resources/graders/subresources/grader_models/index.md +0 −1082 deleted
File Deleted View Diff
1# Grader Models
2
3## Domain Types
4
5### Grader Inputs
6
7- `GraderInputs = array of string or ResponseInputText or object { text, type } or 2 more`
8
9 A list of inputs, each of which may be either an input text, output text, input
10 image, or input audio object.
11
12 - `TextInput = string`
13
14 A text input to the model.
15
16 - `ResponseInputText object { text, type, prompt_cache_breakpoint }`
17
18 A text input to the model.
19
20 - `text: string`
21
22 The text input to the model.
23
24 - `type: "input_text"`
25
26 The type of the input item. Always `input_text`.
27
28 - `"input_text"`
29
30 - `prompt_cache_breakpoint: optional object { mode }`
31
32 Marks the exact end of a reusable prompt prefix. The breakpoint inherits its TTL from the request's `prompt_cache_options.ttl`; the boundary is not rounded to a token block.
33
34 - `mode: "explicit"`
35
36 The breakpoint mode. Always `explicit`.
37
38 - `"explicit"`
39
40 - `OutputText object { text, type }`
41
42 A text output from the model.
43
44 - `text: string`
45
46 The text output from the model.
47
48 - `type: "output_text"`
49
50 The type of the output text. Always `output_text`.
51
52 - `"output_text"`
53
54 - `InputImage object { image_url, type, detail }`
55
56 An image input block used within EvalItem content arrays.
57
58 - `image_url: string`
59
60 The URL of the image input.
61
62 - `type: "input_image"`
63
64 The type of the image input. Always `input_image`.
65
66 - `"input_image"`
67
68 - `detail: optional string`
69
70 The detail level of the image to be sent to the model. One of `high`, `low`, or `auto`. Defaults to `auto`.
71
72 - `ResponseInputAudio object { input_audio, type }`
73
74 An audio input to the model.
75
76 - `input_audio: object { data, format }`
77
78 - `data: string`
79
80 Base64-encoded audio data.
81
82 - `format: "mp3" or "wav"`
83
84 The format of the audio data. Currently supported formats are `mp3` and
85 `wav`.
86
87 - `"mp3"`
88
89 - `"wav"`
90
91 - `type: "input_audio"`
92
93 The type of the input item. Always `input_audio`.
94
95 - `"input_audio"`
96
97### Label Model Grader
98
99- `LabelModelGrader object { input, labels, model, 3 more }`
100
101 A LabelModelGrader object which uses a model to assign labels to each item
102 in the evaluation.
103
104 - `input: array of object { content, role, type }`
105
106 - `content: string or ResponseInputText or object { text, type } or 3 more`
107
108 Inputs to the model - can contain template strings. Supports text, output text, input images, and input audio, either as a single item or an array of items.
109
110 - `TextInput = string`
111
112 A text input to the model.
113
114 - `ResponseInputText object { text, type, prompt_cache_breakpoint }`
115
116 A text input to the model.
117
118 - `text: string`
119
120 The text input to the model.
121
122 - `type: "input_text"`
123
124 The type of the input item. Always `input_text`.
125
126 - `"input_text"`
127
128 - `prompt_cache_breakpoint: optional object { mode }`
129
130 Marks the exact end of a reusable prompt prefix. The breakpoint inherits its TTL from the request's `prompt_cache_options.ttl`; the boundary is not rounded to a token block.
131
132 - `mode: "explicit"`
133
134 The breakpoint mode. Always `explicit`.
135
136 - `"explicit"`
137
138 - `OutputText object { text, type }`
139
140 A text output from the model.
141
142 - `text: string`
143
144 The text output from the model.
145
146 - `type: "output_text"`
147
148 The type of the output text. Always `output_text`.
149
150 - `"output_text"`
151
152 - `InputImage object { image_url, type, detail }`
153
154 An image input block used within EvalItem content arrays.
155
156 - `image_url: string`
157
158 The URL of the image input.
159
160 - `type: "input_image"`
161
162 The type of the image input. Always `input_image`.
163
164 - `"input_image"`
165
166 - `detail: optional string`
167
168 The detail level of the image to be sent to the model. One of `high`, `low`, or `auto`. Defaults to `auto`.
169
170 - `ResponseInputAudio object { input_audio, type }`
171
172 An audio input to the model.
173
174 - `input_audio: object { data, format }`
175
176 - `data: string`
177
178 Base64-encoded audio data.
179
180 - `format: "mp3" or "wav"`
181
182 The format of the audio data. Currently supported formats are `mp3` and
183 `wav`.
184
185 - `"mp3"`
186
187 - `"wav"`
188
189 - `type: "input_audio"`
190
191 The type of the input item. Always `input_audio`.
192
193 - `"input_audio"`
194
195 - `GraderInputs = array of string or ResponseInputText or object { text, type } or 2 more`
196
197 A list of inputs, each of which may be either an input text, output text, input
198 image, or input audio object.
199
200 - `TextInput = string`
201
202 A text input to the model.
203
204 - `ResponseInputText object { text, type, prompt_cache_breakpoint }`
205
206 A text input to the model.
207
208 - `OutputText object { text, type }`
209
210 A text output from the model.
211
212 - `text: string`
213
214 The text output from the model.
215
216 - `type: "output_text"`
217
218 The type of the output text. Always `output_text`.
219
220 - `"output_text"`
221
222 - `InputImage object { image_url, type, detail }`
223
224 An image input block used within EvalItem content arrays.
225
226 - `image_url: string`
227
228 The URL of the image input.
229
230 - `type: "input_image"`
231
232 The type of the image input. Always `input_image`.
233
234 - `"input_image"`
235
236 - `detail: optional string`
237
238 The detail level of the image to be sent to the model. One of `high`, `low`, or `auto`. Defaults to `auto`.
239
240 - `ResponseInputAudio object { input_audio, type }`
241
242 An audio input to the model.
243
244 - `role: "user" or "assistant" or "system" or "developer"`
245
246 The role of the message input. One of `user`, `assistant`, `system`, or
247 `developer`.
248
249 - `"user"`
250
251 - `"assistant"`
252
253 - `"system"`
254
255 - `"developer"`
256
257 - `type: optional "message"`
258
259 The type of the message input. Always `message`.
260
261 - `"message"`
262
263 - `labels: array of string`
264
265 The labels to assign to each item in the evaluation.
266
267 - `model: string`
268
269 The model to use for the evaluation. Must support structured outputs.
270
271 - `name: string`
272
273 The name of the grader.
274
275 - `passing_labels: array of string`
276
277 The labels that indicate a passing result. Must be a subset of labels.
278
279 - `type: "label_model"`
280
281 The object type, which is always `label_model`.
282
283 - `"label_model"`
284
285### Multi Grader
286
287- `MultiGrader object { calculate_output, graders, name, type }`
288
289 A MultiGrader object combines the output of multiple graders to produce a single score.
290
291 - `calculate_output: string`
292
293 A formula to calculate the output based on grader results.
294
295 - `graders: StringCheckGrader or TextSimilarityGrader or PythonGrader or 2 more`
296
297 A StringCheckGrader object that performs a string comparison between input and reference using a specified operation.
298
299 - `StringCheckGrader object { input, name, operation, 2 more }`
300
301 A StringCheckGrader object that performs a string comparison between input and reference using a specified operation.
302
303 - `input: string`
304
305 The input text. This may include template strings.
306
307 - `name: string`
308
309 The name of the grader.
310
311 - `operation: "eq" or "ne" or "like" or "ilike"`
312
313 The string check operation to perform. One of `eq`, `ne`, `like`, or `ilike`.
314
315 - `"eq"`
316
317 - `"ne"`
318
319 - `"like"`
320
321 - `"ilike"`
322
323 - `reference: string`
324
325 The reference text. This may include template strings.
326
327 - `type: "string_check"`
328
329 The object type, which is always `string_check`.
330
331 - `"string_check"`
332
333 - `TextSimilarityGrader object { evaluation_metric, input, name, 2 more }`
334
335 A TextSimilarityGrader object which grades text based on similarity metrics.
336
337 - `evaluation_metric: "cosine" or "fuzzy_match" or "bleu" or 8 more`
338
339 The evaluation metric to use. One of `cosine`, `fuzzy_match`, `bleu`,
340 `gleu`, `meteor`, `rouge_1`, `rouge_2`, `rouge_3`, `rouge_4`, `rouge_5`,
341 or `rouge_l`.
342
343 - `"cosine"`
344
345 - `"fuzzy_match"`
346
347 - `"bleu"`
348
349 - `"gleu"`
350
351 - `"meteor"`
352
353 - `"rouge_1"`
354
355 - `"rouge_2"`
356
357 - `"rouge_3"`
358
359 - `"rouge_4"`
360
361 - `"rouge_5"`
362
363 - `"rouge_l"`
364
365 - `input: string`
366
367 The text being graded.
368
369 - `name: string`
370
371 The name of the grader.
372
373 - `reference: string`
374
375 The text being graded against.
376
377 - `type: "text_similarity"`
378
379 The type of grader.
380
381 - `"text_similarity"`
382
383 - `PythonGrader object { name, source, type, image_tag }`
384
385 A PythonGrader object that runs a python script on the input.
386
387 - `name: string`
388
389 The name of the grader.
390
391 - `source: string`
392
393 The source code of the python script.
394
395 - `type: "python"`
396
397 The object type, which is always `python`.
398
399 - `"python"`
400
401 - `image_tag: optional string`
402
403 The image tag to use for the python script.
404
405 - `ScoreModelGrader object { input, model, name, 3 more }`
406
407 A ScoreModelGrader object that uses a model to assign a score to the input.
408
409 - `input: array of object { content, role, type }`
410
411 The input messages evaluated by the grader. Supports text, output text, input image, and input audio content blocks, and may include template strings.
412
413 - `content: string or ResponseInputText or object { text, type } or 3 more`
414
415 Inputs to the model - can contain template strings. Supports text, output text, input images, and input audio, either as a single item or an array of items.
416
417 - `TextInput = string`
418
419 A text input to the model.
420
421 - `ResponseInputText object { text, type, prompt_cache_breakpoint }`
422
423 A text input to the model.
424
425 - `text: string`
426
427 The text input to the model.
428
429 - `type: "input_text"`
430
431 The type of the input item. Always `input_text`.
432
433 - `"input_text"`
434
435 - `prompt_cache_breakpoint: optional object { mode }`
436
437 Marks the exact end of a reusable prompt prefix. The breakpoint inherits its TTL from the request's `prompt_cache_options.ttl`; the boundary is not rounded to a token block.
438
439 - `mode: "explicit"`
440
441 The breakpoint mode. Always `explicit`.
442
443 - `"explicit"`
444
445 - `OutputText object { text, type }`
446
447 A text output from the model.
448
449 - `text: string`
450
451 The text output from the model.
452
453 - `type: "output_text"`
454
455 The type of the output text. Always `output_text`.
456
457 - `"output_text"`
458
459 - `InputImage object { image_url, type, detail }`
460
461 An image input block used within EvalItem content arrays.
462
463 - `image_url: string`
464
465 The URL of the image input.
466
467 - `type: "input_image"`
468
469 The type of the image input. Always `input_image`.
470
471 - `"input_image"`
472
473 - `detail: optional string`
474
475 The detail level of the image to be sent to the model. One of `high`, `low`, or `auto`. Defaults to `auto`.
476
477 - `ResponseInputAudio object { input_audio, type }`
478
479 An audio input to the model.
480
481 - `input_audio: object { data, format }`
482
483 - `data: string`
484
485 Base64-encoded audio data.
486
487 - `format: "mp3" or "wav"`
488
489 The format of the audio data. Currently supported formats are `mp3` and
490 `wav`.
491
492 - `"mp3"`
493
494 - `"wav"`
495
496 - `type: "input_audio"`
497
498 The type of the input item. Always `input_audio`.
499
500 - `"input_audio"`
501
502 - `GraderInputs = array of string or ResponseInputText or object { text, type } or 2 more`
503
504 A list of inputs, each of which may be either an input text, output text, input
505 image, or input audio object.
506
507 - `TextInput = string`
508
509 A text input to the model.
510
511 - `ResponseInputText object { text, type, prompt_cache_breakpoint }`
512
513 A text input to the model.
514
515 - `OutputText object { text, type }`
516
517 A text output from the model.
518
519 - `text: string`
520
521 The text output from the model.
522
523 - `type: "output_text"`
524
525 The type of the output text. Always `output_text`.
526
527 - `"output_text"`
528
529 - `InputImage object { image_url, type, detail }`
530
531 An image input block used within EvalItem content arrays.
532
533 - `image_url: string`
534
535 The URL of the image input.
536
537 - `type: "input_image"`
538
539 The type of the image input. Always `input_image`.
540
541 - `"input_image"`
542
543 - `detail: optional string`
544
545 The detail level of the image to be sent to the model. One of `high`, `low`, or `auto`. Defaults to `auto`.
546
547 - `ResponseInputAudio object { input_audio, type }`
548
549 An audio input to the model.
550
551 - `role: "user" or "assistant" or "system" or "developer"`
552
553 The role of the message input. One of `user`, `assistant`, `system`, or
554 `developer`.
555
556 - `"user"`
557
558 - `"assistant"`
559
560 - `"system"`
561
562 - `"developer"`
563
564 - `type: optional "message"`
565
566 The type of the message input. Always `message`.
567
568 - `"message"`
569
570 - `model: string`
571
572 The model to use for the evaluation.
573
574 - `name: string`
575
576 The name of the grader.
577
578 - `type: "score_model"`
579
580 The object type, which is always `score_model`.
581
582 - `"score_model"`
583
584 - `range: optional array of number`
585
586 The range of the score. Defaults to `[0, 1]`.
587
588 - `sampling_params: optional object { max_completions_tokens, reasoning_effort, seed, 2 more }`
589
590 The sampling parameters for the model.
591
592 - `max_completions_tokens: optional number or null`
593
594 The maximum number of tokens the grader model may generate in its response.
595
596 - `reasoning_effort: optional ReasoningEffort or null`
597
598 Constrains effort on reasoning for reasoning models. Currently supported
599 values are `none`, `minimal`, `low`, `medium`, `high`, `xhigh`, and `max`.
600 Reducing reasoning effort can result in faster responses and fewer tokens
601 used on reasoning in a response. Not all reasoning models support every
602 value. See the
603 [reasoning guide](/api/docs/guides/reasoning)
604 for model-specific support.
605
606 - `"none"`
607
608 - `"minimal"`
609
610 - `"low"`
611
612 - `"medium"`
613
614 - `"high"`
615
616 - `"xhigh"`
617
618 - `"max"`
619
620 - `seed: optional number or null`
621
622 A seed value to initialize the randomness, during sampling.
623
624 - `temperature: optional number or null`
625
626 A higher temperature increases randomness in the outputs.
627
628 - `top_p: optional number or null`
629
630 An alternative to temperature for nucleus sampling; 1.0 includes all tokens.
631
632 - `LabelModelGrader object { input, labels, model, 3 more }`
633
634 A LabelModelGrader object which uses a model to assign labels to each item
635 in the evaluation.
636
637 - `input: array of object { content, role, type }`
638
639 - `content: string or ResponseInputText or object { text, type } or 3 more`
640
641 Inputs to the model - can contain template strings. Supports text, output text, input images, and input audio, either as a single item or an array of items.
642
643 - `TextInput = string`
644
645 A text input to the model.
646
647 - `ResponseInputText object { text, type, prompt_cache_breakpoint }`
648
649 A text input to the model.
650
651 - `OutputText object { text, type }`
652
653 A text output from the model.
654
655 - `text: string`
656
657 The text output from the model.
658
659 - `type: "output_text"`
660
661 The type of the output text. Always `output_text`.
662
663 - `"output_text"`
664
665 - `InputImage object { image_url, type, detail }`
666
667 An image input block used within EvalItem content arrays.
668
669 - `image_url: string`
670
671 The URL of the image input.
672
673 - `type: "input_image"`
674
675 The type of the image input. Always `input_image`.
676
677 - `"input_image"`
678
679 - `detail: optional string`
680
681 The detail level of the image to be sent to the model. One of `high`, `low`, or `auto`. Defaults to `auto`.
682
683 - `ResponseInputAudio object { input_audio, type }`
684
685 An audio input to the model.
686
687 - `GraderInputs = array of string or ResponseInputText or object { text, type } or 2 more`
688
689 A list of inputs, each of which may be either an input text, output text, input
690 image, or input audio object.
691
692 - `role: "user" or "assistant" or "system" or "developer"`
693
694 The role of the message input. One of `user`, `assistant`, `system`, or
695 `developer`.
696
697 - `"user"`
698
699 - `"assistant"`
700
701 - `"system"`
702
703 - `"developer"`
704
705 - `type: optional "message"`
706
707 The type of the message input. Always `message`.
708
709 - `"message"`
710
711 - `labels: array of string`
712
713 The labels to assign to each item in the evaluation.
714
715 - `model: string`
716
717 The model to use for the evaluation. Must support structured outputs.
718
719 - `name: string`
720
721 The name of the grader.
722
723 - `passing_labels: array of string`
724
725 The labels that indicate a passing result. Must be a subset of labels.
726
727 - `type: "label_model"`
728
729 The object type, which is always `label_model`.
730
731 - `"label_model"`
732
733 - `name: string`
734
735 The name of the grader.
736
737 - `type: "multi"`
738
739 The object type, which is always `multi`.
740
741 - `"multi"`
742
743### Python Grader
744
745- `PythonGrader object { name, source, type, image_tag }`
746
747 A PythonGrader object that runs a python script on the input.
748
749 - `name: string`
750
751 The name of the grader.
752
753 - `source: string`
754
755 The source code of the python script.
756
757 - `type: "python"`
758
759 The object type, which is always `python`.
760
761 - `"python"`
762
763 - `image_tag: optional string`
764
765 The image tag to use for the python script.
766
767### Score Model Grader
768
769- `ScoreModelGrader object { input, model, name, 3 more }`
770
771 A ScoreModelGrader object that uses a model to assign a score to the input.
772
773 - `input: array of object { content, role, type }`
774
775 The input messages evaluated by the grader. Supports text, output text, input image, and input audio content blocks, and may include template strings.
776
777 - `content: string or ResponseInputText or object { text, type } or 3 more`
778
779 Inputs to the model - can contain template strings. Supports text, output text, input images, and input audio, either as a single item or an array of items.
780
781 - `TextInput = string`
782
783 A text input to the model.
784
785 - `ResponseInputText object { text, type, prompt_cache_breakpoint }`
786
787 A text input to the model.
788
789 - `text: string`
790
791 The text input to the model.
792
793 - `type: "input_text"`
794
795 The type of the input item. Always `input_text`.
796
797 - `"input_text"`
798
799 - `prompt_cache_breakpoint: optional object { mode }`
800
801 Marks the exact end of a reusable prompt prefix. The breakpoint inherits its TTL from the request's `prompt_cache_options.ttl`; the boundary is not rounded to a token block.
802
803 - `mode: "explicit"`
804
805 The breakpoint mode. Always `explicit`.
806
807 - `"explicit"`
808
809 - `OutputText object { text, type }`
810
811 A text output from the model.
812
813 - `text: string`
814
815 The text output from the model.
816
817 - `type: "output_text"`
818
819 The type of the output text. Always `output_text`.
820
821 - `"output_text"`
822
823 - `InputImage object { image_url, type, detail }`
824
825 An image input block used within EvalItem content arrays.
826
827 - `image_url: string`
828
829 The URL of the image input.
830
831 - `type: "input_image"`
832
833 The type of the image input. Always `input_image`.
834
835 - `"input_image"`
836
837 - `detail: optional string`
838
839 The detail level of the image to be sent to the model. One of `high`, `low`, or `auto`. Defaults to `auto`.
840
841 - `ResponseInputAudio object { input_audio, type }`
842
843 An audio input to the model.
844
845 - `input_audio: object { data, format }`
846
847 - `data: string`
848
849 Base64-encoded audio data.
850
851 - `format: "mp3" or "wav"`
852
853 The format of the audio data. Currently supported formats are `mp3` and
854 `wav`.
855
856 - `"mp3"`
857
858 - `"wav"`
859
860 - `type: "input_audio"`
861
862 The type of the input item. Always `input_audio`.
863
864 - `"input_audio"`
865
866 - `GraderInputs = array of string or ResponseInputText or object { text, type } or 2 more`
867
868 A list of inputs, each of which may be either an input text, output text, input
869 image, or input audio object.
870
871 - `TextInput = string`
872
873 A text input to the model.
874
875 - `ResponseInputText object { text, type, prompt_cache_breakpoint }`
876
877 A text input to the model.
878
879 - `OutputText object { text, type }`
880
881 A text output from the model.
882
883 - `text: string`
884
885 The text output from the model.
886
887 - `type: "output_text"`
888
889 The type of the output text. Always `output_text`.
890
891 - `"output_text"`
892
893 - `InputImage object { image_url, type, detail }`
894
895 An image input block used within EvalItem content arrays.
896
897 - `image_url: string`
898
899 The URL of the image input.
900
901 - `type: "input_image"`
902
903 The type of the image input. Always `input_image`.
904
905 - `"input_image"`
906
907 - `detail: optional string`
908
909 The detail level of the image to be sent to the model. One of `high`, `low`, or `auto`. Defaults to `auto`.
910
911 - `ResponseInputAudio object { input_audio, type }`
912
913 An audio input to the model.
914
915 - `role: "user" or "assistant" or "system" or "developer"`
916
917 The role of the message input. One of `user`, `assistant`, `system`, or
918 `developer`.
919
920 - `"user"`
921
922 - `"assistant"`
923
924 - `"system"`
925
926 - `"developer"`
927
928 - `type: optional "message"`
929
930 The type of the message input. Always `message`.
931
932 - `"message"`
933
934 - `model: string`
935
936 The model to use for the evaluation.
937
938 - `name: string`
939
940 The name of the grader.
941
942 - `type: "score_model"`
943
944 The object type, which is always `score_model`.
945
946 - `"score_model"`
947
948 - `range: optional array of number`
949
950 The range of the score. Defaults to `[0, 1]`.
951
952 - `sampling_params: optional object { max_completions_tokens, reasoning_effort, seed, 2 more }`
953
954 The sampling parameters for the model.
955
956 - `max_completions_tokens: optional number or null`
957
958 The maximum number of tokens the grader model may generate in its response.
959
960 - `reasoning_effort: optional ReasoningEffort or null`
961
962 Constrains effort on reasoning for reasoning models. Currently supported
963 values are `none`, `minimal`, `low`, `medium`, `high`, `xhigh`, and `max`.
964 Reducing reasoning effort can result in faster responses and fewer tokens
965 used on reasoning in a response. Not all reasoning models support every
966 value. See the
967 [reasoning guide](/api/docs/guides/reasoning)
968 for model-specific support.
969
970 - `"none"`
971
972 - `"minimal"`
973
974 - `"low"`
975
976 - `"medium"`
977
978 - `"high"`
979
980 - `"xhigh"`
981
982 - `"max"`
983
984 - `seed: optional number or null`
985
986 A seed value to initialize the randomness, during sampling.
987
988 - `temperature: optional number or null`
989
990 A higher temperature increases randomness in the outputs.
991
992 - `top_p: optional number or null`
993
994 An alternative to temperature for nucleus sampling; 1.0 includes all tokens.
995
996### String Check Grader
997
998- `StringCheckGrader object { input, name, operation, 2 more }`
999
1000 A StringCheckGrader object that performs a string comparison between input and reference using a specified operation.
1001
1002 - `input: string`
1003
1004 The input text. This may include template strings.
1005
1006 - `name: string`
1007
1008 The name of the grader.
1009
1010 - `operation: "eq" or "ne" or "like" or "ilike"`
1011
1012 The string check operation to perform. One of `eq`, `ne`, `like`, or `ilike`.
1013
1014 - `"eq"`
1015
1016 - `"ne"`
1017
1018 - `"like"`
1019
1020 - `"ilike"`
1021
1022 - `reference: string`
1023
1024 The reference text. This may include template strings.
1025
1026 - `type: "string_check"`
1027
1028 The object type, which is always `string_check`.
1029
1030 - `"string_check"`
1031
1032### Text Similarity Grader
1033
1034- `TextSimilarityGrader object { evaluation_metric, input, name, 2 more }`
1035
1036 A TextSimilarityGrader object which grades text based on similarity metrics.
1037
1038 - `evaluation_metric: "cosine" or "fuzzy_match" or "bleu" or 8 more`
1039
1040 The evaluation metric to use. One of `cosine`, `fuzzy_match`, `bleu`,
1041 `gleu`, `meteor`, `rouge_1`, `rouge_2`, `rouge_3`, `rouge_4`, `rouge_5`,
1042 or `rouge_l`.
1043
1044 - `"cosine"`
1045
1046 - `"fuzzy_match"`
1047
1048 - `"bleu"`
1049
1050 - `"gleu"`
1051
1052 - `"meteor"`
1053
1054 - `"rouge_1"`
1055
1056 - `"rouge_2"`
1057
1058 - `"rouge_3"`
1059
1060 - `"rouge_4"`
1061
1062 - `"rouge_5"`
1063
1064 - `"rouge_l"`
1065
1066 - `input: string`
1067
1068 The text being graded.
1069
1070 - `name: string`
1071
1072 The name of the grader.
1073
1074 - `reference: string`
1075
1076 The text being graded against.
1077
1078 - `type: "text_similarity"`
1079
1080 The type of grader.
1081
1082 - `"text_similarity"`