Datasets:

Languages:
English
ArXiv:
License:
Daniel O'Connell commited on
Commit
83a96a0
·
1 Parent(s): b27cc8f

update metadata

Browse files
README.md CHANGED
@@ -8,219 +8,6 @@ task_categories:
8
  - question-answering
9
  pretty_name: alignment-research-dataset
10
  dataset_info:
11
- - config_name: agentmodels
12
- features:
13
- - name: id
14
- dtype: string
15
- - name: source
16
- dtype: string
17
- - name: title
18
- dtype: string
19
- - name: text
20
- dtype: large_string
21
- - name: url
22
- dtype: string
23
- - name: date_published
24
- dtype: string
25
- - name: authors
26
- sequence: string
27
- - name: summary
28
- sequence: string
29
- - name: source_type
30
- dtype: string
31
- - name: book_title
32
- dtype: string
33
- splits:
34
- - name: train
35
- num_bytes: 386001
36
- num_examples: 21
37
- download_size: 802384
38
- dataset_size: 386001
39
- - config_name: aiimpacts
40
- features:
41
- - name: id
42
- dtype: string
43
- - name: source
44
- dtype: string
45
- - name: title
46
- dtype: string
47
- - name: text
48
- dtype: large_string
49
- - name: url
50
- dtype: string
51
- - name: date_published
52
- dtype: string
53
- - name: authors
54
- sequence: string
55
- - name: summary
56
- sequence: string
57
- - name: source_type
58
- dtype: string
59
- splits:
60
- - name: train
61
- num_bytes: 2213777
62
- num_examples: 253
63
- download_size: 2263756
64
- dataset_size: 2213777
65
- - config_name: aipulse.org
66
- features:
67
- - name: id
68
- dtype: string
69
- - name: source
70
- dtype: string
71
- - name: title
72
- dtype: string
73
- - name: text
74
- dtype: large_string
75
- - name: url
76
- dtype: string
77
- - name: date_published
78
- dtype: string
79
- - name: authors
80
- sequence: string
81
- - name: summary
82
- sequence: string
83
- - name: source_type
84
- dtype: string
85
- splits:
86
- - name: train
87
- num_bytes: 651777
88
- num_examples: 23
89
- download_size: 657165
90
- dataset_size: 651777
91
- - config_name: aisafety.camp
92
- features:
93
- - name: id
94
- dtype: string
95
- - name: source
96
- dtype: string
97
- - name: title
98
- dtype: string
99
- - name: text
100
- dtype: large_string
101
- - name: url
102
- dtype: string
103
- - name: date_published
104
- dtype: string
105
- - name: authors
106
- sequence: string
107
- - name: summary
108
- sequence: string
109
- - name: source_type
110
- dtype: string
111
- splits:
112
- - name: train
113
- num_bytes: 83937
114
- num_examples: 9
115
- download_size: 85765
116
- dataset_size: 83937
117
- - config_name: aisafety.info
118
- features:
119
- - name: id
120
- dtype: string
121
- - name: source
122
- dtype: string
123
- - name: title
124
- dtype: string
125
- - name: text
126
- dtype: large_string
127
- - name: url
128
- dtype: string
129
- - name: date_published
130
- dtype: string
131
- - name: authors
132
- sequence: string
133
- - name: summary
134
- sequence: string
135
- - name: source_type
136
- dtype: string
137
- splits:
138
- - name: train
139
- num_bytes: 890524
140
- num_examples: 315
141
- download_size: 929839
142
- dataset_size: 890524
143
- - config_name: alignment_newsletter
144
- features:
145
- - name: id
146
- dtype: string
147
- - name: source
148
- dtype: string
149
- - name: title
150
- dtype: string
151
- - name: text
152
- dtype: large_string
153
- - name: url
154
- dtype: string
155
- - name: date_published
156
- dtype: string
157
- - name: authors
158
- sequence: string
159
- - name: summary
160
- sequence: string
161
- - name: source_type
162
- dtype: string
163
- - name: venue
164
- dtype: string
165
- - name: newsletter_category
166
- dtype: string
167
- - name: highlight
168
- dtype: int32
169
- - name: newsletter_number
170
- dtype: string
171
- - name: summarizer
172
- dtype: string
173
- - name: opinion
174
- dtype: string
175
- - name: prerequisites
176
- dtype: string
177
- - name: read_more
178
- dtype: string
179
- splits:
180
- - name: train
181
- num_bytes: 3193636
182
- num_examples: 1504
183
- download_size: 3576332
184
- dataset_size: 3193636
185
- - config_name: alignmentforum
186
- features:
187
- - name: id
188
- dtype: string
189
- - name: source
190
- dtype: string
191
- - name: title
192
- dtype: string
193
- - name: text
194
- dtype: large_string
195
- - name: url
196
- dtype: string
197
- - name: date_published
198
- dtype: string
199
- - name: authors
200
- sequence: string
201
- - name: summary
202
- sequence: string
203
- - name: source_type
204
- dtype: string
205
- - name: karma
206
- dtype: int32
207
- - name: votes
208
- dtype: int32
209
- - name: words
210
- dtype: int32
211
- - name: comment_count
212
- dtype: int32
213
- - name: tags
214
- sequence: string
215
- - name: modified_at
216
- dtype: string
217
- splits:
218
- - name: train
219
- num_bytes: 74882445
220
- num_examples: 2698
221
- download_size: 76214133
222
- dataset_size: 74882445
223
- - config_name: all
224
  features:
225
  - name: id
226
  dtype: string
@@ -254,22 +41,6 @@ dataset_info:
254
  sequence: string
255
  - name: modified_at
256
  dtype: string
257
- - name: venue
258
- dtype: string
259
- - name: newsletter_category
260
- dtype: string
261
- - name: highlight
262
- dtype: int32
263
- - name: newsletter_number
264
- dtype: string
265
- - name: summarizer
266
- dtype: string
267
- - name: opinion
268
- dtype: string
269
- - name: prerequisites
270
- dtype: string
271
- - name: read_more
272
- dtype: string
273
  - name: alias
274
  dtype: string
275
  - name: data_last_modified
@@ -290,676 +61,15 @@ dataset_info:
290
  sequence:
291
  - name: title
292
  dtype: string
293
- - name: docx_name
294
- dtype: string
295
  - name: description
296
  dtype: string
297
- - name: filename
298
- dtype: string
299
- splits:
300
- - name: train
301
- num_bytes: 312674982
302
- num_examples: 9687
303
- download_size: 359804141
304
- dataset_size: 312674982
305
- - config_name: arbital
306
- features:
307
- - name: id
308
- dtype: string
309
- - name: source
310
- dtype: string
311
- - name: title
312
- dtype: string
313
- - name: text
314
- dtype: large_string
315
- - name: url
316
- dtype: string
317
- - name: date_published
318
- dtype: string
319
- - name: authors
320
- sequence: string
321
- - name: summary
322
- sequence: string
323
- - name: source_type
324
- dtype: string
325
- - name: alias
326
- dtype: string
327
- - name: tags
328
- sequence: string
329
- splits:
330
- - name: train
331
- num_bytes: 3592118
332
- num_examples: 730
333
- download_size: 4077961
334
- dataset_size: 3592118
335
- - config_name: arxiv
336
- features:
337
- - name: id
338
- dtype: string
339
- - name: source
340
- dtype: string
341
- - name: title
342
- dtype: string
343
- - name: text
344
- dtype: large_string
345
- - name: url
346
- dtype: string
347
- - name: date_published
348
- dtype: string
349
- - name: authors
350
- sequence: string
351
- - name: summary
352
- sequence: string
353
- - name: source_type
354
- dtype: string
355
- - name: data_last_modified
356
- dtype: string
357
- - name: abstract
358
- dtype: string
359
- - name: author_comment
360
- dtype: string
361
- - name: journal_ref
362
- dtype: string
363
- - name: doi
364
- dtype: string
365
- - name: primary_category
366
- dtype: string
367
- - name: categories
368
- sequence: string
369
- splits:
370
- - name: train
371
- num_bytes: 140115174
372
- num_examples: 1144
373
- download_size: 155841593
374
- dataset_size: 140115174
375
- - config_name: audio_transcripts
376
- features:
377
- - name: id
378
- dtype: string
379
- - name: source
380
- dtype: string
381
- - name: title
382
- dtype: string
383
- - name: text
384
- dtype: large_string
385
- - name: url
386
- dtype: string
387
- - name: date_published
388
- dtype: string
389
- - name: authors
390
- sequence: string
391
- - name: summary
392
- sequence: string
393
- - name: source_type
394
- dtype: string
395
- splits:
396
- - name: train
397
- num_bytes: 1312631
398
- num_examples: 37
399
- download_size: 1335397
400
- dataset_size: 1312631
401
- - config_name: carado.moe
402
- features:
403
- - name: id
404
- dtype: string
405
- - name: source
406
- dtype: string
407
- - name: title
408
- dtype: string
409
- - name: text
410
- dtype: large_string
411
- - name: url
412
- dtype: string
413
- - name: date_published
414
- dtype: string
415
- - name: authors
416
- sequence: string
417
- - name: summary
418
- sequence: string
419
- - name: source_type
420
- dtype: string
421
- splits:
422
- - name: train
423
- num_bytes: 1048269
424
- num_examples: 248
425
- download_size: 1083144
426
- dataset_size: 1048269
427
- - config_name: cold_takes
428
- features:
429
- - name: id
430
- dtype: string
431
- - name: source
432
- dtype: string
433
- - name: title
434
- dtype: string
435
- - name: text
436
- dtype: large_string
437
- - name: url
438
- dtype: string
439
- - name: date_published
440
- dtype: string
441
- - name: authors
442
- sequence: string
443
- - name: summary
444
- sequence: string
445
- - name: source_type
446
- dtype: string
447
- splits:
448
- - name: train
449
- num_bytes: 1633384
450
- num_examples: 102
451
- download_size: 2560052
452
- dataset_size: 1633384
453
- - config_name: deepmind_blog
454
- features:
455
- - name: id
456
- dtype: string
457
- - name: source
458
- dtype: string
459
- - name: title
460
- dtype: string
461
- - name: text
462
- dtype: large_string
463
- - name: url
464
- dtype: string
465
- - name: date_published
466
- dtype: string
467
- - name: authors
468
- sequence: string
469
- - name: summary
470
- sequence: string
471
- - name: source_type
472
- dtype: string
473
- splits:
474
- - name: train
475
- num_bytes: 111439
476
- num_examples: 10
477
- download_size: 112339
478
- dataset_size: 111439
479
- - config_name: distill
480
- features:
481
- - name: id
482
- dtype: string
483
- - name: source
484
- dtype: string
485
- - name: title
486
- dtype: string
487
- - name: text
488
- dtype: large_string
489
- - name: url
490
- dtype: string
491
- - name: date_published
492
- dtype: string
493
- - name: authors
494
- sequence: string
495
- - name: summary
496
- sequence: string
497
- - name: source_type
498
- dtype: string
499
- - name: abstract
500
- dtype: string
501
- - name: journal_ref
502
- dtype: string
503
- - name: doi
504
- dtype: string
505
- - name: bibliography_bib
506
- sequence:
507
- - name: title
508
- dtype: string
509
- splits:
510
- - name: train
511
- num_bytes: 2251584
512
- num_examples: 51
513
- download_size: 2435649
514
- dataset_size: 2251584
515
- - config_name: eaforum
516
- features:
517
- - name: id
518
- dtype: string
519
- - name: source
520
- dtype: string
521
- - name: title
522
- dtype: string
523
- - name: text
524
- dtype: large_string
525
- - name: url
526
- dtype: string
527
- - name: date_published
528
- dtype: string
529
- - name: authors
530
- sequence: string
531
- - name: summary
532
- sequence: string
533
- - name: source_type
534
- dtype: string
535
- - name: karma
536
- dtype: int32
537
- - name: votes
538
- dtype: int32
539
- - name: words
540
- dtype: int32
541
- - name: comment_count
542
- dtype: int32
543
- - name: tags
544
- sequence: string
545
- - name: modified_at
546
- dtype: string
547
- splits:
548
- - name: train
549
- num_bytes: 26436402
550
- num_examples: 1557
551
- download_size: 26902036
552
- dataset_size: 26436402
553
- - config_name: gdocs
554
- features:
555
- - name: id
556
- dtype: string
557
- - name: source
558
- dtype: string
559
- - name: title
560
- dtype: string
561
- - name: text
562
- dtype: large_string
563
- - name: url
564
- dtype: string
565
- - name: date_published
566
- dtype: string
567
- - name: authors
568
- sequence: string
569
- - name: summary
570
- sequence: string
571
- - name: source_type
572
- dtype: string
573
- - name: docx_name
574
- dtype: string
575
- splits:
576
- - name: train
577
- num_bytes: 449996
578
- num_examples: 12
579
- download_size: 454227
580
- dataset_size: 449996
581
- - config_name: generative.ink
582
- features:
583
- - name: id
584
- dtype: string
585
- - name: source
586
- dtype: string
587
- - name: title
588
- dtype: string
589
- - name: text
590
- dtype: large_string
591
- - name: url
592
- dtype: string
593
- - name: date_published
594
- dtype: string
595
- - name: authors
596
- sequence: string
597
- - name: summary
598
- sequence: string
599
- - name: source_type
600
- dtype: string
601
- splits:
602
- - name: train
603
- num_bytes: 446939
604
- num_examples: 19
605
- download_size: 449941
606
- dataset_size: 446939
607
- - config_name: gwern_blog
608
- features:
609
- - name: id
610
- dtype: string
611
- - name: source
612
- dtype: string
613
- - name: title
614
- dtype: string
615
- - name: text
616
- dtype: large_string
617
- - name: url
618
- dtype: string
619
- - name: date_published
620
- dtype: string
621
- - name: authors
622
- sequence: string
623
- - name: summary
624
- sequence: string
625
- - name: source_type
626
- dtype: string
627
- - name: description
628
- dtype: string
629
- splits:
630
- - name: train
631
- num_bytes: 686185
632
- num_examples: 7
633
- download_size: 692693
634
- dataset_size: 686185
635
- - config_name: html_articles
636
- features:
637
- - name: id
638
- dtype: string
639
- - name: source
640
- dtype: string
641
- - name: title
642
- dtype: string
643
- - name: text
644
- dtype: large_string
645
- - name: url
646
- dtype: string
647
- - name: date_published
648
- dtype: string
649
- - name: authors
650
- sequence: string
651
- - name: summary
652
- sequence: string
653
- - name: source_type
654
- dtype: string
655
- splits:
656
- - name: train
657
- num_bytes: 6356426
658
- num_examples: 132
659
- download_size: 6423995
660
- dataset_size: 6356426
661
- - config_name: importai
662
- features:
663
- - name: id
664
- dtype: string
665
- - name: source
666
- dtype: string
667
- - name: title
668
- dtype: string
669
- - name: text
670
- dtype: large_string
671
- - name: url
672
- dtype: string
673
- - name: date_published
674
- dtype: string
675
- - name: authors
676
- sequence: string
677
- - name: summary
678
- sequence: string
679
- - name: source_type
680
- dtype: string
681
- splits:
682
- - name: train
683
- num_bytes: 243316
684
- num_examples: 15
685
- download_size: 246831
686
- dataset_size: 243316
687
- - config_name: jsteinhardt_blog
688
- features:
689
- - name: id
690
- dtype: string
691
- - name: source
692
- dtype: string
693
- - name: title
694
- dtype: string
695
- - name: text
696
- dtype: large_string
697
- - name: url
698
- dtype: string
699
- - name: date_published
700
- dtype: string
701
- - name: authors
702
- sequence: string
703
- - name: summary
704
- sequence: string
705
- - name: source_type
706
- dtype: string
707
- splits:
708
- - name: train
709
- num_bytes: 374939
710
- num_examples: 39
711
- download_size: 382468
712
- dataset_size: 374939
713
- - config_name: lesswrong
714
- features:
715
- - name: id
716
- dtype: string
717
- - name: source
718
- dtype: string
719
- - name: title
720
- dtype: string
721
- - name: text
722
- dtype: large_string
723
- - name: url
724
- dtype: string
725
- - name: date_published
726
- dtype: string
727
- - name: authors
728
- sequence: string
729
- - name: summary
730
- sequence: string
731
- - name: source_type
732
- dtype: string
733
- - name: karma
734
- dtype: int32
735
- - name: votes
736
- dtype: int32
737
- - name: words
738
- dtype: int32
739
- - name: comment_count
740
- dtype: int32
741
- - name: tags
742
- sequence: string
743
- - name: modified_at
744
- dtype: string
745
- splits:
746
- - name: train
747
- num_bytes: 96928036
748
- num_examples: 4163
749
- download_size: 98817334
750
- dataset_size: 96928036
751
- - config_name: miri
752
- features:
753
- - name: id
754
- dtype: string
755
- - name: source
756
- dtype: string
757
- - name: title
758
- dtype: string
759
- - name: text
760
- dtype: large_string
761
- - name: url
762
- dtype: string
763
- - name: date_published
764
- dtype: string
765
- - name: authors
766
- sequence: string
767
- - name: summary
768
- sequence: string
769
- - name: source_type
770
- dtype: string
771
- splits:
772
- - name: train
773
- num_bytes: 5685488
774
- num_examples: 492
775
- download_size: 5799863
776
- dataset_size: 5685488
777
- - config_name: ml_safety_newsletter
778
- features:
779
- - name: id
780
- dtype: string
781
- - name: source
782
- dtype: string
783
- - name: title
784
- dtype: string
785
- - name: text
786
- dtype: large_string
787
- - name: url
788
- dtype: string
789
- - name: date_published
790
- dtype: string
791
- - name: authors
792
- sequence: string
793
- - name: summary
794
- sequence: string
795
- - name: source_type
796
- dtype: string
797
- splits:
798
- - name: train
799
- num_bytes: 110320
800
- num_examples: 9
801
- download_size: 112191
802
- dataset_size: 110320
803
- - config_name: nonarxiv_papers
804
- features:
805
- - name: id
806
- dtype: string
807
- - name: source
808
- dtype: string
809
- - name: title
810
- dtype: string
811
- - name: text
812
- dtype: large_string
813
- - name: url
814
- dtype: string
815
- - name: date_published
816
- dtype: string
817
- - name: authors
818
- sequence: string
819
- - name: summary
820
- sequence: string
821
- - name: source_type
822
- dtype: string
823
- - name: abstract
824
- dtype: string
825
- splits:
826
- - name: train
827
- num_bytes: 10099354
828
- num_examples: 200
829
- download_size: 10146094
830
- dataset_size: 10099354
831
- - config_name: pdfs
832
- features:
833
- - name: id
834
- dtype: string
835
- - name: source
836
- dtype: string
837
- - name: title
838
- dtype: string
839
- - name: text
840
- dtype: large_string
841
- - name: url
842
- dtype: string
843
- - name: date_published
844
- dtype: string
845
- - name: authors
846
- sequence: string
847
- - name: summary
848
- sequence: string
849
- - name: source_type
850
- dtype: string
851
- splits:
852
- - name: train
853
- num_bytes: 20818258
854
- num_examples: 233
855
- download_size: 21813824
856
- dataset_size: 20818258
857
- - config_name: qualiacomputing
858
- features:
859
- - name: id
860
- dtype: string
861
- - name: source
862
- dtype: string
863
- - name: title
864
- dtype: string
865
- - name: text
866
- dtype: large_string
867
- - name: url
868
- dtype: string
869
- - name: date_published
870
- dtype: string
871
- - name: authors
872
- sequence: string
873
- - name: summary
874
- sequence: string
875
- - name: source_type
876
- dtype: string
877
- splits:
878
- - name: train
879
- num_bytes: 4397009
880
- num_examples: 302
881
- download_size: 4458760
882
- dataset_size: 4397009
883
- - config_name: reports
884
- features:
885
- - name: id
886
- dtype: string
887
- - name: source
888
- dtype: string
889
- - name: title
890
- dtype: string
891
- - name: text
892
- dtype: large_string
893
- - name: url
894
- dtype: string
895
- - name: date_published
896
- dtype: string
897
- - name: authors
898
- sequence: string
899
- - name: summary
900
- sequence: string
901
- - name: source_type
902
- dtype: string
903
- - name: filename
904
- dtype: string
905
- splits:
906
- - name: train
907
- num_bytes: 5508374
908
- num_examples: 74
909
- download_size: 5528295
910
- dataset_size: 5508374
911
- - config_name: vkrakovna_blog
912
- features:
913
- - name: id
914
- dtype: string
915
- - name: source
916
- dtype: string
917
- - name: title
918
- dtype: string
919
- - name: text
920
- dtype: large_string
921
- - name: url
922
- dtype: string
923
- - name: date_published
924
- dtype: string
925
- - name: authors
926
- sequence: string
927
- - name: summary
928
- sequence: string
929
- - name: source_type
930
- dtype: string
931
- splits:
932
- - name: train
933
- num_bytes: 301626
934
- num_examples: 46
935
- download_size: 310543
936
- dataset_size: 301626
937
- - config_name: yudkowsky_blog
938
- features:
939
- - name: id
940
- dtype: string
941
- - name: source
942
- dtype: string
943
- - name: title
944
- dtype: string
945
- - name: text
946
- dtype: large_string
947
- - name: url
948
- dtype: string
949
- - name: date_published
950
- dtype: string
951
- - name: authors
952
- sequence: string
953
- - name: summary
954
- sequence: string
955
- - name: source_type
956
- dtype: string
957
  splits:
958
  - name: train
959
- num_bytes: 572669
960
- num_examples: 23
961
- download_size: 578093
962
- dataset_size: 572669
963
  ---
964
  # AI Alignment Research Dataset
965
 
@@ -983,19 +93,14 @@ The following list of sources may change and items may be renamed:
983
  - [deepmind_blog](https://deepmindsafetyresearch.medium.com/)
984
  - [distill](https://distill.pub/)
985
  - [eaforum](https://forum.effectivealtruism.org/) - selected posts
986
- - gdocs
987
- - gdrive_ebooks - books include [Superintelligence](https://www.goodreads.com/book/show/20527133-superintelligence), [Human Compatible](https://www.goodreads.com/book/show/44767248-human-compatible), [Life 3.0](https://www.goodreads.com/book/show/34272565-life-3-0), [The Precipice](https://www.goodreads.com/book/show/50485582-the-precipice), and others
988
  - [generative.ink](https://generative.ink/posts/)
989
  - [gwern_blog](https://gwern.net/)
990
  - [importai](https://importai.substack.com)
991
  - [jsteinhardt_blog](https://jsteinhardt.wordpress.com/)
992
  - [lesswrong](https://www.lesswrong.com/) - selected posts
993
- - markdown.ebooks
994
  - [miri](https://intelligence.org/) - MIRI
995
  - [ml_safety_newsletter](https://newsletter.mlsafety.org)
996
- - nonarxiv_papers - other alignment research papers
997
- - [qualiacomputing](https://qualiacomputing.com/)
998
- - reports
999
  - [vkrakovna_blog](https://vkrakovna.wordpress.com)
1000
  - [waitbutwhy](https://waitbutwhy.com/)
1001
  - [yudkowsky_blog](https://www.yudkowsky.net/)
@@ -1049,4 +154,4 @@ This README contains info about the number of rows and their features which shou
1049
 
1050
  For more information, here is the [paper](https://arxiv.org/abs/2206.02841) and [LessWrong](https://www.lesswrong.com/posts/FgjcHiWvADgsocE34/a-descriptive-not-prescriptive-overview-of-current-ai) post. Please use the following citation when using the dataset:
1051
 
1052
- Kirchner, J. H., Smith, L., Thibodeau, J., McDonnell, K., and Reynolds, L. "Understanding AI alignment research: A Systematic Analysis." arXiv preprint arXiv:2022.4338861 (2022).
 
8
  - question-answering
9
  pretty_name: alignment-research-dataset
10
  dataset_info:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
11
  features:
12
  - name: id
13
  dtype: string
 
41
  sequence: string
42
  - name: modified_at
43
  dtype: string
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
44
  - name: alias
45
  dtype: string
46
  - name: data_last_modified
 
61
  sequence:
62
  - name: title
63
  dtype: string
 
 
64
  - name: description
65
  dtype: string
66
+ config_name: all
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
67
  splits:
68
  - name: train
69
+ num_bytes: 411471024
70
+ num_examples: 14163
71
+ download_size: 423573134
72
+ dataset_size: 411471024
73
  ---
74
  # AI Alignment Research Dataset
75
 
 
93
  - [deepmind_blog](https://deepmindsafetyresearch.medium.com/)
94
  - [distill](https://distill.pub/)
95
  - [eaforum](https://forum.effectivealtruism.org/) - selected posts
 
 
96
  - [generative.ink](https://generative.ink/posts/)
97
  - [gwern_blog](https://gwern.net/)
98
  - [importai](https://importai.substack.com)
99
  - [jsteinhardt_blog](https://jsteinhardt.wordpress.com/)
100
  - [lesswrong](https://www.lesswrong.com/) - selected posts
 
101
  - [miri](https://intelligence.org/) - MIRI
102
  - [ml_safety_newsletter](https://newsletter.mlsafety.org)
103
+ - special_docs - This is a catch-all of various article, papers etc. that don't come from a single source.
 
 
104
  - [vkrakovna_blog](https://vkrakovna.wordpress.com)
105
  - [waitbutwhy](https://waitbutwhy.com/)
106
  - [yudkowsky_blog](https://www.yudkowsky.net/)
 
154
 
155
  For more information, here is the [paper](https://arxiv.org/abs/2206.02841) and [LessWrong](https://www.lesswrong.com/posts/FgjcHiWvADgsocE34/a-descriptive-not-prescriptive-overview-of-current-ai) post. Please use the following citation when using the dataset:
156
 
157
+ Kirchner, J. H., Smith, L., Thibodeau, J., McDonnell, K., and Reynolds, L. "Understanding AI alignment research: A Systematic Analysis." arXiv preprint arXiv:2022.4338861 (2022).
alignment-research-dataset.py CHANGED
@@ -100,8 +100,12 @@ DATASOURCES = {
100
  'agentmodels': {
101
  'book_title': Value(dtype='string'),
102
  },
 
 
 
 
 
103
  'aiimpacts': {},
104
- 'aipulse.org': {},
105
  'aisafety.camp': {},
106
  'aisafety.info': {},
107
  'alignmentforum': {
@@ -112,16 +116,6 @@ DATASOURCES = {
112
  'tags': Sequence(feature=Value(dtype='string')),
113
  'modified_at': Value(dtype='string'),
114
  },
115
- 'alignment_newsletter': {
116
- 'venue': Value(dtype='string'),
117
- 'newsletter_category': Value(dtype='string'),
118
- 'highlight': Value(dtype='int32'),
119
- 'newsletter_number': Value(dtype='string'),
120
- 'summarizer': Value(dtype='string'),
121
- 'opinion': Value(dtype='string'),
122
- 'prerequisites': Value(dtype='string'),
123
- 'read_more': Value(dtype='string'),
124
- },
125
  'arbital': {
126
  'alias': Value(dtype='string'),
127
  'tags': Sequence(feature=Value(dtype='string')),
@@ -135,10 +129,10 @@ DATASOURCES = {
135
  'primary_category': Value(dtype='string'),
136
  'categories': Sequence(feature=Value(dtype='string'), length=-1),
137
  },
138
- 'audio_transcripts': {},
139
  'carado.moe': {},
140
  'cold_takes': {},
141
  'deepmind_blog': {},
 
142
  'distill': {
143
  'abstract': Value(dtype='string'),
144
  'journal_ref': Value(dtype='string'),
@@ -153,14 +147,11 @@ DATASOURCES = {
153
  'tags': Sequence(feature=Value(dtype='string')),
154
  'modified_at': Value(dtype='string'),
155
  },
156
- 'gdocs': {
157
- 'docx_name': Value(dtype='string')
158
- },
159
  'generative.ink': {},
160
  'gwern_blog': {
161
  'description': Value(dtype='string'),
162
  },
163
- 'html_articles': {},
164
  'importai': {},
165
  'jsteinhardt_blog': {},
166
  'lesswrong': {
@@ -173,14 +164,9 @@ DATASOURCES = {
173
  },
174
  'miri': {},
175
  'ml_safety_newsletter': {},
176
- 'nonarxiv_papers': {
177
- 'abstract': Value(dtype='string'),
178
- },
179
- 'pdfs': {},
180
- 'qualiacomputing': {},
181
- 'reports': {
182
- 'filename': Value(dtype='string')
183
- },
184
  'vkrakovna_blog': {},
185
  'yudkowsky_blog': {},
186
  }
 
100
  'agentmodels': {
101
  'book_title': Value(dtype='string'),
102
  },
103
+ 'ai_alignment_playlist': {},
104
+ 'ai_explained': {},
105
+ 'ai_safety_reading_group': {},
106
+ 'ai_safety_talks': {},
107
+ 'ai_tech_tu_delft': {},
108
  'aiimpacts': {},
 
109
  'aisafety.camp': {},
110
  'aisafety.info': {},
111
  'alignmentforum': {
 
116
  'tags': Sequence(feature=Value(dtype='string')),
117
  'modified_at': Value(dtype='string'),
118
  },
 
 
 
 
 
 
 
 
 
 
119
  'arbital': {
120
  'alias': Value(dtype='string'),
121
  'tags': Sequence(feature=Value(dtype='string')),
 
129
  'primary_category': Value(dtype='string'),
130
  'categories': Sequence(feature=Value(dtype='string'), length=-1),
131
  },
 
132
  'carado.moe': {},
133
  'cold_takes': {},
134
  'deepmind_blog': {},
135
+ 'deepmind_technical_blog': {},
136
  'distill': {
137
  'abstract': Value(dtype='string'),
138
  'journal_ref': Value(dtype='string'),
 
147
  'tags': Sequence(feature=Value(dtype='string')),
148
  'modified_at': Value(dtype='string'),
149
  },
150
+ 'eleuther.ai': {},
 
 
151
  'generative.ink': {},
152
  'gwern_blog': {
153
  'description': Value(dtype='string'),
154
  },
 
155
  'importai': {},
156
  'jsteinhardt_blog': {},
157
  'lesswrong': {
 
164
  },
165
  'miri': {},
166
  'ml_safety_newsletter': {},
167
+ 'openai.research': {},
168
+ 'rob_miles_ai_safety': {},
169
+ 'special_docs': {},
 
 
 
 
 
170
  'vkrakovna_blog': {},
171
  'yudkowsky_blog': {},
172
  }
gdocs.jsonl DELETED
The diff for this file is too large to render. See raw diff
 
html_articles.jsonl DELETED
The diff for this file is too large to render. See raw diff
 
nonarxiv_papers.jsonl DELETED
The diff for this file is too large to render. See raw diff
 
pdfs.jsonl DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:399171f9ea7675679eb22513c88d161dc86c628e8eb98d61867438132737a035
3
- size 21813824
 
 
 
 
qualiacomputing.jsonl DELETED
The diff for this file is too large to render. See raw diff
 
reports.jsonl DELETED
The diff for this file is too large to render. See raw diff