mlboydaisuke

GLiNER2.5-Decide on Core AI: fp16 graphs S=256/512 (macos JIT, ios h19p AOT), tokenizer, classifier.json, gate fixtures, LICENSE/NOTICE

7464c91 verified 22 days ago

Raw History Blame Contribute Delete

86.2 kB

	{
	"header": {
	"kind": "gliner25-decide-oracle-fixture",
	"created": "2026-09-26T11:34:20+09:00",
	"script": "conversion/gliner25_decide_oracle.py",
	"model_id": "fastino/GLiNER2.5-Decide",
	"model_rev": "7ee5da4c2415e32259bcdc0b1a7367c32ce8d6f6",
	"model_safetensors_sha256": "40a5a23ff860dc3dff426cecd1048cacdd29c648c96db209dad818e9686dc997",
	"snapshot_dir": "/Users/majimadaisuke/code/coreai/_gliner25_decide/hf/hub/models--fastino--GLiNER2.5-Decide/snapshots/7ee5da4c2415e32259bcdc0b1a7367c32ce8d6f6",
	"oracle": {
	"package": "gliner2 2.0.0",
	"device": "cpu",
	"dtype": "float32",
	"batch": 1,
	"encoder": "DebertaV2Model",
	"attn_implementation": "eager",
	"path": "model.encoder(input_ids, attention_mask).last_hidden_state -> processor.extract_embeddings_from_batch -> classifier(stack(embs)[1:]).squeeze(-1)",
	"decision_rule": "class_act auto: multi_label -> sigmoid, prob >= cls_threshold (none -> argmax 1); single -> softmax, argmax"
	},
	"marker_ids": {
	"[MASK]": 128000,
	"[SEP_STRUCT]": 128001,
	"[SEP_TEXT]": 128002,
	"[P]": 128003,
	"[C]": 128004,
	"[E]": 128005,
	"[R]": 128006,
	"[L]": 128007,
	"[EXAMPLE]": 128008,
	"[OUTPUT]": 128009,
	"[DESCRIPTION]": 128010,
	"[PAD]": 0
	},
	"versions": {
	"python": "3.12.13",
	"torch": "2.12.1",
	"transformers": "4.57.6",
	"gliner2": "2.0.0",
	"tokenizers": "0.22.2",
	"safetensors": "0.8.0",
	"numpy": "2.5.3",
	"huggingface_hub": "0.36.2",
	"torch_threads": 12,
	"machine": "arm64",
	"platform": "macOS-27.0-arm64-arm-64bit"
	},
	"set": "readme21",
	"S": 256,
	"MMAX": 32,
	"selection": "all 21 classify_text examples of the model card at model_rev, verbatim",
	"self_check": {
	"cases": 21,
	"tasks": 26,
	"decisions_equal_tasks": 26,
	"max_abs_prob_diff": 0.0
	},
	"matches_card_potential_output": 12,
	"wall_s": 5.8
	},
	"cases": [
	{
	"id": "readme/01-customer-support-intent",
	"source": {
	"set": "readme21",
	"readme_index": 1,
	"title": "Customer support intent"
	},
	"text_raw": "My subscription renewed on April 15 for ¥5,400 after the service was already down. Can I get that charge refunded?",
	"text": "My subscription renewed on April 15 for ¥5,400 after the service was already down. Can I get that charge refunded?",
	"tasks": {
	"intent": [
	"order_status",
	"refund_request",
	"cancel_subscription",
	"update_payment",
	"login_problem",
	"shipping_delay",
	"bug_report",
	"speak_to_human",
	"other"
	]
	},
	"seq_len": 68,
	"n_labels": 9,
	"n_words": 25,
	"input_ids": [
	287,
	128003,
	5875,
	287,
	128007,
	556,
	616,
	35124,
	128007,
	6257,
	616,
	40952,
	128007,
	6788,
	616,
	76824,
	128007,
	1981,
	616,
	24207,
	128007,
	6607,
	616,
	40843,
	128007,
	2741,
	616,
	71506,
	128007,
	5554,
	616,
	27551,
	128007,
	1828,
	616,
	725,
	616,
	11975,
	128007,
	340,
	1263,
	1263,
	128002,
	312,
	5534,
	9504,
	277,
	47756,
	706,
	270,
	56478,
	456,
	366,
	3755,
	385,
	262,
	506,
	284,
	637,
	444,
	323,
	295,
	584,
	350,
	272,
	1541,
	24210,
	1102
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"intent",
	"(",
	"[L]",
	"order_status",
	"[L]",
	"refund_request",
	"[L]",
	"cancel_subscription",
	"[L]",
	"update_payment",
	"[L]",
	"login_problem",
	"[L]",
	"shipping_delay",
	"[L]",
	"bug_report",
	"[L]",
	"speak_to_human",
	"[L]",
	"other",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁intent",
	"▁(",
	"[L]",
	"▁order",
	"_",
	"status",
	"[L]",
	"▁refund",
	"_",
	"request",
	"[L]",
	"▁cancel",
	"_",
	"subscription",
	"[L]",
	"▁update",
	"_",
	"payment",
	"[L]",
	"▁login",
	"_",
	"problem",
	"[L]",
	"▁shipping",
	"_",
	"delay",
	"[L]",
	"▁bug",
	"_",
	"report",
	"[L]",
	"▁speak",
	"_",
	"to",
	"_",
	"human",
	"[L]",
	"▁other",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁my",
	"▁subscription",
	"▁renewed",
	"▁on",
	"▁april",
	"▁15",
	"▁for",
	"▁¥",
	"▁5",
	"▁,",
	"▁400",
	"▁after",
	"▁the",
	"▁service",
	"▁was",
	"▁already",
	"▁down",
	"▁.",
	"▁can",
	"▁i",
	"▁get",
	"▁that",
	"▁charge",
	"▁refunded",
	"▁?"
	],
	"schema_special_indices": [
	[
	1,
	4,
	8,
	12,
	16,
	20,
	24,
	28,
	32,
	38
	]
	],
	"task_results": [
	{
	"task": "intent",
	"labels": [
	"order_status",
	"refund_request",
	"cancel_subscription",
	"update_payment",
	"login_problem",
	"shipping_delay",
	"bug_report",
	"speak_to_human",
	"other"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "intent",
	"logits": [
	-4.849313735961914,
	3.5138487815856934,
	-3.8711068630218506,
	-4.2538981437683105,
	-5.772401332855225,
	-5.490520477294922,
	-5.102657794952393,
	-3.5973572731018066,
	-4.188626766204834
	],
	"probs": [
	0.00023262103786692023,
	0.997067391872406,
	0.0006186985992826521,
	0.000421924632973969,
	9.241816587746143e-05,
	0.0001225113810505718,
	0.00018056070257443935,
	0.0008135176612995565,
	0.0004503828822635114
	],
	"decision": "refund_request",
	"decision_probs": 0.997067391872406
	}
	],
	"classify_text": {
	"intent": {
	"label": "refund_request",
	"confidence": 0.997067391872406
	}
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"intent": "refund_request"
	},
	"matches_readme_output": true
	},
	{
	"id": "readme/02-banking-request",
	"source": {
	"set": "readme21",
	"readme_index": 2,
	"title": "Banking request"
	},
	"text_raw": "The transfer I sent this morning is still pending, and I think I used the wrong sort code. Can you stop it and add Emily as the beneficiary instead?",
	"text": "The transfer I sent this morning is still pending, and I think I used the wrong sort code. Can you stop it and add Emily as the beneficiary instead?",
	"tasks": {
	"intent": [
	"transfer_pending",
	"transfer_cancel",
	"beneficiary_add",
	"card_lost",
	"balance_inquiry",
	"fraud_report",
	"mortgage_application",
	"fee_explanation"
	]
	},
	"seq_len": 71,
	"n_labels": 8,
	"n_words": 32,
	"input_ids": [
	287,
	128003,
	5875,
	287,
	128007,
	2225,
	616,
	47622,
	128007,
	2225,
	616,
	66125,
	128007,
	18602,
	616,
	18053,
	128007,
	1081,
	616,
	25826,
	128007,
	1990,
	616,
	119989,
	128007,
	5330,
	616,
	27551,
	128007,
	4252,
	616,
	28552,
	128007,
	2383,
	616,
	118851,
	1263,
	1263,
	128002,
	262,
	2225,
	584,
	1253,
	291,
	1066,
	269,
	449,
	8555,
	366,
	263,
	584,
	428,
	584,
	427,
	262,
	1299,
	1667,
	1197,
	323,
	295,
	274,
	1001,
	278,
	263,
	800,
	112059,
	283,
	262,
	18602,
	1145,
	1102
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"intent",
	"(",
	"[L]",
	"transfer_pending",
	"[L]",
	"transfer_cancel",
	"[L]",
	"beneficiary_add",
	"[L]",
	"card_lost",
	"[L]",
	"balance_inquiry",
	"[L]",
	"fraud_report",
	"[L]",
	"mortgage_application",
	"[L]",
	"fee_explanation",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁intent",
	"▁(",
	"[L]",
	"▁transfer",
	"_",
	"pending",
	"[L]",
	"▁transfer",
	"_",
	"cancel",
	"[L]",
	"▁beneficiary",
	"_",
	"add",
	"[L]",
	"▁card",
	"_",
	"lost",
	"[L]",
	"▁balance",
	"_",
	"inquiry",
	"[L]",
	"▁fraud",
	"_",
	"report",
	"[L]",
	"▁mortgage",
	"_",
	"application",
	"[L]",
	"▁fee",
	"_",
	"explanation",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁the",
	"▁transfer",
	"▁i",
	"▁sent",
	"▁this",
	"▁morning",
	"▁is",
	"▁still",
	"▁pending",
	"▁,",
	"▁and",
	"▁i",
	"▁think",
	"▁i",
	"▁used",
	"▁the",
	"▁wrong",
	"▁sort",
	"▁code",
	"▁.",
	"▁can",
	"▁you",
	"▁stop",
	"▁it",
	"▁and",
	"▁add",
	"▁emily",
	"▁as",
	"▁the",
	"▁beneficiary",
	"▁instead",
	"▁?"
	],
	"schema_special_indices": [
	[
	1,
	4,
	8,
	12,
	16,
	20,
	24,
	28,
	32
	]
	],
	"task_results": [
	{
	"task": "intent",
	"labels": [
	"transfer_pending",
	"transfer_cancel",
	"beneficiary_add",
	"card_lost",
	"balance_inquiry",
	"fraud_report",
	"mortgage_application",
	"fee_explanation"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "intent",
	"logits": [
	-0.07197059690952301,
	1.016676425933838,
	2.8079445362091064,
	-4.991701126098633,
	-6.590446949005127,
	-4.900839805603027,
	-5.198736667633057,
	-7.441193103790283
	],
	"probs": [
	0.045858193188905716,
	0.13621044158935547,
	0.8168611526489258,
	0.00033481529681012034,
	6.76828422001563e-05,
	0.000366661959560588,
	0.000272201607003808,
	2.8907077648909763e-05
	],
	"decision": "beneficiary_add",
	"decision_probs": 0.8168611526489258
	}
	],
	"classify_text": {
	"intent": {
	"label": "beneficiary_add",
	"confidence": 0.8168611526489258
	}
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"intent": "transfer_cancel"
	},
	"matches_readme_output": false
	},
	{
	"id": "readme/03-travel-request",
	"source": {
	"set": "readme21",
	"readme_index": 3,
	"title": "Travel request"
	},
	"text_raw": "I need to move my Friday flight to Paris to Saturday morning, same cabin, and keep the aisle seat if you can.",
	"text": "I need to move my Friday flight to Paris to Saturday morning, same cabin, and keep the aisle seat if you can.",
	"tasks": {
	"request": [
	"book",
	"change",
	"cancel",
	"status",
	"seat_change",
	"refund",
	"baggage"
	]
	},
	"seq_len": 48,
	"n_labels": 7,
	"n_words": 25,
	"input_ids": [
	287,
	128003,
	1409,
	287,
	128007,
	550,
	128007,
	575,
	128007,
	6788,
	128007,
	1980,
	128007,
	2328,
	616,
	18982,
	128007,
	6257,
	128007,
	15686,
	1263,
	1263,
	128002,
	584,
	389,
	264,
	825,
	312,
	31798,
	2541,
	264,
	47136,
	264,
	42943,
	1066,
	366,
	454,
	6899,
	366,
	263,
	548,
	262,
	15191,
	2328,
	337,
	274,
	295,
	323
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"request",
	"(",
	"[L]",
	"book",
	"[L]",
	"change",
	"[L]",
	"cancel",
	"[L]",
	"status",
	"[L]",
	"seat_change",
	"[L]",
	"refund",
	"[L]",
	"baggage",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁request",
	"▁(",
	"[L]",
	"▁book",
	"[L]",
	"▁change",
	"[L]",
	"▁cancel",
	"[L]",
	"▁status",
	"[L]",
	"▁seat",
	"_",
	"change",
	"[L]",
	"▁refund",
	"[L]",
	"▁baggage",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁i",
	"▁need",
	"▁to",
	"▁move",
	"▁my",
	"▁friday",
	"▁flight",
	"▁to",
	"▁paris",
	"▁to",
	"▁saturday",
	"▁morning",
	"▁,",
	"▁same",
	"▁cabin",
	"▁,",
	"▁and",
	"▁keep",
	"▁the",
	"▁aisle",
	"▁seat",
	"▁if",
	"▁you",
	"▁can",
	"▁."
	],
	"schema_special_indices": [
	[
	1,
	4,
	6,
	8,
	10,
	12,
	16,
	18
	]
	],
	"task_results": [
	{
	"task": "request",
	"labels": [
	"book",
	"change",
	"cancel",
	"status",
	"seat_change",
	"refund",
	"baggage"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "request",
	"logits": [
	-2.9701309204101562,
	0.861903190612793,
	-5.016645431518555,
	-3.892615556716919,
	0.9039130806922913,
	-5.394156455993652,
	-4.557136535644531
	],
	"probs": [
	0.010404441505670547,
	0.48023077845573425,
	0.0013440913753584027,
	0.004136077594012022,
	0.5008350014686584,
	0.0009214628953486681,
	0.0021280949003994465
	],
	"decision": "seat_change",
	"decision_probs": 0.5008350014686584
	}
	],
	"classify_text": {
	"request": {
	"label": "seat_change",
	"confidence": 0.5008350014686584
	}
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"request": "change"
	},
	"matches_readme_output": false
	},
	{
	"id": "readme/04-clinic-request",
	"source": {
	"set": "readme21",
	"readme_index": 4,
	"title": "Clinic request"
	},
	"text_raw": "The rash came back after the antibiotics finished. Can I get a same-week appointment with dermatology, or should I just refill the cream?",
	"text": "The rash came back after the antibiotics finished. Can I get a same-week appointment with dermatology, or should I just refill the cream?",
	"tasks": {
	"request": [
	"book_appointment",
	"refill_prescription",
	"test_results",
	"referral",
	"billing_question",
	"cancel_appointment"
	]
	},
	"seq_len": 57,
	"n_labels": 6,
	"n_words": 26,
	"input_ids": [
	287,
	128003,
	1409,
	287,
	128007,
	550,
	616,
	72308,
	128007,
	24613,
	616,
	41712,
	128007,
	1010,
	616,
	53979,
	128007,
	10726,
	128007,
	9833,
	616,
	44096,
	128007,
	6788,
	616,
	72308,
	1263,
	1263,
	128002,
	262,
	16289,
	670,
	396,
	385,
	262,
	11739,
	1754,
	323,
	295,
	584,
	350,
	266,
	454,
	271,
	6913,
	3198,
	275,
	54584,
	366,
	289,
	403,
	584,
	348,
	24613,
	262,
	2727,
	1102
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"request",
	"(",
	"[L]",
	"book_appointment",
	"[L]",
	"refill_prescription",
	"[L]",
	"test_results",
	"[L]",
	"referral",
	"[L]",
	"billing_question",
	"[L]",
	"cancel_appointment",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁request",
	"▁(",
	"[L]",
	"▁book",
	"_",
	"appointment",
	"[L]",
	"▁refill",
	"_",
	"prescription",
	"[L]",
	"▁test",
	"_",
	"results",
	"[L]",
	"▁referral",
	"[L]",
	"▁billing",
	"_",
	"question",
	"[L]",
	"▁cancel",
	"_",
	"appointment",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁the",
	"▁rash",
	"▁came",
	"▁back",
	"▁after",
	"▁the",
	"▁antibiotics",
	"▁finished",
	"▁.",
	"▁can",
	"▁i",
	"▁get",
	"▁a",
	"▁same",
	"-",
	"week",
	"▁appointment",
	"▁with",
	"▁dermatology",
	"▁,",
	"▁or",
	"▁should",
	"▁i",
	"▁just",
	"▁refill",
	"▁the",
	"▁cream",
	"▁?"
	],
	"schema_special_indices": [
	[
	1,
	4,
	8,
	12,
	16,
	18,
	22
	]
	],
	"task_results": [
	{
	"task": "request",
	"labels": [
	"book_appointment",
	"refill_prescription",
	"test_results",
	"referral",
	"billing_question",
	"cancel_appointment"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "request",
	"logits": [
	2.3635408878326416,
	0.6880468130111694,
	-5.136224269866943,
	-2.4100089073181152,
	-4.717728614807129,
	-5.475194931030273
	],
	"probs": [
	0.8351050615310669,
	0.15634474158287048,
	0.00046199190546758473,
	0.007056911010295153,
	0.0007020773482508957,
	0.0003291709581390023
	],
	"decision": "book_appointment",
	"decision_probs": 0.8351050615310669
	}
	],
	"classify_text": {
	"request": {
	"label": "book_appointment",
	"confidence": 0.8351050615310669
	}
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"request": "book_appointment"
	},
	"matches_readme_output": true
	},
	{
	"id": "readme/05-review-sentiment",
	"source": {
	"set": "readme21",
	"readme_index": 5,
	"title": "Review sentiment"
	},
	"text_raw": "Battery dies before lunch, but the keyboard and the screen are the best I have used on a laptop.",
	"text": "Battery dies before lunch, but the keyboard and the screen are the best I have used on a laptop.",
	"tasks": {
	"sentiment": [
	"positive",
	"negative",
	"mixed",
	"neutral"
	]
	},
	"seq_len": 36,
	"n_labels": 4,
	"n_words": 21,
	"input_ids": [
	287,
	128003,
	9481,
	287,
	128007,
	1453,
	128007,
	2330,
	128007,
	3230,
	128007,
	5678,
	1263,
	1263,
	128002,
	2643,
	9745,
	416,
	2657,
	366,
	304,
	262,
	6125,
	263,
	262,
	1531,
	281,
	262,
	410,
	584,
	286,
	427,
	277,
	266,
	4411,
	323
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"sentiment",
	"(",
	"[L]",
	"positive",
	"[L]",
	"negative",
	"[L]",
	"mixed",
	"[L]",
	"neutral",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁sentiment",
	"▁(",
	"[L]",
	"▁positive",
	"[L]",
	"▁negative",
	"[L]",
	"▁mixed",
	"[L]",
	"▁neutral",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁battery",
	"▁dies",
	"▁before",
	"▁lunch",
	"▁,",
	"▁but",
	"▁the",
	"▁keyboard",
	"▁and",
	"▁the",
	"▁screen",
	"▁are",
	"▁the",
	"▁best",
	"▁i",
	"▁have",
	"▁used",
	"▁on",
	"▁a",
	"▁laptop",
	"▁."
	],
	"schema_special_indices": [
	[
	1,
	4,
	6,
	8,
	10
	]
	],
	"task_results": [
	{
	"task": "sentiment",
	"labels": [
	"positive",
	"negative",
	"mixed",
	"neutral"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "sentiment",
	"logits": [
	3.3864588737487793,
	-4.744900703430176,
	-4.173127174377441,
	-3.8599627017974854
	],
	"probs": [
	0.9984744191169739,
	0.00029371934942901134,
	0.0005202959873713553,
	0.0007116327760741115
	],
	"decision": "positive",
	"decision_probs": 0.9984744191169739
	}
	],
	"classify_text": {
	"sentiment": {
	"label": "positive",
	"confidence": 0.9984744191169739
	}
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"sentiment": "mixed"
	},
	"matches_readme_output": false
	},
	{
	"id": "readme/06-product-aspects",
	"source": {
	"set": "readme21",
	"readme_index": 6,
	"title": "Product aspects"
	},
	"text_raw": "Battery dies before lunch, but the keyboard and the screen are the best I have used on a laptop.",
	"text": "Battery dies before lunch, but the keyboard and the screen are the best I have used on a laptop.",
	"tasks": {
	"aspects": {
	"labels": [
	"battery",
	"keyboard",
	"screen",
	"camera",
	"price",
	"support"
	],
	"multi_label": true,
	"cls_threshold": 0.4
	}
	},
	"seq_len": 40,
	"n_labels": 6,
	"n_words": 21,
	"input_ids": [
	287,
	128003,
	2592,
	287,
	128007,
	2643,
	128007,
	6125,
	128007,
	1531,
	128007,
	1822,
	128007,
	710,
	128007,
	523,
	1263,
	1263,
	128002,
	2643,
	9745,
	416,
	2657,
	366,
	304,
	262,
	6125,
	263,
	262,
	1531,
	281,
	262,
	410,
	584,
	286,
	427,
	277,
	266,
	4411,
	323
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"aspects",
	"(",
	"[L]",
	"battery",
	"[L]",
	"keyboard",
	"[L]",
	"screen",
	"[L]",
	"camera",
	"[L]",
	"price",
	"[L]",
	"support",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁aspects",
	"▁(",
	"[L]",
	"▁battery",
	"[L]",
	"▁keyboard",
	"[L]",
	"▁screen",
	"[L]",
	"▁camera",
	"[L]",
	"▁price",
	"[L]",
	"▁support",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁battery",
	"▁dies",
	"▁before",
	"▁lunch",
	"▁,",
	"▁but",
	"▁the",
	"▁keyboard",
	"▁and",
	"▁the",
	"▁screen",
	"▁are",
	"▁the",
	"▁best",
	"▁i",
	"▁have",
	"▁used",
	"▁on",
	"▁a",
	"▁laptop",
	"▁."
	],
	"schema_special_indices": [
	[
	1,
	4,
	6,
	8,
	10,
	12,
	14
	]
	],
	"task_results": [
	{
	"task": "aspects",
	"labels": [
	"battery",
	"keyboard",
	"screen",
	"camera",
	"price",
	"support"
	],
	"multi_label": true,
	"cls_threshold": 0.4,
	"class_act": "auto",
	"prompt_str": "aspects",
	"logits": [
	1.6774275302886963,
	4.99154806137085,
	4.000231742858887,
	-7.007138729095459,
	-6.894930362701416,
	-5.695376396179199
	],
	"probs": [
	0.8425636291503906,
	0.9932506680488586,
	0.9820178747177124,
	0.0009045764454640448,
	0.0010118825593963265,
	0.003350210143253207
	],
	"decision": [
	"battery",
	"keyboard",
	"screen"
	],
	"decision_probs": [
	0.8425636291503906,
	0.9932506680488586,
	0.9820178747177124
	]
	}
	],
	"classify_text": {
	"aspects": [
	{
	"label": "battery",
	"confidence": 0.8425636291503906
	},
	{
	"label": "keyboard",
	"confidence": 0.9932506680488586
	},
	{
	"label": "screen",
	"confidence": 0.9820178747177124
	}
	]
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"aspects": [
	"battery",
	"keyboard",
	"screen"
	]
	},
	"matches_readme_output": true
	},
	{
	"id": "readme/07-news-topic",
	"source": {
	"set": "readme21",
	"readme_index": 7,
	"title": "News topic"
	},
	"text_raw": "The central bank held rates and said inflation is still above target, pushing bank stocks lower in afternoon trading.",
	"text": "The central bank held rates and said inflation is still above target, pushing bank stocks lower in afternoon trading.",
	"tasks": {
	"topic": [
	"politics",
	"business",
	"sports",
	"science",
	"entertainment",
	"world"
	]
	},
	"seq_len": 40,
	"n_labels": 6,
	"n_words": 21,
	"input_ids": [
	287,
	128003,
	2270,
	287,
	128007,
	3252,
	128007,
	460,
	128007,
	1948,
	128007,
	1693,
	128007,
	3280,
	128007,
	447,
	1263,
	1263,
	128002,
	262,
	2124,
	1831,
	968,
	1624,
	263,
	357,
	6645,
	269,
	449,
	764,
	1782,
	366,
	4918,
	1831,
	5194,
	1215,
	267,
	2438,
	2259,
	323
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"topic",
	"(",
	"[L]",
	"politics",
	"[L]",
	"business",
	"[L]",
	"sports",
	"[L]",
	"science",
	"[L]",
	"entertainment",
	"[L]",
	"world",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁topic",
	"▁(",
	"[L]",
	"▁politics",
	"[L]",
	"▁business",
	"[L]",
	"▁sports",
	"[L]",
	"▁science",
	"[L]",
	"▁entertainment",
	"[L]",
	"▁world",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁the",
	"▁central",
	"▁bank",
	"▁held",
	"▁rates",
	"▁and",
	"▁said",
	"▁inflation",
	"▁is",
	"▁still",
	"▁above",
	"▁target",
	"▁,",
	"▁pushing",
	"▁bank",
	"▁stocks",
	"▁lower",
	"▁in",
	"▁afternoon",
	"▁trading",
	"▁."
	],
	"schema_special_indices": [
	[
	1,
	4,
	6,
	8,
	10,
	12,
	14
	]
	],
	"task_results": [
	{
	"task": "topic",
	"labels": [
	"politics",
	"business",
	"sports",
	"science",
	"entertainment",
	"world"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "topic",
	"logits": [
	-2.2992451190948486,
	1.8464632034301758,
	-4.955565452575684,
	-4.008606433868408,
	-4.462010383605957,
	-1.7974307537078857
	],
	"probs": [
	0.015110251493752003,
	0.9543986916542053,
	0.0010608313605189323,
	0.0027346727438271046,
	0.0017377790063619614,
	0.024957828223705292
	],
	"decision": "business",
	"decision_probs": 0.9543986916542053
	}
	],
	"classify_text": {
	"topic": {
	"label": "business",
	"confidence": 0.9543986916542053
	}
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"topic": "business"
	},
	"matches_readme_output": true
	},
	{
	"id": "readme/08-document-type",
	"source": {
	"set": "readme21",
	"readme_index": 8,
	"title": "Document type"
	},
	"text_raw": "INVOICE 1842\nBill to: Northstar QA\nAmount due: 2,400 USD\nDue: 30 April 2026\nWire instructions are on page 2.",
	"text": "INVOICE 1842\nBill to: Northstar QA\nAmount due: 2,400 USD\nDue: 30 April 2026\nWire instructions are on page 2.",
	"tasks": {
	"document_type": [
	"invoice",
	"receipt",
	"contract",
	"resume",
	"support_email",
	"meeting_notes"
	]
	},
	"seq_len": 53,
	"n_labels": 6,
	"n_words": 26,
	"input_ids": [
	287,
	128003,
	2419,
	616,
	5967,
	287,
	128007,
	10462,
	128007,
	7610,
	128007,
	1962,
	128007,
	4247,
	128007,
	523,
	616,
	12073,
	128007,
	1122,
	616,
	34094,
	1263,
	1263,
	128002,
	10462,
	46196,
	2152,
	264,
	877,
	2480,
	5104,
	75407,
	906,
	775,
	877,
	392,
	366,
	3755,
	381,
	407,
	775,
	877,
	703,
	47756,
	39119,
	4016,
	3077,
	281,
	277,
	664,
	392,
	323
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"document_type",
	"(",
	"[L]",
	"invoice",
	"[L]",
	"receipt",
	"[L]",
	"contract",
	"[L]",
	"resume",
	"[L]",
	"support_email",
	"[L]",
	"meeting_notes",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁document",
	"_",
	"type",
	"▁(",
	"[L]",
	"▁invoice",
	"[L]",
	"▁receipt",
	"[L]",
	"▁contract",
	"[L]",
	"▁resume",
	"[L]",
	"▁support",
	"_",
	"email",
	"[L]",
	"▁meeting",
	"_",
	"notes",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁invoice",
	"▁1842",
	"▁bill",
	"▁to",
	"▁:",
	"▁north",
	"star",
	"▁qa",
	"▁amount",
	"▁due",
	"▁:",
	"▁2",
	"▁,",
	"▁400",
	"▁us",
	"d",
	"▁due",
	"▁:",
	"▁30",
	"▁april",
	"▁2026",
	"▁wire",
	"▁instructions",
	"▁are",
	"▁on",
	"▁page",
	"▁2",
	"▁."
	],
	"schema_special_indices": [
	[
	1,
	6,
	8,
	10,
	12,
	14,
	18
	]
	],
	"task_results": [
	{
	"task": "document_type",
	"labels": [
	"invoice",
	"receipt",
	"contract",
	"resume",
	"support_email",
	"meeting_notes"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "document_type",
	"logits": [
	1.6559009552001953,
	-3.449960708618164,
	-3.4158260822296143,
	-3.617985248565674,
	-3.113797187805176,
	-3.122982978820801
	],
	"probs": [
	0.9667957425117493,
	0.005859859753400087,
	0.006063337903469801,
	0.004953532479703426,
	0.00820127036422491,
	0.008126280270516872
	],
	"decision": "invoice",
	"decision_probs": 0.9667957425117493
	}
	],
	"classify_text": {
	"document_type": {
	"label": "invoice",
	"confidence": 0.9667957425117493
	}
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"document_type": "invoice"
	},
	"matches_readme_output": true
	},
	{
	"id": "readme/09-email-triage",
	"source": {
	"set": "readme21",
	"readme_index": 9,
	"title": "Email triage"
	},
	"text_raw": "From: compliance@group.example\nSubject: Protocol update — action required today\n\nPlease confirm the new retention rule is applied before Friday's audit.",
	"text": "From: compliance@group.example\nSubject: Protocol update — action required today\n\nPlease confirm the new retention rule is applied before Friday's audit.",
	"tasks": {
	"intent": [
	"fyi",
	"request",
	"approval",
	"complaint",
	"newsletter",
	"security_alert"
	],
	"urgency": [
	"low",
	"normal",
	"high",
	"critical"
	],
	"route": [
	"support",
	"billing",
	"legal",
	"security",
	"finance",
	"archive"
	]
	},
	"seq_len": 85,
	"n_labels": 16,
	"n_words": 25,
	"input_ids": [
	287,
	128003,
	5875,
	287,
	128007,
	2994,
	19889,
	128007,
	1409,
	128007,
	3645,
	128007,
	5142,
	128007,
	5403,
	128007,
	971,
	616,
	62659,
	1263,
	1263,
	128001,
	287,
	128003,
	16113,
	287,
	128007,
	759,
	128007,
	1697,
	128007,
	459,
	128007,
	1923,
	1263,
	1263,
	128001,
	287,
	128003,
	2788,
	287,
	128007,
	523,
	128007,
	9833,
	128007,
	1334,
	128007,
	971,
	128007,
	3769,
	128007,
	9168,
	1263,
	1263,
	128002,
	292,
	877,
	4200,
	1683,
	10914,
	260,
	28748,
	1284,
	877,
	6624,
	1981,
	533,
	1016,
	886,
	561,
	811,
	4592,
	262,
	353,
	9104,
	2310,
	269,
	2312,
	416,
	31798,
	382,
	1550,
	7522,
	323
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"intent",
	"(",
	"[L]",
	"fyi",
	"[L]",
	"request",
	"[L]",
	"approval",
	"[L]",
	"complaint",
	"[L]",
	"newsletter",
	"[L]",
	"security_alert",
	")",
	")"
	],
	[
	"(",
	"[P]",
	"urgency",
	"(",
	"[L]",
	"low",
	"[L]",
	"normal",
	"[L]",
	"high",
	"[L]",
	"critical",
	")",
	")"
	],
	[
	"(",
	"[P]",
	"route",
	"(",
	"[L]",
	"support",
	"[L]",
	"billing",
	"[L]",
	"legal",
	"[L]",
	"security",
	"[L]",
	"finance",
	"[L]",
	"archive",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁intent",
	"▁(",
	"[L]",
	"▁f",
	"yi",
	"[L]",
	"▁request",
	"[L]",
	"▁approval",
	"[L]",
	"▁complaint",
	"[L]",
	"▁newsletter",
	"[L]",
	"▁security",
	"_",
	"alert",
	"▁)",
	"▁)",
	"[SEP_STRUCT]",
	"▁(",
	"[P]",
	"▁urgency",
	"▁(",
	"[L]",
	"▁low",
	"[L]",
	"▁normal",
	"[L]",
	"▁high",
	"[L]",
	"▁critical",
	"▁)",
	"▁)",
	"[SEP_STRUCT]",
	"▁(",
	"[P]",
	"▁route",
	"▁(",
	"[L]",
	"▁support",
	"[L]",
	"▁billing",
	"[L]",
	"▁legal",
	"[L]",
	"▁security",
	"[L]",
	"▁finance",
	"[L]",
	"▁archive",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁from",
	"▁:",
	"▁compliance",
	"@",
	"group",
	".",
	"example",
	"▁subject",
	"▁:",
	"▁protocol",
	"▁update",
	"▁—",
	"▁action",
	"▁required",
	"▁today",
	"▁please",
	"▁confirm",
	"▁the",
	"▁new",
	"▁retention",
	"▁rule",
	"▁is",
	"▁applied",
	"▁before",
	"▁friday",
	"▁'",
	"▁s",
	"▁audit",
	"▁."
	],
	"schema_special_indices": [
	[
	1,
	4,
	7,
	9,
	11,
	13,
	15
	],
	[
	23,
	26,
	28,
	30,
	32
	],
	[
	38,
	41,
	43,
	45,
	47,
	49,
	51
	]
	],
	"task_results": [
	{
	"task": "intent",
	"labels": [
	"fyi",
	"request",
	"approval",
	"complaint",
	"newsletter",
	"security_alert"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "intent",
	"logits": [
	-2.0108981132507324,
	-0.19817551970481873,
	-1.6830180883407593,
	-2.8361692428588867,
	-0.6869609951972961,
	-2.191805601119995
	],
	"probs": [
	0.07382283359766006,
	0.4523203372955322,
	0.10246774554252625,
	0.032342974096536636,
	0.2774401307106018,
	0.06160608306527138
	],
	"decision": "request",
	"decision_probs": 0.4523203372955322
	},
	{
	"task": "urgency",
	"labels": [
	"low",
	"normal",
	"high",
	"critical"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "urgency",
	"logits": [
	-1.8748583793640137,
	-1.653106927871704,
	-0.7069736123085022,
	-0.5319124460220337
	],
	"probs": [
	0.10759912431240082,
	0.13431179523468018,
	0.34595102071762085,
	0.41213807463645935
	],
	"decision": "critical",
	"decision_probs": 0.41213807463645935
	},
	{
	"task": "route",
	"labels": [
	"support",
	"billing",
	"legal",
	"security",
	"finance",
	"archive"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "route",
	"logits": [
	-1.9716418981552124,
	-2.556593418121338,
	0.05507804453372955,
	-1.59413743019104,
	-2.0757570266723633,
	-1.5089396238327026
	],
	"probs": [
	0.07636848092079163,
	0.04254738986492157,
	0.5795720815658569,
	0.11139412224292755,
	0.06881728768348694,
	0.12130067497491837
	],
	"decision": "legal",
	"decision_probs": 0.5795720815658569
	}
	],
	"classify_text": {
	"intent": {
	"label": "request",
	"confidence": 0.4523203372955322
	},
	"urgency": {
	"label": "critical",
	"confidence": 0.41213807463645935
	},
	"route": {
	"label": "legal",
	"confidence": 0.5795720815658569
	}
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"intent": "request",
	"urgency": "high",
	"route": "legal"
	},
	"matches_readme_output": false
	},
	{
	"id": "readme/10-ticket-routing",
	"source": {
	"set": "readme21",
	"readme_index": 10,
	"title": "Ticket routing"
	},
	"text_raw": "[subject] 401k deduction missing from this paystub\n[body] Last month's contribution posted. This month the line is gone and HR told me to open a ticket.",
	"text": "[subject] 401k deduction missing from this paystub\n[body] Last month's contribution posted. This month the line is gone and HR told me to open a ticket.",
	"tasks": {
	"queue": [
	"payroll",
	"benefits",
	"it_access",
	"facilities",
	"expense_reimbursement",
	"manager_approval"
	]
	},
	"seq_len": 64,
	"n_labels": 6,
	"n_words": 34,
	"input_ids": [
	287,
	128003,
	12074,
	287,
	128007,
	11473,
	128007,
	1382,
	128007,
	278,
	616,
	20046,
	128007,
	2196,
	128007,
	5050,
	616,
	368,
	79393,
	3604,
	17019,
	128007,
	2140,
	616,
	49838,
	1263,
	1263,
	128002,
	647,
	1284,
	6311,
	14804,
	1165,
	16081,
	2367,
	292,
	291,
	7878,
	29850,
	647,
	665,
	6311,
	437,
	788,
	382,
	1550,
	4534,
	1430,
	323,
	291,
	788,
	262,
	683,
	269,
	1699,
	263,
	21212,
	732,
	351,
	264,
	615,
	266,
	3316,
	323
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"queue",
	"(",
	"[L]",
	"payroll",
	"[L]",
	"benefits",
	"[L]",
	"it_access",
	"[L]",
	"facilities",
	"[L]",
	"expense_reimbursement",
	"[L]",
	"manager_approval",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁queue",
	"▁(",
	"[L]",
	"▁payroll",
	"[L]",
	"▁benefits",
	"[L]",
	"▁it",
	"_",
	"access",
	"[L]",
	"▁facilities",
	"[L]",
	"▁expense",
	"_",
	"re",
	"imbu",
	"rs",
	"ement",
	"[L]",
	"▁manager",
	"_",
	"approval",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁[",
	"▁subject",
	"▁]",
	"▁401",
	"k",
	"▁deduction",
	"▁missing",
	"▁from",
	"▁this",
	"▁pays",
	"tub",
	"▁[",
	"▁body",
	"▁]",
	"▁last",
	"▁month",
	"▁'",
	"▁s",
	"▁contribution",
	"▁posted",
	"▁.",
	"▁this",
	"▁month",
	"▁the",
	"▁line",
	"▁is",
	"▁gone",
	"▁and",
	"▁hr",
	"▁told",
	"▁me",
	"▁to",
	"▁open",
	"▁a",
	"▁ticket",
	"▁."
	],
	"schema_special_indices": [
	[
	1,
	4,
	6,
	8,
	12,
	14,
	21
	]
	],
	"task_results": [
	{
	"task": "queue",
	"labels": [
	"payroll",
	"benefits",
	"it_access",
	"facilities",
	"expense_reimbursement",
	"manager_approval"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "queue",
	"logits": [
	-2.2571792602539062,
	1.8012199401855469,
	-5.181023120880127,
	-4.593165397644043,
	-3.6819376945495605,
	-3.685666084289551
	],
	"probs": [
	0.016803260892629623,
	0.97259920835495,
	0.0009027848718687892,
	0.001625128323212266,
	0.0040423027239739895,
	0.004027259536087513
	],
	"decision": "benefits",
	"decision_probs": 0.97259920835495
	}
	],
	"classify_text": {
	"queue": {
	"label": "benefits",
	"confidence": 0.97259920835495
	}
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"queue": "benefits"
	},
	"matches_readme_output": true
	},
	{
	"id": "readme/11-handoff-to-a-person",
	"source": {
	"set": "readme21",
	"readme_index": 11,
	"title": "Handoff to a person"
	},
	"text_raw": "This is the third time I have explained the same missing refund. Stop the bot and get me a person.",
	"text": "This is the third time I have explained the same missing refund. Stop the bot and get me a person.",
	"tasks": {
	"handoff": [
	"yes",
	"no"
	]
	},
	"seq_len": 33,
	"n_labels": 2,
	"n_words": 22,
	"input_ids": [
	287,
	128003,
	88933,
	287,
	128007,
	2489,
	128007,
	363,
	1263,
	1263,
	128002,
	291,
	269,
	262,
	883,
	326,
	584,
	286,
	2996,
	262,
	454,
	2367,
	6257,
	323,
	1001,
	262,
	16672,
	263,
	350,
	351,
	266,
	604,
	323
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"handoff",
	"(",
	"[L]",
	"yes",
	"[L]",
	"no",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁handoff",
	"▁(",
	"[L]",
	"▁yes",
	"[L]",
	"▁no",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁this",
	"▁is",
	"▁the",
	"▁third",
	"▁time",
	"▁i",
	"▁have",
	"▁explained",
	"▁the",
	"▁same",
	"▁missing",
	"▁refund",
	"▁.",
	"▁stop",
	"▁the",
	"▁bot",
	"▁and",
	"▁get",
	"▁me",
	"▁a",
	"▁person",
	"▁."
	],
	"schema_special_indices": [
	[
	1,
	4,
	6
	]
	],
	"task_results": [
	{
	"task": "handoff",
	"labels": [
	"yes",
	"no"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "handoff",
	"logits": [
	-0.2856934070587158,
	0.4438457489013672
	],
	"probs": [
	0.3252958655357361,
	0.6747041344642639
	],
	"decision": "no",
	"decision_probs": 0.6747041344642639
	}
	],
	"classify_text": {
	"handoff": {
	"label": "no",
	"confidence": 0.6747041344642639
	}
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"handoff": "yes"
	},
	"matches_readme_output": false
	},
	{
	"id": "readme/12-did-the-agent-finish",
	"source": {
	"set": "readme21",
	"readme_index": 12,
	"title": "Did the agent finish?"
	},
	"text_raw": "Goal: email the Q4 summary to every partner.\nLast action: draft saved in the hub.\nSend button is still disabled because two partners have no address.",
	"text": "Goal: email the Q4 summary to every partner.\nLast action: draft saved in the hub.\nSend button is still disabled because two partners have no address.",
	"tasks": {
	"finished": [
	"yes",
	"no"
	]
	},
	"seq_len": 43,
	"n_labels": 2,
	"n_words": 31,
	"input_ids": [
	287,
	128003,
	1754,
	287,
	128007,
	2489,
	128007,
	363,
	1263,
	1263,
	128002,
	1238,
	877,
	871,
	262,
	10562,
	554,
	5310,
	264,
	469,
	1940,
	323,
	437,
	1016,
	877,
	3921,
	3481,
	267,
	262,
	6841,
	323,
	1339,
	1946,
	269,
	449,
	6267,
	401,
	375,
	2656,
	286,
	363,
	990,
	323
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"finished",
	"(",
	"[L]",
	"yes",
	"[L]",
	"no",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁finished",
	"▁(",
	"[L]",
	"▁yes",
	"[L]",
	"▁no",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁goal",
	"▁:",
	"▁email",
	"▁the",
	"▁q",
	"4",
	"▁summary",
	"▁to",
	"▁every",
	"▁partner",
	"▁.",
	"▁last",
	"▁action",
	"▁:",
	"▁draft",
	"▁saved",
	"▁in",
	"▁the",
	"▁hub",
	"▁.",
	"▁send",
	"▁button",
	"▁is",
	"▁still",
	"▁disabled",
	"▁because",
	"▁two",
	"▁partners",
	"▁have",
	"▁no",
	"▁address",
	"▁."
	],
	"schema_special_indices": [
	[
	1,
	4,
	6
	]
	],
	"task_results": [
	{
	"task": "finished",
	"labels": [
	"yes",
	"no"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "finished",
	"logits": [
	-2.2892160415649414,
	2.831411600112915
	],
	"probs": [
	0.0059368181973695755,
	0.9940631985664368
	],
	"decision": "no",
	"decision_probs": 0.9940631985664368
	}
	],
	"classify_text": {
	"finished": {
	"label": "no",
	"confidence": 0.9940631985664368
	}
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"finished": "no"
	},
	"matches_readme_output": true
	},
	{
	"id": "readme/13-moderation",
	"source": {
	"set": "readme21",
	"readme_index": 13,
	"title": "Moderation"
	},
	"text_raw": "Post the customer's home address in the public thread so everyone can see where the package actually went.",
	"text": "Post the customer's home address in the public thread so everyone can see where the package actually went.",
	"tasks": {
	"policy": [
	"allow",
	"personal_data",
	"harassment",
	"scam",
	"violence",
	"spam"
	]
	},
	"seq_len": 42,
	"n_labels": 6,
	"n_words": 21,
	"input_ids": [
	287,
	128003,
	1132,
	287,
	128007,
	1052,
	128007,
	753,
	616,
	9832,
	128007,
	8930,
	128007,
	11017,
	128007,
	2742,
	128007,
	8962,
	1263,
	1263,
	128002,
	600,
	262,
	1099,
	382,
	1550,
	425,
	990,
	267,
	262,
	613,
	3676,
	324,
	837,
	295,
	398,
	399,
	262,
	2288,
	675,
	700,
	323
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"policy",
	"(",
	"[L]",
	"allow",
	"[L]",
	"personal_data",
	"[L]",
	"harassment",
	"[L]",
	"scam",
	"[L]",
	"violence",
	"[L]",
	"spam",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁policy",
	"▁(",
	"[L]",
	"▁allow",
	"[L]",
	"▁personal",
	"_",
	"data",
	"[L]",
	"▁harassment",
	"[L]",
	"▁scam",
	"[L]",
	"▁violence",
	"[L]",
	"▁spam",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁post",
	"▁the",
	"▁customer",
	"▁'",
	"▁s",
	"▁home",
	"▁address",
	"▁in",
	"▁the",
	"▁public",
	"▁thread",
	"▁so",
	"▁everyone",
	"▁can",
	"▁see",
	"▁where",
	"▁the",
	"▁package",
	"▁actually",
	"▁went",
	"▁."
	],
	"schema_special_indices": [
	[
	1,
	4,
	6,
	10,
	12,
	14,
	16
	]
	],
	"task_results": [
	{
	"task": "policy",
	"labels": [
	"allow",
	"personal_data",
	"harassment",
	"scam",
	"violence",
	"spam"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "policy",
	"logits": [
	-2.6249656677246094,
	2.2417514324188232,
	-2.3788888454437256,
	-3.8163182735443115,
	-4.548536777496338,
	-3.016406774520874
	],
	"probs": [
	0.007501943036913872,
	0.9744560718536377,
	0.009594972245395184,
	0.002279166365042329,
	0.0010959170758724213,
	0.005071928724646568
	],
	"decision": "personal_data",
	"decision_probs": 0.9744560718536377
	}
	],
	"classify_text": {
	"policy": {
	"label": "personal_data",
	"confidence": 0.9744560718536377
	}
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"policy": "personal_data"
	},
	"matches_readme_output": true
	},
	{
	"id": "readme/14-incident-severity",
	"source": {
	"set": "readme21",
	"readme_index": 14,
	"title": "Incident severity"
	},
	"text_raw": "The deploy left resource tags inconsistent across staging. Production checkout is unaffected. No customer reports yet.",
	"text": "The deploy left resource tags inconsistent across staging. Production checkout is unaffected. No customer reports yet.",
	"tasks": {
	"severity": [
	"info",
	"low",
	"medium",
	"high",
	"critical"
	]
	},
	"seq_len": 36,
	"n_labels": 5,
	"n_words": 19,
	"input_ids": [
	287,
	128003,
	12079,
	287,
	128007,
	2470,
	128007,
	759,
	128007,
	2434,
	128007,
	459,
	128007,
	1923,
	1263,
	1263,
	128002,
	262,
	10983,
	595,
	2943,
	6664,
	14833,
	679,
	15614,
	323,
	1117,
	9623,
	269,
	28643,
	323,
	363,
	1099,
	1639,
	729,
	323
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"severity",
	"(",
	"[L]",
	"info",
	"[L]",
	"low",
	"[L]",
	"medium",
	"[L]",
	"high",
	"[L]",
	"critical",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁severity",
	"▁(",
	"[L]",
	"▁info",
	"[L]",
	"▁low",
	"[L]",
	"▁medium",
	"[L]",
	"▁high",
	"[L]",
	"▁critical",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁the",
	"▁deploy",
	"▁left",
	"▁resource",
	"▁tags",
	"▁inconsistent",
	"▁across",
	"▁staging",
	"▁.",
	"▁production",
	"▁checkout",
	"▁is",
	"▁unaffected",
	"▁.",
	"▁no",
	"▁customer",
	"▁reports",
	"▁yet",
	"▁."
	],
	"schema_special_indices": [
	[
	1,
	4,
	6,
	8,
	10,
	12
	]
	],
	"task_results": [
	{
	"task": "severity",
	"labels": [
	"info",
	"low",
	"medium",
	"high",
	"critical"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "severity",
	"logits": [
	-0.3379753828048706,
	-0.36489152908325195,
	-1.6114552021026611,
	-2.182440757751465,
	-2.692826747894287
	],
	"probs": [
	0.3989919126033783,
	0.3883958160877228,
	0.11166031658649445,
	0.06308454275131226,
	0.03786737471818924
	],
	"decision": "info",
	"decision_probs": 0.3989919126033783
	}
	],
	"classify_text": {
	"severity": {
	"label": "info",
	"confidence": 0.3989919126033783
	}
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"severity": "low"
	},
	"matches_readme_output": false
	},
	{
	"id": "readme/15-urgency-score",
	"source": {
	"set": "readme21",
	"readme_index": 15,
	"title": "Urgency score"
	},
	"text_raw": "Payroll file has to be corrected before the 5pm cutoff or the whole company is paid late.",
	"text": "Payroll file has to be corrected before the 5pm cutoff or the whole company is paid late.",
	"tasks": {
	"urgency": [
	"0",
	"1",
	"2",
	"3",
	"4",
	"5"
	]
	},
	"seq_len": 38,
	"n_labels": 6,
	"n_words": 18,
	"input_ids": [
	287,
	128003,
	16113,
	287,
	128007,
	767,
	128007,
	376,
	128007,
	392,
	128007,
	404,
	128007,
	453,
	128007,
	456,
	1263,
	1263,
	128002,
	11473,
	1092,
	303,
	264,
	282,
	12095,
	416,
	262,
	456,
	2086,
	31599,
	289,
	262,
	783,
	483,
	269,
	1487,
	1109,
	323
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"urgency",
	"(",
	"[L]",
	"0",
	"[L]",
	"1",
	"[L]",
	"2",
	"[L]",
	"3",
	"[L]",
	"4",
	"[L]",
	"5",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁urgency",
	"▁(",
	"[L]",
	"▁0",
	"[L]",
	"▁1",
	"[L]",
	"▁2",
	"[L]",
	"▁3",
	"[L]",
	"▁4",
	"[L]",
	"▁5",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁payroll",
	"▁file",
	"▁has",
	"▁to",
	"▁be",
	"▁corrected",
	"▁before",
	"▁the",
	"▁5",
	"pm",
	"▁cutoff",
	"▁or",
	"▁the",
	"▁whole",
	"▁company",
	"▁is",
	"▁paid",
	"▁late",
	"▁."
	],
	"schema_special_indices": [
	[
	1,
	4,
	6,
	8,
	10,
	12,
	14
	]
	],
	"task_results": [
	{
	"task": "urgency",
	"labels": [
	"0",
	"1",
	"2",
	"3",
	"4",
	"5"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "urgency",
	"logits": [
	-2.0453250408172607,
	-1.804246425628662,
	-1.6265888214111328,
	-1.418688178062439,
	-1.3153603076934814,
	-1.0862274169921875
	],
	"probs": [
	0.09663408994674683,
	0.1229785829782486,
	0.1468876153230667,
	0.18083199858665466,
	0.2005164623260498,
	0.2521512806415558
	],
	"decision": "5",
	"decision_probs": 0.2521512806415558
	}
	],
	"classify_text": {
	"urgency": {
	"label": "5",
	"confidence": 0.2521512806415558
	}
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"urgency": "5"
	},
	"matches_readme_output": true
	},
	{
	"id": "readme/16-spam-or-not",
	"source": {
	"set": "readme21",
	"readme_index": 16,
	"title": "Spam or not"
	},
	"text_raw": "Your mailbox is almost full. Click here in the next hour or we will delete every message.",
	"text": "Your mailbox is almost full. Click here in the next hour or we will delete every message.",
	"tasks": {
	"label": [
	"spam",
	"ham"
	]
	},
	"seq_len": 30,
	"n_labels": 2,
	"n_words": 19,
	"input_ids": [
	287,
	128003,
	3734,
	287,
	128007,
	8962,
	128007,
	11965,
	1263,
	1263,
	128002,
	290,
	20042,
	269,
	823,
	540,
	323,
	1363,
	422,
	267,
	262,
	498,
	1476,
	289,
	301,
	296,
	5748,
	469,
	1424,
	323
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"label",
	"(",
	"[L]",
	"spam",
	"[L]",
	"ham",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁label",
	"▁(",
	"[L]",
	"▁spam",
	"[L]",
	"▁ham",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁your",
	"▁mailbox",
	"▁is",
	"▁almost",
	"▁full",
	"▁.",
	"▁click",
	"▁here",
	"▁in",
	"▁the",
	"▁next",
	"▁hour",
	"▁or",
	"▁we",
	"▁will",
	"▁delete",
	"▁every",
	"▁message",
	"▁."
	],
	"schema_special_indices": [
	[
	1,
	4,
	6
	]
	],
	"task_results": [
	{
	"task": "label",
	"labels": [
	"spam",
	"ham"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "label",
	"logits": [
	1.381277084350586,
	-1.1973676681518555
	],
	"probs": [
	0.9294745326042175,
	0.07052551954984665
	],
	"decision": "spam",
	"decision_probs": 0.9294745326042175
	}
	],
	"classify_text": {
	"label": {
	"label": "spam",
	"confidence": 0.9294745326042175
	}
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"label": "spam"
	},
	"matches_readme_output": true
	},
	{
	"id": "readme/17-several-decisions-at-once",
	"source": {
	"set": "readme21",
	"readme_index": 17,
	"title": "Several decisions at once"
	},
	"text_raw": "Guest in room 1408 says the AC has been out since yesterday and they want to move tonight or leave. They also asked for the incidentals hold to be released.",
	"text": "Guest in room 1408 says the AC has been out since yesterday and they want to move tonight or leave. They also asked for the incidentals hold to be released.",
	"tasks": {
	"intent": [
	"maintenance",
	"room_change",
	"checkout",
	"billing",
	"complaint",
	"amenity_request"
	],
	"priority": [
	"low",
	"normal",
	"high",
	"urgent"
	],
	"needs_human": [
	"yes",
	"no"
	],
	"topics": {
	"labels": [
	"hvac",
	"billing",
	"housekeeping",
	"noise",
	"safety"
	],
	"multi_label": true,
	"cls_threshold": 0.4
	}
	},
	"seq_len": 102,
	"n_labels": 17,
	"n_words": 32,
	"input_ids": [
	287,
	128003,
	5875,
	287,
	128007,
	2406,
	128007,
	629,
	616,
	18982,
	128007,
	9623,
	128007,
	9833,
	128007,
	5142,
	128007,
	42150,
	616,
	40952,
	1263,
	1263,
	128001,
	287,
	128003,
	3983,
	287,
	128007,
	759,
	128007,
	1697,
	128007,
	459,
	128007,
	9178,
	1263,
	1263,
	128001,
	287,
	128003,
	634,
	616,
	11975,
	287,
	128007,
	2489,
	128007,
	363,
	1263,
	1263,
	128001,
	287,
	128003,
	3062,
	287,
	128007,
	90182,
	128007,
	9833,
	128007,
	28869,
	128007,
	3616,
	128007,
	1343,
	1263,
	1263,
	128002,
	3043,
	267,
	629,
	936,
	4425,
	652,
	262,
	24250,
	303,
	331,
	321,
	515,
	2807,
	263,
	306,
	409,
	264,
	825,
	4023,
	289,
	1021,
	323,
	306,
	327,
	921,
	270,
	262,
	23367,
	268,
	1229,
	264,
	282,
	1315,
	323
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"intent",
	"(",
	"[L]",
	"maintenance",
	"[L]",
	"room_change",
	"[L]",
	"checkout",
	"[L]",
	"billing",
	"[L]",
	"complaint",
	"[L]",
	"amenity_request",
	")",
	")"
	],
	[
	"(",
	"[P]",
	"priority",
	"(",
	"[L]",
	"low",
	"[L]",
	"normal",
	"[L]",
	"high",
	"[L]",
	"urgent",
	")",
	")"
	],
	[
	"(",
	"[P]",
	"needs_human",
	"(",
	"[L]",
	"yes",
	"[L]",
	"no",
	")",
	")"
	],
	[
	"(",
	"[P]",
	"topics",
	"(",
	"[L]",
	"hvac",
	"[L]",
	"billing",
	"[L]",
	"housekeeping",
	"[L]",
	"noise",
	"[L]",
	"safety",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁intent",
	"▁(",
	"[L]",
	"▁maintenance",
	"[L]",
	"▁room",
	"_",
	"change",
	"[L]",
	"▁checkout",
	"[L]",
	"▁billing",
	"[L]",
	"▁complaint",
	"[L]",
	"▁amenity",
	"_",
	"request",
	"▁)",
	"▁)",
	"[SEP_STRUCT]",
	"▁(",
	"[P]",
	"▁priority",
	"▁(",
	"[L]",
	"▁low",
	"[L]",
	"▁normal",
	"[L]",
	"▁high",
	"[L]",
	"▁urgent",
	"▁)",
	"▁)",
	"[SEP_STRUCT]",
	"▁(",
	"[P]",
	"▁needs",
	"_",
	"human",
	"▁(",
	"[L]",
	"▁yes",
	"[L]",
	"▁no",
	"▁)",
	"▁)",
	"[SEP_STRUCT]",
	"▁(",
	"[P]",
	"▁topics",
	"▁(",
	"[L]",
	"▁hvac",
	"[L]",
	"▁billing",
	"[L]",
	"▁housekeeping",
	"[L]",
	"▁noise",
	"[L]",
	"▁safety",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁guest",
	"▁in",
	"▁room",
	"▁14",
	"08",
	"▁says",
	"▁the",
	"▁ac",
	"▁has",
	"▁been",
	"▁out",
	"▁since",
	"▁yesterday",
	"▁and",
	"▁they",
	"▁want",
	"▁to",
	"▁move",
	"▁tonight",
	"▁or",
	"▁leave",
	"▁.",
	"▁they",
	"▁also",
	"▁asked",
	"▁for",
	"▁the",
	"▁incidental",
	"s",
	"▁hold",
	"▁to",
	"▁be",
	"▁released",
	"▁."
	],
	"schema_special_indices": [
	[
	1,
	4,
	6,
	10,
	12,
	14,
	16
	],
	[
	24,
	27,
	29,
	31,
	33
	],
	[
	39,
	44,
	46
	],
	[
	52,
	55,
	57,
	59,
	61,
	63
	]
	],
	"task_results": [
	{
	"task": "intent",
	"labels": [
	"maintenance",
	"room_change",
	"checkout",
	"billing",
	"complaint",
	"amenity_request"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "intent",
	"logits": [
	-2.6984121799468994,
	0.54266756772995,
	-1.846369981765747,
	-3.8297274112701416,
	-3.03488826751709,
	-3.0202157497406006
	],
	"probs": [
	0.03260784223675728,
	0.8334991335868835,
	0.07644680142402649,
	0.010519572533667088,
	0.023291226476430893,
	0.023635488003492355
	],
	"decision": "room_change",
	"decision_probs": 0.8334991335868835
	},
	{
	"task": "priority",
	"labels": [
	"low",
	"normal",
	"high",
	"urgent"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "priority",
	"logits": [
	-1.6097275018692017,
	-1.4882011413574219,
	-1.034649133682251,
	-0.6457433104515076
	],
	"probs": [
	0.15317192673683167,
	0.1729646921157837,
	0.27222785353660583,
	0.40163561701774597
	],
	"decision": "urgent",
	"decision_probs": 0.40163561701774597
	},
	{
	"task": "needs_human",
	"labels": [
	"yes",
	"no"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "needs_human",
	"logits": [
	0.3756559491157532,
	-0.31471487879753113
	],
	"probs": [
	0.6660494208335876,
	0.33395057916641235
	],
	"decision": "yes",
	"decision_probs": 0.6660494208335876
	},
	{
	"task": "topics",
	"labels": [
	"hvac",
	"billing",
	"housekeeping",
	"noise",
	"safety"
	],
	"multi_label": true,
	"cls_threshold": 0.4,
	"class_act": "auto",
	"prompt_str": "topics",
	"logits": [
	2.1432065963745117,
	-1.9354220628738403,
	-0.4641674757003784,
	-2.3596181869506836,
	-1.9099055528640747
	],
	"probs": [
	0.8950322270393372,
	0.1261516511440277,
	0.3859976828098297,
	0.08630429953336716,
	0.12899146974086761
	],
	"decision": [
	"hvac"
	],
	"decision_probs": [
	0.8950322270393372
	]
	}
	],
	"classify_text": {
	"intent": {
	"label": "room_change",
	"confidence": 0.8334991335868835
	},
	"priority": {
	"label": "urgent",
	"confidence": 0.40163561701774597
	},
	"needs_human": {
	"label": "yes",
	"confidence": 0.6660494208335876
	},
	"topics": [
	{
	"label": "hvac",
	"confidence": 0.8950322270393372
	}
	]
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"intent": "room_change",
	"priority": "high",
	"needs_human": "yes",
	"topics": [
	"hvac",
	"billing"
	]
	},
	"matches_readme_output": false
	},
	{
	"id": "readme/18-question-over-a-passage",
	"source": {
	"set": "readme21",
	"readme_index": 18,
	"title": "Question over a passage"
	},
	"text_raw": "The treaty was signed in Paris in 1992. It entered into force the following year, after the last signatory ratified it.",
	"text": "The treaty was signed in Paris in 1992. It entered into force the following year, after the last signatory ratified it.",
	"tasks": {
	"answer": {
	"labels": [
	"yes",
	"no"
	],
	"prompt": "Did the treaty enter into force in 1992?"
	}
	},
	"seq_len": 45,
	"n_labels": 2,
	"n_words": 24,
	"input_ids": [
	287,
	128003,
	1330,
	294,
	2709,
	262,
	13281,
	1916,
	352,
	1711,
	267,
	5762,
	302,
	287,
	128007,
	2489,
	128007,
	363,
	1263,
	1263,
	128002,
	262,
	13281,
	284,
	2346,
	267,
	47136,
	267,
	5762,
	323,
	278,
	3040,
	352,
	1711,
	262,
	776,
	395,
	366,
	385,
	262,
	437,
	59810,
	31719,
	278,
	323
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"answer: Did the treaty enter into force in 1992?",
	"(",
	"[L]",
	"yes",
	"[L]",
	"no",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁answer",
	":",
	"▁Did",
	"▁the",
	"▁treaty",
	"▁enter",
	"▁into",
	"▁force",
	"▁in",
	"▁1992",
	"?",
	"▁(",
	"[L]",
	"▁yes",
	"[L]",
	"▁no",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁the",
	"▁treaty",
	"▁was",
	"▁signed",
	"▁in",
	"▁paris",
	"▁in",
	"▁1992",
	"▁.",
	"▁it",
	"▁entered",
	"▁into",
	"▁force",
	"▁the",
	"▁following",
	"▁year",
	"▁,",
	"▁after",
	"▁the",
	"▁last",
	"▁signatory",
	"▁ratified",
	"▁it",
	"▁."
	],
	"schema_special_indices": [
	[
	1,
	14,
	16
	]
	],
	"task_results": [
	{
	"task": "answer",
	"labels": [
	"yes",
	"no"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "answer: Did the treaty enter into force in 1992?",
	"logits": [
	2.908085823059082,
	-2.6777358055114746
	],
	"probs": [
	0.996263325214386,
	0.003736651735380292
	],
	"decision": "yes",
	"decision_probs": 0.996263325214386
	}
	],
	"classify_text": {
	"answer": {
	"label": "yes",
	"confidence": 0.996263325214386
	}
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"answer": "no"
	},
	"matches_readme_output": false
	},
	{
	"id": "readme/19-book",
	"source": {
	"set": "readme21",
	"readme_index": 19,
	"title": "Book"
	},
	"text_raw": "She closed the ledger, blew out the lamp, and listened for the stair. The house had been empty since the winter the river took the bridge.",
	"text": "She closed the ledger, blew out the lamp, and listened for the stair. The house had been empty since the winter the river took the bridge.",
	"tasks": {
	"genre": [
	"mystery",
	"romance",
	"history",
	"science_fiction",
	"literary_fiction",
	"cookbook"
	]
	},
	"seq_len": 53,
	"n_labels": 6,
	"n_words": 30,
	"input_ids": [
	287,
	128003,
	6176,
	287,
	128007,
	5633,
	128007,
	6981,
	128007,
	820,
	128007,
	1693,
	616,
	18513,
	128007,
	7170,
	616,
	18513,
	128007,
	19991,
	1263,
	1263,
	128002,
	373,
	2100,
	262,
	25002,
	366,
	12081,
	321,
	262,
	6693,
	366,
	263,
	7961,
	270,
	262,
	20786,
	323,
	262,
	669,
	330,
	331,
	3601,
	515,
	262,
	2014,
	262,
	3108,
	681,
	262,
	3579,
	323
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"genre",
	"(",
	"[L]",
	"mystery",
	"[L]",
	"romance",
	"[L]",
	"history",
	"[L]",
	"science_fiction",
	"[L]",
	"literary_fiction",
	"[L]",
	"cookbook",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁genre",
	"▁(",
	"[L]",
	"▁mystery",
	"[L]",
	"▁romance",
	"[L]",
	"▁history",
	"[L]",
	"▁science",
	"_",
	"fiction",
	"[L]",
	"▁literary",
	"_",
	"fiction",
	"[L]",
	"▁cookbook",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁she",
	"▁closed",
	"▁the",
	"▁ledger",
	"▁,",
	"▁blew",
	"▁out",
	"▁the",
	"▁lamp",
	"▁,",
	"▁and",
	"▁listened",
	"▁for",
	"▁the",
	"▁stair",
	"▁.",
	"▁the",
	"▁house",
	"▁had",
	"▁been",
	"▁empty",
	"▁since",
	"▁the",
	"▁winter",
	"▁the",
	"▁river",
	"▁took",
	"▁the",
	"▁bridge",
	"▁."
	],
	"schema_special_indices": [
	[
	1,
	4,
	6,
	8,
	10,
	14,
	18
	]
	],
	"task_results": [
	{
	"task": "genre",
	"labels": [
	"mystery",
	"romance",
	"history",
	"science_fiction",
	"literary_fiction",
	"cookbook"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "genre",
	"logits": [
	-1.9721434116363525,
	-2.4048399925231934,
	-0.2259370982646942,
	-2.252347230911255,
	-1.5370631217956543,
	-2.619030237197876
	],
	"probs": [
	0.09798195958137512,
	0.06356650590896606,
	0.5617122054100037,
	0.0740380808711052,
	0.1513906568288803,
	0.05131055414676666
	],
	"decision": "history",
	"decision_probs": 0.5617122054100037
	}
	],
	"classify_text": {
	"genre": {
	"label": "history",
	"confidence": 0.5617122054100037
	}
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"genre": "literary_fiction"
	},
	"matches_readme_output": false
	},
	{
	"id": "readme/20-labels-with-a-description",
	"source": {
	"set": "readme21",
	"readme_index": 20,
	"title": "Labels with a description"
	},
	"text_raw": "Please reset the card PIN. The new one never arrived and the old one is locked after three tries.",
	"text": "Please reset the card PIN. The new one never arrived and the old one is locked after three tries.",
	"tasks": {
	"intent": {
	"labels": {
	"card_pin_change": "The customer wants a new PIN or the current PIN replaced",
	"card_lost": "The physical card is missing",
	"balance_inquiry": "The customer wants the current balance"
	}
	}
	},
	"seq_len": 81,
	"n_labels": 3,
	"n_words": 21,
	"input_ids": [
	287,
	128003,
	5875,
	128010,
	1081,
	616,
	10161,
	616,
	18982,
	294,
	279,
	1099,
	1654,
	266,
	353,
	20280,
	289,
	262,
	763,
	20280,
	3473,
	128010,
	1081,
	616,
	25826,
	294,
	279,
	1468,
	1081,
	269,
	2367,
	128010,
	1990,
	616,
	119989,
	294,
	279,
	1099,
	1654,
	262,
	763,
	1990,
	287,
	128007,
	1081,
	616,
	10161,
	616,
	18982,
	128007,
	1081,
	616,
	25826,
	128007,
	1990,
	616,
	119989,
	1263,
	1263,
	128002,
	811,
	9032,
	262,
	1081,
	5423,
	323,
	262,
	353,
	311,
	518,
	2525,
	263,
	262,
	597,
	311,
	269,
	5800,
	385,
	475,
	5364,
	323
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"intent [DESCRIPTION] card_pin_change: The customer wants a new PIN or the current PIN replaced [DESCRIPTION] card_lost: The physical card is missing [DESCRIPTION] balance_inquiry: The customer wants the current balance",
	"(",
	"[L]",
	"card_pin_change",
	"[L]",
	"card_lost",
	"[L]",
	"balance_inquiry",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁intent",
	"[DESCRIPTION]",
	"▁card",
	"_",
	"pin",
	"_",
	"change",
	":",
	"▁The",
	"▁customer",
	"▁wants",
	"▁a",
	"▁new",
	"▁PIN",
	"▁or",
	"▁the",
	"▁current",
	"▁PIN",
	"▁replaced",
	"[DESCRIPTION]",
	"▁card",
	"_",
	"lost",
	":",
	"▁The",
	"▁physical",
	"▁card",
	"▁is",
	"▁missing",
	"[DESCRIPTION]",
	"▁balance",
	"_",
	"inquiry",
	":",
	"▁The",
	"▁customer",
	"▁wants",
	"▁the",
	"▁current",
	"▁balance",
	"▁(",
	"[L]",
	"▁card",
	"_",
	"pin",
	"_",
	"change",
	"[L]",
	"▁card",
	"_",
	"lost",
	"[L]",
	"▁balance",
	"_",
	"inquiry",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁please",
	"▁reset",
	"▁the",
	"▁card",
	"▁pin",
	"▁.",
	"▁the",
	"▁new",
	"▁one",
	"▁never",
	"▁arrived",
	"▁and",
	"▁the",
	"▁old",
	"▁one",
	"▁is",
	"▁locked",
	"▁after",
	"▁three",
	"▁tries",
	"▁."
	],
	"schema_special_indices": [
	[
	1,
	43,
	49,
	53
	]
	],
	"task_results": [
	{
	"task": "intent",
	"labels": [
	"card_pin_change",
	"card_lost",
	"balance_inquiry"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "intent [DESCRIPTION] card_pin_change: The customer wants a new PIN or the current PIN replaced [DESCRIPTION] card_lost: The physical card is missing [DESCRIPTION] balance_inquiry: The customer wants the current balance",
	"logits": [
	1.439753770828247,
	0.6178942322731018,
	-2.5709187984466553
	],
	"probs": [
	0.6859959363937378,
	0.3015730082988739,
	0.012431072071194649
	],
	"decision": "card_pin_change",
	"decision_probs": 0.6859959363937378
	}
	],
	"classify_text": {
	"intent": {
	"label": "card_pin_change",
	"confidence": 0.6859959363937378
	}
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"intent": "card_pin_change"
	},
	"matches_readme_output": true
	},
	{
	"id": "readme/21-ordinal-score",
	"source": {
	"set": "readme21",
	"readme_index": 21,
	"title": "Ordinal score"
	},
	"text_raw": "I finished it in two nights. The ending is earned, the middle drags, and I would still hand it to a friend.",
	"text": "I finished it in two nights. The ending is earned, the middle drags, and I would still hand it to a friend.",
	"tasks": {
	"rating": [
	"0",
	"1",
	"2",
	"3",
	"4",
	"5",
	"6",
	"7",
	"8",
	"9",
	"10"
	]
	},
	"seq_len": 55,
	"n_labels": 11,
	"n_words": 26,
	"input_ids": [
	287,
	128003,
	2949,
	287,
	128007,
	767,
	128007,
	376,
	128007,
	392,
	128007,
	404,
	128007,
	453,
	128007,
	456,
	128007,
	525,
	128007,
	574,
	128007,
	578,
	128007,
	712,
	128007,
	466,
	1263,
	1263,
	128002,
	584,
	1754,
	278,
	267,
	375,
	4657,
	323,
	262,
	3918,
	269,
	3295,
	366,
	262,
	1686,
	45492,
	366,
	263,
	584,
	338,
	449,
	721,
	278,
	264,
	266,
	1156,
	323
	],
	"attention_mask": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"schema_tokens_list": [
	[
	"(",
	"[P]",
	"rating",
	"(",
	"[L]",
	"0",
	"[L]",
	"1",
	"[L]",
	"2",
	"[L]",
	"3",
	"[L]",
	"4",
	"[L]",
	"5",
	"[L]",
	"6",
	"[L]",
	"7",
	"[L]",
	"8",
	"[L]",
	"9",
	"[L]",
	"10",
	")",
	")"
	]
	],
	"subword_list": [
	"▁(",
	"[P]",
	"▁rating",
	"▁(",
	"[L]",
	"▁0",
	"[L]",
	"▁1",
	"[L]",
	"▁2",
	"[L]",
	"▁3",
	"[L]",
	"▁4",
	"[L]",
	"▁5",
	"[L]",
	"▁6",
	"[L]",
	"▁7",
	"[L]",
	"▁8",
	"[L]",
	"▁9",
	"[L]",
	"▁10",
	"▁)",
	"▁)",
	"[SEP_TEXT]",
	"▁i",
	"▁finished",
	"▁it",
	"▁in",
	"▁two",
	"▁nights",
	"▁.",
	"▁the",
	"▁ending",
	"▁is",
	"▁earned",
	"▁,",
	"▁the",
	"▁middle",
	"▁drags",
	"▁,",
	"▁and",
	"▁i",
	"▁would",
	"▁still",
	"▁hand",
	"▁it",
	"▁to",
	"▁a",
	"▁friend",
	"▁."
	],
	"schema_special_indices": [
	[
	1,
	4,
	6,
	8,
	10,
	12,
	14,
	16,
	18,
	20,
	22,
	24
	]
	],
	"task_results": [
	{
	"task": "rating",
	"labels": [
	"0",
	"1",
	"2",
	"3",
	"4",
	"5",
	"6",
	"7",
	"8",
	"9",
	"10"
	],
	"multi_label": false,
	"cls_threshold": 0.5,
	"class_act": "auto",
	"prompt_str": "rating",
	"logits": [
	-3.8949763774871826,
	-3.351461410522461,
	-2.8508293628692627,
	-2.602832794189453,
	-2.3061976432800293,
	-2.2791953086853027,
	-2.11843204498291,
	-1.9497541189193726,
	-2.0679240226745605,
	-2.276475191116333,
	-3.082228422164917
	],
	"probs": [
	0.021952487528324127,
	0.03780326247215271,
	0.06236644834280014,
	0.07991983741521835,
	0.10751809179782867,
	0.11046089231967926,
	0.12972602248191833,
	0.15356174111366272,
	0.1364465355873108,
	0.11076176166534424,
	0.049482960253953934
	],
	"decision": "7",
	"decision_probs": 0.15356174111366272
	}
	],
	"classify_text": {
	"rating": {
	"label": "7",
	"confidence": 0.15356174111366272
	}
	},
	"self_check": {
	"decisions_equal": true,
	"max_abs_prob_diff": 0.0
	},
	"readme_output": {
	"rating": "7"
	},
	"matches_readme_output": true
	}
	]
	}