Datasets:
Confirm-Labs
/
pythia-12b-neuron-dataset-examples

Dataset card Data Studio Files Community
layer
uint8
0
35
neuron
uint16
0
20.5k
rank
uint8
0
63
activation
float32
-0
61.8
position
uint8
0
15
text
stringlengths
1
1.74k
id0
uint16
0
50.3k
id1
uint16
0
50.3k
id2
uint16
0
50.3k
id3
uint16
0
50.3k
id4
uint16
0
50.3k
id5
uint16
0
50.3k
id6
uint16
0
50.3k
id7
uint16
0
50.3k
id8
uint16
0
50.3k
id9
uint16
0
50.3k
id10
uint16
0
50.3k
id11
uint16
0
50.3k
id12
uint16
0
50.3k
id13
uint16
0
50.3k
id14
uint16
0
50.3k
id15
uint16
0
50.3k
35
20,437
32
6.636719
0
oof
49,912
5,903
254
14,637
281
14,741
20,477
1,738
187
186
1,033
49,912
5,903
254
64
15,905
35
20,437
33
6.636719
0
oof
49,912
432
19,040
13,260
347
281
697
3,753
13
7,866
1,480
432
247
1,643
1,077
2,087
35
20,437
34
6.636719
0
oof
49,912
3,354
546
810
254
723
282
5,852
372
8,363
85
423
3,889
372
13,660
13,616
35
20,437
35
6.636719
0
oof
49,912
275
362
23,151
22,252
546
14,374
489
282
294
43,340
15
12,793
310
259
9,314
35
20,437
36
6.636719
0
oof
49,912
615
28,045
2,072
295
9,623
395
2,856
16,465
10,814
30,687
3,889
1,499
6,533
1,182
276
35
20,437
37
6.636719
0
oof
49,912
33,533
6,105
13
19,605
1,024
13
651
417
7,156
273
19,693
247
2,329
90
275
35
20,437
38
6.636719
0
oof
49,912
13
816
751
326
15
309
452
23,928
898
1,107
273
731
7,963
9,809
327
35
20,437
39
6.636719
0
oof
49,912
1,255
285
3,809
301
35,715
285
385
26,500
6,988
37,171
25,443
13
369
671
13
35
20,437
40
6.636719
0
oof
49,912
15
24,531
279
273
253
278
486
38,731
736
21,733
76
13
18,588
273
253
35
20,437
41
6.636719
0
oof
49,912
264
15
388
2,310
2,205
4,571
626
3,553
352
3,815
15
496
2,314
8,098
13
35
20,437
42
6.554688
1
"[
50,254
39,258
683
19,589
15
20,226
5,023
25,804
5,023
29,064
9,502
995
187
50,254
3
17
35
20,437
43
6.539063
1
http
50,254
2,413
6,825
6,074
15
1,178
6,825
6,942
1,587
9,278
3,287
187
50,254
2,413
6,825
6,074
35
20,437
44
6.453125
0
,,
19,396
367
606
3,358
13
18,633
2,462
8,662
6
310
5,816
2
187
50,266
15,968
187
35
20,437
45
6.453125
0
,,
19,396
89
578
480
74
1,990
81
889
10
559
3,353
3,005
3,202
3,830
3,202
2,162
35
20,437
46
6.453125
0
,,
19,396
89
578
891
75
1,990
81
2,023
89
578
480
74
1,990
18
39,115
7,718
35
20,437
47
6.453125
0
,,
19,396
19,396
13
187
187
48,144
434
9,792
9,870
496
1,157
19,396
19,396
19,396
38,742
8
35
20,437
48
6.453125
1
,,,,
19,396
19,396
13
187
187
48,144
434
9,792
9,870
496
1,157
19,396
19,396
19,396
38,742
8
35
20,437
49
6.453125
0
,,
19,396
13
187
187
13,745
16,560
264
187
187
1,394
476
8,046
4,318
13,896
281
789
35
20,437
50
6.453125
0
,,
19,396
533
1,384
1,107
329
10,993
537
2,598
7,422
310
3,563
374
8,007
537
249
436
35
20,437
51
6.453125
0
,,
19,396
513
64
39,894
64
14,382
9
12,540
13
331
6,917
1,228
187
187
24,582
4,228
35
20,437
52
6.453125
0
,,
19,396
835
247
1,566
310
327
3,148
15
380
4,376
273
253
17,052
398
310
370
35
20,437
53
6.453125
0
,,
19,396
13
22,641
27
309
651
4,489
436
11,959
969
13
533
3,164
1,912
626
452
35
20,437
54
6.453125
0
,,
19,396
50,276
29
283
209
24,893
27,896
41,351
28,154
20,326
346
12,361
3
187
50,274
2,364
35
20,437
55
6.453125
0
,,
19,396
18,365
473
5,211
9,331
1,157
473
3,309
6,868
273
13,187
2,866
1,425
1,108
27,477
1,889
35
20,437
56
6.453125
0
,,
19,396
13
20
15
26
187
187
29,255
20,837
2,866
1,012
2,904
15
4,196
5,770
2,866
35
20,437
57
6.453125
0
,,
19,396
1,397
76
8,340
62
187
186
15
7,173
277
3,319
68
13
66
186
28
35
20,437
58
6.453125
0
,,
19,396
19,921
60
21
3,291
5,264
3,233
187
50,272
859
27
2,548
568
4,756
1,138
4
35
20,437
59
6.453125
0
,,
19,396
27
26,673
322
15
56
15
20
69
25,305
50,256
20
187
92
10,685
15
35
20,437
60
6.453125
0
,,
19,396
4,117
3,291
187
62
187
187
7,229
19
25,265
64
42,429
426
544
187
50,274
35
20,437
61
6.453125
0
,,
19,396
597
452
2,218
2,717
533
4,710
625
285
625
952
562
273
789
13
762
35
20,437
62
6.453125
0
,,
19,396
368
476
897
32,911
12,636
290
281
1,078
20,864
747
34,413
40,840
10,865
15
1,737
35
20,437
63
6.453125
0
,,
19,396
329
48,583
19,039
13
38,859
11,759
27
831
5,382
310
417
12,482
949
253
13,126
35
20,438
0
7.730469
13
}$ and, in particular, holds with the sensitivities $\widetilde{\
724
285
13
275
1,798
13
6,556
342
253
21,750
16,762
669
6,796
464
6,165
4,689
35
20,438
1
7.308594
7
\] $\operatorname{lip}\bigl(\
696
669
4,820
92
20,844
889
17,896
1,035
6,526
1,126
2,407
92
89
599
393
437
35
20,438
2
7.207031
14
}$ $\hat{\beta}_{u,\widehat{J}}$ $\
724
50,275
1,202
700
464
2,461
2,026
86
1,337
8,752
92
43
4,018
50,275
1,202
700
35
20,438
3
7.195313
6
that $I(\hat A,\
326
370
42
1,035
700
329
1,337
700
378
10
21,846
50,276
29
21,846
50,276
18
35
20,438
4
7.164063
13
mathbb{I}}_{\textnormal{loc}}}^{-1} \left(\
1,991
92
42
14,024
28,975
92
9,450
599
11,444
18
94
393
1,274
1,035
1,124
92
35
20,438
5
7.148438
7
beta^*)\right)\ge 1-\
2,461
3,503
1,572
918
1,572
463
337
2,249
2,733
64
20
4,700
187
187
1,231
873
35
20,438
6
7.144531
11
using the estimator of the gradient in, of the form $$\
970
253
29,107
273
253
11,786
275
1,157
273
253
830
1,764
2,043
92
2,132
94
35
20,438
7
7.144531
11
From this, using the definition of the sensitivity $\widetilde{\
4,325
436
13
970
253
5,426
273
253
7,340
669
6,796
464
6,165
2,026
81
13
35
20,438
8
7.078125
15
1:jhat} J(\theta^*) \subseteq J(\
18
27
75
700
94
187
50,274
43
1,035
3,124
3,503
10
393
11,861
500
1,035
35
20,438
9
7.0625
14
_0\subset\{1,\hdots,K\}$ $$\widetilde{\
64
17
61
6,040
6,921
18
1,337
73
6,768
13
44
10,952
1,764
6,796
464
6,165
35
20,438
10
7.054688
10
D_X}^{-1} \left(\widehat{\
399
64
57
11,444
18
94
393
1,274
1,035
8,752
464
2,461
10,780
2,461
24,638
918
35
20,438
11
7.050781
15
{y})$, $\widehat{\mathsf{S}}_T(\theta')$ and $\
92
90
19,446
669
8,752
464
9,501
92
52
5,000
53
1,035
3,124
31,807
285
669
35
20,438
12
7.050781
11
^{- U(\hat y)/2} e^{-T(\
2,497
530
1,035
700
340
1,933
19
94
299
2,497
53
1,035
700
465
3,117
299
35
20,438
13
7.050781
14
\log\frac{1}{\epsilon}(1+\lambda_{1}(\
61
2,808
61
1,124
92
18
2,704
4,259
1,603
18
2,766
2,260
578
18
3,713
4,153
35
20,438
14
7.027344
8
Then, $$\operatorname{lip}\bigl(\
2,635
13
1,764
4,820
92
20,844
889
17,896
1,035
6,526
2,253
2,407
92
89
94
393
35
20,438
15
7.019531
7
E\left[\Phi\left(\
38
61
1,274
5,709
6,065
61
1,274
1,035
1,124
92
54
18,958
47
64
19
2,497
35
20,438
16
7.015625
10
U}-\theta^*\right|_\infty +\left|\
54
10,780
3,124
24,638
918
93
2,253
3,259
14,030
1,274
3,577
1,124
18
79
464
3,342
35
20,438
17
7.007813
14
obtain that $$\begin{aligned} \Delta_r \left(\
4,044
326
1,764
2,043
92
2,132
94
187
61
3,442
64
83
393
1,274
1,035
1,124
35
20,438
18
7.007813
5
L}}}(\hat{\beta},\
45
39,111
700
464
2,461
5,548
2,461
3,503
11,127
13,658
13
359
3,091
1,908
253
669
35
20,438
19
7.003906
14
{\mathbb{V}}_{\textnormal{loc}}}^{-1} \left(\
464
1,991
92
55
14,024
28,975
92
9,450
599
11,444
18
94
393
1,274
1,035
1,274
35
20,438
20
6.992188
10
\\ &\le r\left(\widehat{\
3,353
187
5,977
282
187
83
61
1,274
1,035
8,752
464
2,592
9,213
2,609
464
8,752
35
20,438
21
6.988281
4
{conv}}\left\{\
92
13,118
3,080
1,274
17,567
2,204
578
74
393
249
378
94
299
64
74
28,511
35
20,438
22
6.988281
8
}\left\{d_1\left(\
889
1,274
6,921
69
64
18
61
1,274
1,035
6,796
92
54
9,213
1,124
92
18
35
20,438
23
6.984375
12
]&\leq\frac{36L_{\lambda}\left(\
62
5,977
3,040
61
1,124
92
1,812
45
1,126
2,260
889
1,274
1,035
85
6,065
1,126
35
20,438
24
6.984375
8
}|_2} \ge\widetilde{\
8,589
64
19
94
393
463
61
6,796
464
6,165
2,138
323
690
669
6,796
464
35
20,438
25
6.976563
11
V}}_{\textnormal{loc}}}^{-1} \left(\
55
14,024
28,975
92
9,450
599
11,444
18
94
393
1,274
1,035
1,124
464
3,214
2,704
35
20,438
26
6.96875
10
begin{aligned}&\Vert \mathrm {sdist}(\
2,043
92
2,132
41,915
7,994
393
2,690
551
84
8,155
3,713
3,830
19,049
2,592
8,454
83
35
20,438
27
6.957031
7
I(\hat S_{1},\
187
42
1,035
700
322
578
18
5,548
700
322
578
19
2,311
708
426
708
35
20,438
28
6.957031
14
{J}}$ $\hat{\beta}_{u,SC}$ $\
92
43
4,018
50,275
1,202
700
464
2,461
2,026
86
13
4,061
724
50,275
1,202
6,165
35
20,438
29
6.957031
4
empirical risk function as $$\
16,774
2,495
1,159
347
1,764
1,968
92
2,574
27
358
5,378
474
14
16,272
14
7,265
35
20,438
30
6.957031
7
neq\emptyset],$$ $$1-\
9,540
61
20,760
1,092
1,890
3,318
18
2,249
4,347
1,035
2,260
9
90
17,990
2,260
6,263
35
20,438
31
6.953125
12
estimator of, $$\begin{aligned} \widehat{\
29,107
273
1,157
1,764
2,043
92
2,132
94
187
50,274
61
8,752
464
9,501
92
52
35
20,438
32
6.953125
14
{\sigma}+\sqrt{\widehat{Q}(\beta^*)}-\sqrt{\
464
2,592
9,213
2,609
464
8,752
92
50
3,713
2,461
3,503
3,117
2,249
2,609
464
8,752
35
20,438
33
6.945313
6
),\\ \left(\widetilde{\
46,865
187
61
1,274
1,035
6,796
464
4,535
11,110
70
12
90
2,850
505
428
187
35
20,438
34
6.9375
11
^{- U(\hat x)/2} e^{- T(\
2,497
530
1,035
700
1,269
1,933
19
94
299
2,497
308
1,035
700
268
3,117
299
35
20,438
35
6.9375
11
} e^{- T(\hat p)} e^{- U(\
94
299
2,497
308
1,035
700
268
3,117
299
2,497
530
1,035
700
1,269
1,933
19
35
20,438
36
6.9375
5
,\epsilon} \left(\
1,337
4,259
94
393
1,274
1,035
3,582
444
63
79
1,126
2,461
1,337
2,059
61
2,059
35
20,438
37
6.9375
13
sqrt{1-\delta} \le \sigma_{\min} (\
2,609
92
18
2,249
3,005
94
393
282
393
2,592
1,126
1,222
94
5,081
6,065
64
35
20,438
38
6.9375
12
admissible perturbation of $\Omega$; 2. $\
324
13,038
20,452
273
669
4,153
16,446
187
187
19
15
50,276
1,202
4,153
5
310
35
20,438
39
6.925781
11
have: $$J(\beta^*) \subseteq J(\
452
27
3,318
43
1,035
2,461
3,503
10
393
11,861
500
1,035
8,752
61
2,461
11,127
35
20,438
40
6.921875
15
)) &\geq \min\limits_{j\geq 1} \Big\{\
1,228
9,443
5,090
393
1,222
61
10,423
578
75
61
5,090
337
94
393
5,178
17,567
35
20,438
41
6.914063
14
t:6\] above) the upper bound on $H_\theta (\
85
27
23
696
1,840
10
253
5,170
3,033
327
370
41
2,253
3,124
5,081
2,592
35
20,438
42
6.910156
11
\Delta_n^{m-1} \left (\
61
3,442
64
79
768
78
14
18
94
393
1,274
5,081
1,124
92
34
64
35
20,438
43
6.90625
7
where $I(\hat X^{+},\
835
370
42
1,035
700
1,594
13,566
5,548
700
1,594
2,497
8,395
2
2,029
2
42
35
20,438
44
6.90625
13
can replace there $|J(\theta^*)|$ by $|J(\
476
8,171
627
10,493
43
1,035
3,124
3,503
8,579
5
407
10,493
43
1,035
8,752
61
35
20,438
45
6.902344
4
\mu}}\left (\
61
1,906
3,080
1,274
5,081
2,043
92
3,728
1,217
68
94
1,926
1,109
309
2,000
81
35
20,438
46
6.898438
10
\\ &\le \operatorname{lip}\bigl(\
3,202
187
5,977
282
393
4,820
92
20,844
889
17,896
1,035
6,526
2,253
2,407
92
89
35
20,438
47
6.894531
11
= Nr(\hat{\theta})'[R(\
426
427
83
1,035
700
464
3,124
2,311
8
60
51
1,035
700
464
3,124
2,311
35
20,438
48
6.886719
13
\right|_{\infty}\quad {\rm and}\quad \widetilde{\
61
918
34,813
3,259
889
3,362
1,926
1,109
285
889
3,362
393
6,796
464
6,165
2,026
35
20,438
49
6.886719
6
;t}+\max \left(\
28
85
9,213
4,090
393
1,274
1,035
20,999
464
2,204
889
10,423
578
74
30
18
35
20,438
50
6.882813
10
estimator of, $$\begin{aligned} \
29,107
273
1,157
1,764
2,043
92
2,132
94
187
50,274
61
8,752
464
9,501
92
52
35
20,438
51
6.882813
8
}^T\left(\bold{Y}-\
2,306
53
61
1,274
1,035
12,509
92
58
10,780
12,509
92
57
889
2,461
61
918
35
20,438
52
6.875
8
}}}^{-1} \left(\left(\
599
11,444
18
94
393
1,274
1,035
1,274
1,035
1,124
464
3,214
2,704
3,214
391
94
35
20,438
53
6.871094
3
left(\widetilde{\
1,274
1,035
6,796
464
4,535
11,110
70
12
90
2,850
505
428
393
6,796
92
90
35
20,438
54
6.867188
10
{loc}}}^{-1} \left(\left(\
92
9,450
599
11,444
18
94
393
1,274
1,035
1,274
1,035
1,124
464
3,214
2,704
3,214
35
20,438
55
6.867188
10
begin{aligned} \Phi\left(\
2,043
92
2,132
94
187
50,274
61
6,065
61
1,274
1,035
11,765
92
83
889
918
35
20,438
56
6.863281
2
}\left (\
889
1,274
5,081
1,588
92
38
35,420
38,013
6,394
87
22,254
79
599
1,157
2,386
38,013
35
20,438
57
6.847656
9
^{k_1}\\ &(\widetilde{\
768
76
64
18
11,054
187
7
1,035
6,796
464
11,920
92
51
25
8
27,927
35
20,438
58
6.847656
9
P}}({\mathcal{Y}}^n)}\left(\
49
19,753
1,588
92
58
7,294
79
7,398
1,274
1,035
2,808
393
2,859
2,249
2,808
393
35
20,438
59
6.84375
7
mathcal{L}}}(\hat{\beta},\
1,588
92
45
39,111
700
464
2,461
5,548
2,461
3,503
4,244
534
6,125
253
2,957
23,122
35
20,438
60
6.84375
15
{aligned}$$ then $$\begin{aligned} \Gamma(\theta,\
92
2,132
2,138
840
1,764
2,043
92
2,132
94
187
50,270
61
4,220
1,035
3,124
1,337
35
20,438
61
6.839844
11
}_{\theta_0}(\mathbf{y})$, $\widehat{\
4,689
3,124
64
17
3,713
2,407
92
90
19,446
669
8,752
464
9,501
92
52
5,000
35
20,438
62
6.839844
5
, $(\widehat{\beta},\
13
9,722
8,752
464
2,461
5,548
8,752
464
2,592
6,580
46,926
253
17,705
669
1,274
21,837
35
20,438
63
6.839844
14
then such optimal diffusion control is given by $$\begin{aligned} \
840
824
8,654
12,393
1,453
310
1,677
407
1,764
2,043
92
2,132
94
187
61
2,009
35
20,439
0
7.304688
13
armies and created _dynasties_ (DY-nuh-
29,894
285
3,562
795
42,927
505
447
64
313
32,201
14
3,023
73
14
296
1,796
35
20,439
1
7.136719
9
Feeling Nostalgic nos·
7,510
8,855
427
493
13,256
280
187
187
23,047
6,256
22,559
6,256
45,795
187
187
19
35
20,439
2
6.695313
9
1ska 1ske 1ski 1
18
32,518
187
18
40,993
187
18
9,327
187
18
3,319
80
187
18
3,319
86
35
20,439
3
6.679688
7
kind to and is not portrayed sympathe
2,238
281
285
310
417
30,804
12,514
4,349
85
1,037
15
24,697
13
352
812
320