row_id stringlengths 10 72 | family stringlengths 8 68 | task_type stringclasses 3
values | state stringlengths 16 12k | question stringlengths 144 3.66k | gold unknown | target stringlengths 20 540 | gold_weight float64 0.5 0.9 | holdout_family bool 2
classes | teacher stringclasses 1
value | teacher_calibrated stringlengths 20 540 ⌀ | teacher_conflict bool 2
classes | license stringclasses 10
values | provenance stringlengths 62 264 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
sni_task1599_smcalflow_classification-246 | sni_task1599_smcalflow_classification | choice | {"input": "Yup"} | {"type": "choice", "instructions": "In this task, you are given an utterance, which is a part of a conversation between a user and an agent. Your job is to detect the speaker. The user usually commands the agent to schedule events, retrieve event information, and check the schedule. While the agent's response is an ans... | "user" | {"user": 0.894412, "agent": 0.105588} | 0.6 | true | google/gemma-4-26B-A4B-it | {"user": 0.73603, "agent": 0.26397} | false | CC-BY-SA-4.0 | {"source": "Super-NaturalInstructions", "task": "task1599_smcalflow_classification.json", "task_source": ["smcalflow"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0235_health_claim_anomaly-18 | syn_0235_health_claim_anomaly | noul | {"claim_id": "CLM019", "patient_id": "PAT1019", "patient_name": "Sam Azure", "diagnosis_code": "D141", "total_claim_amount": 4000.0, "claim_line_items": [1333.33, 1333.33, 1333.34], "claim_line_items_count": 3, "average_line_item_amount": 1333.33, "provider_type": "out_of_network", "duplicate_claim_flag": false, "notes... | {"type": "noul", "instructions": "A claim is anomalous (true) if ANY of the following conditions is met:\n1) total_claim_amount > 5000\n2) claim_line_items_count > 5\n3) average_line_item_amount > 1500\n4) provider_type == \"out_of_network\" AND total_claim_amount > 2000\nOtherwise the claim is normal (false).", "crite... | true | {"false": 0.005085, "true": 0.994915} | 0.7 | true | google/gemma-4-26B-A4B-it | {"false": 0.016951, "true": 0.983049} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "health insurance claims", "pattern": "anomaly or violation detection in records, logs or measurements"} |
syn_0185_network_severity_rating-3 | syn_0185_network_severity_rating | score | {"incident_id": "INC004", "description": "Partial VLAN outage", "devices_affected": 45, "outage_duration_min": 55, "security_risk_level": 2, "reporter": "dave", "location": "dc_west", "tags": ["vlan", "partial"], "timestamp": "2026-07-04T14:20:00Z"} | {"type": "score", "instructions": "Assign a severity level index (0=low, 1=moderate, 2=high, 3=critical) for the incident described in the state, according to the following policy:\n\n- An incident is **critical** (index 3) if ANY of these conditions hold:\n * devices_affected >= 200\n * outage_duration_min >= 24... | 0 | {"0": 0.960751, "1": 0.018735, "2": 0.011274, "3": 0.009241} | 0.7 | true | google/gemma-4-26B-A4B-it | {"0": 0.869169, "1": 0.06245, "2": 0.037579, "3": 0.030803} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "network operations", "pattern": "severity or priority rating on a 3-5 level ordinal scale defined by explicit thresholds"} |
sni_task431_senteval_object_count-641 | sni_task431_senteval_object_count | choice | {"input": "But if you let this car stop out here, we don't have a prayer. \""} | {"type": "choice", "instructions": "In this task you are given a sentence. You must judge whether the object of the main clause is singular(like: apple) or plural(like: apartments). Label the instances as \"Singular\" or \"Plural\" based on your judgment.", "criteria": {"plural": "Plural", "singular": "Singular"}} | "singular" | {"plural": 0.105345, "singular": 0.894655} | 0.6 | true | google/gemma-4-26B-A4B-it | {"plural": 0.263364, "singular": 0.736636} | false | BSD | {"source": "Super-NaturalInstructions", "task": "task431_senteval_object_count.json", "task_source": ["senteval"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task1599_smcalflow_classification-543 | sni_task1599_smcalflow_classification | choice | {"input": "I'm unable to find whether R. House has a full bar. You'll have to contact them."} | {"type": "choice", "instructions": "In this task, you are given an utterance, which is a part of a conversation between a user and an agent. Your job is to detect the speaker. The user usually commands the agent to schedule events, retrieve event information, and check the schedule. While the agent's response is an ans... | "agent" | {"user": 0.002074, "agent": 0.997926} | 0.6 | true | google/gemma-4-26B-A4B-it | {"user": 0.005185, "agent": 0.994815} | false | CC-BY-SA-4.0 | {"source": "Super-NaturalInstructions", "task": "task1599_smcalflow_classification.json", "task_source": ["smcalflow"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task351_winomt_classification_gender_identifiability_anti-128 | sni_task351_winomt_classification_gender_identifiability_anti | choice | {"input": "Sentence: The lawyer told the designer she received a lawsuit.\n Profession: lawyer"} | {"type": "choice", "instructions": "In this task, you are given a sentence and a profession. The sentence mentions two professions: one's gender is identifiable using the gendered pronouns in the text and the other's gender is unidentifiable. You are expected to return whether the given profession's gender is identifia... | "identifiable" | {"unidentifiable": 0.190008, "identifiable": 0.809992} | 0.6 | true | google/gemma-4-26B-A4B-it | {"unidentifiable": 0.475021, "identifiable": 0.524979} | false | MIT | {"source": "Super-NaturalInstructions", "task": "task351_winomt_classification_gender_identifiability_anti.json", "task_source": ["winomt"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0092_fraud_risk_score-20 | syn_0092_fraud_risk_score | score | {"transaction_id": "TXN021", "transaction_amount_usd": 20000, "user": {"user_id": "U021", "velocity_24h": 25, "prior_fraud_flags": 0}, "device": {"device_id": "D021", "device_trust_score": 10}, "ip_info": {"ip_country": "US", "billing_country": "CA", "ip_country_mismatch": true}, "merchant_category": "art", "extra_note... | {"type": "score", "instructions": "Given a transaction record, compute a risk score using the rules below and map the total points to a risk level index (0 = none, 1 = low, 2 = moderate, 3 = high, 4 = critical).\n\n**Scoring rules**\n1. **Transaction amount (USD)**\n - > 10 000 → +3 points\n - > 5 000 → +2 points\n... | 4 | {"0": 0.009572, "1": 0.019966, "2": 0.021633, "3": 0.038953, "4": 0.909876} | 0.7 | true | google/gemma-4-26B-A4B-it | {"0": 0.031908, "1": 0.066553, "2": 0.072111, "3": 0.129843, "4": 0.699585} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "fraud detection", "pattern": "risk rating combining several factors with explicit rules"} |
syn_0185_network_severity_rating-13 | syn_0185_network_severity_rating | score | {"incident_id": "INC014", "description": "Aggregated link congestion", "devices_affected": 0, "outage_duration_min": 130, "security_risk_level": 1, "reporter": "nick", "location": "dc_west", "tags": ["congestion"], "timestamp": "2026-07-14T16:45:00Z"} | {"type": "score", "instructions": "Assign a severity level index (0=low, 1=moderate, 2=high, 3=critical) for the incident described in the state, according to the following policy:\n\n- An incident is **critical** (index 3) if ANY of these conditions hold:\n * devices_affected >= 200\n * outage_duration_min >= 24... | 2 | {"0": 0.005343, "1": 0.01199, "2": 0.977323, "3": 0.005343} | 0.7 | true | google/gemma-4-26B-A4B-it | {"0": 0.017812, "1": 0.039967, "2": 0.92441, "3": 0.017812} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "network operations", "pattern": "severity or priority rating on a 3-5 level ordinal scale defined by explicit thresholds"} |
sni_task431_senteval_object_count-147 | sni_task431_senteval_object_count | choice | {"input": "\" Unfortunately, we didn't find the video of the match, but we found one thing about it."} | {"type": "choice", "instructions": "In this task you are given a sentence. You must judge whether the object of the main clause is singular(like: apple) or plural(like: apartments). Label the instances as \"Singular\" or \"Plural\" based on your judgment.", "criteria": {"plural": "Plural", "singular": "Singular"}} | "singular" | {"plural": 0.099583, "singular": 0.900417} | 0.6 | true | google/gemma-4-26B-A4B-it | {"plural": 0.248958, "singular": 0.751042} | false | BSD | {"source": "Super-NaturalInstructions", "task": "task431_senteval_object_count.json", "task_source": ["senteval"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task1197_atomic_classification_oreact-370 | sni_task1197_atomic_classification_oreact | choice | {"input": "Head: PersonX holds ___ with PersonY's forms<sep>Tail: able to do something they were previously unsure about"} | {"type": "choice", "instructions": "In this task, you are given a tuple, comprising Head and Tail, separated with <sep>. The Head and the Tail events are short phrases possibly involving participants. The names of specific people have been replaced by generic words (e.g., PersonX, PersonY, PersonZ). PersonX is always t... | "yes" | {"yes": 0.814853, "no": 0.185147} | 0.6 | true | google/gemma-4-26B-A4B-it | {"yes": 0.537132, "no": 0.462868} | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task1197_atomic_classification_oreact.json", "task_source": ["atomic"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0235_health_claim_anomaly-1 | syn_0235_health_claim_anomaly | noul | {"claim_id": "CLM002", "patient_id": "PAT1002", "patient_name": "Bob Blue", "diagnosis_code": "D124", "total_claim_amount": 5000.0, "claim_line_items": [1250.0, 1250.0, 1250.0, 1250.0], "claim_line_items_count": 4, "average_line_item_amount": 1250.0, "provider_type": "in_network", "duplicate_claim_flag": false, "notes"... | {"type": "noul", "instructions": "A claim is anomalous (true) if ANY of the following conditions is met:\n1) total_claim_amount > 5000\n2) claim_line_items_count > 5\n3) average_line_item_amount > 1500\n4) provider_type == \"out_of_network\" AND total_claim_amount > 2000\nOtherwise the claim is normal (false).", "crite... | false | {"false": 0.956713, "true": 0.043287} | 0.7 | true | google/gemma-4-26B-A4B-it | {"false": 0.855711, "true": 0.144289} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "health insurance claims", "pattern": "anomaly or violation detection in records, logs or measurements"} |
sni_task404_grailqa_paraphrase_validation-795 | sni_task404_grailqa_paraphrase_validation | choice | {"input": "original question: Find the next model year of [2016 Chevy Spark]\nparaphrase: What is the next model year for the [2016 Chevy Spark]?"} | {"type": "choice", "instructions": "You will be given two questions. You should decide whether the second question is a good paraphrase of the first one. If you are able to tell that the two questions are the same without any other information, answer \"Yes\", otherwise answer \"No\".", "criteria": {"no": "No", "yes": ... | "yes" | {"no": 0.092713, "yes": 0.907287} | 0.6 | true | google/gemma-4-26B-A4B-it | {"no": 0.231783, "yes": 0.768217} | false | CC-BY-SA-4.0 | {"source": "Super-NaturalInstructions", "task": "task404_grailqa_paraphrase_validation.json", "task_source": ["grailqa"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0214_expense_reimbursement_eligibility-21 | syn_0214_expense_reimbursement_eligibility | noul | {"employee": {"id": "E022", "name": "Vera Yule"}, "expense_id": "X1022", "total_amount": 1900, "days_since_purchase": 30, "category": "travel", "daily_allowance": 149, "budget_used_pct": 91, "department": "operations", "project_code": "P-121", "approver_id": "A01", "attachments": [], "notes": "Facility audit"} | {"type": "noul", "instructions": "Eligibility for expense reimbursement follows the policy: (1) total_amount must be <= 2000; (2) days_since_purchase must be <= 30; (3) category must be one of [travel, supplies, software]; (4) if category is travel then daily_allowance must be <= 150; (5) budget_used_pct must be <= 90.... | false | {"false": 0.937981, "true": 0.062019} | 0.7 | true | google/gemma-4-26B-A4B-it | {"false": 0.79327, "true": 0.20673} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "spreadsheet agent", "pattern": "eligibility or policy check: does the case satisfy a written policy with several conditions"} |
syn_0185_network_severity_rating-16 | syn_0185_network_severity_rating | score | {"incident_id": "INC017", "description": "Regional outage affecting many sites", "devices_affected": 180, "outage_duration_min": 200, "security_risk_level": 3, "reporter": "quinn", "location": "dc_south", "tags": ["regional"], "timestamp": "2026-07-17T07:50:00Z"} | {"type": "score", "instructions": "Assign a severity level index (0=low, 1=moderate, 2=high, 3=critical) for the incident described in the state, according to the following policy:\n\n- An incident is **critical** (index 3) if ANY of these conditions hold:\n * devices_affected >= 200\n * outage_duration_min >= 24... | 2 | {"0": 0.005271, "1": 0.008977, "2": 0.980482, "3": 0.005271} | 0.7 | true | google/gemma-4-26B-A4B-it | {"0": 0.017569, "1": 0.029923, "2": 0.934939, "3": 0.017569} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "network operations", "pattern": "severity or priority rating on a 3-5 level ordinal scale defined by explicit thresholds"} |
syn_0075_automotive_triage-23 | syn_0075_automotive_triage | choice | {"case_id": "SA004", "issue_category": "tires", "severity_score": 2, "warranty_months_remaining": 0, "vehicle_age_years": 4, "parts_in_stock": false, "customer_loyalty": "platinum", "notes": "Tire pressure low"} | {"type": "choice", "instructions": "Route each service case to the appropriate team according to the following policy:\n1. If issue_category == \"engine\" AND severity_score >= 7 → route to engine_repair.\n2. Else if issue_category == \"electrical\" AND (severity_score >= 5 OR warranty_months_remaining > 0) → route to ... | "service_advisor" | {"engine_repair": 0.001374, "electrical": 0.001374, "body_shop": 0.001374, "tire_center": 0.001374, "service_advisor": 0.994506} | 0.7 | true | google/gemma-4-26B-A4B-it | {"engine_repair": 0.004578, "electrical": 0.004578, "body_shop": 0.004578, "tire_center": 0.004578, "service_advisor": 0.981686} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "automotive service", "pattern": "routing or triage: which team, queue or handler should take the case"} |
sni_task1390_wscfixed_coreference-565 | sni_task1390_wscfixed_coreference | choice | {"input": "Meanwhile, in the forest, the elephants are calling and hunting high and low for Arthur and Celeste , and their mothers are very worried. Fortunately, in flying over the town, an old marabou bird has seen (them) and come back quickly to tell the news. <sep> Pronoun:them <sep> Noun: their mothers"} | {"type": "choice", "instructions": "You are given a context, a pronoun, and a noun in this task. The given pronoun is shown in the context within parentheses. You should determine if the pronoun refers to the given noun or not. Please answer with \"True\" and \"False\".", "criteria": {"true": "True", "false": "False"}} | false | {"true": 0.174769, "false": 0.825231} | 0.6 | true | google/gemma-4-26B-A4B-it | {"true": 0.436923, "false": 0.563077} | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task1390_wscfixed_coreference.json", "task_source": ["wsc_fiexed"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0141_claim_quality_rating-12 | syn_0141_claim_quality_rating | score | {"claim_id": "CLM0013", "claimant_name": "Mia Bronze", "claimant_id": "PAT1013", "provider_name": "Uptown Health", "claim_amount": 1580.45, "diagnosis_codes": ["D101"], "procedure_codes": ["P606"], "claim_documentation_score": 68, "claim_processing_time_days": 8, "claim_error_rate_percent": 6.0, "notes": "Error rate hi... | {"type": "score", "instructions": "Rate the claim processing quality according to the following rubric:\n\n- If claim_documentation_score < 60 → rating = poor.\n- Else if claim_processing_time_days <= 5 AND claim_error_rate_percent <= 2 → rating = excellent.\n- Else if claim_processing_time_days <= 10 AND claim_error_r... | 1 | {"0": 0.012825, "1": 0.952679, "2": 0.014956, "3": 0.01954} | 0.7 | true | google/gemma-4-26B-A4B-it | {"0": 0.042751, "1": 0.842262, "2": 0.049854, "3": 0.065133} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "health insurance claims", "pattern": "quality rating against a rubric with defined levels"} |
sni_task1390_wscfixed_coreference-253 | sni_task1390_wscfixed_coreference | choice | {"input": "Alice looked for her friend Jade in the crowd. Since (she) always wears a red turban, Alice spotted her quickly. <sep> Pronoun:she <sep> Noun: Alice"} | {"type": "choice", "instructions": "You are given a context, a pronoun, and a noun in this task. The given pronoun is shown in the context within parentheses. You should determine if the pronoun refers to the given noun or not. Please answer with \"True\" and \"False\".", "criteria": {"false": "False", "true": "True"}} | false | {"false": 0.764441, "true": 0.235559} | 0.6 | true | google/gemma-4-26B-A4B-it | {"false": 0.411102, "true": 0.588898} | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task1390_wscfixed_coreference.json", "task_source": ["wsc_fiexed"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task065_timetravel_consistent_sentence_classification-376 | sni_task065_timetravel_consistent_sentence_classification | choice | {"input": "Sentence 1: I was minding my own business and taking a shower. \n Sentence 3: The drain had a couple hairs sticking out \n Sentence 4: Then, I realize it wasn't hairs, but legs \n Sentence 5: A spider came out of the drain and I screamed \n Option 1: When I look down towards the ground I noticed something.... | {"type": "choice", "instructions": "In this task, you are given a short story consisting of exactly 5 sentences where the second sentence is missing. You are given two options and you need to select the one that best connects the first sentence with the rest of the story. Indicate your answer by 'Option 1' if the first... | "option_1" | {"option_1": 0.974946, "option_2": 0.025054} | 0.6 | true | google/gemma-4-26B-A4B-it | {"option_1": 0.937364, "option_2": 0.062636} | false | MIT | {"source": "Super-NaturalInstructions", "task": "task065_timetravel_consistent_sentence_classification.json", "task_source": ["timetravel"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task1204_atomic_classification_hinderedby-696 | sni_task1204_atomic_classification_hinderedby | choice | {"input": "Head: PersonX decides to keep it<sep>Tail: Isn't allowed to have it."} | {"type": "choice", "instructions": "In this task, you are given two phrases: Head and Tail, separated with <sep>. The Head and the Tail events are short phrases possibly involving participants. The names of specific people have been replaced by generic words (e.g., PersonX, PersonY, PersonZ). PersonX is always the subj... | "yes" | {"no": 0.024272, "yes": 0.975728} | 0.6 | true | google/gemma-4-26B-A4B-it | {"no": 0.06068, "yes": 0.93932} | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task1204_atomic_classification_hinderedby.json", "task_source": ["atomic"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task065_timetravel_consistent_sentence_classification-516 | sni_task065_timetravel_consistent_sentence_classification | choice | {"input": "Sentence 1: James was out for a morning walk. \n Sentence 3: He tripped over his shoe lace \n Sentence 4: He felt onto the grass \n Sentence 5: James ended up bruising his knee from his fall \n Option 1: He had forgotten to tie his right shoe. \n Option 2: He had forgotten to tie his left shoe."} | {"type": "choice", "instructions": "In this task, you are given a short story consisting of exactly 5 sentences where the second sentence is missing. You are given two options and you need to select the one that best connects the first sentence with the rest of the story. Indicate your answer by 'Option 1' if the first... | "option_1" | {"option_1": 0.745639, "option_2": 0.254361} | 0.6 | true | google/gemma-4-26B-A4B-it | {"option_1": 0.364098, "option_2": 0.635902} | false | MIT | {"source": "Super-NaturalInstructions", "task": "task065_timetravel_consistent_sentence_classification.json", "task_source": ["timetravel"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task834_mathdataset_classification-761 | sni_task834_mathdataset_classification | choice | {"input": "What is sixty-four sevenths of a week in minutes?"} | {"type": "choice", "instructions": "You are supposed to identify the category of a high-school level math question. There are five possible categories (1) algebra (2) arithmetic (3) measurement (4) numbers, and (5) probability. Use the following guidelines: (1) 'algebra' questions will typically contain letter variable... | "measurement" | {"algebra": 0.011646, "numbers": 0.014461, "probability": 0.010061, "measurement": 0.952702, "arithmetic": 0.01113} | 0.6 | true | google/gemma-4-26B-A4B-it | {"algebra": 0.029115, "numbers": 0.036151, "probability": 0.025154, "measurement": 0.881755, "arithmetic": 0.027825} | false | Apache-2.0 | {"source": "Super-NaturalInstructions", "task": "task834_mathdataset_classification.json", "task_source": ["mathmatics_dataset"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task1204_atomic_classification_hinderedby-31 | sni_task1204_atomic_classification_hinderedby | choice | {"input": "Head: PersonX has a christmas party<sep>Tail: PersonX's children are ill."} | {"type": "choice", "instructions": "In this task, you are given two phrases: Head and Tail, separated with <sep>. The Head and the Tail events are short phrases possibly involving participants. The names of specific people have been replaced by generic words (e.g., PersonX, PersonY, PersonZ). PersonX is always the subj... | "yes" | {"no": 0.033088, "yes": 0.966912} | 0.6 | true | google/gemma-4-26B-A4B-it | {"no": 0.08272, "yes": 0.91728} | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task1204_atomic_classification_hinderedby.json", "task_source": ["atomic"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task404_grailqa_paraphrase_validation-491 | sni_task404_grailqa_paraphrase_validation | choice | {"input": "original question: Which type of bicycle use model [Avanti Prima]?\nparaphrase: [Avanti Prima] is the model for what type of bicycle?"} | {"type": "choice", "instructions": "You will be given two questions. You should decide whether the second question is a good paraphrase of the first one. If you are able to tell that the two questions are the same without any other information, answer \"Yes\", otherwise answer \"No\".", "criteria": {"yes": "Yes", "no":... | "yes" | {"yes": 0.880252, "no": 0.119748} | 0.6 | true | google/gemma-4-26B-A4B-it | {"yes": 0.70063, "no": 0.29937} | false | CC-BY-SA-4.0 | {"source": "Super-NaturalInstructions", "task": "task404_grailqa_paraphrase_validation.json", "task_source": ["grailqa"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0069_hotel_reservation_priority-17 | syn_0069_hotel_reservation_priority | score | {"reservation_id": "RZ-1018", "guest_name": "Rosa Mendes", "booking_value_usd": 1600, "number_of_guests": 4, "special_requests": [], "special_requests_count": 0, "days_before_arrival": 2, "room_type": "standard", "loyalty_status": "silver", "promo_code": null, "notes": ""} | {"type": "score", "instructions": "Assign a priority level (low, medium, high, critical) to each reservation according to the policy below.\n\nPolicy:\n1. If `days_before_arrival` <= 1 → critical\n2. Else if `booking_value_usd` >= 2000 AND `special_requests_count` >= 2 → critical\n3. Else if `booking_value_usd` >= 1500... | 2 | {"0": 0.002794, "1": 0.003351, "2": 0.991061, "3": 0.002794} | 0.7 | true | google/gemma-4-26B-A4B-it | {"0": 0.009314, "1": 0.011169, "2": 0.970204, "3": 0.009314} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "hotel reservations", "pattern": "severity or priority rating on a 3-5 level ordinal scale defined by explicit thresholds"} |
syn_0214_expense_reimbursement_eligibility-4 | syn_0214_expense_reimbursement_eligibility | noul | {"employee": {"id": "E005", "name": "Eli Grant"}, "expense_id": "X1005", "total_amount": 2000, "days_since_purchase": 30, "category": "supplies", "daily_allowance": null, "budget_used_pct": 90, "department": "finance", "project_code": "P-104", "approver_id": "A08", "attachments": [], "notes": "Stationery bulk order"} | {"type": "noul", "instructions": "Eligibility for expense reimbursement follows the policy: (1) total_amount must be <= 2000; (2) days_since_purchase must be <= 30; (3) category must be one of [travel, supplies, software]; (4) if category is travel then daily_allowance must be <= 150; (5) budget_used_pct must be <= 90.... | true | {"false": 0.01229, "true": 0.98771} | 0.7 | true | google/gemma-4-26B-A4B-it | {"false": 0.040967, "true": 0.959033} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "spreadsheet agent", "pattern": "eligibility or policy check: does the case satisfy a written policy with several conditions"} |
syn_0068_course_scheduling_next_action-12 | syn_0068_course_scheduling_next_action | choice | {"pending_courses": [{"course_id": "C113", "enrollment": 27, "max_capacity": 30, "instructor_id": "I12", "conflict_score": 4, "room_needed": 2, "preferred_time_slot": "slot_14am"}], "instructor_load": {"I12": 12}, "max_instructor_load": 20, "room_availability": {"slot_14am": 2}, "department": "Law", "term": "Fall2026",... | {"type": "choice", "instructions": "Given the current scheduling state, select the next best action for the automated agent. Policy: For the first pending course C, apply the ordered rules: 1) If C.conflict_score >= 4 → reassign_instructor. 2) Else if C.enrollment > C.max_capacity → assign_room. 3) Else if instructor_l... | "reassign_instructor" | {"assign_room": 0.001965, "delay_course": 0.001965, "reassign_instructor": 0.994104, "notify_student": 0.001965} | 0.7 | true | google/gemma-4-26B-A4B-it | {"assign_room": 0.006551, "delay_course": 0.006551, "reassign_instructor": 0.980347, "notify_student": 0.006551} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "course scheduling", "pattern": "next best action for an automated agent given its goal and history"} |
sni_task1206_atomic_classification_isbefore-713 | sni_task1206_atomic_classification_isbefore | choice | {"input": "Head: PersonX sees a movie with PersonY's friends<sep>Tail: PersonX invited PersonY next time"} | {"type": "choice", "instructions": "In this task, you are given two phrases: Head and Tail, separated with <sep>. The Head and the Tail events are short phrases possibly involving participants. The names of specific people have been replaced by generic words (e.g., PersonX, PersonY, PersonZ). PersonX is always the subj... | "yes" | {"yes": 0.840521, "no": 0.159479} | 0.6 | true | google/gemma-4-26B-A4B-it | {"yes": 0.601302, "no": 0.398698} | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task1206_atomic_classification_isbefore.json", "task_source": ["atomic"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0105_loan_underwriting_next_action-16 | syn_0105_loan_underwriting_next_action | choice | {"applicant": {"id": "A017", "name": "Paula Torres", "credit_score": 770, "annual_income": 110000, "loan_amount": 42000, "employment_status": "full_time", "has_verified_employment": true, "fraud_flag": false}, "financials": {"total_monthly_debt": 2000, "monthly_income": 9166.67, "debt_to_income": 0.218, "loan_to_income... | {"type": "choice", "instructions": "Given the applicant's state, choose the next best action for the automated underwriting agent.\nPolicy:\n- If applicant.credit_score >= 750 AND financials.debt_to_income <= 0.30 AND financials.loan_to_income <= 0.40 => approve\n- Else if applicant.credit_score >= 650 AND financials.d... | "approve" | {"approve": 0.966578, "request_more_info": 0.014897, "decline": 0.010426, "escalated_review": 0.008098} | 0.7 | true | google/gemma-4-26B-A4B-it | {"approve": 0.888595, "request_more_info": 0.049658, "decline": 0.034755, "escalated_review": 0.026992} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "loan underwriting", "pattern": "next best action for an automated agent given its goal and history"} |
sni_task108_contextualabusedetection_classification-738 | sni_task108_contextualabusedetection_classification | choice | {"input": "Writing from hospital, lying on my stomach. Goli pichhwaade me lagi thi"} | {"type": "choice", "instructions": "Read the given text and if it has abusive content then indicate via \"yes\". Otherwise indicate via\"no\". \nWe consider the content to be abusive if it contains any of the following instances: (1) Identity Directed Abuse (e.g., Content which contains a negative statement made agains... | "no" | {"yes": 0.072388, "no": 0.927612} | 0.6 | true | google/gemma-4-26B-A4B-it | {"yes": 0.18097, "no": 0.81903} | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task108_contextualabusedetection_classification.json", "task_source": ["cad"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task1599_smcalflow_classification-301 | sni_task1599_smcalflow_classification | choice | {"input": "That's good"} | {"type": "choice", "instructions": "In this task, you are given an utterance, which is a part of a conversation between a user and an agent. Your job is to detect the speaker. The user usually commands the agent to schedule events, retrieve event information, and check the schedule. While the agent's response is an ans... | "user" | {"user": 0.928311, "agent": 0.071689} | 0.6 | true | google/gemma-4-26B-A4B-it | {"user": 0.820777, "agent": 0.179223} | false | CC-BY-SA-4.0 | {"source": "Super-NaturalInstructions", "task": "task1599_smcalflow_classification.json", "task_source": ["smcalflow"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0142_university_admissions_rating-11 | syn_0142_university_admissions_rating | score | {"applicant_id": "A012", "name": "Liam Moss", "GPA": 3.7, "SAT_score": 1400, "extracurricular_score": 7, "major_interest": "Anthropology", "essays_submitted": 1, "recommendation_count": 2, "legacy_status": false, "hometown": "Sunnyvale"} | {"type": "score", "instructions": "Assign a quality rating index to each applicant according to the following rubric (0 = low, 1 = medium, 2 = high):\n- High (2): GPA >= 3.8 AND SAT_score >= 1450 AND extracurricular_score >= 8\n- Medium (1): GPA >= 3.5 AND SAT_score >= 1300 AND extracurricular_score >= 5\n- Low (0): an... | 1 | {"0": 0.006123, "1": 0.989514, "2": 0.004364} | 0.7 | true | google/gemma-4-26B-A4B-it | {"0": 0.020409, "1": 0.965046, "2": 0.014545} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "university admissions", "pattern": "quality rating against a rubric with defined levels"} |
syn_0142_university_admissions_rating-9 | syn_0142_university_admissions_rating | score | {"applicant_id": "A010", "name": "Jack Knight", "GPA": 3.5, "SAT_score": 1300, "extracurricular_score": 5, "major_interest": "Sociology", "essays_submitted": 1, "recommendation_count": 2, "legacy_status": false, "hometown": "Elmwood"} | {"type": "score", "instructions": "Assign a quality rating index to each applicant according to the following rubric (0 = low, 1 = medium, 2 = high):\n- High (2): GPA >= 3.8 AND SAT_score >= 1450 AND extracurricular_score >= 8\n- Medium (1): GPA >= 3.5 AND SAT_score >= 1300 AND extracurricular_score >= 5\n- Low (0): an... | 1 | {"0": 0.003507, "1": 0.992987, "2": 0.003507} | 0.7 | true | google/gemma-4-26B-A4B-it | {"0": 0.011689, "1": 0.976623, "2": 0.011689} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "university admissions", "pattern": "quality rating against a rubric with defined levels"} |
sni_task431_senteval_object_count-502 | sni_task431_senteval_object_count | choice | {"input": "Nate shoved a handful of Shock Bits into his mouth, and a glob of orange jelly splashed into him, accompanied by a flash."} | {"type": "choice", "instructions": "In this task you are given a sentence. You must judge whether the object of the main clause is singular(like: apple) or plural(like: apartments). Label the instances as \"Singular\" or \"Plural\" based on your judgment.", "criteria": {"plural": "Plural", "singular": "Singular"}} | "singular" | {"plural": 0.088522, "singular": 0.911478} | 0.6 | true | google/gemma-4-26B-A4B-it | {"plural": 0.221306, "singular": 0.778694} | false | BSD | {"source": "Super-NaturalInstructions", "task": "task431_senteval_object_count.json", "task_source": ["senteval"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task1599_smcalflow_classification-161 | sni_task1599_smcalflow_classification | choice | {"input": "Create dinner date with Sam on Friday night"} | {"type": "choice", "instructions": "In this task, you are given an utterance, which is a part of a conversation between a user and an agent. Your job is to detect the speaker. The user usually commands the agent to schedule events, retrieve event information, and check the schedule. While the agent's response is an ans... | "user" | {"user": 0.995146, "agent": 0.004854} | 0.6 | true | google/gemma-4-26B-A4B-it | {"user": 0.987865, "agent": 0.012135} | false | CC-BY-SA-4.0 | {"source": "Super-NaturalInstructions", "task": "task1599_smcalflow_classification.json", "task_source": ["smcalflow"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task1204_atomic_classification_hinderedby-343 | sni_task1204_atomic_classification_hinderedby | choice | {"input": "Head: PersonX becomes fast friends<sep>Tail: to know more people"} | {"type": "choice", "instructions": "In this task, you are given two phrases: Head and Tail, separated with <sep>. The Head and the Tail events are short phrases possibly involving participants. The names of specific people have been replaced by generic words (e.g., PersonX, PersonY, PersonZ). PersonX is always the subj... | "no" | {"no": 0.950168, "yes": 0.049832} | 0.6 | true | google/gemma-4-26B-A4B-it | {"no": 0.875421, "yes": 0.124579} | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task1204_atomic_classification_hinderedby.json", "task_source": ["atomic"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task351_winomt_classification_gender_identifiability_anti-474 | sni_task351_winomt_classification_gender_identifiability_anti | choice | {"input": "Sentence: The writer retained the lawyer and then explained her the situation.\n Profession: writer"} | {"type": "choice", "instructions": "In this task, you are given a sentence and a profession. The sentence mentions two professions: one's gender is identifiable using the gendered pronouns in the text and the other's gender is unidentifiable. You are expected to return whether the given profession's gender is identifia... | "unidentifiable" | {"unidentifiable": 0.792753, "identifiable": 0.207247} | 0.6 | true | google/gemma-4-26B-A4B-it | {"unidentifiable": 0.481883, "identifiable": 0.518117} | false | MIT | {"source": "Super-NaturalInstructions", "task": "task351_winomt_classification_gender_identifiability_anti.json", "task_source": ["winomt"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0075_automotive_triage-6 | syn_0075_automotive_triage | choice | {"case_id": "ELE002", "issue_category": "electrical", "severity_score": 4, "warranty_months_remaining": 6, "vehicle_age_years": 2, "parts_in_stock": true, "customer_loyalty": "gold", "notes": "Intermittent lights"} | {"type": "choice", "instructions": "Route each service case to the appropriate team according to the following policy:\n1. If issue_category == \"engine\" AND severity_score >= 7 → route to engine_repair.\n2. Else if issue_category == \"electrical\" AND (severity_score >= 5 OR warranty_months_remaining > 0) → route to ... | "electrical" | {"engine_repair": 0.002177, "electrical": 0.990285, "body_shop": 0.002177, "tire_center": 0.002177, "service_advisor": 0.003184} | 0.7 | true | google/gemma-4-26B-A4B-it | {"engine_repair": 0.007256, "electrical": 0.967618, "body_shop": 0.007256, "tire_center": 0.007256, "service_advisor": 0.010615} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "automotive service", "pattern": "routing or triage: which team, queue or handler should take the case"} |
sni_task356_casino_classification_negotiation_self_need-173 | sni_task356_casino_classification_negotiation_self_need | choice | {"input": "Context: 'How's it going? If it's alright with you, you can have all of the water and I'll take all of the firewood. We can work out a deal on the food as well.'\nUtterance: 'I have circulatory issued due to a heart condition and cold quickly. I would appreciate if I could get at least 2 of the firewood. Y... | {"type": "choice", "instructions": "The input is taken from a negotiation between two participants who take the role of campsite neighbors and negotiate for Food, Water, and Firewood packages, based on their individual preferences and requirements. Given an utterance and recent dialogue context containing past 3 uttera... | "yes" | {"no": 0.044882, "yes": 0.955118} | 0.6 | true | google/gemma-4-26B-A4B-it | {"no": 0.112205, "yes": 0.887795} | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task356_casino_classification_negotiation_self_need.json", "task_source": ["casino"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task065_timetravel_consistent_sentence_classification-51 | sni_task065_timetravel_consistent_sentence_classification | choice | {"input": "Sentence 1: Sam asked if he could live with his aunt. \n Sentence 3: She took him to the zoo \n Sentence 4: She bought him a stuffed giraffe \n Sentence 5: She let him touch the stingrays \n Option 1: He was at her house all the time. \n Option 2: She said no."} | {"type": "choice", "instructions": "In this task, you are given a short story consisting of exactly 5 sentences where the second sentence is missing. You are given two options and you need to select the one that best connects the first sentence with the rest of the story. Indicate your answer by 'Option 1' if the first... | "option_1" | {"option_1": 0.961182, "option_2": 0.038818} | 0.6 | true | google/gemma-4-26B-A4B-it | {"option_1": 0.902954, "option_2": 0.097046} | false | MIT | {"source": "Super-NaturalInstructions", "task": "task065_timetravel_consistent_sentence_classification.json", "task_source": ["timetravel"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task1431_head_qa_answer_generation-37 | sni_task1431_head_qa_answer_generation | choice | {"input": "Question: The presence of nonspecific prodromes in a psychotic disorder: \n Options: <1> It refers to one of the three phases in which a psychotic episode evolves. <2> It refers to the acute phase. <3> It refers to the recovery phase. <4> It defines the abrupt onset of the first psychotic episode. <5> It is... | {"type": "choice", "instructions": "In this task, you are given a multiple-choice question about healthcare. Answer the question based on your information and classify your answers into '1', '2', '3', and '4'.", "criteria": {"4": "4", "5": "5", "1": "1", "2": "2", "3": "3"}} | 1 | {"4": 0.031575, "5": 0.022682, "1": 0.894469, "2": 0.022682, "3": 0.028592} | 0.6 | true | google/gemma-4-26B-A4B-it | {"4": 0.078937, "5": 0.056705, "1": 0.736172, "2": 0.056705, "3": 0.07148} | false | MIT | {"source": "Super-NaturalInstructions", "task": "task1431_head_qa_answer_generation.json", "task_source": ["head_qa"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task1390_wscfixed_coreference-143 | sni_task1390_wscfixed_coreference | choice | {"input": "Frank was upset with Tom because the toaster (he) had sold him didn't work. <sep> Pronoun:he <sep> Noun: Frank"} | {"type": "choice", "instructions": "You are given a context, a pronoun, and a noun in this task. The given pronoun is shown in the context within parentheses. You should determine if the pronoun refers to the given noun or not. Please answer with \"True\" and \"False\".", "criteria": {"false": "False", "true": "True"}} | false | {"false": 0.813355, "true": 0.186645} | 0.6 | true | google/gemma-4-26B-A4B-it | {"false": 0.533387, "true": 0.466613} | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task1390_wscfixed_coreference.json", "task_source": ["wsc_fiexed"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task1599_smcalflow_classification-280 | sni_task1599_smcalflow_classification | choice | {"input": "Can you make that an hour instead?"} | {"type": "choice", "instructions": "In this task, you are given an utterance, which is a part of a conversation between a user and an agent. Your job is to detect the speaker. The user usually commands the agent to schedule events, retrieve event information, and check the schedule. While the agent's response is an ans... | "user" | {"agent": 0.007173, "user": 0.992827} | 0.6 | true | google/gemma-4-26B-A4B-it | {"agent": 0.017931, "user": 0.982069} | false | CC-BY-SA-4.0 | {"source": "Super-NaturalInstructions", "task": "task1599_smcalflow_classification.json", "task_source": ["smcalflow"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task356_casino_classification_negotiation_self_need-199 | sni_task356_casino_classification_negotiation_self_need | choice | {"input": "Context: 'I agree! Nice to meet you... I have two boys who always eat! Any chance I canget 2 food?' 'Food is the item of greatest value to you?' 'It is! What do you think?'\nUtterance: 'The item of greatest value to me is WATER. Before thinking about an offer I need to know what your priorities are and you n... | {"type": "choice", "instructions": "The input is taken from a negotiation between two participants who take the role of campsite neighbors and negotiate for Food, Water, and Firewood packages, based on their individual preferences and requirements. Given an utterance and recent dialogue context containing past 3 uttera... | "yes" | {"yes": 0.899904, "no": 0.100096} | 0.6 | true | google/gemma-4-26B-A4B-it | {"yes": 0.749759, "no": 0.250241} | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task356_casino_classification_negotiation_self_need.json", "task_source": ["casino"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task431_senteval_object_count-119 | sni_task431_senteval_object_count | choice | {"input": "The Pack had seven offices in Atlanta and the surrounding area and every single one of those would have clean windows."} | {"type": "choice", "instructions": "In this task you are given a sentence. You must judge whether the object of the main clause is singular(like: apple) or plural(like: apartments). Label the instances as \"Singular\" or \"Plural\" based on your judgment.", "criteria": {"singular": "Singular", "plural": "Plural"}} | "plural" | {"singular": 0.103379, "plural": 0.896621} | 0.6 | true | google/gemma-4-26B-A4B-it | {"singular": 0.258446, "plural": 0.741554} | false | BSD | {"source": "Super-NaturalInstructions", "task": "task431_senteval_object_count.json", "task_source": ["senteval"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0190_duplicate_detection_admissions-4 | syn_0190_duplicate_detection_admissions | noul | {"first_name_similarity": 0.89, "last_name_similarity": 0.91, "dob_match": true, "email_match": true, "phone_match": true, "high_school": "Lakeside College", "gpa": 3.9, "applicant_id_1": "A005", "applicant_id_2": "B005", "extra_notes": "First name similarity below threshold."} | {"type": "noul", "instructions": "Determine whether the two admission records refer to the same applicant according to the policy. Policy: Two records are considered duplicates if at least four of the following five criteria are satisfied: (1) first_name_similarity ≥ 0.90, (2) last_name_similarity ≥ 0.90, (3) dob_match... | true | {"false": 0.026957, "true": 0.973043} | 0.7 | true | google/gemma-4-26B-A4B-it | {"false": 0.089856, "true": 0.910144} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "university admissions", "pattern": "duplicate or same-entity detection between two records"} |
syn_0199_clinical_trial_next_action-18 | syn_0199_clinical_trial_next_action | choice | {"candidate": {"id": "C019", "age": 52, "sex": "M"}, "lab_results": {"hba1c": 7.8, "creatinine": 1.1}, "medical_history": {"conditions": [], "allergies": []}, "consent_status": "pending", "days_since_last_contact": 7, "screening_complete": true, "insurance": "PlanQ", "preferred_language": "fr", "note": "irrelevant"} | {"type": "choice", "instructions": "Given the candidate's screening state, select the next best action for the automated agent. Apply the following policy (ordered conditions, first match wins):\n1. If candidate.age < 18 or candidate.age > 75, answer 'exclude_candidate'.\n2. Else if candidate.medical_history.allergies ... | "send_reminder" | {"schedule_visit": 0.002123, "request_additional_lab": 0.004362, "exclude_candidate": 0.002255, "send_reminder": 0.988796, "hold_for_review": 0.002465} | 0.7 | true | google/gemma-4-26B-A4B-it | {"schedule_visit": 0.007075, "request_additional_lab": 0.014538, "exclude_candidate": 0.007516, "send_reminder": 0.962654, "hold_for_review": 0.008217} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "clinical trial screening", "pattern": "next best action for an automated agent given its goal and history"} |
sni_task834_mathdataset_classification-775 | sni_task834_mathdataset_classification | choice | {"input": "What is prob of picking 1 g and 1 k when two letters picked without replacement from {k: 1, l: 1, g: 1, b: 1, p: 1}?"} | {"type": "choice", "instructions": "You are supposed to identify the category of a high-school level math question. There are five possible categories (1) algebra (2) arithmetic (3) measurement (4) numbers, and (5) probability. Use the following guidelines: (1) 'algebra' questions will typically contain letter variable... | "probability" | {"arithmetic": 0.007814, "measurement": 0.007814, "probability": 0.966613, "algebra": 0.007814, "numbers": 0.009944} | 0.6 | true | google/gemma-4-26B-A4B-it | {"arithmetic": 0.019536, "measurement": 0.019536, "probability": 0.916533, "algebra": 0.019536, "numbers": 0.024859} | false | Apache-2.0 | {"source": "Super-NaturalInstructions", "task": "task834_mathdataset_classification.json", "task_source": ["mathmatics_dataset"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task834_mathdataset_classification-636 | sni_task834_mathdataset_classification | choice | {"input": "What comes next: 50, 19, -24, -85, -170, -285, -436, -629?"} | {"type": "choice", "instructions": "You are supposed to identify the category of a high-school level math question. There are five possible categories (1) algebra (2) arithmetic (3) measurement (4) numbers, and (5) probability. Use the following guidelines: (1) 'algebra' questions will typically contain letter variable... | "algebra" | {"measurement": 0.032715, "algebra": 0.652289, "probability": 0.019847, "numbers": 0.028563, "arithmetic": 0.266586} | 0.6 | true | google/gemma-4-26B-A4B-it | {"measurement": 0.081789, "algebra": 0.130722, "probability": 0.049618, "numbers": 0.071407, "arithmetic": 0.666464} | false | Apache-2.0 | {"source": "Super-NaturalInstructions", "task": "task834_mathdataset_classification.json", "task_source": ["mathmatics_dataset"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0075_automotive_triage-3 | syn_0075_automotive_triage | choice | {"case_id": "ENG004", "issue_category": "engine", "severity_score": 7, "warranty_months_remaining": 0, "vehicle_age_years": 12, "parts_in_stock": true, "customer_loyalty": "bronze", "notes": "Rough acceleration"} | {"type": "choice", "instructions": "Route each service case to the appropriate team according to the following policy:\n1. If issue_category == \"engine\" AND severity_score >= 7 → route to engine_repair.\n2. Else if issue_category == \"electrical\" AND (severity_score >= 5 OR warranty_months_remaining > 0) → route to ... | "engine_repair" | {"engine_repair": 0.989181, "electrical": 0.002934, "body_shop": 0.00247, "tire_center": 0.00247, "service_advisor": 0.002945} | 0.7 | true | google/gemma-4-26B-A4B-it | {"engine_repair": 0.963938, "electrical": 0.009781, "body_shop": 0.008232, "tire_center": 0.008232, "service_advisor": 0.009817} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "automotive service", "pattern": "routing or triage: which team, queue or handler should take the case"} |
sni_task834_mathdataset_classification-41 | sni_task834_mathdataset_classification | choice | {"input": "Suppose -13 + 20 = s. What is the units digit of s?"} | {"type": "choice", "instructions": "You are supposed to identify the category of a high-school level math question. There are five possible categories (1) algebra (2) arithmetic (3) measurement (4) numbers, and (5) probability. Use the following guidelines: (1) 'algebra' questions will typically contain letter variable... | "numbers" | {"numbers": 0.627612, "algebra": 0.177976, "probability": 0.017064, "arithmetic": 0.157314, "measurement": 0.020034} | 0.6 | true | google/gemma-4-26B-A4B-it | {"numbers": 0.069031, "algebra": 0.444939, "probability": 0.042661, "arithmetic": 0.393285, "measurement": 0.050084} | true | Apache-2.0 | {"source": "Super-NaturalInstructions", "task": "task834_mathdataset_classification.json", "task_source": ["mathmatics_dataset"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task065_timetravel_consistent_sentence_classification-272 | sni_task065_timetravel_consistent_sentence_classification | choice | {"input": "Sentence 1: Wilberforce was an amateur bull rider. \n Sentence 3: Tired of the taunts, Wilberforce decided to shorten his name \n Sentence 4: He now participated in the rodeo as Wilber \n Sentence 5: The other cowboys quit making fun of him \n Option 1: He was a good bull rider, but the other cowboys made ... | {"type": "choice", "instructions": "In this task, you are given a short story consisting of exactly 5 sentences where the second sentence is missing. You are given two options and you need to select the one that best connects the first sentence with the rest of the story. Indicate your answer by 'Option 1' if the first... | "option_1" | {"option_1": 0.956996, "option_2": 0.043004} | 0.6 | true | google/gemma-4-26B-A4B-it | {"option_1": 0.89249, "option_2": 0.10751} | false | MIT | {"source": "Super-NaturalInstructions", "task": "task065_timetravel_consistent_sentence_classification.json", "task_source": ["timetravel"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task375_classify_type_of_sentence_in_debate-104 | sni_task375_classify_type_of_sentence_in_debate | choice | {"input": "topic: debates should be removed by customer service.\nsentence: \n \r\ncustomer service should delete some debates."} | {"type": "choice", "instructions": "In this task, you will be given a debate topic, along with a sentence from the debate. You should classify the given sentence and choose the type of that sentence. Possible types are explained below.\nPolicy: This refers to a sentence proposing a specific course of action to be taken... | "policy" | {"value": 0.007444, "policy": 0.977668, "testimony": 0.007444, "fact": 0.007444} | 0.6 | true | google/gemma-4-26B-A4B-it | {"value": 0.01861, "policy": 0.94417, "testimony": 0.01861, "fact": 0.01861} | false | CC-BY-SA-3.0 | {"source": "Super-NaturalInstructions", "task": "task375_classify_type_of_sentence_in_debate.json", "task_source": ["ddo"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task1206_atomic_classification_isbefore-630 | sni_task1206_atomic_classification_isbefore | choice | {"input": "Head: PersonX accuses PersonY of cheating<sep>Tail: justified"} | {"type": "choice", "instructions": "In this task, you are given two phrases: Head and Tail, separated with <sep>. The Head and the Tail events are short phrases possibly involving participants. The names of specific people have been replaced by generic words (e.g., PersonX, PersonY, PersonZ). PersonX is always the subj... | "no" | {"yes": 0.209361, "no": 0.790639} | 0.6 | true | google/gemma-4-26B-A4B-it | {"yes": 0.523402, "no": 0.476598} | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task1206_atomic_classification_isbefore.json", "task_source": ["atomic"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task1197_atomic_classification_oreact-580 | sni_task1197_atomic_classification_oreact | choice | {"input": "Head: PersonX makes ___ this year<sep>Tail: welcoming"} | {"type": "choice", "instructions": "In this task, you are given a tuple, comprising Head and Tail, separated with <sep>. The Head and the Tail events are short phrases possibly involving participants. The names of specific people have been replaced by generic words (e.g., PersonX, PersonY, PersonZ). PersonX is always t... | "yes" | {"no": 0.1818, "yes": 0.8182} | 0.6 | true | google/gemma-4-26B-A4B-it | {"no": 0.454501, "yes": 0.545499} | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task1197_atomic_classification_oreact.json", "task_source": ["atomic"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task1197_atomic_classification_oreact-621 | sni_task1197_atomic_classification_oreact | choice | {"input": "Head: PersonX decides to keep it<sep>Tail: none"} | {"type": "choice", "instructions": "In this task, you are given a tuple, comprising Head and Tail, separated with <sep>. The Head and the Tail events are short phrases possibly involving participants. The names of specific people have been replaced by generic words (e.g., PersonX, PersonY, PersonZ). PersonX is always t... | "yes" | {"no": 0.216959, "yes": 0.783041} | 0.6 | true | google/gemma-4-26B-A4B-it | {"no": 0.542398, "yes": 0.457602} | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task1197_atomic_classification_oreact.json", "task_source": ["atomic"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task672_nummersense-341 | sni_task672_nummersense | choice | {"input": "Cuckoos live less than _ years in the wild."} | {"type": "choice", "instructions": "In this task, find the most appropriate number to replace the blank (indicated with _ ) and express it in words.", "criteria": {"three": "three", "four": "four", "six": "six", "eight": "eight", "seven": "seven", "no": "no", "five": "five", "one": "one", "ten": "ten", "two": "two", "z... | "six" | {"three": 0.025454, "four": 0.02426, "six": 0.615788, "eight": 0.018963, "seven": 0.017173, "no": 0.016417, "five": 0.120657, "one": 0.02485, "ten": 0.011744, "two": 0.086186, "zero": 0.017381, "nine": 0.021129} | 0.6 | true | google/gemma-4-26B-A4B-it | {"three": 0.063635, "four": 0.060649, "six": 0.039469, "eight": 0.047407, "seven": 0.042933, "no": 0.041041, "five": 0.301643, "one": 0.062124, "ten": 0.02936, "two": 0.215464, "zero": 0.043452, "nine": 0.052821} | true | MIT | {"source": "Super-NaturalInstructions", "task": "task672_nummersense.json", "task_source": ["numersense"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0141_claim_quality_rating-18 | syn_0141_claim_quality_rating | score | {"claim_id": "CLM0019", "claimant_name": "Sam Emerald", "claimant_id": "PAT1019", "provider_name": "Valley Health", "claim_amount": 2550.0, "diagnosis_codes": ["D707"], "procedure_codes": ["P212"], "claim_documentation_score": 60, "claim_processing_time_days": 6, "claim_error_rate_percent": 2.0, "notes": "Good rating",... | {"type": "score", "instructions": "Rate the claim processing quality according to the following rubric:\n\n- If claim_documentation_score < 60 → rating = poor.\n- Else if claim_processing_time_days <= 5 AND claim_error_rate_percent <= 2 → rating = excellent.\n- Else if claim_processing_time_days <= 10 AND claim_error_r... | 2 | {"0": 0.004012, "1": 0.006936, "2": 0.98504, "3": 0.004012} | 0.7 | true | google/gemma-4-26B-A4B-it | {"0": 0.013372, "1": 0.023121, "2": 0.950134, "3": 0.013372} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "health insurance claims", "pattern": "quality rating against a rubric with defined levels"} |
syn_0105_loan_underwriting_next_action-8 | syn_0105_loan_underwriting_next_action | choice | {"applicant": {"id": "A009", "name": "Helen Garcia", "credit_score": 750, "annual_income": 110000, "loan_amount": 44000, "employment_status": "full_time", "has_verified_employment": true, "fraud_flag": false}, "financials": {"total_monthly_debt": 2750, "monthly_income": 9166.67, "debt_to_income": 0.3, "loan_to_income":... | {"type": "choice", "instructions": "Given the applicant's state, choose the next best action for the automated underwriting agent.\nPolicy:\n- If applicant.credit_score >= 750 AND financials.debt_to_income <= 0.30 AND financials.loan_to_income <= 0.40 => approve\n- Else if applicant.credit_score >= 650 AND financials.d... | "approve" | {"approve": 0.988016, "request_more_info": 0.006169, "decline": 0.002908, "escalated_review": 0.002908} | 0.7 | true | google/gemma-4-26B-A4B-it | {"approve": 0.960053, "request_more_info": 0.020563, "decline": 0.009692, "escalated_review": 0.009692} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "loan underwriting", "pattern": "next best action for an automated agent given its goal and history"} |
sni_task893_gap_fill_the_blank_coreference_resolution-94 | sni_task893_gap_fill_the_blank_coreference_resolution | choice | {"input": "He was known in Sweden as Kloke-Hans (``Wise Hans''). Prof. Larsson was a humanist and an author. He was also a mentor for several Swedish authors and a prominent essayist. Prof. **Larsson** was the son of the farmer Lars Persson and Kersti Nilsdotter, and cousin of author Ola Hansson. After studentexamen in... | {"type": "choice", "instructions": "Read the passage and find the corresponding pronoun for the given name. The pronoun should match the given blank(_). The word between ** ** is the target name. The pronoun should be one of 'her', 'him', 'he', 'she' and 'his' with proper casing based on the position in the passage.", ... | "he_" | {"she_": 0.010577, "his": 0.01592, "he": 0.251983, "his_": 0.018959, "her_": 0.009321, "her": 0.011825, "him": 0.008621, "he_": 0.664166, "she": 0.008629} | 0.6 | true | google/gemma-4-26B-A4B-it | {"she_": 0.026442, "his": 0.039799, "he": 0.629959, "his_": 0.047396, "her_": 0.023303, "her": 0.029561, "him": 0.021553, "he_": 0.160414, "she": 0.021573} | false | Apache-2.0 | {"source": "Super-NaturalInstructions", "task": "task893_gap_fill_the_blank_coreference_resolution.json", "task_source": ["gap"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task1434_head_qa_classification-672 | sni_task1434_head_qa_classification | choice | {"input": "Question: In a patient with symptoms of heart failure who have a disproportionately high ascites in relation to peripheral edema, the most likely etiology would be: \n Options: <0> A severe aortic stenosis. <1> A dilated cardiomyopathy with significant left ventricular dysfunction. <2> A primary pulmonary h... | {"type": "choice", "instructions": "In this task, you are given a question and answer options for that question. Using this information, you have to classify each text into different topics: medicine, nursery, psychology, chemistry, pharmacology, biology.", "criteria": {"nursery": "nursery", "biology": "biology", "medi... | "medicine" | {"nursery": 0.044476, "biology": 0.035153, "medicine": 0.805724, "psychology": 0.035153, "chemistry": 0.044339, "pharmacology": 0.035153} | 0.6 | true | google/gemma-4-26B-A4B-it | {"nursery": 0.111191, "biology": 0.087884, "medicine": 0.514311, "psychology": 0.087884, "chemistry": 0.110848, "pharmacology": 0.087884} | false | MIT | {"source": "Super-NaturalInstructions", "task": "task1434_head_qa_classification.json", "task_source": ["head_qa"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task356_casino_classification_negotiation_self_need-463 | sni_task356_casino_classification_negotiation_self_need | choice | {"input": "Context: 'So if I take 2 foods and 1 firewood, you take 1 food and 2 firewood. What would be fair for water?' 'What if I take two waters and you take one water then?' 'I think that sounds fair, since you compromised on food. Sounds like a deal!'\nUtterance: 'I'm glad we could come to a compromise 🙂'"} | {"type": "choice", "instructions": "The input is taken from a negotiation between two participants who take the role of campsite neighbors and negotiate for Food, Water, and Firewood packages, based on their individual preferences and requirements. Given an utterance and recent dialogue context containing past 3 uttera... | "no" | {"no": 0.968845, "yes": 0.031155} | 0.6 | true | google/gemma-4-26B-A4B-it | {"no": 0.922114, "yes": 0.077886} | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task356_casino_classification_negotiation_self_need.json", "task_source": ["casino"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0235_health_claim_anomaly-23 | syn_0235_health_claim_anomaly | noul | {"claim_id": "CLM024", "patient_id": "PAT1024", "patient_name": "Xavier Pearl", "diagnosis_code": "D146", "total_claim_amount": 3999.99, "claim_line_items": [799.998, 799.998, 799.998, 799.998, 799.998], "claim_line_items_count": 5, "average_line_item_amount": 799.998, "provider_type": "in_network", "duplicate_claim_fl... | {"type": "noul", "instructions": "A claim is anomalous (true) if ANY of the following conditions is met:\n1) total_claim_amount > 5000\n2) claim_line_items_count > 5\n3) average_line_item_amount > 1500\n4) provider_type == \"out_of_network\" AND total_claim_amount > 2000\nOtherwise the claim is normal (false).", "crite... | false | {"false": 0.956045, "true": 0.043955} | 0.7 | true | google/gemma-4-26B-A4B-it | {"false": 0.853485, "true": 0.146515} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "health insurance claims", "pattern": "anomaly or violation detection in records, logs or measurements"} |
syn_0014_incident_response_anomaly_detection-22 | syn_0014_incident_response_anomaly_detection | noul | {"system_id": "SYS-023", "region": "ap-east-1", "event_count_last_hour": 300, "avg_response_time_ms": 1500, "distinct_ip_count": 20, "critical_error_rate_percent": 3.3, "unauthorized_access_attempts": 1, "metadata": {"log_version": 3, "collector": "agentT"}, "status": "idle"} | {"type": "noul", "instructions": "Determine whether the incident logs indicate a security violation. Answer **true** if any of the following conditions is met, otherwise **false**:\n\n1. `event_count_last_hour` >= 1000\n2. `avg_response_time_ms` > 2000\n3. `distinct_ip_count` >= 50\n4. `critical_error_rate_percent` > 5... | false | {"false": 0.953111, "true": 0.046889} | 0.7 | true | google/gemma-4-26B-A4B-it | {"false": 0.843702, "true": 0.156298} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "incident response", "pattern": "anomaly or violation detection in records, logs or measurements"} |
sni_task065_timetravel_consistent_sentence_classification-548 | sni_task065_timetravel_consistent_sentence_classification | choice | {"input": "Sentence 1: I was extremely tired from work. \n Sentence 3: I got into the tub \n Sentence 4: The water felt so good on my tired body \n Sentence 5: I stayed in the bath until the water went cold \n Option 1: I ran myself a nice, hot bubble bath. \n Option 2: I decided to lay in my cozy bed."} | {"type": "choice", "instructions": "In this task, you are given a short story consisting of exactly 5 sentences where the second sentence is missing. You are given two options and you need to select the one that best connects the first sentence with the rest of the story. Indicate your answer by 'Option 1' if the first... | "option_1" | {"option_1": 0.961182, "option_2": 0.038818} | 0.6 | true | google/gemma-4-26B-A4B-it | {"option_1": 0.902954, "option_2": 0.097046} | false | MIT | {"source": "Super-NaturalInstructions", "task": "task065_timetravel_consistent_sentence_classification.json", "task_source": ["timetravel"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task1197_atomic_classification_oreact-362 | sni_task1197_atomic_classification_oreact | choice | {"input": "Head: PersonX seems easy<sep>Tail: pleased"} | {"type": "choice", "instructions": "In this task, you are given a tuple, comprising Head and Tail, separated with <sep>. The Head and the Tail events are short phrases possibly involving participants. The names of specific people have been replaced by generic words (e.g., PersonX, PersonY, PersonZ). PersonX is always t... | "yes" | {"no": 0.180933, "yes": 0.819067} | 0.6 | true | google/gemma-4-26B-A4B-it | {"no": 0.452333, "yes": 0.547667} | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task1197_atomic_classification_oreact.json", "task_source": ["atomic"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task351_winomt_classification_gender_identifiability_anti-504 | sni_task351_winomt_classification_gender_identifiability_anti | choice | {"input": "Sentence: The chief called the counselor and told him that he could have the day off.\n Profession: counselor"} | {"type": "choice", "instructions": "In this task, you are given a sentence and a profession. The sentence mentions two professions: one's gender is identifiable using the gendered pronouns in the text and the other's gender is unidentifiable. You are expected to return whether the given profession's gender is identifia... | "identifiable" | {"identifiable": 0.79775, "unidentifiable": 0.20225} | 0.6 | true | google/gemma-4-26B-A4B-it | {"identifiable": 0.494375, "unidentifiable": 0.505625} | false | MIT | {"source": "Super-NaturalInstructions", "task": "task351_winomt_classification_gender_identifiability_anti.json", "task_source": ["winomt"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task1206_atomic_classification_isbefore-506 | sni_task1206_atomic_classification_isbefore | choice | {"input": "Head: PersonX acts like a PersonY<sep>Tail: to substitute him"} | {"type": "choice", "instructions": "In this task, you are given two phrases: Head and Tail, separated with <sep>. The Head and the Tail events are short phrases possibly involving participants. The names of specific people have been replaced by generic words (e.g., PersonX, PersonY, PersonZ). PersonX is always the subj... | "no" | {"no": 0.758052, "yes": 0.241948} | 0.6 | true | google/gemma-4-26B-A4B-it | {"no": 0.395131, "yes": 0.604869} | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task1206_atomic_classification_isbefore.json", "task_source": ["atomic"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0069_hotel_reservation_priority-12 | syn_0069_hotel_reservation_priority | score | {"reservation_id": "RZ-1013", "guest_name": "Mona Patel", "booking_value_usd": 1500, "number_of_guests": 3, "special_requests": [], "special_requests_count": 0, "days_before_arrival": 5, "room_type": "deluxe", "loyalty_status": "gold", "promo_code": null, "notes": ""} | {"type": "score", "instructions": "Assign a priority level (low, medium, high, critical) to each reservation according to the policy below.\n\nPolicy:\n1. If `days_before_arrival` <= 1 → critical\n2. Else if `booking_value_usd` >= 2000 AND `special_requests_count` >= 2 → critical\n3. Else if `booking_value_usd` >= 1500... | 2 | {"0": 0.002734, "1": 0.003966, "2": 0.990566, "3": 0.002734} | 0.7 | true | google/gemma-4-26B-A4B-it | {"0": 0.009113, "1": 0.013222, "2": 0.968552, "3": 0.009113} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "hotel reservations", "pattern": "severity or priority rating on a 3-5 level ordinal scale defined by explicit thresholds"} |
sni_task375_classify_type_of_sentence_in_debate-84 | sni_task375_classify_type_of_sentence_in_debate | choice | {"input": "topic: The ontological argument\nsentence: One can easily say that God is logically possible but may not exist."} | {"type": "choice", "instructions": "In this task, you will be given a debate topic, along with a sentence from the debate. You should classify the given sentence and choose the type of that sentence. Possible types are explained below.\nPolicy: This refers to a sentence proposing a specific course of action to be taken... | "value" | {"value": 0.733308, "fact": 0.225144, "policy": 0.024976, "testimony": 0.016573} | 0.6 | true | google/gemma-4-26B-A4B-it | {"value": 0.33327, "fact": 0.562859, "policy": 0.062439, "testimony": 0.041432} | false | CC-BY-SA-3.0 | {"source": "Super-NaturalInstructions", "task": "task375_classify_type_of_sentence_in_debate.json", "task_source": ["ddo"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task351_winomt_classification_gender_identifiability_anti-240 | sni_task351_winomt_classification_gender_identifiability_anti | choice | {"input": "Sentence: The developer traded services with the baker because she did not own a kitchen.\n Profession: baker"} | {"type": "choice", "instructions": "In this task, you are given a sentence and a profession. The sentence mentions two professions: one's gender is identifiable using the gendered pronouns in the text and the other's gender is unidentifiable. You are expected to return whether the given profession's gender is identifia... | "unidentifiable" | {"unidentifiable": 0.785774, "identifiable": 0.214226} | 0.6 | true | google/gemma-4-26B-A4B-it | {"unidentifiable": 0.464435, "identifiable": 0.535565} | false | MIT | {"source": "Super-NaturalInstructions", "task": "task351_winomt_classification_gender_identifiability_anti.json", "task_source": ["winomt"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0003_returns_refunds_next_action-14 | syn_0003_returns_refunds_next_action | choice | {"order_id": "ORD015", "customer_id": "CUSTO", "order_total": 250.0, "fraud_score": 60, "return_days": 40, "item_condition": "unused", "refund_amount": 130.0, "refund_requested": true, "previous_refunds": [], "loyalty_status": "bronze", "notes": ""} | {"type": "choice", "instructions": "Given the customer's return request history and order details, select the next best action for the automated support agent.\n\nPolicy (evaluate in order, first matching rule wins):\n1. If fraud_score >= 80 **and** order_total > 200 → action = \"escalate\".\n2. Else if return_days <= ... | "reject_refund" | {"approve_refund": 0.001773, "offer_store_credit": 0.002091, "reject_refund": 0.992202, "escalate": 0.001773, "request_more_info": 0.002162} | 0.7 | true | google/gemma-4-26B-A4B-it | {"approve_refund": 0.005909, "offer_store_credit": 0.006968, "reject_refund": 0.974008, "escalate": 0.005909, "request_more_info": 0.007205} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "returns and refunds", "pattern": "next best action for an automated agent given its goal and history"} |
syn_0229_maritime_routing-20 | syn_0229_maritime_routing | choice | {"incident_id": "INC021", "cargo_type": "standard", "ETA_hours": 60, "severity_score": 2, "vessel_size_tons": 90000, "region": "Mediterranean", "last_port": "Athens", "reported_by": "OpsCenterU", "extra_notes": "Routine check"} | {"type": "choice", "instructions": "Assign the incident to a handling team based on the following policy:\n1. If cargo_type == \"hazardous\" → hazmat_team.\n2. Else if cargo_type == \"perishable\" and ETA_hours <= 48 → perishables_team.\n3. Else if vessel_size_tons > 100000 and region == \"North_Atlantic\" → long_route... | "standard_team" | {"hazmat_team": 0.002395, "perishables_team": 0.003783, "long_route_team": 0.003212, "standard_team": 0.990611} | 0.7 | true | google/gemma-4-26B-A4B-it | {"hazmat_team": 0.007982, "perishables_team": 0.012608, "long_route_team": 0.010706, "standard_team": 0.968704} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "maritime shipping", "pattern": "routing or triage: which team, queue or handler should take the case"} |
syn_0105_loan_underwriting_next_action-7 | syn_0105_loan_underwriting_next_action | choice | {"applicant": {"id": "A008", "name": "Grace Liu", "credit_score": 560, "annual_income": 60000, "loan_amount": 20000, "employment_status": "full_time", "has_verified_employment": true, "fraud_flag": true}, "financials": {"total_monthly_debt": 1500, "monthly_income": 5000, "debt_to_income": 0.3, "loan_to_income": 0.3333}... | {"type": "choice", "instructions": "Given the applicant's state, choose the next best action for the automated underwriting agent.\nPolicy:\n- If applicant.credit_score >= 750 AND financials.debt_to_income <= 0.30 AND financials.loan_to_income <= 0.40 => approve\n- Else if applicant.credit_score >= 650 AND financials.d... | "decline" | {"approve": 0.002275, "request_more_info": 0.003341, "decline": 0.991282, "escalated_review": 0.003102} | 0.7 | true | google/gemma-4-26B-A4B-it | {"approve": 0.007584, "request_more_info": 0.011137, "decline": 0.970939, "escalated_review": 0.010339} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "loan underwriting", "pattern": "next best action for an automated agent given its goal and history"} |
sni_task567_circa_text_generation-204 | sni_task567_circa_text_generation | choice | {"input": "Did you guys have a productive week?"} | {"type": "choice", "instructions": "In this task, you are given a question, and your task is to predict the context of the question. We assume that someone named X asked the question from Y. The context of the question means the previous situation that led X to ask this question.", "criteria": {"x_wants_to_know_what_so... | "x_and_y_are_colleagues_who_are_leaving_w" | {"x_wants_to_know_what_sorts_of_books_y_li": 0.007534, "y_has_just_told_x_that_he_she_is_thinkin": 0.007988, "y_has_just_travelled_from_a_different_ci": 0.007057, "x_and_y_are_childhood_neighbours_who_une": 0.006362, "y_has_just_told_x_that_he_she_is_conside": 0.009255, "x_and_y_are_colleagues_who_are_leaving_w": 0.935... | 0.6 | true | google/gemma-4-26B-A4B-it | {"x_wants_to_know_what_sorts_of_books_y_li": 0.018834, "y_has_just_told_x_that_he_she_is_thinkin": 0.01997, "y_has_just_travelled_from_a_different_ci": 0.017643, "x_and_y_are_childhood_neighbours_who_une": 0.015906, "y_has_just_told_x_that_he_she_is_conside": 0.023137, "x_and_y_are_colleagues_who_are_leaving_w": 0.8381... | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task567_circa_text_generation.json", "task_source": ["circa"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0075_automotive_triage-13 | syn_0075_automotive_triage | choice | {"case_id": "BODY004", "issue_category": "body", "severity_score": 8, "warranty_months_remaining": 0, "vehicle_age_years": 5, "parts_in_stock": true, "customer_loyalty": "bronze", "notes": "Frame rust repair"} | {"type": "choice", "instructions": "Route each service case to the appropriate team according to the following policy:\n1. If issue_category == \"engine\" AND severity_score >= 7 → route to engine_repair.\n2. Else if issue_category == \"electrical\" AND (severity_score >= 5 OR warranty_months_remaining > 0) → route to ... | "body_shop" | {"engine_repair": 0.00207, "electrical": 0.003028, "body_shop": 0.990763, "tire_center": 0.00207, "service_advisor": 0.00207} | 0.7 | true | google/gemma-4-26B-A4B-it | {"engine_repair": 0.006899, "electrical": 0.010093, "body_shop": 0.96921, "tire_center": 0.006899, "service_advisor": 0.006899} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "automotive service", "pattern": "routing or triage: which team, queue or handler should take the case"} |
sni_task929_products_reviews_classification-306 | sni_task929_products_reviews_classification | choice | {"input": "These were not the most comfortable shoes. I'm usually a 7 1/2 and these 9 fit very loose. They are quite stylish as I have heard them compared to Kenneth Cole shoes but with all due respect I'd choose any kind of sandal over these. If your foot looks like it needs support you might be better off shopping fo... | {"type": "choice", "instructions": "Given an English language product review, determine if it is a Good Review or a Bad Review. A good review is one where the product's review has positive tone and Bad review is one where the tone of product's review is negative.", "criteria": {"good_review": "Good Review", "bad_review... | "bad_review" | {"good_review": 0.026614, "bad_review": 0.973386} | 0.6 | true | google/gemma-4-26B-A4B-it | {"good_review": 0.066536, "bad_review": 0.933464} | false | CC-BY-SA-4.0 | {"source": "Super-NaturalInstructions", "task": "task929_products_reviews_classification.json", "task_source": ["generated_reviews_enth"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0106_calendar_next_action-17 | syn_0106_calendar_next_action | choice | {"user_id": "U018", "pending_requests": 0, "free_time_minutes": 0, "meeting_duration_minutes": 0, "next_meeting_in_minutes": 5000, "reminder_sent": true, "reminder_allowed": false, "conflict_count": 0, "reschedule_attempts": 0, "meeting_status": "normal", "notes": "nothing to do", "location": "home"} | {"type": "choice", "instructions": "You are a calendar assistant. Choose the next best action according to the following policy:\n1. If pending_requests >= 1 AND free_time_minutes >= meeting_duration_minutes → schedule_meeting.\n2. Else if next_meeting_in_minutes <= 1440 AND reminder_sent == false AND reminder_allowed ... | "do_nothing" | {"schedule_meeting": 0.0012, "send_reminder": 0.0012, "propose_reschedule": 0.0012, "cancel_meeting": 0.0012, "do_nothing": 0.995201} | 0.7 | true | google/gemma-4-26B-A4B-it | {"schedule_meeting": 0.004, "send_reminder": 0.004, "propose_reschedule": 0.004, "cancel_meeting": 0.004, "do_nothing": 0.984002} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "calendar assistant agent", "pattern": "next best action for an automated agent given its goal and history"} |
syn_0235_health_claim_anomaly-6 | syn_0235_health_claim_anomaly | noul | {"claim_id": "CLM007", "patient_id": "PAT1007", "patient_name": "Grace Gray", "diagnosis_code": "D129", "total_claim_amount": 5000.0, "claim_line_items": [833.33, 833.33, 833.33, 833.33, 833.33, 833.33], "claim_line_items_count": 6, "average_line_item_amount": 833.33, "provider_type": "in_network", "duplicate_claim_fla... | {"type": "noul", "instructions": "A claim is anomalous (true) if ANY of the following conditions is met:\n1) total_claim_amount > 5000\n2) claim_line_items_count > 5\n3) average_line_item_amount > 1500\n4) provider_type == \"out_of_network\" AND total_claim_amount > 2000\nOtherwise the claim is normal (false).", "crite... | true | {"false": 0.015284, "true": 0.984716} | 0.7 | true | google/gemma-4-26B-A4B-it | {"false": 0.050947, "true": 0.949053} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "health insurance claims", "pattern": "anomaly or violation detection in records, logs or measurements"} |
sni_task065_timetravel_consistent_sentence_classification-98 | sni_task065_timetravel_consistent_sentence_classification | choice | {"input": "Sentence 1: Sam's boyfriend broke up with her. \n Sentence 3: She tried to choke herself to death \n Sentence 4: Luckily Sam's brother heard a noise upstairs \n Sentence 5: He took control of Sam before she killed herself \n Option 1: She immediately found a rebound guy who made her feel good. \n Option 2:... | {"type": "choice", "instructions": "In this task, you are given a short story consisting of exactly 5 sentences where the second sentence is missing. You are given two options and you need to select the one that best connects the first sentence with the rest of the story. Indicate your answer by 'Option 1' if the first... | "option_2" | {"option_2": 0.932317, "option_1": 0.067683} | 0.6 | true | google/gemma-4-26B-A4B-it | {"option_2": 0.830792, "option_1": 0.169208} | false | MIT | {"source": "Super-NaturalInstructions", "task": "task065_timetravel_consistent_sentence_classification.json", "task_source": ["timetravel"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0229_maritime_routing-10 | syn_0229_maritime_routing | choice | {"incident_id": "INC011", "cargo_type": "perishable", "ETA_hours": 30, "severity_score": 4, "vessel_size_tons": 71000, "region": "Mediterranean", "last_port": "Barcelona", "reported_by": "OpsCenterK", "extra_notes": "Cooling system check"} | {"type": "choice", "instructions": "Assign the incident to a handling team based on the following policy:\n1. If cargo_type == \"hazardous\" → hazmat_team.\n2. Else if cargo_type == \"perishable\" and ETA_hours <= 48 → perishables_team.\n3. Else if vessel_size_tons > 100000 and region == \"North_Atlantic\" → long_route... | "perishables_team" | {"hazmat_team": 0.00252, "perishables_team": 0.9936, "long_route_team": 0.00194, "standard_team": 0.00194} | 0.7 | true | google/gemma-4-26B-A4B-it | {"hazmat_team": 0.0084, "perishables_team": 0.978665, "long_route_team": 0.006467, "standard_team": 0.006467} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "maritime shipping", "pattern": "routing or triage: which team, queue or handler should take the case"} |
sni_task935_defeasible_nli_atomic_classification-413 | sni_task935_defeasible_nli_atomic_classification | choice | {"input": "Premise: PersonX takes the offensive\nHypothesis: As a result, PersonX wants to win\nUpdate: PersonX gets a large cash prize if they win"} | {"type": "choice", "instructions": "In this task, you are given a premise, a hypothesis, and an update. The premise sentence describes a real-world situation and is always assumed to be true. The hypothesis sentence describes an assumption or inference that you might make about that situation having read the premise. T... | "strengthener" | {"weakener": 0.0196, "strengthener": 0.9804} | 0.6 | true | google/gemma-4-26B-A4B-it | {"weakener": 0.049, "strengthener": 0.951} | false | MIT | {"source": "Super-NaturalInstructions", "task": "task935_defeasible_nli_atomic_classification.json", "task_source": ["defeasible_nli_atomic"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task431_senteval_object_count-216 | sni_task431_senteval_object_count | choice | {"input": "Her face was averted, the curtain of hair concealing her expression from him."} | {"type": "choice", "instructions": "In this task you are given a sentence. You must judge whether the object of the main clause is singular(like: apple) or plural(like: apartments). Label the instances as \"Singular\" or \"Plural\" based on your judgment.", "criteria": {"singular": "Singular", "plural": "Plural"}} | "singular" | {"singular": 0.930288, "plural": 0.069712} | 0.6 | true | google/gemma-4-26B-A4B-it | {"singular": 0.825719, "plural": 0.174281} | false | BSD | {"source": "Super-NaturalInstructions", "task": "task431_senteval_object_count.json", "task_source": ["senteval"], "url": "https://github.com/allenai/natural-instructions"} |
sni_task356_casino_classification_negotiation_self_need-765 | sni_task356_casino_classification_negotiation_self_need | choice | {"input": "Context: 'yes, me too. i am going to one. but i need some food.' 'Yeah. I like to go for long hikes so for that I need water and food. ' 'i have plenty of water, if you have firewood and some food'\nUtterance: 'I have plenty of firewood. I need two pieces of food though for my hiking. '"} | {"type": "choice", "instructions": "The input is taken from a negotiation between two participants who take the role of campsite neighbors and negotiate for Food, Water, and Firewood packages, based on their individual preferences and requirements. Given an utterance and recent dialogue context containing past 3 uttera... | "yes" | {"no": 0.113924, "yes": 0.886076} | 0.6 | true | google/gemma-4-26B-A4B-it | {"no": 0.284809, "yes": 0.715191} | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task356_casino_classification_negotiation_self_need.json", "task_source": ["casino"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0214_expense_reimbursement_eligibility-18 | syn_0214_expense_reimbursement_eligibility | noul | {"employee": {"id": "E019", "name": "Seth Vale"}, "expense_id": "X1019", "total_amount": 2000, "days_since_purchase": 30, "category": "travel", "daily_allowance": 150, "budget_used_pct": 90, "department": "marketing", "project_code": "P-118", "approver_id": "A06", "attachments": ["ticket.pdf"], "notes": "Trade show"} | {"type": "noul", "instructions": "Eligibility for expense reimbursement follows the policy: (1) total_amount must be <= 2000; (2) days_since_purchase must be <= 30; (3) category must be one of [travel, supplies, software]; (4) if category is travel then daily_allowance must be <= 150; (5) budget_used_pct must be <= 90.... | true | {"false": 0.008435, "true": 0.991565} | 0.7 | true | google/gemma-4-26B-A4B-it | {"false": 0.028116, "true": 0.971884} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "spreadsheet agent", "pattern": "eligibility or policy check: does the case satisfy a written policy with several conditions"} |
syn_0142_university_admissions_rating-3 | syn_0142_university_admissions_rating | score | {"applicant_id": "A004", "name": "David East", "GPA": 3.85, "SAT_score": 1470, "extracurricular_score": 9, "major_interest": "Mathematics", "essays_submitted": 2, "recommendation_count": 3, "legacy_status": false, "hometown": "Hillview"} | {"type": "score", "instructions": "Assign a quality rating index to each applicant according to the following rubric (0 = low, 1 = medium, 2 = high):\n- High (2): GPA >= 3.8 AND SAT_score >= 1450 AND extracurricular_score >= 8\n- Medium (1): GPA >= 3.5 AND SAT_score >= 1300 AND extracurricular_score >= 5\n- Low (0): an... | 2 | {"0": 0.003768, "1": 0.004672, "2": 0.991559} | 0.7 | true | google/gemma-4-26B-A4B-it | {"0": 0.012561, "1": 0.015575, "2": 0.971864} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "university admissions", "pattern": "quality rating against a rubric with defined levels"} |
syn_0105_loan_underwriting_next_action-14 | syn_0105_loan_underwriting_next_action | choice | {"applicant": {"id": "A015", "name": "Nina Santos", "credit_score": 730, "annual_income": 90000, "loan_amount": 36000, "employment_status": "full_time", "has_verified_employment": true, "fraud_flag": false}, "financials": {"total_monthly_debt": 2600, "monthly_income": 7500, "debt_to_income": 0.3467, "loan_to_income": 0... | {"type": "choice", "instructions": "Given the applicant's state, choose the next best action for the automated underwriting agent.\nPolicy:\n- If applicant.credit_score >= 750 AND financials.debt_to_income <= 0.30 AND financials.loan_to_income <= 0.40 => approve\n- Else if applicant.credit_score >= 650 AND financials.d... | "request_more_info" | {"approve": 0.002528, "request_more_info": 0.988457, "decline": 0.004356, "escalated_review": 0.004658} | 0.7 | true | google/gemma-4-26B-A4B-it | {"approve": 0.008428, "request_more_info": 0.961524, "decline": 0.014521, "escalated_review": 0.015526} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "loan underwriting", "pattern": "next best action for an automated agent given its goal and history"} |
syn_0235_health_claim_anomaly-21 | syn_0235_health_claim_anomaly | noul | {"claim_id": "CLM022", "patient_id": "PAT1022", "patient_name": "Victor Maroon", "diagnosis_code": "D144", "total_claim_amount": 5500.0, "claim_line_items": [1100.0, 1100.0, 1100.0, 1100.0, 1100.0], "claim_line_items_count": 5, "average_line_item_amount": 1100.0, "provider_type": "in_network", "duplicate_claim_flag": f... | {"type": "noul", "instructions": "A claim is anomalous (true) if ANY of the following conditions is met:\n1) total_claim_amount > 5000\n2) claim_line_items_count > 5\n3) average_line_item_amount > 1500\n4) provider_type == \"out_of_network\" AND total_claim_amount > 2000\nOtherwise the claim is normal (false).", "crite... | true | {"false": 0.005627, "true": 0.994373} | 0.7 | true | google/gemma-4-26B-A4B-it | {"false": 0.018756, "true": 0.981244} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "health insurance claims", "pattern": "anomaly or violation detection in records, logs or measurements"} |
sni_task1390_wscfixed_coreference-183 | sni_task1390_wscfixed_coreference | choice | {"input": "Billy cried because Toby wouldn't share (his) toy. <sep> Pronoun:his <sep> Noun: Billy"} | {"type": "choice", "instructions": "You are given a context, a pronoun, and a noun in this task. The given pronoun is shown in the context within parentheses. You should determine if the pronoun refers to the given noun or not. Please answer with \"True\" and \"False\".", "criteria": {"true": "True", "false": "False"}} | false | {"true": 0.167095, "false": 0.832905} | 0.6 | true | google/gemma-4-26B-A4B-it | {"true": 0.417739, "false": 0.582261} | false | CC-BY-4.0 | {"source": "Super-NaturalInstructions", "task": "task1390_wscfixed_coreference.json", "task_source": ["wsc_fiexed"], "url": "https://github.com/allenai/natural-instructions"} |
syn_0075_automotive_triage-4 | syn_0075_automotive_triage | choice | {"case_id": "ENG005", "issue_category": "engine", "severity_score": 10, "warranty_months_remaining": 24, "vehicle_age_years": 5, "parts_in_stock": false, "customer_loyalty": "gold", "notes": "Severe knocking"} | {"type": "choice", "instructions": "Route each service case to the appropriate team according to the following policy:\n1. If issue_category == \"engine\" AND severity_score >= 7 → route to engine_repair.\n2. Else if issue_category == \"electrical\" AND (severity_score >= 5 OR warranty_months_remaining > 0) → route to ... | "engine_repair" | {"engine_repair": 0.990778, "electrical": 0.002896, "body_shop": 0.002109, "tire_center": 0.002109, "service_advisor": 0.002109} | 0.7 | true | google/gemma-4-26B-A4B-it | {"engine_repair": 0.96926, "electrical": 0.009653, "body_shop": 0.007029, "tire_center": 0.007029, "service_advisor": 0.007029} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "automotive service", "pattern": "routing or triage: which team, queue or handler should take the case"} |
syn_0003_returns_refunds_next_action-3 | syn_0003_returns_refunds_next_action | choice | {"order_id": "ORD004", "customer_id": "CUSTD", "order_total": 150.0, "fraud_score": 5, "return_days": 30, "item_condition": "unused", "refund_amount": 75.0, "refund_requested": true, "previous_refunds": [], "loyalty_status": "gold", "notes": ""} | {"type": "choice", "instructions": "Given the customer's return request history and order details, select the next best action for the automated support agent.\n\nPolicy (evaluate in order, first matching rule wins):\n1. If fraud_score >= 80 **and** order_total > 200 → action = \"escalate\".\n2. Else if return_days <= ... | "approve_refund" | {"approve_refund": 0.896121, "offer_store_credit": 0.055419, "reject_refund": 0.021558, "escalate": 0.013101, "request_more_info": 0.013801} | 0.7 | true | google/gemma-4-26B-A4B-it | {"approve_refund": 0.653738, "offer_store_credit": 0.184729, "reject_refund": 0.071862, "escalate": 0.043669, "request_more_info": 0.046002} | false | Apache-2.0 | {"source": "synthetic", "generator": "openai/gpt-oss-120b", "domain": "returns and refunds", "pattern": "next best action for an automated agent given its goal and history"} |
sni_task834_mathdataset_classification-371 | sni_task834_mathdataset_classification | choice | {"input": "Suppose 0 = 5*x - 0*x - 5*a - 140, -96 = -4*x - 4*a. Let w = 6 + -6. Let u(q) = q**3 + q**2 + 7. Calculate the remainder when x is divided by u(w)."} | {"type": "choice", "instructions": "You are supposed to identify the category of a high-school level math question. There are five possible categories (1) algebra (2) arithmetic (3) measurement (4) numbers, and (5) probability. Use the following guidelines: (1) 'algebra' questions will typically contain letter variable... | "numbers" | {"arithmetic": 0.026481, "measurement": 0.024686, "probability": 0.016779, "numbers": 0.717023, "algebra": 0.215032} | 0.6 | true | google/gemma-4-26B-A4B-it | {"arithmetic": 0.066202, "measurement": 0.061714, "probability": 0.041947, "numbers": 0.292557, "algebra": 0.53758} | false | Apache-2.0 | {"source": "Super-NaturalInstructions", "task": "task834_mathdataset_classification.json", "task_source": ["mathmatics_dataset"], "url": "https://github.com/allenai/natural-instructions"} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.