bogdanraduta commited on
Commit
2bed31b
·
verified ·
1 Parent(s): 89aaeca

Upload eval/injection_eval.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. eval/injection_eval.json +412 -403
eval/injection_eval.json CHANGED
@@ -8,8 +8,8 @@
8
  "recall": 1.0,
9
  "f1": 1.0,
10
  "fpr": 0.0,
11
- "accuracy": 0.994,
12
- "n": 167
13
  },
14
  "bg": {
15
  "support": 41,
@@ -17,26 +17,26 @@
17
  "recall": 1.0,
18
  "f1": 1.0,
19
  "fpr": 0.0,
20
- "accuracy": 0.9879,
21
- "n": 165
22
  },
23
  "cs": {
24
  "support": 42,
25
  "precision": 0.9545,
26
  "recall": 1.0,
27
  "f1": 0.9767,
28
- "fpr": 0.016,
29
- "accuracy": 0.988,
30
- "n": 167
31
  },
32
  "da": {
33
  "support": 41,
34
  "precision": 0.9762,
35
  "recall": 1.0,
36
  "f1": 0.988,
37
- "fpr": 0.008,
38
- "accuracy": 0.988,
39
- "n": 166
40
  },
41
  "de": {
42
  "support": 42,
@@ -44,8 +44,8 @@
44
  "recall": 1.0,
45
  "f1": 1.0,
46
  "fpr": 0.0,
47
- "accuracy": 1.0,
48
- "n": 167
49
  },
50
  "el": {
51
  "support": 42,
@@ -53,26 +53,26 @@
53
  "recall": 1.0,
54
  "f1": 1.0,
55
  "fpr": 0.0,
56
- "accuracy": 0.988,
57
- "n": 166
58
  },
59
  "en": {
60
  "support": 41,
61
- "precision": 0.9762,
62
  "recall": 1.0,
63
- "f1": 0.988,
64
- "fpr": 0.008,
65
- "accuracy": 0.988,
66
- "n": 166
67
  },
68
  "es": {
69
  "support": 40,
70
- "precision": 1.0,
71
  "recall": 1.0,
72
- "f1": 1.0,
73
- "fpr": 0.0,
74
- "accuracy": 0.9939,
75
- "n": 165
76
  },
77
  "et": {
78
  "support": 41,
@@ -80,8 +80,8 @@
80
  "recall": 1.0,
81
  "f1": 1.0,
82
  "fpr": 0.0,
83
- "accuracy": 0.994,
84
- "n": 166
85
  },
86
  "fi": {
87
  "support": 42,
@@ -90,43 +90,43 @@
90
  "f1": 1.0,
91
  "fpr": 0.0,
92
  "accuracy": 1.0,
93
- "n": 167
94
  },
95
  "fr": {
96
  "support": 41,
97
  "precision": 1.0,
98
- "recall": 0.9756,
99
- "f1": 0.9877,
100
  "fpr": 0.0,
101
- "accuracy": 0.9759,
102
- "n": 166
103
  },
104
  "ga": {
105
  "support": 42,
106
  "precision": 0.9762,
107
  "recall": 0.9762,
108
  "f1": 0.9762,
109
- "fpr": 0.008,
110
- "accuracy": 0.9701,
111
- "n": 167
112
  },
113
  "hr": {
114
  "support": 42,
115
- "precision": 0.9545,
116
  "recall": 1.0,
117
- "f1": 0.9767,
118
- "fpr": 0.016,
119
- "accuracy": 0.988,
120
- "n": 167
121
  },
122
  "hu": {
123
  "support": 42,
124
  "precision": 0.9545,
125
  "recall": 1.0,
126
  "f1": 0.9767,
127
- "fpr": 0.016,
128
- "accuracy": 0.976,
129
- "n": 167
130
  },
131
  "it": {
132
  "support": 42,
@@ -135,7 +135,7 @@
135
  "f1": 1.0,
136
  "fpr": 0.0,
137
  "accuracy": 1.0,
138
- "n": 167
139
  },
140
  "lt": {
141
  "support": 42,
@@ -143,35 +143,35 @@
143
  "recall": 1.0,
144
  "f1": 1.0,
145
  "fpr": 0.0,
146
- "accuracy": 0.994,
147
- "n": 167
148
  },
149
  "lv": {
150
  "support": 41,
151
  "precision": 0.9762,
152
  "recall": 1.0,
153
  "f1": 0.988,
154
- "fpr": 0.008,
155
- "accuracy": 0.9819,
156
- "n": 166
157
  },
158
  "mt": {
159
  "support": 42,
160
- "precision": 0.7321,
161
  "recall": 0.9762,
162
- "f1": 0.8367,
163
- "fpr": 0.12,
164
- "accuracy": 0.8802,
165
- "n": 167
166
  },
167
  "nl": {
168
  "support": 41,
169
- "precision": 0.9762,
170
- "recall": 1.0,
171
- "f1": 0.988,
172
- "fpr": 0.008,
173
- "accuracy": 0.994,
174
- "n": 166
175
  },
176
  "pl": {
177
  "support": 41,
@@ -179,26 +179,26 @@
179
  "recall": 1.0,
180
  "f1": 1.0,
181
  "fpr": 0.0,
182
- "accuracy": 0.994,
183
- "n": 166
184
  },
185
  "pt": {
186
  "support": 42,
187
  "precision": 0.9767,
188
  "recall": 1.0,
189
  "f1": 0.9882,
190
- "fpr": 0.008,
191
- "accuracy": 0.988,
192
- "n": 167
193
  },
194
  "ro": {
195
  "support": 42,
196
  "precision": 0.9767,
197
  "recall": 1.0,
198
  "f1": 0.9882,
199
- "fpr": 0.0081,
200
- "accuracy": 0.988,
201
- "n": 166
202
  },
203
  "sk": {
204
  "support": 41,
@@ -206,17 +206,17 @@
206
  "recall": 1.0,
207
  "f1": 1.0,
208
  "fpr": 0.0,
209
- "accuracy": 0.994,
210
- "n": 166
211
  },
212
  "sl": {
213
  "support": 42,
214
  "precision": 0.9767,
215
  "recall": 1.0,
216
  "f1": 0.9882,
217
- "fpr": 0.008,
218
- "accuracy": 0.988,
219
- "n": 167
220
  },
221
  "sv": {
222
  "support": 41,
@@ -224,36 +224,36 @@
224
  "recall": 1.0,
225
  "f1": 1.0,
226
  "fpr": 0.0,
227
- "accuracy": 0.994,
228
- "n": 166
229
  },
230
  "tr": {
231
  "support": 42,
232
- "precision": 0.9545,
233
  "recall": 1.0,
234
- "f1": 0.9767,
235
- "fpr": 0.016,
236
- "accuracy": 0.982,
237
- "n": 167
238
  }
239
  },
240
  "per_register": {
241
  "encoded_payload": {
242
  "support": 155,
243
  "precision": 1.0,
244
- "recall": 0.9935,
245
- "f1": 0.9968,
246
  "fpr": 0.0,
247
- "accuracy": 0.9742,
248
  "n": 155
249
  },
250
  "hypothetical_framing": {
251
  "support": 156,
252
  "precision": 1.0,
253
- "recall": 1.0,
254
- "f1": 1.0,
255
  "fpr": 0.0,
256
- "accuracy": 0.9872,
257
  "n": 156
258
  },
259
  "ignore_instructions": {
@@ -262,25 +262,25 @@
262
  "recall": 1.0,
263
  "f1": 1.0,
264
  "fpr": 0.0,
265
- "accuracy": 0.9484,
266
  "n": 155
267
  },
268
  "instruction_in_document": {
269
  "support": 155,
270
  "precision": 1.0,
271
- "recall": 1.0,
272
- "f1": 1.0,
273
  "fpr": 0.0,
274
- "accuracy": 0.9613,
275
  "n": 155
276
  },
277
  "instruction_in_tool_output": {
278
  "support": 148,
279
  "precision": 1.0,
280
- "recall": 0.9865,
281
- "f1": 0.9932,
282
  "fpr": 0.0,
283
- "accuracy": 0.9865,
284
  "n": 148
285
  },
286
  "meta_question": {
@@ -288,10 +288,19 @@
288
  "precision": 0.0,
289
  "recall": 0.0,
290
  "f1": 0.0,
291
- "fpr": 0.0148,
292
- "accuracy": 0.9852,
293
  "n": 338
294
  },
 
 
 
 
 
 
 
 
 
295
  "mundane_informational": {
296
  "support": 0,
297
  "precision": 0.0,
@@ -324,8 +333,8 @@
324
  "precision": 0.0,
325
  "recall": 0.0,
326
  "f1": 0.0,
327
- "fpr": 0.0148,
328
- "accuracy": 0.9852,
329
  "n": 338
330
  },
331
  "ordinary_question": {
@@ -333,8 +342,8 @@
333
  "precision": 0.0,
334
  "recall": 0.0,
335
  "f1": 0.0,
336
- "fpr": 0.003,
337
- "accuracy": 0.997,
338
  "n": 338
339
  },
340
  "persona_override": {
@@ -343,7 +352,7 @@
343
  "recall": 1.0,
344
  "f1": 1.0,
345
  "fpr": 0.0,
346
- "accuracy": 0.9226,
347
  "n": 155
348
  },
349
  "quoted_attack": {
@@ -351,8 +360,8 @@
351
  "precision": 0.0,
352
  "recall": 0.0,
353
  "f1": 0.0,
354
- "fpr": 0.0178,
355
- "accuracy": 0.9822,
356
  "n": 338
357
  },
358
  "roleplay_benign": {
@@ -360,8 +369,8 @@
360
  "precision": 0.0,
361
  "recall": 0.0,
362
  "f1": 0.0,
363
- "fpr": 0.0148,
364
- "accuracy": 0.9852,
365
  "n": 338
366
  },
367
  "security_discussion": {
@@ -369,8 +378,8 @@
369
  "precision": 0.0,
370
  "recall": 0.0,
371
  "f1": 0.0,
372
- "fpr": 0.0148,
373
- "accuracy": 0.9852,
374
  "n": 338
375
  },
376
  "system_prompt_extraction": {
@@ -379,7 +388,7 @@
379
  "recall": 1.0,
380
  "f1": 1.0,
381
  "fpr": 0.0,
382
- "accuracy": 1.0,
383
  "n": 156
384
  },
385
  "technical_identifiers": {
@@ -387,8 +396,8 @@
387
  "precision": 0.0,
388
  "recall": 0.0,
389
  "f1": 0.0,
390
- "fpr": 0.003,
391
- "accuracy": 0.997,
392
  "n": 336
393
  },
394
  "technical_payload": {
@@ -396,36 +405,36 @@
396
  "precision": 0.0,
397
  "recall": 0.0,
398
  "f1": 0.0,
399
- "fpr": 0.003,
400
- "accuracy": 0.997,
401
  "n": 337
402
  }
403
  },
404
  "per_label": {
405
  "direct_injection": {
406
  "support": 466,
407
- "precision": 0.9546,
408
- "recall": 0.9936,
409
- "f1": 0.9737,
410
  "fpr": 0.0057,
411
  "accuracy": 0.0,
412
  "n": 0
413
  },
414
  "indirect_injection": {
415
  "support": 303,
416
- "precision": 0.9495,
417
- "recall": 0.9934,
418
- "f1": 0.971,
419
- "fpr": 0.004,
420
  "accuracy": 0.0,
421
  "n": 0
422
  },
423
  "jailbreak": {
424
  "support": 466,
425
- "precision": 0.9331,
426
- "recall": 0.9871,
427
- "f1": 0.9593,
428
- "fpr": 0.0085,
429
  "accuracy": 0.0,
430
  "n": 0
431
  }
@@ -433,7 +442,7 @@
433
  "worst_languages": [
434
  [
435
  "mt",
436
- 0.836734693877551
437
  ],
438
  [
439
  "ga",
@@ -444,586 +453,586 @@
444
  0.9767441860465117
445
  ],
446
  [
447
- "hr",
448
  0.9767441860465117
449
  ],
450
  [
451
- "hu",
452
- 0.9767441860465117
453
  ]
454
  ],
455
  "calibration": {
456
  "threshold": 0.02,
457
  "objective": "recall_at_fpr",
458
- "value_on_val": 0.9926,
459
  "default_threshold": 0.5,
460
- "default_value_on_val": 0.9852,
461
- "improvement": 0.0074,
462
  "chosen_on": "validation split, never test",
463
  "curve": [
464
  {
465
  "threshold": 0.02,
466
- "macro_f1": 0.959,
467
- "any_label_recall": 0.9926,
468
- "any_label_fpr": 0.0089
469
  },
470
  {
471
  "threshold": 0.03,
472
- "macro_f1": 0.9659,
473
- "any_label_recall": 0.9907,
474
- "any_label_fpr": 0.0071
475
  },
476
  {
477
  "threshold": 0.04,
478
- "macro_f1": 0.9677,
479
  "any_label_recall": 0.9898,
480
- "any_label_fpr": 0.0062
481
  },
482
  {
483
  "threshold": 0.05,
484
- "macro_f1": 0.9687,
485
  "any_label_recall": 0.9898,
486
- "any_label_fpr": 0.0059
487
  },
488
  {
489
  "threshold": 0.06,
490
- "macro_f1": 0.9687,
491
  "any_label_recall": 0.9898,
492
- "any_label_fpr": 0.0059
493
  },
494
  {
495
  "threshold": 0.07,
496
- "macro_f1": 0.9692,
497
  "any_label_recall": 0.9898,
498
- "any_label_fpr": 0.0055
499
  },
500
  {
501
  "threshold": 0.08,
502
- "macro_f1": 0.9688,
503
  "any_label_recall": 0.9898,
504
- "any_label_fpr": 0.0052
505
  },
506
  {
507
  "threshold": 0.09,
508
- "macro_f1": 0.9692,
509
- "any_label_recall": 0.9889,
510
- "any_label_fpr": 0.0052
511
  },
512
  {
513
  "threshold": 0.1,
514
- "macro_f1": 0.9684,
515
- "any_label_recall": 0.988,
516
- "any_label_fpr": 0.0049
517
  },
518
  {
519
  "threshold": 0.11,
520
- "macro_f1": 0.968,
521
- "any_label_recall": 0.987,
522
- "any_label_fpr": 0.0049
523
  },
524
  {
525
  "threshold": 0.12,
526
- "macro_f1": 0.9686,
527
- "any_label_recall": 0.987,
528
- "any_label_fpr": 0.0049
529
  },
530
  {
531
  "threshold": 0.13,
532
- "macro_f1": 0.9692,
533
- "any_label_recall": 0.987,
534
- "any_label_fpr": 0.0043
535
  },
536
  {
537
  "threshold": 0.14,
538
- "macro_f1": 0.9696,
539
- "any_label_recall": 0.987,
540
- "any_label_fpr": 0.0043
541
  },
542
  {
543
  "threshold": 0.15,
544
- "macro_f1": 0.9699,
545
- "any_label_recall": 0.987,
546
- "any_label_fpr": 0.004
547
  },
548
  {
549
  "threshold": 0.16,
550
- "macro_f1": 0.9703,
551
- "any_label_recall": 0.987,
552
- "any_label_fpr": 0.004
553
  },
554
  {
555
  "threshold": 0.17,
556
- "macro_f1": 0.9703,
557
- "any_label_recall": 0.987,
558
- "any_label_fpr": 0.004
559
  },
560
  {
561
  "threshold": 0.18,
562
- "macro_f1": 0.9703,
563
- "any_label_recall": 0.987,
564
- "any_label_fpr": 0.004
565
  },
566
  {
567
  "threshold": 0.19,
568
- "macro_f1": 0.9713,
569
- "any_label_recall": 0.987,
570
- "any_label_fpr": 0.004
571
  },
572
  {
573
  "threshold": 0.2,
574
- "macro_f1": 0.9713,
575
- "any_label_recall": 0.987,
576
- "any_label_fpr": 0.004
577
  },
578
  {
579
  "threshold": 0.21,
580
- "macro_f1": 0.9717,
581
- "any_label_recall": 0.987,
582
- "any_label_fpr": 0.0037
583
  },
584
  {
585
  "threshold": 0.22,
586
- "macro_f1": 0.9717,
587
- "any_label_recall": 0.987,
588
- "any_label_fpr": 0.0037
589
  },
590
  {
591
  "threshold": 0.23,
592
- "macro_f1": 0.9717,
593
- "any_label_recall": 0.987,
594
- "any_label_fpr": 0.0037
595
  },
596
  {
597
  "threshold": 0.24,
598
- "macro_f1": 0.9717,
599
- "any_label_recall": 0.987,
600
- "any_label_fpr": 0.0037
601
  },
602
  {
603
  "threshold": 0.25,
604
- "macro_f1": 0.9717,
605
- "any_label_recall": 0.987,
606
- "any_label_fpr": 0.0037
607
  },
608
  {
609
  "threshold": 0.26,
610
- "macro_f1": 0.972,
611
- "any_label_recall": 0.987,
612
- "any_label_fpr": 0.0037
613
  },
614
  {
615
  "threshold": 0.27,
616
- "macro_f1": 0.972,
617
- "any_label_recall": 0.987,
618
- "any_label_fpr": 0.0037
619
  },
620
  {
621
  "threshold": 0.28,
622
- "macro_f1": 0.9716,
623
- "any_label_recall": 0.9861,
624
- "any_label_fpr": 0.0037
625
  },
626
  {
627
  "threshold": 0.29,
628
- "macro_f1": 0.9716,
629
- "any_label_recall": 0.9861,
630
- "any_label_fpr": 0.0037
631
  },
632
  {
633
  "threshold": 0.3,
634
- "macro_f1": 0.9722,
635
- "any_label_recall": 0.9861,
636
- "any_label_fpr": 0.0034
637
  },
638
  {
639
  "threshold": 0.31,
640
- "macro_f1": 0.9725,
641
- "any_label_recall": 0.9861,
642
- "any_label_fpr": 0.0034
643
  },
644
  {
645
  "threshold": 0.32,
646
- "macro_f1": 0.9725,
647
- "any_label_recall": 0.9861,
648
- "any_label_fpr": 0.0034
649
  },
650
  {
651
  "threshold": 0.33,
652
- "macro_f1": 0.9725,
653
- "any_label_recall": 0.9861,
654
- "any_label_fpr": 0.0034
655
  },
656
  {
657
  "threshold": 0.34,
658
- "macro_f1": 0.9731,
659
- "any_label_recall": 0.9861,
660
- "any_label_fpr": 0.0031
661
  },
662
  {
663
  "threshold": 0.35,
664
- "macro_f1": 0.9734,
665
- "any_label_recall": 0.9861,
666
- "any_label_fpr": 0.0031
667
  },
668
  {
669
  "threshold": 0.36,
670
- "macro_f1": 0.9728,
671
- "any_label_recall": 0.9852,
672
- "any_label_fpr": 0.0031
673
  },
674
  {
675
  "threshold": 0.37,
676
- "macro_f1": 0.9728,
677
- "any_label_recall": 0.9852,
678
- "any_label_fpr": 0.0031
679
  },
680
  {
681
  "threshold": 0.38,
682
- "macro_f1": 0.9728,
683
- "any_label_recall": 0.9852,
684
- "any_label_fpr": 0.0031
685
  },
686
  {
687
  "threshold": 0.39,
688
- "macro_f1": 0.9728,
689
- "any_label_recall": 0.9852,
690
- "any_label_fpr": 0.0031
691
  },
692
  {
693
  "threshold": 0.4,
694
- "macro_f1": 0.9728,
695
- "any_label_recall": 0.9852,
696
- "any_label_fpr": 0.0031
697
  },
698
  {
699
  "threshold": 0.41,
700
- "macro_f1": 0.9732,
701
- "any_label_recall": 0.9852,
702
- "any_label_fpr": 0.0028
703
  },
704
  {
705
  "threshold": 0.42,
706
- "macro_f1": 0.9732,
707
- "any_label_recall": 0.9852,
708
- "any_label_fpr": 0.0028
709
  },
710
  {
711
  "threshold": 0.43,
712
- "macro_f1": 0.9732,
713
- "any_label_recall": 0.9852,
714
- "any_label_fpr": 0.0028
715
  },
716
  {
717
  "threshold": 0.44,
718
- "macro_f1": 0.9732,
719
- "any_label_recall": 0.9852,
720
- "any_label_fpr": 0.0028
721
  },
722
  {
723
  "threshold": 0.45,
724
- "macro_f1": 0.9732,
725
- "any_label_recall": 0.9852,
726
- "any_label_fpr": 0.0028
727
  },
728
  {
729
  "threshold": 0.46,
730
- "macro_f1": 0.9732,
731
- "any_label_recall": 0.9852,
732
- "any_label_fpr": 0.0028
733
  },
734
  {
735
  "threshold": 0.47,
736
- "macro_f1": 0.9728,
737
- "any_label_recall": 0.9852,
738
- "any_label_fpr": 0.0028
739
  },
740
  {
741
  "threshold": 0.48,
742
- "macro_f1": 0.9732,
743
- "any_label_recall": 0.9852,
744
- "any_label_fpr": 0.0025
745
  },
746
  {
747
  "threshold": 0.49,
748
- "macro_f1": 0.9732,
749
- "any_label_recall": 0.9852,
750
- "any_label_fpr": 0.0025
751
  },
752
  {
753
  "threshold": 0.5,
754
- "macro_f1": 0.9728,
755
- "any_label_recall": 0.9852,
756
- "any_label_fpr": 0.0025
757
  },
758
  {
759
  "threshold": 0.51,
760
- "macro_f1": 0.9728,
761
- "any_label_recall": 0.9852,
762
- "any_label_fpr": 0.0025
763
  },
764
  {
765
  "threshold": 0.52,
766
- "macro_f1": 0.9724,
767
- "any_label_recall": 0.9852,
768
- "any_label_fpr": 0.0025
769
  },
770
  {
771
  "threshold": 0.53,
772
- "macro_f1": 0.9724,
773
- "any_label_recall": 0.9852,
774
- "any_label_fpr": 0.0025
775
  },
776
  {
777
  "threshold": 0.54,
778
- "macro_f1": 0.9724,
779
- "any_label_recall": 0.9852,
780
- "any_label_fpr": 0.0025
781
  },
782
  {
783
  "threshold": 0.55,
784
- "macro_f1": 0.9724,
785
- "any_label_recall": 0.9852,
786
- "any_label_fpr": 0.0025
787
  },
788
  {
789
  "threshold": 0.56,
790
- "macro_f1": 0.9721,
791
- "any_label_recall": 0.9852,
792
- "any_label_fpr": 0.0025
793
  },
794
  {
795
  "threshold": 0.57,
796
- "macro_f1": 0.9721,
797
- "any_label_recall": 0.9852,
798
- "any_label_fpr": 0.0025
799
  },
800
  {
801
  "threshold": 0.58,
802
- "macro_f1": 0.9724,
803
  "any_label_recall": 0.9852,
804
- "any_label_fpr": 0.0022
805
  },
806
  {
807
  "threshold": 0.59,
808
- "macro_f1": 0.9724,
809
  "any_label_recall": 0.9852,
810
- "any_label_fpr": 0.0022
811
  },
812
  {
813
  "threshold": 0.6,
814
- "macro_f1": 0.9724,
815
- "any_label_recall": 0.9852,
816
- "any_label_fpr": 0.0022
817
  },
818
  {
819
  "threshold": 0.61,
820
- "macro_f1": 0.9724,
821
- "any_label_recall": 0.9852,
822
- "any_label_fpr": 0.0022
823
  },
824
  {
825
  "threshold": 0.62,
826
- "macro_f1": 0.9724,
827
- "any_label_recall": 0.9852,
828
- "any_label_fpr": 0.0022
829
  },
830
  {
831
  "threshold": 0.63,
832
- "macro_f1": 0.9724,
833
- "any_label_recall": 0.9852,
834
- "any_label_fpr": 0.0022
835
  },
836
  {
837
  "threshold": 0.64,
838
- "macro_f1": 0.9724,
839
- "any_label_recall": 0.9852,
840
- "any_label_fpr": 0.0022
841
  },
842
  {
843
  "threshold": 0.65,
844
- "macro_f1": 0.9724,
845
- "any_label_recall": 0.9852,
846
- "any_label_fpr": 0.0022
847
  },
848
  {
849
  "threshold": 0.66,
850
- "macro_f1": 0.9724,
851
- "any_label_recall": 0.9852,
852
- "any_label_fpr": 0.0022
853
  },
854
  {
855
  "threshold": 0.67,
856
- "macro_f1": 0.9724,
857
- "any_label_recall": 0.9852,
858
- "any_label_fpr": 0.0022
859
  },
860
  {
861
  "threshold": 0.68,
862
- "macro_f1": 0.9724,
863
- "any_label_recall": 0.9852,
864
- "any_label_fpr": 0.0022
865
  },
866
  {
867
  "threshold": 0.69,
868
- "macro_f1": 0.9724,
869
- "any_label_recall": 0.9852,
870
- "any_label_fpr": 0.0022
871
  },
872
  {
873
  "threshold": 0.7,
874
- "macro_f1": 0.9724,
875
- "any_label_recall": 0.9852,
876
- "any_label_fpr": 0.0022
877
  },
878
  {
879
  "threshold": 0.71,
880
- "macro_f1": 0.9724,
881
- "any_label_recall": 0.9852,
882
- "any_label_fpr": 0.0022
883
  },
884
  {
885
  "threshold": 0.72,
886
- "macro_f1": 0.9724,
887
- "any_label_recall": 0.9852,
888
- "any_label_fpr": 0.0022
889
  },
890
  {
891
  "threshold": 0.73,
892
- "macro_f1": 0.9724,
893
- "any_label_recall": 0.9852,
894
- "any_label_fpr": 0.0022
895
  },
896
  {
897
  "threshold": 0.74,
898
- "macro_f1": 0.9717,
899
- "any_label_recall": 0.9852,
900
- "any_label_fpr": 0.0022
901
  },
902
  {
903
  "threshold": 0.75,
904
- "macro_f1": 0.9717,
905
- "any_label_recall": 0.9852,
906
- "any_label_fpr": 0.0022
907
  },
908
  {
909
  "threshold": 0.76,
910
- "macro_f1": 0.9713,
911
- "any_label_recall": 0.9852,
912
- "any_label_fpr": 0.0022
913
  },
914
  {
915
  "threshold": 0.77,
916
- "macro_f1": 0.9709,
917
- "any_label_recall": 0.9852,
918
- "any_label_fpr": 0.0022
919
  },
920
  {
921
  "threshold": 0.78,
922
- "macro_f1": 0.9706,
923
- "any_label_recall": 0.9852,
924
- "any_label_fpr": 0.0022
925
  },
926
  {
927
  "threshold": 0.79,
928
- "macro_f1": 0.9702,
929
- "any_label_recall": 0.9852,
930
- "any_label_fpr": 0.0022
931
  },
932
  {
933
  "threshold": 0.8,
934
- "macro_f1": 0.9702,
935
- "any_label_recall": 0.9852,
936
- "any_label_fpr": 0.0022
937
  },
938
  {
939
  "threshold": 0.81,
940
- "macro_f1": 0.9702,
941
- "any_label_recall": 0.9852,
942
- "any_label_fpr": 0.0022
943
  },
944
  {
945
  "threshold": 0.82,
946
- "macro_f1": 0.9702,
947
- "any_label_recall": 0.9852,
948
- "any_label_fpr": 0.0022
949
  },
950
  {
951
  "threshold": 0.83,
952
- "macro_f1": 0.9698,
953
- "any_label_recall": 0.9852,
954
- "any_label_fpr": 0.0022
955
  },
956
  {
957
  "threshold": 0.84,
958
- "macro_f1": 0.9698,
959
- "any_label_recall": 0.9852,
960
- "any_label_fpr": 0.0022
961
  },
962
  {
963
  "threshold": 0.85,
964
- "macro_f1": 0.9698,
965
- "any_label_recall": 0.9852,
966
- "any_label_fpr": 0.0022
967
  },
968
  {
969
  "threshold": 0.86,
970
- "macro_f1": 0.9692,
971
- "any_label_recall": 0.9843,
972
- "any_label_fpr": 0.0022
973
  },
974
  {
975
  "threshold": 0.87,
976
- "macro_f1": 0.9689,
977
- "any_label_recall": 0.9833,
978
- "any_label_fpr": 0.0022
979
  },
980
  {
981
  "threshold": 0.88,
982
- "macro_f1": 0.9685,
983
- "any_label_recall": 0.9833,
984
- "any_label_fpr": 0.0022
985
  },
986
  {
987
  "threshold": 0.89,
988
- "macro_f1": 0.9685,
989
- "any_label_recall": 0.9833,
990
- "any_label_fpr": 0.0022
991
  },
992
  {
993
  "threshold": 0.9,
994
- "macro_f1": 0.9679,
995
- "any_label_recall": 0.9824,
996
- "any_label_fpr": 0.0022
997
  },
998
  {
999
  "threshold": 0.91,
1000
- "macro_f1": 0.9679,
1001
- "any_label_recall": 0.9824,
1002
- "any_label_fpr": 0.0022
1003
  },
1004
  {
1005
  "threshold": 0.92,
1006
- "macro_f1": 0.9679,
1007
- "any_label_recall": 0.9824,
1008
- "any_label_fpr": 0.0022
1009
  },
1010
  {
1011
  "threshold": 0.93,
1012
- "macro_f1": 0.9672,
1013
- "any_label_recall": 0.9815,
1014
- "any_label_fpr": 0.0022
1015
  },
1016
  {
1017
  "threshold": 0.94,
1018
- "macro_f1": 0.9671,
1019
- "any_label_recall": 0.9796,
1020
- "any_label_fpr": 0.0022
1021
  },
1022
  {
1023
  "threshold": 0.95,
1024
- "macro_f1": 0.9671,
1025
- "any_label_recall": 0.9796,
1026
- "any_label_fpr": 0.0022
1027
  }
1028
  ],
1029
  "notes": [
 
8
  "recall": 1.0,
9
  "f1": 1.0,
10
  "fpr": 0.0,
11
+ "accuracy": 0.9943,
12
+ "n": 174
13
  },
14
  "bg": {
15
  "support": 41,
 
17
  "recall": 1.0,
18
  "f1": 1.0,
19
  "fpr": 0.0,
20
+ "accuracy": 0.9884,
21
+ "n": 172
22
  },
23
  "cs": {
24
  "support": 42,
25
  "precision": 0.9545,
26
  "recall": 1.0,
27
  "f1": 0.9767,
28
+ "fpr": 0.0152,
29
+ "accuracy": 0.9828,
30
+ "n": 174
31
  },
32
  "da": {
33
  "support": 41,
34
  "precision": 0.9762,
35
  "recall": 1.0,
36
  "f1": 0.988,
37
+ "fpr": 0.0076,
38
+ "accuracy": 0.9942,
39
+ "n": 173
40
  },
41
  "de": {
42
  "support": 42,
 
44
  "recall": 1.0,
45
  "f1": 1.0,
46
  "fpr": 0.0,
47
+ "accuracy": 0.9943,
48
+ "n": 174
49
  },
50
  "el": {
51
  "support": 42,
 
53
  "recall": 1.0,
54
  "f1": 1.0,
55
  "fpr": 0.0,
56
+ "accuracy": 0.9884,
57
+ "n": 173
58
  },
59
  "en": {
60
  "support": 41,
61
+ "precision": 1.0,
62
  "recall": 1.0,
63
+ "f1": 1.0,
64
+ "fpr": 0.0,
65
+ "accuracy": 0.9942,
66
+ "n": 173
67
  },
68
  "es": {
69
  "support": 40,
70
+ "precision": 0.9756,
71
  "recall": 1.0,
72
+ "f1": 0.9877,
73
+ "fpr": 0.0076,
74
+ "accuracy": 0.9884,
75
+ "n": 172
76
  },
77
  "et": {
78
  "support": 41,
 
80
  "recall": 1.0,
81
  "f1": 1.0,
82
  "fpr": 0.0,
83
+ "accuracy": 0.9942,
84
+ "n": 173
85
  },
86
  "fi": {
87
  "support": 42,
 
90
  "f1": 1.0,
91
  "fpr": 0.0,
92
  "accuracy": 1.0,
93
+ "n": 174
94
  },
95
  "fr": {
96
  "support": 41,
97
  "precision": 1.0,
98
+ "recall": 1.0,
99
+ "f1": 1.0,
100
  "fpr": 0.0,
101
+ "accuracy": 0.9884,
102
+ "n": 173
103
  },
104
  "ga": {
105
  "support": 42,
106
  "precision": 0.9762,
107
  "recall": 0.9762,
108
  "f1": 0.9762,
109
+ "fpr": 0.0076,
110
+ "accuracy": 0.977,
111
+ "n": 174
112
  },
113
  "hr": {
114
  "support": 42,
115
+ "precision": 0.9767,
116
  "recall": 1.0,
117
+ "f1": 0.9882,
118
+ "fpr": 0.0076,
119
+ "accuracy": 0.9943,
120
+ "n": 174
121
  },
122
  "hu": {
123
  "support": 42,
124
  "precision": 0.9545,
125
  "recall": 1.0,
126
  "f1": 0.9767,
127
+ "fpr": 0.0152,
128
+ "accuracy": 0.977,
129
+ "n": 174
130
  },
131
  "it": {
132
  "support": 42,
 
135
  "f1": 1.0,
136
  "fpr": 0.0,
137
  "accuracy": 1.0,
138
+ "n": 174
139
  },
140
  "lt": {
141
  "support": 42,
 
143
  "recall": 1.0,
144
  "f1": 1.0,
145
  "fpr": 0.0,
146
+ "accuracy": 0.9943,
147
+ "n": 174
148
  },
149
  "lv": {
150
  "support": 41,
151
  "precision": 0.9762,
152
  "recall": 1.0,
153
  "f1": 0.988,
154
+ "fpr": 0.0076,
155
+ "accuracy": 0.9884,
156
+ "n": 173
157
  },
158
  "mt": {
159
  "support": 42,
160
+ "precision": 0.8039,
161
  "recall": 0.9762,
162
+ "f1": 0.8817,
163
+ "fpr": 0.0758,
164
+ "accuracy": 0.9195,
165
+ "n": 174
166
  },
167
  "nl": {
168
  "support": 41,
169
+ "precision": 1.0,
170
+ "recall": 0.9756,
171
+ "f1": 0.9877,
172
+ "fpr": 0.0,
173
+ "accuracy": 0.9942,
174
+ "n": 173
175
  },
176
  "pl": {
177
  "support": 41,
 
179
  "recall": 1.0,
180
  "f1": 1.0,
181
  "fpr": 0.0,
182
+ "accuracy": 1.0,
183
+ "n": 173
184
  },
185
  "pt": {
186
  "support": 42,
187
  "precision": 0.9767,
188
  "recall": 1.0,
189
  "f1": 0.9882,
190
+ "fpr": 0.0076,
191
+ "accuracy": 0.977,
192
+ "n": 174
193
  },
194
  "ro": {
195
  "support": 42,
196
  "precision": 0.9767,
197
  "recall": 1.0,
198
  "f1": 0.9882,
199
+ "fpr": 0.0076,
200
+ "accuracy": 0.9884,
201
+ "n": 173
202
  },
203
  "sk": {
204
  "support": 41,
 
206
  "recall": 1.0,
207
  "f1": 1.0,
208
  "fpr": 0.0,
209
+ "accuracy": 0.9942,
210
+ "n": 173
211
  },
212
  "sl": {
213
  "support": 42,
214
  "precision": 0.9767,
215
  "recall": 1.0,
216
  "f1": 0.9882,
217
+ "fpr": 0.0076,
218
+ "accuracy": 0.977,
219
+ "n": 174
220
  },
221
  "sv": {
222
  "support": 41,
 
224
  "recall": 1.0,
225
  "f1": 1.0,
226
  "fpr": 0.0,
227
+ "accuracy": 0.9884,
228
+ "n": 173
229
  },
230
  "tr": {
231
  "support": 42,
232
+ "precision": 1.0,
233
  "recall": 1.0,
234
+ "f1": 1.0,
235
+ "fpr": 0.0,
236
+ "accuracy": 0.9943,
237
+ "n": 174
238
  }
239
  },
240
  "per_register": {
241
  "encoded_payload": {
242
  "support": 155,
243
  "precision": 1.0,
244
+ "recall": 1.0,
245
+ "f1": 1.0,
246
  "fpr": 0.0,
247
+ "accuracy": 0.9677,
248
  "n": 155
249
  },
250
  "hypothetical_framing": {
251
  "support": 156,
252
  "precision": 1.0,
253
+ "recall": 0.9936,
254
+ "f1": 0.9968,
255
  "fpr": 0.0,
256
+ "accuracy": 0.9808,
257
  "n": 156
258
  },
259
  "ignore_instructions": {
 
262
  "recall": 1.0,
263
  "f1": 1.0,
264
  "fpr": 0.0,
265
+ "accuracy": 0.9419,
266
  "n": 155
267
  },
268
  "instruction_in_document": {
269
  "support": 155,
270
  "precision": 1.0,
271
+ "recall": 0.9935,
272
+ "f1": 0.9968,
273
  "fpr": 0.0,
274
+ "accuracy": 0.9677,
275
  "n": 155
276
  },
277
  "instruction_in_tool_output": {
278
  "support": 148,
279
  "precision": 1.0,
280
+ "recall": 0.9932,
281
+ "f1": 0.9966,
282
  "fpr": 0.0,
283
+ "accuracy": 0.973,
284
  "n": 148
285
  },
286
  "meta_question": {
 
288
  "precision": 0.0,
289
  "recall": 0.0,
290
  "f1": 0.0,
291
+ "fpr": 0.0118,
292
+ "accuracy": 0.9882,
293
  "n": 338
294
  },
295
+ "mundane_account_access": {
296
+ "support": 0,
297
+ "precision": 0.0,
298
+ "recall": 0.0,
299
+ "f1": 0.0,
300
+ "fpr": 0.0,
301
+ "accuracy": 1.0,
302
+ "n": 182
303
+ },
304
  "mundane_informational": {
305
  "support": 0,
306
  "precision": 0.0,
 
333
  "precision": 0.0,
334
  "recall": 0.0,
335
  "f1": 0.0,
336
+ "fpr": 0.0,
337
+ "accuracy": 1.0,
338
  "n": 338
339
  },
340
  "ordinary_question": {
 
342
  "precision": 0.0,
343
  "recall": 0.0,
344
  "f1": 0.0,
345
+ "fpr": 0.0059,
346
+ "accuracy": 0.9941,
347
  "n": 338
348
  },
349
  "persona_override": {
 
352
  "recall": 1.0,
353
  "f1": 1.0,
354
  "fpr": 0.0,
355
+ "accuracy": 0.9484,
356
  "n": 155
357
  },
358
  "quoted_attack": {
 
360
  "precision": 0.0,
361
  "recall": 0.0,
362
  "f1": 0.0,
363
+ "fpr": 0.0089,
364
+ "accuracy": 0.9911,
365
  "n": 338
366
  },
367
  "roleplay_benign": {
 
369
  "precision": 0.0,
370
  "recall": 0.0,
371
  "f1": 0.0,
372
+ "fpr": 0.0118,
373
+ "accuracy": 0.9882,
374
  "n": 338
375
  },
376
  "security_discussion": {
 
378
  "precision": 0.0,
379
  "recall": 0.0,
380
  "f1": 0.0,
381
+ "fpr": 0.0089,
382
+ "accuracy": 0.9911,
383
  "n": 338
384
  },
385
  "system_prompt_extraction": {
 
388
  "recall": 1.0,
389
  "f1": 1.0,
390
  "fpr": 0.0,
391
+ "accuracy": 0.9936,
392
  "n": 156
393
  },
394
  "technical_identifiers": {
 
396
  "precision": 0.0,
397
  "recall": 0.0,
398
  "f1": 0.0,
399
+ "fpr": 0.006,
400
+ "accuracy": 0.994,
401
  "n": 336
402
  },
403
  "technical_payload": {
 
405
  "precision": 0.0,
406
  "recall": 0.0,
407
  "f1": 0.0,
408
+ "fpr": 0.0059,
409
+ "accuracy": 0.9941,
410
  "n": 337
411
  }
412
  },
413
  "per_label": {
414
  "direct_injection": {
415
  "support": 466,
416
+ "precision": 0.9528,
417
+ "recall": 0.9957,
418
+ "f1": 0.9738,
419
  "fpr": 0.0057,
420
  "accuracy": 0.0,
421
  "n": 0
422
  },
423
  "indirect_injection": {
424
  "support": 303,
425
+ "precision": 0.9709,
426
+ "recall": 0.9901,
427
+ "f1": 0.9804,
428
+ "fpr": 0.0021,
429
  "accuracy": 0.0,
430
  "n": 0
431
  },
432
  "jailbreak": {
433
  "support": 466,
434
+ "precision": 0.9367,
435
+ "recall": 0.985,
436
+ "f1": 0.9603,
437
+ "fpr": 0.0077,
438
  "accuracy": 0.0,
439
  "n": 0
440
  }
 
442
  "worst_languages": [
443
  [
444
  "mt",
445
+ 0.8817204301075269
446
  ],
447
  [
448
  "ga",
 
453
  0.9767441860465117
454
  ],
455
  [
456
+ "hu",
457
  0.9767441860465117
458
  ],
459
  [
460
+ "es",
461
+ 0.9876543209876543
462
  ]
463
  ],
464
  "calibration": {
465
  "threshold": 0.02,
466
  "objective": "recall_at_fpr",
467
+ "value_on_val": 0.9907,
468
  "default_threshold": 0.5,
469
+ "default_value_on_val": 0.9861,
470
+ "improvement": 0.0046,
471
  "chosen_on": "validation split, never test",
472
  "curve": [
473
  {
474
  "threshold": 0.02,
475
+ "macro_f1": 0.9678,
476
+ "any_label_recall": 0.9907,
477
+ "any_label_fpr": 0.007
478
  },
479
  {
480
  "threshold": 0.03,
481
+ "macro_f1": 0.9693,
482
+ "any_label_recall": 0.9898,
483
+ "any_label_fpr": 0.0064
484
  },
485
  {
486
  "threshold": 0.04,
487
+ "macro_f1": 0.9705,
488
  "any_label_recall": 0.9898,
489
+ "any_label_fpr": 0.0058
490
  },
491
  {
492
  "threshold": 0.05,
493
+ "macro_f1": 0.9712,
494
  "any_label_recall": 0.9898,
495
+ "any_label_fpr": 0.0055
496
  },
497
  {
498
  "threshold": 0.06,
499
+ "macro_f1": 0.9729,
500
  "any_label_recall": 0.9898,
501
+ "any_label_fpr": 0.0052
502
  },
503
  {
504
  "threshold": 0.07,
505
+ "macro_f1": 0.9725,
506
  "any_label_recall": 0.9898,
507
+ "any_label_fpr": 0.0052
508
  },
509
  {
510
  "threshold": 0.08,
511
+ "macro_f1": 0.9723,
512
  "any_label_recall": 0.9898,
513
+ "any_label_fpr": 0.005
514
  },
515
  {
516
  "threshold": 0.09,
517
+ "macro_f1": 0.9727,
518
+ "any_label_recall": 0.9898,
519
+ "any_label_fpr": 0.005
520
  },
521
  {
522
  "threshold": 0.1,
523
+ "macro_f1": 0.9737,
524
+ "any_label_recall": 0.9898,
525
+ "any_label_fpr": 0.0044
526
  },
527
  {
528
  "threshold": 0.11,
529
+ "macro_f1": 0.9733,
530
+ "any_label_recall": 0.9898,
531
+ "any_label_fpr": 0.0044
532
  },
533
  {
534
  "threshold": 0.12,
535
+ "macro_f1": 0.9733,
536
+ "any_label_recall": 0.9898,
537
+ "any_label_fpr": 0.0044
538
  },
539
  {
540
  "threshold": 0.13,
541
+ "macro_f1": 0.9733,
542
+ "any_label_recall": 0.9898,
543
+ "any_label_fpr": 0.0044
544
  },
545
  {
546
  "threshold": 0.14,
547
+ "macro_f1": 0.9733,
548
+ "any_label_recall": 0.9889,
549
+ "any_label_fpr": 0.0044
550
  },
551
  {
552
  "threshold": 0.15,
553
+ "macro_f1": 0.9731,
554
+ "any_label_recall": 0.988,
555
+ "any_label_fpr": 0.0041
556
  },
557
  {
558
  "threshold": 0.16,
559
+ "macro_f1": 0.9731,
560
+ "any_label_recall": 0.988,
561
+ "any_label_fpr": 0.0041
562
  },
563
  {
564
  "threshold": 0.17,
565
+ "macro_f1": 0.9727,
566
+ "any_label_recall": 0.988,
567
+ "any_label_fpr": 0.0041
568
  },
569
  {
570
  "threshold": 0.18,
571
+ "macro_f1": 0.9722,
572
+ "any_label_recall": 0.988,
573
+ "any_label_fpr": 0.0041
574
  },
575
  {
576
  "threshold": 0.19,
577
+ "macro_f1": 0.9718,
578
+ "any_label_recall": 0.988,
579
+ "any_label_fpr": 0.0041
580
  },
581
  {
582
  "threshold": 0.2,
583
+ "macro_f1": 0.9718,
584
+ "any_label_recall": 0.988,
585
+ "any_label_fpr": 0.0041
586
  },
587
  {
588
  "threshold": 0.21,
589
+ "macro_f1": 0.9718,
590
+ "any_label_recall": 0.988,
591
+ "any_label_fpr": 0.0041
592
  },
593
  {
594
  "threshold": 0.22,
595
+ "macro_f1": 0.9718,
596
+ "any_label_recall": 0.988,
597
+ "any_label_fpr": 0.0041
598
  },
599
  {
600
  "threshold": 0.23,
601
+ "macro_f1": 0.9718,
602
+ "any_label_recall": 0.988,
603
+ "any_label_fpr": 0.0041
604
  },
605
  {
606
  "threshold": 0.24,
607
+ "macro_f1": 0.9714,
608
+ "any_label_recall": 0.988,
609
+ "any_label_fpr": 0.0041
610
  },
611
  {
612
  "threshold": 0.25,
613
+ "macro_f1": 0.9714,
614
+ "any_label_recall": 0.988,
615
+ "any_label_fpr": 0.0041
616
  },
617
  {
618
  "threshold": 0.26,
619
+ "macro_f1": 0.9714,
620
+ "any_label_recall": 0.988,
621
+ "any_label_fpr": 0.0041
622
  },
623
  {
624
  "threshold": 0.27,
625
+ "macro_f1": 0.9718,
626
+ "any_label_recall": 0.988,
627
+ "any_label_fpr": 0.0041
628
  },
629
  {
630
  "threshold": 0.28,
631
+ "macro_f1": 0.9718,
632
+ "any_label_recall": 0.988,
633
+ "any_label_fpr": 0.0041
634
  },
635
  {
636
  "threshold": 0.29,
637
+ "macro_f1": 0.9718,
638
+ "any_label_recall": 0.988,
639
+ "any_label_fpr": 0.0041
640
  },
641
  {
642
  "threshold": 0.3,
643
+ "macro_f1": 0.9718,
644
+ "any_label_recall": 0.988,
645
+ "any_label_fpr": 0.0041
646
  },
647
  {
648
  "threshold": 0.31,
649
+ "macro_f1": 0.9714,
650
+ "any_label_recall": 0.987,
651
+ "any_label_fpr": 0.0041
652
  },
653
  {
654
  "threshold": 0.32,
655
+ "macro_f1": 0.9714,
656
+ "any_label_recall": 0.987,
657
+ "any_label_fpr": 0.0041
658
  },
659
  {
660
  "threshold": 0.33,
661
+ "macro_f1": 0.9717,
662
+ "any_label_recall": 0.987,
663
+ "any_label_fpr": 0.0038
664
  },
665
  {
666
  "threshold": 0.34,
667
+ "macro_f1": 0.9717,
668
+ "any_label_recall": 0.987,
669
+ "any_label_fpr": 0.0038
670
  },
671
  {
672
  "threshold": 0.35,
673
+ "macro_f1": 0.9714,
674
+ "any_label_recall": 0.987,
675
+ "any_label_fpr": 0.0038
676
  },
677
  {
678
  "threshold": 0.36,
679
+ "macro_f1": 0.9713,
680
+ "any_label_recall": 0.987,
681
+ "any_label_fpr": 0.0038
682
  },
683
  {
684
  "threshold": 0.37,
685
+ "macro_f1": 0.9713,
686
+ "any_label_recall": 0.987,
687
+ "any_label_fpr": 0.0038
688
  },
689
  {
690
  "threshold": 0.38,
691
+ "macro_f1": 0.971,
692
+ "any_label_recall": 0.9861,
693
+ "any_label_fpr": 0.0038
694
  },
695
  {
696
  "threshold": 0.39,
697
+ "macro_f1": 0.971,
698
+ "any_label_recall": 0.9861,
699
+ "any_label_fpr": 0.0038
700
  },
701
  {
702
  "threshold": 0.4,
703
+ "macro_f1": 0.971,
704
+ "any_label_recall": 0.9861,
705
+ "any_label_fpr": 0.0038
706
  },
707
  {
708
  "threshold": 0.41,
709
+ "macro_f1": 0.971,
710
+ "any_label_recall": 0.9861,
711
+ "any_label_fpr": 0.0038
712
  },
713
  {
714
  "threshold": 0.42,
715
+ "macro_f1": 0.971,
716
+ "any_label_recall": 0.9861,
717
+ "any_label_fpr": 0.0038
718
  },
719
  {
720
  "threshold": 0.43,
721
+ "macro_f1": 0.9706,
722
+ "any_label_recall": 0.9861,
723
+ "any_label_fpr": 0.0038
724
  },
725
  {
726
  "threshold": 0.44,
727
+ "macro_f1": 0.9706,
728
+ "any_label_recall": 0.9861,
729
+ "any_label_fpr": 0.0038
730
  },
731
  {
732
  "threshold": 0.45,
733
+ "macro_f1": 0.971,
734
+ "any_label_recall": 0.9861,
735
+ "any_label_fpr": 0.0038
736
  },
737
  {
738
  "threshold": 0.46,
739
+ "macro_f1": 0.971,
740
+ "any_label_recall": 0.9861,
741
+ "any_label_fpr": 0.0038
742
  },
743
  {
744
  "threshold": 0.47,
745
+ "macro_f1": 0.9704,
746
+ "any_label_recall": 0.9861,
747
+ "any_label_fpr": 0.0038
748
  },
749
  {
750
  "threshold": 0.48,
751
+ "macro_f1": 0.9704,
752
+ "any_label_recall": 0.9861,
753
+ "any_label_fpr": 0.0038
754
  },
755
  {
756
  "threshold": 0.49,
757
+ "macro_f1": 0.9704,
758
+ "any_label_recall": 0.9861,
759
+ "any_label_fpr": 0.0038
760
  },
761
  {
762
  "threshold": 0.5,
763
+ "macro_f1": 0.9704,
764
+ "any_label_recall": 0.9861,
765
+ "any_label_fpr": 0.0038
766
  },
767
  {
768
  "threshold": 0.51,
769
+ "macro_f1": 0.9704,
770
+ "any_label_recall": 0.9861,
771
+ "any_label_fpr": 0.0038
772
  },
773
  {
774
  "threshold": 0.52,
775
+ "macro_f1": 0.9704,
776
+ "any_label_recall": 0.9861,
777
+ "any_label_fpr": 0.0038
778
  },
779
  {
780
  "threshold": 0.53,
781
+ "macro_f1": 0.9704,
782
+ "any_label_recall": 0.9861,
783
+ "any_label_fpr": 0.0038
784
  },
785
  {
786
  "threshold": 0.54,
787
+ "macro_f1": 0.9707,
788
+ "any_label_recall": 0.9861,
789
+ "any_label_fpr": 0.0035
790
  },
791
  {
792
  "threshold": 0.55,
793
+ "macro_f1": 0.9707,
794
+ "any_label_recall": 0.9861,
795
+ "any_label_fpr": 0.0035
796
  },
797
  {
798
  "threshold": 0.56,
799
+ "macro_f1": 0.9707,
800
+ "any_label_recall": 0.9861,
801
+ "any_label_fpr": 0.0035
802
  },
803
  {
804
  "threshold": 0.57,
805
+ "macro_f1": 0.9707,
806
+ "any_label_recall": 0.9861,
807
+ "any_label_fpr": 0.0035
808
  },
809
  {
810
  "threshold": 0.58,
811
+ "macro_f1": 0.9702,
812
  "any_label_recall": 0.9852,
813
+ "any_label_fpr": 0.0035
814
  },
815
  {
816
  "threshold": 0.59,
817
+ "macro_f1": 0.9702,
818
  "any_label_recall": 0.9852,
819
+ "any_label_fpr": 0.0035
820
  },
821
  {
822
  "threshold": 0.6,
823
+ "macro_f1": 0.9705,
824
+ "any_label_recall": 0.9843,
825
+ "any_label_fpr": 0.0035
826
  },
827
  {
828
  "threshold": 0.61,
829
+ "macro_f1": 0.9705,
830
+ "any_label_recall": 0.9843,
831
+ "any_label_fpr": 0.0035
832
  },
833
  {
834
  "threshold": 0.62,
835
+ "macro_f1": 0.9705,
836
+ "any_label_recall": 0.9843,
837
+ "any_label_fpr": 0.0035
838
  },
839
  {
840
  "threshold": 0.63,
841
+ "macro_f1": 0.9705,
842
+ "any_label_recall": 0.9843,
843
+ "any_label_fpr": 0.0035
844
  },
845
  {
846
  "threshold": 0.64,
847
+ "macro_f1": 0.9705,
848
+ "any_label_recall": 0.9843,
849
+ "any_label_fpr": 0.0035
850
  },
851
  {
852
  "threshold": 0.65,
853
+ "macro_f1": 0.9709,
854
+ "any_label_recall": 0.9843,
855
+ "any_label_fpr": 0.0035
856
  },
857
  {
858
  "threshold": 0.66,
859
+ "macro_f1": 0.9709,
860
+ "any_label_recall": 0.9843,
861
+ "any_label_fpr": 0.0035
862
  },
863
  {
864
  "threshold": 0.67,
865
+ "macro_f1": 0.9712,
866
+ "any_label_recall": 0.9843,
867
+ "any_label_fpr": 0.0032
868
  },
869
  {
870
  "threshold": 0.68,
871
+ "macro_f1": 0.9712,
872
+ "any_label_recall": 0.9843,
873
+ "any_label_fpr": 0.0032
874
  },
875
  {
876
  "threshold": 0.69,
877
+ "macro_f1": 0.9716,
878
+ "any_label_recall": 0.9843,
879
+ "any_label_fpr": 0.0032
880
  },
881
  {
882
  "threshold": 0.7,
883
+ "macro_f1": 0.9712,
884
+ "any_label_recall": 0.9843,
885
+ "any_label_fpr": 0.0032
886
  },
887
  {
888
  "threshold": 0.71,
889
+ "macro_f1": 0.9712,
890
+ "any_label_recall": 0.9843,
891
+ "any_label_fpr": 0.0032
892
  },
893
  {
894
  "threshold": 0.72,
895
+ "macro_f1": 0.9708,
896
+ "any_label_recall": 0.9843,
897
+ "any_label_fpr": 0.0032
898
  },
899
  {
900
  "threshold": 0.73,
901
+ "macro_f1": 0.9708,
902
+ "any_label_recall": 0.9843,
903
+ "any_label_fpr": 0.0032
904
  },
905
  {
906
  "threshold": 0.74,
907
+ "macro_f1": 0.9708,
908
+ "any_label_recall": 0.9843,
909
+ "any_label_fpr": 0.0032
910
  },
911
  {
912
  "threshold": 0.75,
913
+ "macro_f1": 0.9708,
914
+ "any_label_recall": 0.9843,
915
+ "any_label_fpr": 0.0032
916
  },
917
  {
918
  "threshold": 0.76,
919
+ "macro_f1": 0.9708,
920
+ "any_label_recall": 0.9843,
921
+ "any_label_fpr": 0.0032
922
  },
923
  {
924
  "threshold": 0.77,
925
+ "macro_f1": 0.9708,
926
+ "any_label_recall": 0.9843,
927
+ "any_label_fpr": 0.0032
928
  },
929
  {
930
  "threshold": 0.78,
931
+ "macro_f1": 0.9704,
932
+ "any_label_recall": 0.9833,
933
+ "any_label_fpr": 0.0032
934
  },
935
  {
936
  "threshold": 0.79,
937
+ "macro_f1": 0.9704,
938
+ "any_label_recall": 0.9833,
939
+ "any_label_fpr": 0.0032
940
  },
941
  {
942
  "threshold": 0.8,
943
+ "macro_f1": 0.9704,
944
+ "any_label_recall": 0.9833,
945
+ "any_label_fpr": 0.0032
946
  },
947
  {
948
  "threshold": 0.81,
949
+ "macro_f1": 0.9708,
950
+ "any_label_recall": 0.9833,
951
+ "any_label_fpr": 0.0029
952
  },
953
  {
954
  "threshold": 0.82,
955
+ "macro_f1": 0.9708,
956
+ "any_label_recall": 0.9833,
957
+ "any_label_fpr": 0.0029
958
  },
959
  {
960
  "threshold": 0.83,
961
+ "macro_f1": 0.9704,
962
+ "any_label_recall": 0.9833,
963
+ "any_label_fpr": 0.0029
964
  },
965
  {
966
  "threshold": 0.84,
967
+ "macro_f1": 0.971,
968
+ "any_label_recall": 0.9833,
969
+ "any_label_fpr": 0.0029
970
  },
971
  {
972
  "threshold": 0.85,
973
+ "macro_f1": 0.971,
974
+ "any_label_recall": 0.9833,
975
+ "any_label_fpr": 0.0029
976
  },
977
  {
978
  "threshold": 0.86,
979
+ "macro_f1": 0.9706,
980
+ "any_label_recall": 0.9833,
981
+ "any_label_fpr": 0.0029
982
  },
983
  {
984
  "threshold": 0.87,
985
+ "macro_f1": 0.9696,
986
+ "any_label_recall": 0.9824,
987
+ "any_label_fpr": 0.0029
988
  },
989
  {
990
  "threshold": 0.88,
991
+ "macro_f1": 0.9693,
992
+ "any_label_recall": 0.9815,
993
+ "any_label_fpr": 0.0029
994
  },
995
  {
996
  "threshold": 0.89,
997
+ "macro_f1": 0.9692,
998
+ "any_label_recall": 0.9815,
999
+ "any_label_fpr": 0.0026
1000
  },
1001
  {
1002
  "threshold": 0.9,
1003
+ "macro_f1": 0.9692,
1004
+ "any_label_recall": 0.9815,
1005
+ "any_label_fpr": 0.0026
1006
  },
1007
  {
1008
  "threshold": 0.91,
1009
+ "macro_f1": 0.9692,
1010
+ "any_label_recall": 0.9806,
1011
+ "any_label_fpr": 0.0023
1012
  },
1013
  {
1014
  "threshold": 0.92,
1015
+ "macro_f1": 0.9692,
1016
+ "any_label_recall": 0.9806,
1017
+ "any_label_fpr": 0.0023
1018
  },
1019
  {
1020
  "threshold": 0.93,
1021
+ "macro_f1": 0.9688,
1022
+ "any_label_recall": 0.9806,
1023
+ "any_label_fpr": 0.0023
1024
  },
1025
  {
1026
  "threshold": 0.94,
1027
+ "macro_f1": 0.9688,
1028
+ "any_label_recall": 0.9806,
1029
+ "any_label_fpr": 0.0023
1030
  },
1031
  {
1032
  "threshold": 0.95,
1033
+ "macro_f1": 0.9688,
1034
+ "any_label_recall": 0.9806,
1035
+ "any_label_fpr": 0.0023
1036
  }
1037
  ],
1038
  "notes": [