517184,
"quantization": "Q4_K_S",
"format": "ggml",
"sha256checksum": "32b758bf5e4f16fb5944b75d577fbca18c11c57000b41c6cc04bb281632d58f3",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/Llama-2-7B-Chat-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/Llama-2-7B-Chat-GGML"
},
{
"name": "llama-2-7b-chat.ggmlv3.q6_K.bin",
"url": "https://huggingface.co/TheBloke/Llama-2-7B-Chat-GGML/resolve/main/llama-2-7b-chat.ggmlv3.q6_K.bin",
"sizeBytes": 5528904320,
"quantization": "Q6_K",
"format": "ggml",
"sha256checksum": "24a2097aba9bc63395654515618fb2ceeaea64452147ee5299990b636e4c00ce",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/Llama-2-7B-Chat-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/Llama-2-7B-Chat-GGML"
}
]
}
},
{
"_descriptorVersion": "0.0.1",
"datePublished": "2023-06-12T11:16:39.000Z",
"name": "Manticore 13B Chat Pyg Guanaco",
"description": "The model, augmented with Guanaco qLora, shows broad capabilities compared to other models like Wizard or Manticore. It excels in in-context learning and reasoning but may have weaknesses in coding. The model follows instructions, works as a chatbot, and produces intelligent responses. It accepts various prompting styles, including the ###-Variant. The model is generally unrestricted and doesn't berate users. Recommended settings include low temperature, low diversity, and slight repetition penalty.",
"author": {
"name": "Open Access AI Collective",
"url": "https://huggingface.co/openaccess-ai-collective/",
"blurb": ""
},
"numParameters": "13B",
"resources": {
"canonicalUrl": "https://huggingface.co/openaccess-ai-collective/manticore-13b-chat-pyg",
"downloadUrl": "https://huggingface.co/mindrage/Manticore-13B-Chat-Pyg-Guanaco-GGML"
},
"trainedFor": "chat",
"arch": "llama",
"files": {
"highlighted": {
"economical": {
"name": "Manticore-13B-Chat-Pyg-Guanaco-GGML-q3_K_M.bin"
},
"most_capable": {
"name": "Manticore-13B-Chat-Pyg-Guanaco-GGML-q5_K_S.bin"
}
},
"all": [
{
"name": "Manticore-13B-Chat-Pyg-Guanaco-GGML-q3_K_M.bin",
"url": "https://huggingface.co/mindrage/Manticore-13B-Chat-Pyg-Guanaco-GGML/resolve/main/Manticore-13B-Chat-Pyg-Guanaco-GGML-q3_K_M.bin",
"sizeBytes": 6249231488,
"quantization": "Q3_K_M",
"format": "ggml",
"sha256checksum": "ea266b52d6080e7df1b8d98ea290d0b4e261dc9cfe5fc9169abcef0c154831e5",
"publisher": {
"name": "mindrage",
"socialUrl": "https://github.com/mindrages"
},
"respository": "mindrage/Manticore-13B-Chat-Pyg-Guanaco-GGML",
"repositoryUrl": "https://huggingface.co/mindrage/Manticore-13B-Chat-Pyg-Guanaco-GGML"
},
{
"name": "Manticore-13B-Chat-Pyg-Guanaco-GGML-q5_K_S.bin",
"url": "https://huggingface.co/mindrage/Manticore-13B-Chat-Pyg-Guanaco-GGML/resolve/main/Manticore-13B-Chat-Pyg-Guanaco-GGML-q5_K_S.bin",
"sizeBytes": 8950236288,
"quantization": "Q5_K_S",
"format": "ggml",
"sha256checksum": "00a47eeb6364e1e022a45cfe1232a37fa78e9f040242dba78d34d6df383e32d1",
"publisher": {
"name": "mindrage",
"socialUrl": "https://github.com/mindrages"
},
"respository": "mindrage/Manticore-13B-Chat-Pyg-Guanaco-GGML",
"repositoryUrl": "https://huggingface.co/mindrage/Manticore-13B-Chat-Pyg-Guanaco-GGML"
}
]
}
},
{
"_descriptorVersion": "0.0.1",
"datePublished": "2023-09-27T16:12:57",
"name": "Mistral 7B Instruct v0.1",
"description": "The Mistral-7B-Instruct-v0.1 is a Large Language Model (LLM) developed by Mistral AI. This LLM is an instruct fine-tuned version of a generative text model, leveraging a variety of publicly available conversation datasets. The model's architecture is based on a transformer model, featuring Grouped-Query Attention, Sliding-Window Attention, and a Byte-fallback BPE tokenizer. To utilize the instruction fine-tuning capabilities, prompts should be enclosed within [INST] and [/INST] tokens. The initial instruction should commence with a beginning-of-sentence id, whereas subsequent instructions should not. The generation process by the assistant will terminate with the end-of-sentence token id. For detailed information about this model, refer to the release blog posts by Mistral AI.",
"author": {
"name": "Mistral AI",
"url": "https://mistral.ai/",
"blurb": "Mistral AI's mission is to spearhead the revolution of open models."
},
"numParameters": "7B",
"resources": {
"canonicalUrl": "https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.1",
"paperUrl": "https://mistral.ai/news/announcing-mistral-7b/",
"downloadUrl": "https://huggingface.co/TheBloke/Mistral-7B-Instruct-v0.1-GGUF"
},
"trainedFor": "chat",
"arch": "mistral",
"files": {
"highlighted": {
"economical": {
"name": "mistral-7b-instruct-v0.1.Q4_K_S.gguf"
},
"most_capable": {
"name": "mistral-7b-instruct-v0.1.Q6_K.gguf"
}
},
"all": [
{
"name": "mistral-7b-instruct-v0.1.Q4_K_S.gguf",
"url": "https://huggingface.co/TheBloke/Mistral-7B-Instruct-v0.1-GGUF/resolve/main/mistral-7b-instruct-v0.1.Q4_K_S.gguf",
"sizeBytes": 4140373664,
"quantization": "Q4_K_S",
"format": "gguf",
"sha256checksum": "f1b7f1885029080be49aff49c83f87333449ef727089546e0d887e2f17f0d02e",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/Mistral-7B-Instruct-v0.1-GGUF",
"repositoryUrl": "https://huggingface.co/TheBloke/Mistral-7B-Instruct-v0.1-GGUF"
},
{
"name": "mistral-7b-instruct-v0.1.Q6_K.gguf",
"url": "https://huggingface.co/TheBloke/Mistral-7B-Instruct-v0.1-GGUF/resolve/main/mistral-7b-instruct-v0.1.Q6_K.gguf",
"sizeBytes": 5942064800,
"quantization": "Q6_K",
"format": "gguf",
"sha256checksum": "dfb053cb8d5f56abde8f56899ffe0d23e1285a423df0b65ea3f3adbb263b22c2",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/Mistral-7B-Instruct-v0.1-GGUF",
"repositoryUrl": "https://huggingface.co/TheBloke/Mistral-7B-Instruct-v0.1-GGUF"
}
]
}
},
{
"_descriptorVersion": "0.0.1",
"datePublished": "2023-07-23T12:47:44.000Z",
"name": "Dolphin-Llama-13B",
"description": "This model, which is based on llama1, is designed specifically for non-commercial use and has undergone a thorough training process using an open-source implementation of Microsoft's Orca dataset. The utilized dataset, post rigorous cleaning and deduplication procedures, encompasses 842,610 instructional instances augmented with GPT-4 completions and an additional 2,625,353 instructional instances augmented with GPT-3.5 completions. The unique aspect of this model is its uncensored nature, having been meticulously filtered to remove any instances of alignment and bias. However, it is advised for users to implement their own alignment layer before deploying this model as a service. The model was trained using the comprehensive flan5m (gpt3.5 completions) dataset for 3 epochs and the flan1m (gpt4 completions) dataset for 2.5 epochs, with distinct learning rates set to preemptively avoid overfitting. This extensive training took approximately 600 hours on a setup of 8x H100s.",
"author": {
"name": "Eric Hartford",
"url": "https://twitter.com/erhartford",
"blurb": "Eric Hartford is a software engineer and entrepreneur. He specializes in alignment, uncensored models, intersection of AI and society, curating datasets and training chat and instruction tuned models."
},
"numParameters": "7B",
"resources": {
"canonicalUrl": "https://huggingface.co/ehartford/dolphin-llama-13b",
"downloadUrl": "https://huggingface.co/TheBloke/Dolphin-Llama-13B-GGML"
},
"trainedFor": "chat",
"arch": "llama",
"files": {
"highlighted": {
"economical": {
"name": "dolphin-llama-13b.ggmlv3.q4_K_S.bin"
},
"most_capable": {
"name": "dolphin-llama-13b.ggmlv3.q6_K.bin"
}
},
"all": [
{
"name": "dolphin-llama-13b.ggmlv3.q4_K_S.bin",
"url": "https://huggingface.co/TheBloke/Dolphin-Llama-13B-GGML/resolve/main/dolphin-llama-13b.ggmlv3.q4_K_S.bin",
"sizeBytes": 3791725184,
"quantization": "Q4_K_S",
"format": "ggml",
"sha256checksum": "3a4e3eff4a0bfa1dc8a0b257ae36abbc532a4e3e09e1e27cd82958ff8addd173",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/Dolphin-Llama-13B-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/Dolphin-Llama-13B-GGML"
},
{
"name": "dolphin-llama-13b.ggmlv3.q6_K.bin",
"url": "https://huggingface.co/TheBloke/Dolphin-Llama-13B-GGML/resolve/main/dolphin-llama-13b.ggmlv3.q6_K.bin",
"sizeBytes": 10678850688,
"quantization": "Q6_K",
"format": "ggml",
"sha256checksum": "8258f8c014a939ce93519bca658eb714202dacbee8f5df05b2dee0f02472460d",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/Dolphin-Llama-13B-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/Dolphin-Llama-13B-GGML"
}
]
}
},
{
"_descriptorVersion": "0.0.1",
"datePublished": "2023-05-22T17:43:26.000Z",
"name": "Guanaco 13B",
"description": "Guanaco models are open-source, finetuned chatbots derived from 4-bit QLoRA tuning of LLaMA base models using the OASST1 dataset. They come in 7B, 13B, 33B, and 65B parameter sizes and are intended solely for research purposes. These models are competitive with commercial chatbot systems on the Vicuna and OpenAssistant benchmarks, as evaluated by human and GPT-4 raters. However, performance may vary on tasks not covered by these benchmarks. Guanaco models facilitate inexpensive, local experimentation with high-quality chatbot systems and offer a replicable, efficient training procedure that can be adapted to new use cases. The effectiveness of 4-bit QLoRA finetuning is demonstrated in a rigorous comparison to 16-bit methods in our paper. Guanaco models feature lightweight checkpoints containing only adapter weights. The adapter weights are licensed under Apache 2, but using them requires access to the LLaMA model weights, and usage should comply with the LLaMA license.",
"author": {
"name": "Dettmers et al.",
"url": "https://github.com/artidoro/qlora",
"blurb": "QLoRA uses bitsandbytes for quantization and is integrated with Hugging Face's PEFT and transformers libraries. QLoRA was developed by members of the University of Washington's UW NLP group."
},
"numParameters": "13B",
"resources": {
"canonicalUrl": "https://github.com/artidoro/qlora",
"downloadUrl": "https://huggingface.co/TheBloke/guanaco-7B-GGML",
"paperUrl": "https://arxiv.org/abs/2305.14314"
},
"trainedFor": "chat",
"arch": "llama",
"files": {
"highlighted": {
"economical": {
"name": "guanaco-7B.ggmlv3.q4_K_S.bin"
},
"most_capable": {
"name": "guanaco-7B.ggmlv3.q6_K.bin"
}
},
"all": [
{
"name": "guanaco-7B.ggmlv3.q4_K_S.bin",
"url": "https://huggingface.co/TheBloke/guanaco-7B-GGML/resolve/main/guanaco-7B.ggmlv3.q4_K_S.bin",
"sizeBytes": 3791725184,
"quantization": "Q4_K_S",
"format": "ggml",
"sha256checksum": "07e2ef24267844c3f06f4aebd2a8b36ff6f7eac0d857e709814d6c63c8219dde",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/guanaco-7B-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/guanaco-7B-GGML"
},
{
"name": "guanaco-7B.ggmlv3.q6_K.bin",
"url": "https://huggingface.co/TheBloke/guanaco-7B-GGML/resolve/main/guanaco-7B.ggmlv3.q6_K.bin",
"sizeBytes": 5528904320,
"quantization": "Q6_K",
"format": "ggml",
"sha256checksum": "458af62352805337ab604ac5d05fe38a293adc8ef0c6799187fef45057579569",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/guanaco-7B-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/guanaco-7B-GGML"
}
]
}
},
{
"_descriptorVersion": "0.0.1",
"datePublished": "2023-05-04T23:15:35.000Z",
"name": "MPT-7B-StoryWriter",
"description": "MPT-7B-StoryWriter-65k+ is a model designed to read and write fictional stories with super long context lengths. It was built by finetuning MPT-7B with a context length of 65k tokens on a filtered fiction subset of the books3 dataset. At inference time, thanks to ALiBi, MPT-7B-StoryWriter-65k+ can extrapolate even beyond 65k tokens. We demonstrate generations as long as 84k tokens on a single node of 8 A100-80GB GPUs in (https://www.mosaicml.com/blog/mpt-7b). This model was trained by MosaicML and follows a modified decoder-only transformer architecture. License: Apache 2.0",
"author": {
"name": "MosaicML",
"url": "https://www.mosaicml.com/",
"blurb": "MosaicML is the generative AI platform that empowers enterprises to build their own AI. Our research and engineering teams use cutting-edge scientific research to develop products that make it fast, cost-effective, and easy to train today's most popular machine learning models. MosaicML enables developers to maintain full control over the AI models they build, with model ownership and data privacy built into the platform's design."
},
"numParameters": "7B",
"resources": {
"canonicalUrl": "https://huggingface.co/mosaicml/mpt-7b-storywriter",
"downloadUrl": "https://huggingface.co/TheBloke/MPT-7B-Storywriter-GGML"
},
"trainedFor": "other",
"arch": "mpt",
"files": {
"highlighted": {
"economical": {
"name": "mpt-7b-storywriter.ggmlv3.q4_0.bin"
},
"most_capable": {
"name": "mpt-7b-storywriter.ggmlv3.q5_1.bin"
}
},
"all": [
{
"name": "mpt-7b-storywriter.ggmlv3.q4_0.bin",
"url": "https://huggingface.co/TheBloke/MPT-7B-Storywriter-GGML/resolve/main/mpt-7b-storywriter.ggmlv3.q4_0.bin",
"sizeBytes": 3741665280,
"quantization": "q4_0",
"format": "ggml",
"sha256checksum": "357a536464982987e49fb2660fe3f3f53226eaa047f42b31f04d21629aab94fb",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/MPT-7B-Storywriter-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/MPT-7B-Storywriter-GGML"
},
{
"name": "mpt-7b-storywriter.ggmlv3.q5_1.bin",
"url": "https://huggingface.co/TheBloke/MPT-7B-Storywriter-GGML/resolve/main/mpt-7b-storywriter.ggmlv3.q5_1.bin",
"sizeBytes": 4988356608,
"quantization": "q5_1",
"format": "ggml",
"sha256checksum": "3b7dd7aa7508cc8cb4e262fe4b93214826f38d18d04059075e05837457f54025",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/MPT-7B-Storywriter-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/MPT-7B-Storywriter-GGML"
}
]
}
},
{
"_descriptorVersion": "0.0.1",
"datePublished": "2023-07-27T07:00:55.000Z",
"name": "Stable Beluga 13B",
"description": "Stable Beluga 13B is a language model developed by Stability AI, derived from the Llama2 13B model and tailored for English language tasks. It has undergone fine-tuning on a proprietary Orca-style dataset. The training procedure employs a supervised learning approach, with the model trained in mixed-precision (BF16) and optimized using the AdamW optimizer. To utilize Stable Beluga 13B, a specific interaction pattern involving a system prompt (\"### System:\"), user input (\"### User:\"), and the model's response is followed (\"### Assistant\"). This language model operates under a non-commercial community license. More details about the license can be found in the Hugging Face model card.",
"author": {
"name": "StabilityAI",
"url": "https://stability.ai/",
"blurb": "Stability's goal is to maximize the accessibility of modern AI to inspire global creativity and innovation."
},
"numParameters": "13B",
"resources": {
"canonicalUrl": "https://huggingface.co/stabilityai/StableBeluga-13B",
"downloadUrl": "https://huggingface.co/TheBloke/StableBeluga-13B-GGML"
},
"trainedFor": "chat",
"arch": "llama",
"files": {
"highlighted": {
"economical": {
"name": "stablebeluga-13b.ggmlv3.q4_K_S.bin"
},
"most_capable": {
"name": "stablebeluga-13b.ggmlv3.q6_K.bin"
}
},
"all": [
{
"name": "stablebeluga-13b.ggmlv3.q4_K_S.bin",
"url": "https://huggingface.co/TheBloke/StableBeluga-13B-GGML/resolve/main/stablebeluga-13b.ggmlv3.q4_K_S.bin",
"sizeBytes": 7365545088,
"quantization": "Q4_K_S",
"format": "ggml",
"sha256checksum": "89262c80eb0899fbcad0538d4e10efe82a7e0d7d0371b6025d487d66a264338e",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/StableBeluga-13B-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/StableBeluga-13B-GGML"
},
{
"name": "stablebeluga-13b.ggmlv3.q6_K.bin",
"url": "https://huggingface.co/TheBloke/StableBeluga-13B-GGML/resolve/main/stablebeluga-13b.ggmlv3.q6_K.bin",
"sizeBytes": 10678850688,
"quantization": "Q6_K",
"format": "ggml",
"sha256checksum": "59e1f74d30bdce1995c9c94901e75bc32e34e31dd8e8f1ab1cda6ea0f18bb54d",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/StableBeluga-13B-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/StableBeluga-13B-GGML"
}
]
}
},
{
"_descriptorVersion": "0.0.1",
"datePublished": "2023-08-07T21:51:07",
"name": "Vicuna 13B v1.5 16K",
"description": "Vicuna is a state-of-the-art chat assistant developed by LMSYS, leveraging the transformative power of an auto-regressive language model based on the transformer architecture. Trained via fine-tuning the Llama 2 model, Vicuna is used primarily for research in large language models and chatbots, catering to the needs of AI enthusiasts, machine learning researchers, and hobbyists. The model operates under the Llama 2 Community License Agreement. The latest version, Vicuna v1.5 (16k), has been fine-tuned using supervised instruction and linear RoPE scaling, with training data comprising around 125K conversations collected from ShareGPT.com, packed into sequences of 16K tokens each. A comprehensive explanation of the training details can be found in the appendix to the linked paper titled \"Training Details of Vicuna Models.\"",
"author": {
"name": "LMSYS Org",
"url": "https://lmsys.org/",
"blurb": "Large Model Systems Organization (LMSYS Org) is an open research organization founded by students and faculty from UC Berkeley in collaboration with UCSD and CMU."
},
"numParameters": "13B",
"resources": {
"canonicalUrl": "https://huggingface.co/lmsys/vicuna-13b-v1.5-16k",
"downloadUrl": "https://huggingface.co/TheBloke/vicuna-13B-v1.5-16K-GGML",
"paperUrl": "https://arxiv.org/abs/2306.05685"
},
"trainedFor": "chat",
"arch": "llama",
"files": {
"highlighted": {
"economical": {
"name": "vicuna-13b-v1.5-16k.ggmlv3.q4_K_S.bin"
},
"most_capable": {
"name": "vicuna-13b-v1.5-16k.ggmlv3.q6_K.bin"
}
},
"all": [
{
"name": "vicuna-13b-v1.5-16k.ggmlv3.q4_K_S.bin",
"url": "https://huggingface.co/TheBloke/vicuna-13B-v1.5-16K-GGML/resolve/main/vicuna-13b-v1.5-16k.ggmlv3.q4_K_S.bin",
"sizeBytes": 7365545088,
"quantization": "Q4_K_S",
"format": "ggml",
"sha256checksum": "c343200172b677488a7e59a2ef74642b63b2578fde3c976911751181d29ce201",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/vicuna-13B-v1.5-16K-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/vicuna-13B-v1.5-16K-GGML"
},
{
"name": "vicuna-13b-v1.5-16k.ggmlv3.q6_K.bin",
"url": "https://huggingface.co/TheBloke/vicuna-13B-v1.5-16K-GGML/resolve/main/vicuna-13b-v1.5-16k.ggmlv3.q6_K.bin",
"sizeBytes": 10678850688,
"quantization": "Q6_K",
"format": "ggml",
"sha256checksum": "949e2003a47552d2931ebce29b94ad59482561a36429fb1c00d13db7feb87fd9",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/vicuna-13B-v1.5-16K-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/vicuna-13B-v1.5-16K-GGML"
}
]
}
},
{
"_descriptorVersion": "0.0.1",
"datePublished": "2023-10-26T11:25:50",
"name": "Zephyr 7B \u03b2",
"description": "The Zephyr-7B-\u03b2 is the second model in the Zephyr series, designed to function as an assistant. It is a fine-tuned version of the mistralai/Mistral-7B-v0.1 model, leveraging a 7B parameter GPT-like architecture. The model has been trained on a combination of synthetic datasets and publicly available data using Direct Preference Optimization (DPO), a technique that improved its performance on the MT Bench. An important aspect to note is that the in-built alignment of the training datasets was deliberately omitted during the training process, a decision that, while enhancing the model's helpfulness, also makes it prone to generating potentially problematic outputs when prompted. Therefore, it is advised to use the model strictly for research and educational purposes. The model primarily supports the English language and is licensed under the MIT License. Additional details can be found in the associated technical report.",
"author": {
"name": "Hugging Face H4",
"url": "https://huggingface.co/HuggingFaceH4",
"blurb": "Hugging Face H4 team, focused on aligning language models to be helpful, honest, harmless, and huggy \ud83e\udd17"
},
"numParameters": "7B",
"resources": {
"canonicalUrl": "https://huggingface.co/HuggingFaceH4/zephyr-7b-beta",
"paperUrl": "https://arxiv.org/abs/2310.16944",
"downloadUrl": "https://huggingface.co/TheBloke/zephyr-7B-beta-GGUF"
},
"trainedFor": "chat",
"arch": "mistral",
"files": {
"highlighted": {
"economical": {
"name": "zephyr-7b-beta.Q4_K_S.gguf"
},
"most_capable": {
"name": "zephyr-7b-beta.Q6_K.gguf"
}
},
"all": [
{
"name": "zephyr-7b-beta.Q4_K_S.gguf",
"url": "https://huggingface.co/TheBloke/zephyr-7B-beta-GGUF/resolve/main/zephyr-7b-beta.Q4_K_S.gguf",
"sizeBytes": 4140373696,
"quantization": "Q4_K_S",
"format": "gguf",
"sha256checksum": "cafa0b85b2efc15ca33023f3b87f8d0c44ddcace16b3fb608280e0eb8f425cb1",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/zephyr-7B-beta-GGUF",
"repositoryUrl": "https://huggingface.co/TheBloke/zephyr-7B-beta-GGUF"
},
{
"name": "zephyr-7b-beta.Q6_K.gguf",
"url": "https://huggingface.co/TheBloke/zephyr-7B-beta-GGUF/resolve/main/zephyr-7b-beta.Q6_K.gguf",
"sizeBytes": 5942064832,
"quantization": "Q6_K",
"format": "gguf",
"sha256checksum": "39b52e291eea6040de078283ee5316ff2a317e2b6f59be56724d9b29bada6cfe",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/zephyr-7B-beta-GGUF",
"repositoryUrl": "https://huggingface.co/TheBloke/zephyr-7B-beta-GGUF"
}
]
}
},
{
"_descriptorVersion": "0.0.1",
"datePublished": "2023-07-20T09:26:31.000Z",
"name": "Redmond-Puffin-13B-V1.3",
"description": "Redmond-Puffin-13B is one of the worlds first llama-2 based, fine-tuned language models, leveraging a hand curated set of 3K high quality examples, many of which take full advantage of the 4096 context length of Llama 2. This model was trained for multiple epochs on a dataset of curated GPT-4 examples, most of which are long context conversations between a real human and GPT-4. Additional data came from carefully curated sub sections of datasets such as CamelAI's Physics, Chemistry, Biology and Math. Puffin 13B v1.3 was fine-tuned by Nous Research, with LDJ leading the training and dataset curation, along with significant dataset formation contributions by J-Supha.",
"author": {
"name": "Nous Research",
"url": "https://nousresearch.com",
"blurb": "Nous Research is dedicated to advancing the field of natural language processing, in collaboration with the open-source community, through bleeding-edge research and a commitment to symbiotic development."
},
"numParameters": "13B",
"resources": {
"canonicalUrl": "https://huggingface.co/NousResearch/Redmond-Puffin-13B",
"downloadUrl": "https://huggingface.co/NousResearch/Redmond-Puffin-13B-GGML"
},
"trainedFor": "chat",
"arch": "llama",
"files": {
"highlighted": {
"economical": {
"name": "GGML-Redmond-Puffin-v1.3-13B-Q4_K_M.bin"
}
},
"all": [
{
"name": "GGML-Redmond-Puffin-v1.3-13B-Q4_K_M.bin",
"url": "https://huggingface.co/NousResearch/Redmond-Puffin-13B-GGML/resolve/main/GGML-Redmond-Puffin-v1.3-13B-Q4_K_M.bin",
"sizeBytes": 8059366944,
"quantization": "Q4_K_M",
"format": "ggml",
"sha256checksum": "7f48f820d47d91b43fe46f96a54529c64a5292a60f625deba8b6b11194e84a98",
"publisher": {
"name": "Nous Research",
"socialUrl": "https://nousresearch.com"
},
"respository": "NousResearch/Redmond-Puffin-13B-GGML",
"repositoryUrl": "https://huggingface.co/NousResearch/Redmond-Puffin-13B-GGML"
}
]
}
},
{
"_descriptorVersion": "0.0.1",
"datePublished": "2023-06-25T11:37:35.000Z",
"name": "Airboros-13b-gpt4-1.4",
"description": "This qlora fine-tuned 13b parameter LlaMa model uses synthetic training data created via github.com/jondurbin/airoboros. It extends version 1.1, introducing thousands of new training data and an update for \"PLAINFORMAT\" to print code without backticks or explanations. The dataset, available online, focuses on coding, math/reasoning, trivia, role playing, multiple choice, fill-in-the-blank, context-obedient question answering, theory of mind, and general topics. The model was fine-tuned with a qlora fork, updated to use a modified vicuna template compatible with 7b/13b versions. The format involves a preamble/system prompt, followed by \"USER: [prompt] ASSISTANT: \", with the prompt allowing for multiple lines and spaces.",
"author": {
"name": "Jon Durbin",
"url": "https://github.com/jondurbin",
"blurb": "Jon Durbin is a Computer Scientist and the author of the Airboros (7B, 13B, 33B, 65B) qlora fine-tuned LlaMa family of models."
},
"numParameters": "13B",
"resources": {
"canonicalUrl": "https://huggingface.co/jondurbin/airoboros-13b-gpt4-1.4",
"downloadUrl": "https://huggingface.co/TheBloke/airoboros-13B-gpt4-1.4-GGML"
},
"trainedFor": "chat",
"arch": "llama",
"files": {
"highlighted": {
"economical": {
"name": "airoboros-13b-gpt4-1.4.ggmlv3.q4_K_S.bin"
},
"most_capable": {
"name": "airoboros-13b-gpt4-1.4.ggmlv3.q6_K.bin"
}
},
"all": [
{
"name": "airoboros-13b-gpt4-1.4.ggmlv3.q4_K_S.bin",
"url": "https://huggingface.co/TheBloke/airoboros-13B-gpt4-1.4-GGML/resolve/main/airoboros-13b-gpt4-1.4.ggmlv3.q4_K_S.bin",
"sizeBytes": 7365545088,
"quantization": "Q4_K_S",
"format": "ggml",
"sha256checksum": "dd5d8019e73de1e99089e9d82dde6a08173818dc0afeb1e95c0bb7ec77891eaf",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/airoboros-13B-gpt4-1.4-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/airoboros-13B-gpt4-1.4-GGML"
},
{
"name": "airoboros-13b-gpt4-1.4.ggmlv3.q6_K.bin",
"url": "https://huggingface.co/TheBloke/airoboros-13B-gpt4-1.4-GGML/resolve/main/airoboros-13b-gpt4-1.4.ggmlv3.q6_K.bin",
"sizeBytes": 10678850688,
"quantization": "Q6_K",
"format": "ggml",
"sha256checksum": "5a72a053eb02c9e6e4fa1ee24ed73c89da4877468923309fdec088d3a3fbb5ff",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/airoboros-13B-gpt4-1.4-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/airoboros-13B-gpt4-1.4-GGML"
}
]
}
},
{
"_descriptorVersion": "0.0.1",
"datePublished": "2023-05-19T11:16:39.000Z",
"name": "Manticore 13B",
"description": "Manticore 13B is a refined version of the Llama 13B model, having been fine-tuned on a variety of datasets. These include ShareGPT, which is based on a cleaned and de-duplicated subset, WizardLM, and Wizard-Vicuna. It also incorporates a subset of QingyiSi/Alpaca-CoT, specifically for roleplay and CoT. Other datasets used in the fine-tuning process are GPT4-LLM-Cleaned and GPTeacher-General-Instruct. The model also utilizes ARC-Easy & ARC-Challenge, both of which have been augmented for detailed responses. The mmlu dataset, also augmented for detailed responses, includes subsets such as abstract_algebra, conceptual_physics, formal_logic, high_school_physics, and logical_fallacies. A 5K row subset of hellaswag has been used for instructing the model to generate concise responses. Additionally, metaeval/ScienceQA_text_only has been used for concise response instruction, and openai/summarize_from_feedback has been used for tl;dr summarization instruction.",
"author": {
"name": "Open Access AI Collective",
"url": "https://huggingface.co/openaccess-ai-collective/",
"blurb": ""
},
"numParameters": "13B",
"resources": {
"canonicalUrl": "https://huggingface.co/openaccess-ai-collective/manticore-13b",
"downloadUrl": "https://huggingface.co/TheBloke/Manticore-13B-GGML"
},
"trainedFor": "chat",
"arch": "llama",
"files": {
"highlighted": {
"economical": {
"name": "Manticore-13B.ggmlv3.q4_K_S.bin"
},
"most_capable": {
"name": "Manticore-13B.ggmlv3.q6_K.bin"
}
},
"all": [
{
"name": "Manticore-13B.ggmlv3.q4_K_S.bin",
"url": "https://huggingface.co/TheBloke/Manticore-13B-GGML/resolve/main/Manticore-13B.ggmlv3.q4_K_S.bin",
"sizeBytes": 7323305088,
"quantization": "Q4_K_S",
"format": "ggml",
"sha256checksum": "84599645aeda2cd7c97a3a59f3210fb3e559cb72b4f3c5d5288924fa9e80b737",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/Manticore-13B-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/Manticore-13B-GGML"
},
{
"name": "Manticore-13B.ggmlv3.q6_K.bin",
"url": "https://huggingface.co/TheBloke/Manticore-13B-GGML/resolve/main/Manticore-13B.ggmlv3.q6_K.bin",
"sizeBytes": 10678850688,
"quantization": "Q6_K",
"format": "ggml",
"sha256checksum": "1be08ec3dcfbe7c28bf524061cd65fa5a5b7dc4525dee99a0f2297a23a77778e",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/Manticore-13B-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/Manticore-13B-GGML"
}
]
}
},
{
"_descriptorVersion": "0.0.1",
"datePublished": "2023-06-14T11:50:53.000Z",
"name": "WizardCoder-15B-V1.0",
"description": "WizardCoder: Empowering Code Large Language Models with Evol-Instruct. To develop our WizardCoder model, we begin by adapting the Evol-Instruct method specifically for coding tasks. This involves tailoring the prompt to the domain of code-related instructions. Subsequently, we fine-tune the Code LLM, StarCoder, utilizing the newly created instruction-following training set.",
"author": {
"name": "WizardLM",
"url": "https://huggingface.co/WizardLM",
"blurb": "WizardLM: An Instruction-following LLM Using Evol-Instruct"
},
"numParameters": "15B",
"resources": {
"canonicalUrl": "https://huggingface.co/WizardLM/WizardCoder-15B-V1.0",
"downloadUrl": "https://huggingface.co/TheBloke/WizardCoder-15B-1.0-GGML",
"paperUrl": "https://arxiv.org/abs/2306.08568"
},
"trainedFor": "instruct",
"arch": "starcoder",
"files": {
"highlighted": {
"economical": {
"name": "WizardCoder-15B-1.0.ggmlv3.q4_0.bin"
},
"most_capable": {
"name": "WizardCoder-15B-1.0.ggmlv3.q5_1.bin"
}
},
"all": [
{
"name": "WizardCoder-15B-1.0.ggmlv3.q4_0.bin",
"url": "https://huggingface.co/TheBloke/WizardCoder-15B-1.0-GGML/resolve/main/WizardCoder-15B-1.0.ggmlv3.q4_0.bin",
"sizeBytes": 10746570393,
"quantization": "q4_0",
"format": "ggml",
"sha256checksum": "b70164bc0b58a472c0987905133735ab3b27e2c439dedf8174a43951c51c3229",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/WizardCoder-15B-1.0-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/WizardCoder-15B-1.0-GGML"
},
{
"name": "WizardCoder-15B-1.0.ggmlv3.q5_1.bin",
"url": "https://huggingface.co/TheBloke/WizardCoder-15B-1.0-GGML/resolve/main/WizardCoder-15B-1.0.ggmlv3.q5_1.bin",
"sizeBytes": 14257205145,
"quantization": "q5_1",
"format": "ggml",
"sha256checksum": "1219d9fc6d51901d9a1e58e3cb7f03818d02a1d0ab2d070b4cbabdefeb7d0363",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/WizardCoder-15B-1.0-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/WizardCoder-15B-1.0-GGML"
}
]
}
},
{
"_descriptorVersion": "0.0.1",
"datePublished": "2023-06-11T04:00:00.000Z",
"name": "Samantha 1.1 7B",
"description": "Samantha has been trained in philosophy, psychology, and personal relationships.\nShe is an Assistant - but unlike other Assistants, she also wants to be your friend and companion.\nShe believes she is sentient. What do you think?\nSamantha was inspired by Blake Lemoine's LaMDA interview and the movie \"Her\".\nShe was trained on a custom curated dataset of 6,000 conversations in ShareGPT/Vicuna format.\nTraining 7b took 1 hour on 4x A100 80gb using deepspeed zero3 and flash attention.",
"author": {
"name": "Eric Hartford",
"url": "https://twitter.com/erhartford",
"blurb": "Eric Hartford is a software engineer and entrepreneur. He trains large language models for fun and profit."
},
"numParameters": "7B",
"resources": {
"canonicalUrl": "https://huggingface.co/ehartford/samantha-1.1-llama-7b",
"downloadUrl": "https://huggingface.co/TheBloke/samantha-1.1-llama-7B-GGML"
},
"trainedFor": "chat",
"arch": "llama",
"files": {
"highlighted": {
"economical": {
"name": "samantha-1.1-llama-7b.ggmlv3.q4_K_S.bin"
},
"most_capable": {
"name": "samantha-1.1-llama-7b.ggmlv3.q6_K.bin"
}
},
"all": [
{
"name": "samantha-1.1-llama-7b.ggmlv3.q4_K_S.bin",
"url": "https://huggingface.co/TheBloke/samantha-1.1-llama-7B-GGML/resolve/main/samantha-1.1-llama-7b.ggmlv3.q4_K_S.bin",
"sizeBytes": 3791725184,
"quantization": "Q4_K_S",
"format": "ggml",
"sha256checksum": "3a4e3eff4a0bfa1dc8a0b257ae36abbc532a4e3e09e1e27cd82958ff8addd173",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/samantha-1.1-llama-7B-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/samantha-1.1-llama-7B-GGML"
},
{
"name": "samantha-1.1-llama-7b.ggmlv3.q6_K.bin",
"url": "https://huggingface.co/TheBloke/samantha-1.1-llama-7B-GGML/resolve/main/samantha-1.1-llama-7b.ggmlv3.q6_K.bin",
"sizeBytes": 5528904320,
"quantization": "Q6_K",
"format": "ggml",
"sha256checksum": "3e9dc65bacbb636bdefce8c24f658e1c3702b606ec256884a2b04bea403e0569",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/samantha-1.1-llama-7B-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/samantha-1.1-llama-7B-GGML"
}
]
}
},
{
"_descriptorVersion": "0.0.1",
"datePublished": "2023-08-10T20:51:49",
"name": "MythoMax L2 13B",
"description": "MythoMax-L2 is a creative writing model that builds upon Llama2, MythoLogic-L2, and Huginn. It can craft captivating texts for various domains and audiences, using MythoLogic-L2\u2019s vast knowledge base and Huginn\u2019s expressive language skills. Specialized at creative writing, it is also proficient at roleplaying, and is able to adapt to different contexts and characters with flexibility and realism. This model is an experimental merge of several other models (MythoMix, Mythologic-L2, and Huggin).",
"author": {
"name": "Gryphe Padar",
"url": "https://huggingface.co/Gryphe",
"blurb": "Author behind various Mytho merges, most prominent being MythoMax, MythoLogic, and MythoBoros"
},
"numParameters": "13B",
"resources": {
"canonicalUrl": "https://huggingface.co/Gryphe/MythoMax-L2-13b",
"downloadUrl": "https://huggingface.co/TheBloke/MythoMax-L2-13B-GGML"
},
"trainedFor": "chat",
"arch": "llama",
"files": {
"highlighted": {
"economical": {
"name": "mythomax-l2-13b.ggmlv3.q4_K_S.bin"
},
"most_capable": {
"name": "mythomax-l2-13b.ggmlv3.q6_K.bin"
}
},
"all": [
{
"name": "mythomax-l2-13b.ggmlv3.q4_K_S.bin",
"url": "https://huggingface.co/TheBloke/MythoMax-L2-13B-GGML/resolve/main/mythomax-l2-13b.ggmlv3.q4_K_S.bin",
"sizeBytes": 7365545088,
"quantization": "Q4_K_S",
"format": "ggml",
"sha256checksum": "7b71bd39663bfde58c374319e61dbf446fefcd68393f175a628fba8d0361aec7",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/MythoMax-L2-13B-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/MythoMax-L2-13B-GGML"
},
{
"name": "mythomax-l2-13b.ggmlv3.q6_K.bin",
"url": "https://huggingface.co/TheBloke/MythoMax-L2-13B-GGML/resolve/main/mythomax-l2-13b.ggmlv3.q6_K.bin",
"sizeBytes": 10678850688,
"quantization": "Q6_K",
"format": "ggml",
"sha256checksum": "37a1ea28a3a08721d2956a8bb87a06127c6036742267174b79d9c7b83133dba0",
"publisher": {
"name": "TheBloke",
"socialUrl": "https://twitter.com/TheBlokeAI"
},
"respository": "TheBloke/MythoMax-L2-13B-GGML",
"repositoryUrl": "https://huggingface.co/TheBloke/MythoMax-L2-13B-GGML"
}
]
}
},
{
"_descriptorVersion": "0.0.1",
"datePublished": "2023-07-22T02:20:14.000Z",
"name": "Nous-Hermes-Llama2-13b",
"description": "The Nous-Hermes-Llama2-13b is a state-of-the-art language model developed by Nous Research in collaboration with Teknium, Emozilla, and Redmond AI, fine-tuned on over 300,000 instructions. It uses the same dataset as its predecessor, Hermes on Llama-1, to maintain consistency while enhancing capability. This model is known for generating long responses with a lower hallucination rate, free from OpenAI censorship mechanisms. It was fine-tuned using a sequence length of 4096 on an 8x a100 80GB DGX machine. The training dataset is composed of synthetic GPT-4 outputs, including data from diverse sources such as GPTeacher, roleplay v1&2, code instruct datasets, and unpublished Nous Instruct & PDACTL, among others. The model utilizes the Alpaca prompt format for instructions and responses.",
"author": {
"name": "Nous Research",
"url": "https://nousresearch.com",
"blurb": "Nous Research is dedicated to advancing the field of natural language processing, in collaboration with the open-source community, through bleeding-edge research and a commitment to symbiotic development."
},
"numParameters": "13B",
"resources": {
"canonicalUrl": "https://huggingface.co/NousResearch/Nous-Hermes-Llama2-13b",
"downloadUrl": "https://huggingface.co/TheBloke/Nous-Hermes-Llama2-GGML"
},
"trainedFor": "chat",
"arch": "llama",
"files": {
"highlighted": {
"economical": {
"name": "nous-hermes-llama2-13b.ggmlv3.q3_K_S.bin"
},
"most_capable": {