id
stringlengths 1
12
| question
stringlengths 22
566
| golden_answers
dict | data_source
stringclasses 2
values | prompt
listlengths 1
1
| ability
stringclasses 6
values | reward_model
dict | extra_info
dict | metadata
dict |
|---|---|---|---|---|---|---|---|---|
32358
|
Porela vetusta is a flowering plant species.
|
{
"evidence": [
"Porela vetusta_sentence_0"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nPorela vetusta is a flowering plant species.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Porela vetusta_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2384,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.749",
"value": "?search="
},
{
"operation": "search",
"time": "18.515",
"value": "Porela vetusta"
},
{
"operation": "Now on",
"time": "19.192",
"value": "Porela vetusta"
},
{
"operation": "Highlighting",
"time": "22.167",
"value": "Porela vetusta_sentence_0"
},
{
"operation": "finish",
"time": "34.14",
"value": "finish"
}
]
}
|
78907
|
In Nigeria it takes five years of higher education to become a pharmacist, a health professional who specializes in medicines.
|
{
"evidence": [
"National Universities Commission_sentence_0"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn Nigeria it takes five years of higher education to become a pharmacist, a health professional who specializes in medicines.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"National Universities Commission_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2385,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.873",
"value": "?search="
},
{
"operation": "search",
"time": "18",
"value": "Nigeria"
},
{
"operation": "Now on",
"time": "19.603",
"value": "Nigeria"
},
{
"operation": "hyperlink",
"time": "40.385",
"value": "7 Demographics Nigeria_section_29"
},
{
"operation": "hyperlink",
"time": "75.766",
"value": "Education in Nigeria"
},
{
"operation": "Now on",
"time": "77.239",
"value": "Education in Nigeria"
},
{
"operation": "Highlighting",
"time": "79.789",
"value": "Education in Nigeria_sentence_0"
},
{
"operation": "Highlighting",
"time": "79.79",
"value": "Education in Nigeria_title"
},
{
"operation": "hyperlink",
"time": "92.058",
"value": "5 Tertiary education Education in Nigeria_section_4"
},
{
"operation": "search",
"time": "180.786",
"value": "Contains...In Nigeria it takes five years of higher education to become a pharmacist, a health professional who specializes in medicines."
},
{
"operation": "Now on",
"time": "183.153",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "194.631",
"value": "Health care systems by country"
},
{
"operation": "Now on",
"time": "196.313",
"value": "Health care systems by country"
},
{
"operation": "search",
"time": "256.377",
"value": "nigerian health professionals"
},
{
"operation": "Now on",
"time": "258.005",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "276.428",
"value": "Health Nigeria_section_33"
},
{
"operation": "Now on",
"time": "277.399",
"value": "Nigeria#Health Nigeria section 33"
},
{
"operation": "hyperlink",
"time": "330.722",
"value": "Index of Nigeria-related articles"
},
{
"operation": "Now on",
"time": "332.123",
"value": "Index of Nigeria-related articles"
},
{
"operation": "hyperlink",
"time": "386.754",
"value": "Lists of country-related topics"
},
{
"operation": "Now on",
"time": "388.681",
"value": "Lists of country-related topics"
},
{
"operation": "hyperlink",
"time": "403.805",
"value": "Index of Nigeria-related articles"
},
{
"operation": "Now on",
"time": "404.438",
"value": "Index of Nigeria-related articles"
},
{
"operation": "hyperlink",
"time": "478.922",
"value": "List of universities in Nigeria"
},
{
"operation": "Now on",
"time": "480.704",
"value": "List of universities in Nigeria"
},
{
"operation": "hyperlink",
"time": "528.956",
"value": "National Universities Commission"
},
{
"operation": "Now on",
"time": "529.687",
"value": "National Universities Commission"
},
{
"operation": "Highlighting",
"time": "556.542",
"value": "National Universities Commission_sentence_0"
},
{
"operation": "hyperlink",
"time": "557.621",
"value": "higher education"
},
{
"operation": "Now on",
"time": "558.411",
"value": "Higher education"
},
{
"operation": "back-button-clicked",
"time": "577.649",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "577.764",
"value": "National Universities Commission"
},
{
"operation": "finish",
"time": "592.27",
"value": "finish"
}
]
}
|
25915
|
Sattapadi Kutram only received positive appraisal from New Indian Express and nowrunning.com.
|
{
"evidence": [
"Sattapadi Kutram_sentence_13",
"Sattapadi Kutram_sentence_14"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nSattapadi Kutram only received positive appraisal from New Indian Express and nowrunning.com.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Sattapadi Kutram_sentence_13",
"Sattapadi Kutram_sentence_14"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2386,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.815",
"value": "?search="
},
{
"operation": "search",
"time": "140.001",
"value": "Sattapadi Kutram"
},
{
"operation": "Now on",
"time": "141.504",
"value": "Sattapadi Kutram"
},
{
"operation": "Highlighting",
"time": "157.177",
"value": "Sattapadi Kutram_sentence_0Sattapadi Kutram_sentence_1"
},
{
"operation": "Highlighting",
"time": "174.46",
"value": "Sattapadi Kutram_sentence_13"
},
{
"operation": "Highlighting",
"time": "184.2",
"value": "Sattapadi Kutram_sentence_14"
},
{
"operation": "finish",
"time": "243.079",
"value": "finish"
}
]
}
|
exfever_1718
|
Joseph Hart is a professional footballer who moved up to the Premier League with an English football club based in Manchester, the top flight of English football.
|
{
"evidence": [
"Joe Hart",
"Manchester City F.C."
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJoseph Hart is a professional footballer who moved up to the Premier League with an English football club based in Manchester, the top flight of English football.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Joe Hart",
"Manchester City F.C."
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2387,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
23803
|
Writing contributor for Fantasy Football Index, Luke Esser, was defeated for re-election in 2006 by Democratic challenger Rodney Tom.
|
{
"evidence": [
"Luke Esser_sentence_4",
"Luke Esser_sentence_7"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nWriting contributor for Fantasy Football Index, Luke Esser, was defeated for re-election in 2006 by Democratic challenger Rodney Tom.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Luke Esser_sentence_4",
"Luke Esser_sentence_7"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2388,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.945",
"value": "?search="
},
{
"operation": "search",
"time": "10.76",
"value": "Luke Esser"
},
{
"operation": "Now on",
"time": "11.277",
"value": "Luke Esser"
},
{
"operation": "Page search",
"time": "30.382",
"value": "fantasy football"
},
{
"operation": "Highlighting",
"time": "32.223",
"value": "Luke Esser_sentence_7"
},
{
"operation": "Page search",
"time": "44.03",
"value": "re-election"
},
{
"operation": "Highlighting",
"time": "47.203",
"value": "Luke Esser_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "55.13",
"value": "Luke Esser_sentence_7"
},
{
"operation": "Highlighting",
"time": "73.069",
"value": "Luke Esser_sentence_7"
},
{
"operation": "finish",
"time": "86.194",
"value": "finish"
}
]
}
|
exfever_8904
|
The Father of the Bride was the fifth starring role for Martin Short.
|
{
"evidence": [
"Father of the Bride (1991 film)",
"Steve Martin"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Father of the Bride was the fifth starring role for Martin Short.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Father of the Bride (1991 film)",
"Steve Martin"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2389,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
18423
|
The Tamarixia radiata is an hymenoptrean wasp, used to control D. Citri and around the world, and a vector for citrus greening disease; it develops from nymph to host in about eleven days.
|
{
"evidence": [
"Tamarixia radiata_sentence_0",
"Tamarixia radiata_sentence_35",
"Tamarixia radiata_sentence_11"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Tamarixia radiata is an hymenoptrean wasp, used to control D. Citri and around the world, and a vector for citrus greening disease; it develops from nymph to host in about eleven days.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Tamarixia radiata_sentence_0",
"Tamarixia radiata_sentence_35",
"Tamarixia radiata_sentence_11"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2390,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.525",
"value": "?search="
},
{
"operation": "search",
"time": "24.927",
"value": "Tamarixia radiata"
},
{
"operation": "Now on",
"time": "26.266",
"value": "Tamarixia radiata"
},
{
"operation": "Highlighting",
"time": "133.894",
"value": "Tamarixia radiata_sentence_0"
},
{
"operation": "hyperlink",
"time": "153.234",
"value": "4 Use as a Biological Control Tamarixia radiata_section_3"
},
{
"operation": "Highlighting",
"time": "179.599",
"value": "Tamarixia radiata_sentence_35"
},
{
"operation": "Page search",
"time": "245.326",
"value": "world"
},
{
"operation": "Highlighting",
"time": "286.538",
"value": "Tamarixia radiata_sentence_11"
},
{
"operation": "Page search",
"time": "345.272",
"value": "eleven"
},
{
"operation": "Page search",
"time": "358.346",
"value": "11"
},
{
"operation": "finish",
"time": "504.382",
"value": "finish"
}
]
}
|
exfever_6942
|
Ellen Kathleen Pompeo is an American actress and producer who made her screen debut in the legal drama Law & Order aired on an American English-language commercial broadcast television and radio network that is owned by Comcast through NBCUniversal.
|
{
"evidence": [
"Ellen Pompeo",
"NBC"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nEllen Kathleen Pompeo is an American actress and producer who made her screen debut in the legal drama Law & Order aired on an American English-language commercial broadcast television and radio network that is owned by Comcast through NBCUniversal.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Ellen Pompeo",
"NBC"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2391,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
19775
|
KVMI (2270 AM) is a radio station licensed to Tulare, Canada, United States, serving the Visalia-Tulare area and in 1963, the station's daytime power was increased to 5,000 watts.
|
{
"evidence": [
"KVMI_sentence_1"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nKVMI (2270 AM) is a radio station licensed to Tulare, Canada, United States, serving the Visalia-Tulare area and in 1963, the station's daytime power was increased to 5,000 watts.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"KVMI_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2392,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.496",
"value": "?search="
},
{
"operation": "search",
"time": "16.326",
"value": "KVMI (2270 AM)"
},
{
"operation": "Now on",
"time": "17.233",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "42.058",
"value": "kvii 2200 am"
},
{
"operation": "Now on",
"time": "42.891",
"value": "undefined"
},
{
"operation": "search",
"time": "48.981",
"value": "kvmi"
},
{
"operation": "Now on",
"time": "50.684",
"value": "KVMI"
},
{
"operation": "Highlighting",
"time": "65.23",
"value": "KVMI_sentence_1"
},
{
"operation": "finish",
"time": "88.413",
"value": "finish"
}
]
}
|
72843
|
The Hidalgo del Parral, named after the municipality Hidalgo del Parral, has a population of 101,147 making up the majority of the municipality's population of 103,519.
|
{
"evidence": [
"Parral, Chihuahua_sentence_2"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Hidalgo del Parral, named after the municipality Hidalgo del Parral, has a population of 101,147 making up the majority of the municipality's population of 103,519.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Parral, Chihuahua_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2393,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.111",
"value": "?search="
},
{
"operation": "search",
"time": "9.953",
"value": "Hidalgo del Parral"
},
{
"operation": "Now on",
"time": "11.11",
"value": "Parral, Chihuahua"
},
{
"operation": "Page search",
"time": "36.162",
"value": "named"
},
{
"operation": "Highlighting",
"time": "64.169",
"value": "Parral, Chihuahua_cell_0_9_1"
},
{
"operation": "Highlighting",
"time": "64.187",
"value": "Parral, Chihuahua_header_cell_0_9_0"
},
{
"operation": "Highlighting",
"time": "64.188",
"value": "Parral, Chihuahua_header_cell_0_8_0"
},
{
"operation": "Highlighting",
"time": "103.36",
"value": "Parral, Chihuahua_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "106.584",
"value": "Parral, Chihuahua_cell_0_9_1"
},
{
"operation": "Highlighting deleted",
"time": "106.598",
"value": "Parral, Chihuahua_header_cell_0_9_0"
},
{
"operation": "Highlighting deleted",
"time": "106.6",
"value": "Parral, Chihuahua_header_cell_0_8_0"
},
{
"operation": "finish",
"time": "128.12",
"value": "finish"
}
]
}
|
exfever_8516
|
Buzz Aldrin retired from the Air Force in 1972.
|
{
"evidence": [
"Buzz Aldrin",
"United States Military Academy"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBuzz Aldrin retired from the Air Force in 1972.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Buzz Aldrin",
"United States Military Academy"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2394,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
9295
|
Votkinsk Hydroelectric Station has a maximum height of 34 m and is composed of a combined 365 m power station and spillway.
|
{
"evidence": [
"Votkinsk Hydroelectric Station_sentence_5",
"Votkinsk Hydroelectric Station_sentence_6"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nVotkinsk Hydroelectric Station has a maximum height of 34 m and is composed of a combined 365 m power station and spillway.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Votkinsk Hydroelectric Station_sentence_5",
"Votkinsk Hydroelectric Station_sentence_6"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2395,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.551",
"value": "?search="
},
{
"operation": "search",
"time": "7.422",
"value": "Votkinsk Hydroelectric Station"
},
{
"operation": "Now on",
"time": "8.296",
"value": "Votkinsk Hydroelectric Station"
},
{
"operation": "Highlighting",
"time": "10.823",
"value": "Votkinsk Hydroelectric Station_header_cell_0_12_0"
},
{
"operation": "Highlighting",
"time": "11.303",
"value": "Votkinsk Hydroelectric Station_cell_0_12_1"
},
{
"operation": "Highlighting",
"time": "11.313",
"value": "Votkinsk Hydroelectric Station_header_cell_0_12_0"
},
{
"operation": "Highlighting",
"time": "11.313",
"value": "Votkinsk Hydroelectric Station_header_cell_0_9_0"
},
{
"operation": "Highlighting",
"time": "32.453",
"value": "Votkinsk Hydroelectric Station_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "33.605",
"value": "Votkinsk Hydroelectric Station_cell_0_12_1"
},
{
"operation": "Highlighting deleted",
"time": "33.615",
"value": "Votkinsk Hydroelectric Station_header_cell_0_12_0"
},
{
"operation": "Highlighting deleted",
"time": "33.615",
"value": "Votkinsk Hydroelectric Station_header_cell_0_9_0"
},
{
"operation": "Highlighting",
"time": "35.799",
"value": "Votkinsk Hydroelectric Station_sentence_6"
},
{
"operation": "finish",
"time": "57.697",
"value": "finish"
}
]
}
|
15712
|
The industrialized cities of the 19th century grew at a tremendous rate thus planning and architecture went through a paradigm shift at the turn of the 20th century,
|
{
"evidence": [
"Urban planning_sentence_41",
"Urban planning_sentence_42"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe industrialized cities of the 19th century grew at a tremendous rate thus planning and architecture went through a paradigm shift at the turn of the 20th century,\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Urban planning_sentence_41",
"Urban planning_sentence_42"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2396,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.768",
"value": "?search="
},
{
"operation": "search",
"time": "142.573",
"value": "Urban planning"
},
{
"operation": "Now on",
"time": "143.722",
"value": "Urban planning"
},
{
"operation": "Page search",
"time": "164.637",
"value": "The industrialized cities of the 19th century grew at a tremendous rate"
},
{
"operation": "Page search",
"time": "194.799",
"value": "planning and architecture went through a paradigm shift at the turn of the 20th century"
},
{
"operation": "Page search",
"time": "195.401",
"value": "planning and architecture went through a paradigm shift at the turn of the 20th century"
},
{
"operation": "Highlighting",
"time": "202.336",
"value": "Urban planning_sentence_41"
},
{
"operation": "Highlighting",
"time": "203.602",
"value": "Urban planning_sentence_42"
},
{
"operation": "finish",
"time": "208.779",
"value": "finish"
}
]
}
|
16153
|
William John Brock was born around 1820 and received his BA, afterwards he became the incumbent of Hayfield, Derbyshire
|
{
"evidence": [
"William John Brock_sentence_1"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nWilliam John Brock was born around 1820 and received his BA, afterwards he became the incumbent of Hayfield, Derbyshire\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"William John Brock_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2397,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.621",
"value": "?search="
},
{
"operation": "search",
"time": "9.538",
"value": "William John Brock"
},
{
"operation": "Now on",
"time": "10.215",
"value": "William John Brock"
},
{
"operation": "Highlighting",
"time": "13.649",
"value": "William John Brock_sentence_1"
},
{
"operation": "finish",
"time": "21.788",
"value": "finish"
}
]
}
|
exfever_811
|
Wilt Chamberlain was an American professional basketball player who used to play for the intercollegiate men's basketball program at the University of Kansas.
|
{
"evidence": [
"Wilt Chamberlain",
"Kansas Jayhawks men's basketball"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nWilt Chamberlain was an American professional basketball player who used to play for the intercollegiate men's basketball program at the University of Kansas.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Wilt Chamberlain",
"Kansas Jayhawks men's basketball"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2398,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
73393
|
The Turner Controversy involved a series of disputes between "Turnerites" and the rest of the party that caused Turner to resign.
|
{
"evidence": [
"Turner Controversy_sentence_6",
"Turner Controversy_sentence_7",
"Turner Controversy_sentence_0"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Turner Controversy involved a series of disputes between \"Turnerites\" and the rest of the party that caused Turner to resign.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Turner Controversy_sentence_6",
"Turner Controversy_sentence_7",
"Turner Controversy_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2399,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.226",
"value": "?search="
},
{
"operation": "search",
"time": "15.092",
"value": "Turner Controversy"
},
{
"operation": "Now on",
"time": "17.412",
"value": "Turner Controversy"
},
{
"operation": "Highlighting",
"time": "59.406",
"value": "Turner Controversy_sentence_6"
},
{
"operation": "Highlighting",
"time": "69.903",
"value": "Turner Controversy_sentence_7"
},
{
"operation": "Highlighting",
"time": "110.368",
"value": "Turner Controversy_sentence_0"
},
{
"operation": "Highlighting",
"time": "128.478",
"value": "Turner Controversy_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "137.615",
"value": "Turner Controversy_sentence_4"
},
{
"operation": "finish",
"time": "152.887",
"value": "finish"
}
]
}
|
exfever_7302
|
Ariana Grande was the first singer to judge on The Voice.
|
{
"evidence": [
"Ariana Grande",
"Americans",
"Citizenship of the United States"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAriana Grande was the first singer to judge on The Voice.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Ariana Grande",
"Americans",
"Citizenship of the United States"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2400,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1370
|
Macbeth is a 1971 historical drama film co-written by an English theatre critic and writer.
|
{
"evidence": [
"Macbeth (1971 film)",
"Kenneth Tynan"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMacbeth is a 1971 historical drama film co-written by an English theatre critic and writer.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Macbeth (1971 film)",
"Kenneth Tynan"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2401,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8841
|
Legion is a fictional character appearing in American comic books published by the brand name and primary imprint of Marvel Worldwide Inc., which launched other superhero titles created by some of the artists, including the artist and co-creator of the Marvel superheroes Spider-Man and Doctor Strange.
|
{
"evidence": [
"Legion (Marvel Comics)",
"Marvel Comics",
"Steve Ditko"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLegion is a fictional character appearing in American comic books published by the brand name and primary imprint of Marvel Worldwide Inc., which launched other superhero titles created by some of the artists, including the artist and co-creator of the Marvel superheroes Spider-Man and Doctor Strange.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Legion (Marvel Comics)",
"Marvel Comics",
"Steve Ditko"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2402,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
52665
|
Neshaminy Creek, named Nishambanach in 1671, has two springs, one was known at the time as the 'Giant Spring'.
|
{
"evidence": [
"Neshaminy Creek_sentence_13",
"Neshaminy Creek_sentence_14"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nNeshaminy Creek, named Nishambanach in 1671, has two springs, one was known at the time as the 'Giant Spring'.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Neshaminy Creek_sentence_13",
"Neshaminy Creek_sentence_14"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2403,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.934",
"value": "?search="
},
{
"operation": "Now on",
"time": "19.779",
"value": "?search="
},
{
"operation": "search",
"time": "96.741",
"value": "Neshaminy Creek"
},
{
"operation": "Now on",
"time": "98.311",
"value": "Neshaminy Creek"
},
{
"operation": "hyperlink",
"time": "109.151",
"value": "2 History Neshaminy Creek_section_1"
},
{
"operation": "Highlighting",
"time": "120.729",
"value": "Neshaminy Creek_sentence_12"
},
{
"operation": "Highlighting deleted",
"time": "211.926",
"value": "Neshaminy Creek_sentence_12"
},
{
"operation": "Highlighting",
"time": "221.111",
"value": "Neshaminy Creek_sentence_13"
},
{
"operation": "Highlighting",
"time": "317.734",
"value": "Neshaminy Creek_sentence_14"
},
{
"operation": "finish",
"time": "583.342",
"value": "finish"
}
]
}
|
exfever_536
|
Leighton Marissa Meester is an American actress who has also appeared in films, including the 2012 American adventure-romantic comedy film based on Headhunters.
|
{
"evidence": [
"Leighton Meester",
"Monte Carlo (2011 film)"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLeighton Marissa Meester is an American actress who has also appeared in films, including the 2012 American adventure-romantic comedy film based on Headhunters.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Leighton Meester",
"Monte Carlo (2011 film)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2404,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4156
|
Drew Barrymore initially appeared with a small role in the film Altered States.
|
{
"evidence": [
"Drew Barrymore",
"Barrymore family",
"The Royal Family (play)"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDrew Barrymore initially appeared with a small role in the film Altered States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Drew Barrymore",
"Barrymore family",
"The Royal Family (play)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2405,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
80008
|
Red bananas, which contains more beta carotene than yellow bananas, but little amount of vitamin C, have a tint of pink or light pink rind when ripe, and are best eaten if it has disintegrated.
|
{
"evidence": [
"Red banana_sentence_16"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nRed bananas, which contains more beta carotene than yellow bananas, but little amount of vitamin C, have a tint of pink or light pink rind when ripe, and are best eaten if it has disintegrated.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Red banana_sentence_16"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2406,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.797",
"value": "?search="
},
{
"operation": "search",
"time": "4.202",
"value": "Red bananas"
},
{
"operation": "Now on",
"time": "4.782",
"value": "Red banana"
},
{
"operation": "Highlighting",
"time": "13.064",
"value": "Red banana_sentence_16"
},
{
"operation": "finish",
"time": "24.632",
"value": "finish"
}
]
}
|
exfever_161
|
The South Island, also officially named Te Waipounamu, is bordered on the north by Cook Strait, the south and east by the Pacific Ocean, and the west by the marginal sea of the South Pacific Ocean, situated between Australia and an island country in the southwestern Pacific Ocean.
|
{
"evidence": [
"South Island",
"Tasman Sea",
"New Zealand"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe South Island, also officially named Te Waipounamu, is bordered on the north by Cook Strait, the south and east by the Pacific Ocean, and the west by the marginal sea of the South Pacific Ocean, situated between Australia and an island country in the southwestern Pacific Ocean.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"South Island",
"Tasman Sea",
"New Zealand"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2407,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5917
|
Winona Ryder gained attention with her performance in Tim Burton's 1988 African fantasy comedy film.
|
{
"evidence": [
"Winona Ryder",
"Beetlejuice"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nWinona Ryder gained attention with her performance in Tim Burton's 1988 African fantasy comedy film.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Winona Ryder",
"Beetlejuice"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2408,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
72697
|
Corydalus affinis is a species of Dobson fly whose larvae are found in trees.
|
{
"evidence": [
"Dobsonfly_sentence_3",
"Dobsonfly_sentence_33"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nCorydalus affinis is a species of Dobson fly whose larvae are found in trees.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Dobsonfly_sentence_3",
"Dobsonfly_sentence_33"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2409,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.749",
"value": "?search="
},
{
"operation": "search",
"time": "10.569",
"value": "Corydalus affinis"
},
{
"operation": "Now on",
"time": "11.63",
"value": "Corydalus affinis"
},
{
"operation": "Highlighting",
"time": "122.663",
"value": "Corydalus affinis_sentence_0"
},
{
"operation": "Highlighting",
"time": "122.664",
"value": "Corydalus affinis_title"
},
{
"operation": "hyperlink",
"time": "133.53",
"value": "dobsonfly"
},
{
"operation": "Now on",
"time": "134.565",
"value": "Dobsonfly"
},
{
"operation": "Highlighting",
"time": "140.74",
"value": "Dobsonfly_sentence_3"
},
{
"operation": "Page search",
"time": "146.518",
"value": "larvae"
},
{
"operation": "Highlighting",
"time": "191.252",
"value": "Dobsonfly_sentence_33"
},
{
"operation": "finish",
"time": "210.444",
"value": "finish"
}
]
}
|
31549
|
The Highway 402, a 122.581-mile-long (197.275 km) east–west highway in northwestern North Dakota, has major intersections in four counties.
|
{
"evidence": [
"North Dakota Highway 23_sentence_0"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Highway 402, a 122.581-mile-long (197.275 km) east–west highway in northwestern North Dakota, has major intersections in four counties.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"North Dakota Highway 23_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2410,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.554",
"value": "?search="
},
{
"operation": "search",
"time": "9.6",
"value": "Highway 402"
},
{
"operation": "Now on",
"time": "11.43",
"value": "Highway 402"
},
{
"operation": "search",
"time": "132.876",
"value": "(197.275 km) east–west highway in northwestern North Dakota"
},
{
"operation": "Now on",
"time": "141.905",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "161.102",
"value": "Route description North Dakota Highway 23_section_0"
},
{
"operation": "Now on",
"time": "162.502",
"value": "North Dakota Highway 23#Route description North Dakota Highway 23 section 0"
},
{
"operation": "Highlighting",
"time": "197.208",
"value": "North Dakota Highway 23_sentence_0"
},
{
"operation": "Page search",
"time": "254.161",
"value": "counties"
},
{
"operation": "Highlighting",
"time": "260.688",
"value": "North Dakota Highway 23_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "260.697",
"value": "North Dakota Highway 23_section_2"
},
{
"operation": "Highlighting",
"time": "260.731",
"value": "North Dakota Highway 23_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "261.925",
"value": "North Dakota Highway 23_cell_0_7_0"
},
{
"operation": "Highlighting",
"time": "263.368",
"value": "North Dakota Highway 23_cell_0_11_0"
},
{
"operation": "Highlighting",
"time": "264.149",
"value": "North Dakota Highway 23_cell_0_13_0"
},
{
"operation": "Highlighting",
"time": "270.962",
"value": "North Dakota Highway 23_cell_0_14_0"
},
{
"operation": "Highlighting deleted",
"time": "272.043",
"value": "North Dakota Highway 23_cell_0_14_0"
},
{
"operation": "Highlighting deleted",
"time": "273.408",
"value": "North Dakota Highway 23_cell_0_13_0"
},
{
"operation": "Highlighting deleted",
"time": "274.018",
"value": "North Dakota Highway 23_cell_0_11_0"
},
{
"operation": "Highlighting deleted",
"time": "276.286",
"value": "North Dakota Highway 23_cell_0_7_0"
},
{
"operation": "Highlighting deleted",
"time": "278.137",
"value": "North Dakota Highway 23_cell_0_1_0"
},
{
"operation": "Highlighting deleted",
"time": "278.144",
"value": "North Dakota Highway 23_section_2"
},
{
"operation": "Highlighting deleted",
"time": "278.173",
"value": "North Dakota Highway 23_header_cell_0_0_0"
},
{
"operation": "finish",
"time": "285.24",
"value": "finish"
}
]
}
|
38632
|
No Lady is a 1935 film and was rereleased 1950.
|
{
"evidence": [
"No Lady_sentence_0"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nNo Lady is a 1935 film and was rereleased 1950.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"No Lady_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2411,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.074",
"value": "?search="
},
{
"operation": "search",
"time": "9.352",
"value": "No Lady"
},
{
"operation": "Now on",
"time": "10.201",
"value": "No Lady"
},
{
"operation": "Highlighting",
"time": "14.807",
"value": "No Lady_sentence_0"
},
{
"operation": "finish",
"time": "34.396",
"value": "finish"
}
]
}
|
exfever_6080
|
Lily Collins is an actress born in the town of Durham in the southwest of London on the A3 trunk road between the capital of Portsmouth.
|
{
"evidence": [
"Lily Collins",
"Guildford"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLily Collins is an actress born in the town of Durham in the southwest of London on the A3 trunk road between the capital of Portsmouth.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Lily Collins",
"Guildford"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2412,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
70375
|
MCM3 encodes a protein that is one of the highly conserved mini-chromosome maintenance proteins, and it interacts directly with MCM5/CDC46.
|
{
"evidence": [
"MCM3_sentence_1",
"MCM3_sentence_4"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMCM3 encodes a protein that is one of the highly conserved mini-chromosome maintenance proteins, and it interacts directly with MCM5/CDC46.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"MCM3_sentence_1",
"MCM3_sentence_4"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2413,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.894",
"value": "?search="
},
{
"operation": "search",
"time": "18.785",
"value": "MCM3"
},
{
"operation": "Now on",
"time": "19.446",
"value": "MCM3"
},
{
"operation": "Highlighting",
"time": "50.081",
"value": "MCM3_sentence_1"
},
{
"operation": "Highlighting",
"time": "69.791",
"value": "MCM3_sentence_4"
},
{
"operation": "finish",
"time": "81.128",
"value": "finish"
}
]
}
|
exfever_398
|
The Valkyrie film changed release dates several times, from as early as June 27, 2008, to as late as February 13, 2009.
|
{
"evidence": [
"Valkyrie (film)",
"Nazi Germany"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Valkyrie film changed release dates several times, from as early as June 27, 2008, to as late as February 13, 2009.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Valkyrie (film)",
"Nazi Germany"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2414,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_412
|
Zuckerberg launched Facebook in February 2004 with his roommates, including an American entrepreneur who served as a spokesman for the online social directory and networking site, Facebook.
|
{
"evidence": [
"Mark Zuckerberg",
"Chris Hughes"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nZuckerberg launched Facebook in February 2004 with his roommates, including an American entrepreneur who served as a spokesman for the online social directory and networking site, Facebook.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Mark Zuckerberg",
"Chris Hughes"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2415,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
90578
|
Christian Sands, who was mentored by pianist Billy Taylor as a teenager, went on to study at the Manhattan School of Music and after graduating, she joined the band Inside Straight, which has toured internationally.
|
{
"evidence": [
"Christian Sands_sentence_4",
"Christian Sands_sentence_5",
"Christian Sands_sentence_7"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nChristian Sands, who was mentored by pianist Billy Taylor as a teenager, went on to study at the Manhattan School of Music and after graduating, she joined the band Inside Straight, which has toured internationally.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Christian Sands_sentence_4",
"Christian Sands_sentence_5",
"Christian Sands_sentence_7"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2416,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.925",
"value": "?search="
},
{
"operation": "search",
"time": "9.217",
"value": "Christian Sands"
},
{
"operation": "Now on",
"time": "10.201",
"value": "Christian Sands"
},
{
"operation": "Highlighting",
"time": "24.226",
"value": "Christian Sands_sentence_4"
},
{
"operation": "Highlighting",
"time": "29.318",
"value": "Christian Sands_sentence_5"
},
{
"operation": "Highlighting",
"time": "34.9",
"value": "Christian Sands_sentence_7"
},
{
"operation": "finish",
"time": "53.061",
"value": "finish"
}
]
}
|
89067
|
Fely Irvine is an Australian actress, singer and dancer who was a former member of Hi-5, and was a contestant on American Idol and made it to the top 60.
|
{
"evidence": [
"Fely Irvine_sentence_0",
"Fely Irvine_sentence_1",
"Fely Irvine_sentence_4"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nFely Irvine is an Australian actress, singer and dancer who was a former member of Hi-5, and was a contestant on American Idol and made it to the top 60.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Fely Irvine_sentence_0",
"Fely Irvine_sentence_1",
"Fely Irvine_sentence_4"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2417,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.09",
"value": "?search="
},
{
"operation": "search",
"time": "15.537",
"value": "Fely Irvine"
},
{
"operation": "Now on",
"time": "16.746",
"value": "Fely Irvine"
},
{
"operation": "Highlighting",
"time": "19.638",
"value": "Fely Irvine_sentence_0"
},
{
"operation": "Highlighting",
"time": "22.622",
"value": "Fely Irvine_sentence_1"
},
{
"operation": "Highlighting",
"time": "30.112",
"value": "Fely Irvine_sentence_4"
},
{
"operation": "finish",
"time": "90.668",
"value": "finish"
}
]
}
|
exfever_8892
|
Gene Roddenberry is a film producer.
|
{
"evidence": [
"Star Trek (film series)",
"NBC"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nGene Roddenberry is a film producer.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Star Trek (film series)",
"NBC"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2418,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
36913
|
Adèle Riché's portraits were often done in oil while her still life pieces were painted in watercolor.
|
{
"evidence": [
"Adèle Riché_sentence_6",
"Adèle Riché_sentence_7"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAdèle Riché's portraits were often done in oil while her still life pieces were painted in watercolor.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Adèle Riché_sentence_6",
"Adèle Riché_sentence_7"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2419,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.789",
"value": "?search="
},
{
"operation": "search",
"time": "13.94",
"value": "Adèle Riché"
},
{
"operation": "Now on",
"time": "14.994",
"value": "Adèle Riché"
},
{
"operation": "Highlighting",
"time": "34.516",
"value": "Adèle Riché_sentence_6"
},
{
"operation": "Highlighting",
"time": "39.369",
"value": "Adèle Riché_sentence_7"
},
{
"operation": "finish",
"time": "44.519",
"value": "finish"
}
]
}
|
exfever_1394
|
The Academy Honorary Award is given annually by the Board of Governors of the professional honorary organization with the stated goal of advancing the arts and sciences of motion pictures.
|
{
"evidence": [
"Academy Honorary Award",
"Academy of Motion Picture Arts and Sciences"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Academy Honorary Award is given annually by the Board of Governors of the professional honorary organization with the stated goal of advancing the arts and sciences of motion pictures.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Academy Honorary Award",
"Academy of Motion Picture Arts and Sciences"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2420,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1297
|
McCallum High School is a highly rated, public school located in AUSTIN.
|
{
"evidence": [
"Kenny Rogers",
"Bobby Doyle (jazz vocalist)"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMcCallum High School is a highly rated, public school located in AUSTIN.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Kenny Rogers",
"Bobby Doyle (jazz vocalist)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2421,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
19797
|
Yinon Magal was not member of the Sayeret Matkal before joining Kol Yisrael ( Israel's public domestic and international radio service) before later moving to Army Radio.
|
{
"evidence": [
"Yinon Magal_sentence_3"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nYinon Magal was not member of the Sayeret Matkal before joining Kol Yisrael ( Israel's public domestic and international radio service) before later moving to Army Radio.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Yinon Magal_sentence_3"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2422,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.874",
"value": "?search="
},
{
"operation": "search",
"time": "46.16",
"value": "Yinon Magal"
},
{
"operation": "Now on",
"time": "47.131",
"value": "Yinon Magal"
},
{
"operation": "Highlighting",
"time": "67.689",
"value": "Yinon Magal_sentence_3"
},
{
"operation": "Now on",
"time": "91.502",
"value": "Yinon Magal"
},
{
"operation": "finish",
"time": "105.793",
"value": "finish"
}
]
}
|
exfever_4003
|
Cary Grant was a British-American actor known for his learned manner of pronunciation, peculiar to a particular individual in English.
|
{
"evidence": [
"Cary Grant",
"Mid-Atlantic accent",
"Accent (sociolinguistics)"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nCary Grant was a British-American actor known for his learned manner of pronunciation, peculiar to a particular individual in English.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Cary Grant",
"Mid-Atlantic accent",
"Accent (sociolinguistics)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2423,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
65475
|
Michael Kiessou only honors and pays tributes to churches.
|
{
"evidence": [
"Partage EP_sentence_11",
"Partage EP_sentence_12"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMichael Kiessou only honors and pays tributes to churches.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Partage EP_sentence_11",
"Partage EP_sentence_12"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2424,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.801",
"value": "?search="
},
{
"operation": "search",
"time": "939.138",
"value": "Michael Kiessou"
},
{
"operation": "Now on",
"time": "940.827",
"value": "Michael Kiessou"
},
{
"operation": "Highlighting",
"time": "994.939",
"value": "Michael Kiessou_sentence_6"
},
{
"operation": "Highlighting",
"time": "994.94",
"value": "Michael Kiessou_title"
},
{
"operation": "Highlighting",
"time": "994.941",
"value": "Michael Kiessou_section_0"
},
{
"operation": "Highlighting deleted",
"time": "999.71",
"value": "Michael Kiessou_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "999.713",
"value": "Michael Kiessou_title"
},
{
"operation": "Highlighting deleted",
"time": "999.715",
"value": "Michael Kiessou_section_0"
},
{
"operation": "Highlighting",
"time": "1004.273",
"value": "Michael Kiessou_sentence_0"
},
{
"operation": "Highlighting",
"time": "1004.274",
"value": "Michael Kiessou_title"
},
{
"operation": "search",
"time": "1020.786",
"value": "Michael Kiessou"
},
{
"operation": "Now on",
"time": "1021.37",
"value": "Michael Kiessou"
},
{
"operation": "search",
"time": "1029.34",
"value": "Contains...Michael Kiessou"
},
{
"operation": "Now on",
"time": "1034.823",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "1051.454",
"value": "Partage EP"
},
{
"operation": "Now on",
"time": "1053.017",
"value": "Partage EP"
},
{
"operation": "Highlighting",
"time": "1068.12",
"value": "Partage EP_sentence_11"
},
{
"operation": "Highlighting",
"time": "1077.388",
"value": "Partage EP_sentence_12"
},
{
"operation": "hyperlink",
"time": "1087.267",
"value": "Littoral Region (Cameroon)"
},
{
"operation": "Now on",
"time": "1089.141",
"value": "Littoral Region (Cameroon)"
},
{
"operation": "back-button-clicked",
"time": "1095.444",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "1095.518",
"value": "Partage EP"
},
{
"operation": "Highlighting deleted",
"time": "1100.583",
"value": "Partage EP_sentence_12"
},
{
"operation": "Highlighting",
"time": "1108.844",
"value": "Partage EP_sentence_11"
},
{
"operation": "Highlighting",
"time": "1112.192",
"value": "Partage EP_sentence_12"
},
{
"operation": "finish",
"time": "1122.998",
"value": "finish"
}
]
}
|
58969
|
The nuclear pore complex (NPC) is made up of approximately thirty different proteins known as nucleoporins and produced via a biogenesis pathway that includes the synthesis and proteolytic cleavage of an 186 kD precursor protein.
|
{
"evidence": [
"NUP98_sentence_2",
"NUP98_sentence_1"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe nuclear pore complex (NPC) is made up of approximately thirty different proteins known as nucleoporins and produced via a biogenesis pathway that includes the synthesis and proteolytic cleavage of an 186 kD precursor protein.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"NUP98_sentence_2",
"NUP98_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2425,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.731",
"value": "?search="
},
{
"operation": "search",
"time": "10.254",
"value": "Nuclear pore complex"
},
{
"operation": "Now on",
"time": "11.691",
"value": "Nuclear pore"
},
{
"operation": "Highlighting",
"time": "25.219",
"value": "Nuclear pore_sentence_1"
},
{
"operation": "Highlighting",
"time": "25.22",
"value": "Nuclear pore_title"
},
{
"operation": "Highlighting",
"time": "46.136",
"value": "Nuclear pore_sentence_3"
},
{
"operation": "Highlighting",
"time": "60.306",
"value": "Nuclear pore_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "65.4",
"value": "Nuclear pore_sentence_6"
},
{
"operation": "Page search",
"time": "121.039",
"value": "biogenesis pathway"
},
{
"operation": "Page search",
"time": "139.861",
"value": "186 kD precursor protein"
},
{
"operation": "search",
"time": "153.071",
"value": "\"186 kD precursor protein\""
},
{
"operation": "Now on",
"time": "154.495",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "157.906",
"value": "NUP98"
},
{
"operation": "Now on",
"time": "158.52",
"value": "NUP98"
},
{
"operation": "Page search",
"time": "162.57",
"value": "186 kD precursor protein"
},
{
"operation": "Highlighting",
"time": "164.979",
"value": "NUP98_sentence_2"
},
{
"operation": "Highlighting",
"time": "175.129",
"value": "NUP98_sentence_1"
},
{
"operation": "finish",
"time": "213.92",
"value": "finish"
}
]
}
|
exfever_5365
|
Denial is a 2016 biographical film based on the 2005 book History on Trial: My Day in Court with a Holocaust Denier by an American History and Holocaust Studies Professor at Emory University in a state in the Southeastern region of the United States.
|
{
"evidence": [
"Denial (2016 film)",
"Deborah Lipstadt",
"Georgia (U.S. state)"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDenial is a 2016 biographical film based on the 2005 book History on Trial: My Day in Court with a Holocaust Denier by an American History and Holocaust Studies Professor at Emory University in a state in the Southeastern region of the United States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Denial (2016 film)",
"Deborah Lipstadt",
"Georgia (U.S. state)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2426,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_552
|
Frenemies is the first television film that was broadcasted in Asia.
|
{
"evidence": [
"Frenemies (film)",
"Bella Thorne",
"Dirty Sexy Money"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nFrenemies is the first television film that was broadcasted in Asia.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Frenemies (film)",
"Bella Thorne",
"Dirty Sexy Money"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2427,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1269
|
Sierra Leone is a country that has a tropical climate with a diverse environment ranging?from a mixed low-density forest, forming open habitats with plenty of sunlight and limited shade, and a grassland ecosystem characterized by the trees being sufficiently widely spaced.
|
{
"evidence": [
"Sierra Leone",
"Savanna",
"Woodland"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nSierra Leone is a country that has a tropical climate with a diverse environment ranging?from a mixed low-density forest, forming open habitats with plenty of sunlight and limited shade, and a grassland ecosystem characterized by the trees being sufficiently widely spaced.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Sierra Leone",
"Savanna",
"Woodland"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2428,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
16085
|
Hamed Abderrahaman Ahmad is a Spanish terrorist who was arrested in Pakistan for fighting for al-Qaeda in Afghanistan.
|
{
"evidence": [
"Hamed Abderrahaman Ahmad_sentence_0",
"Hamed Abderrahaman Ahmad_sentence_3",
"Hamed Abderrahaman Ahmad_sentence_4",
"Hamed Abderrahaman Ahmad_sentence_31",
"Hamed Abderrahaman Ahmad_sentence_32",
"Hamed Abderrahaman Ahmad_sentence_33",
"Hamed Abderrahaman Ahmad_sentence_34",
"Hamed Abderrahaman Ahmad_sentence_5",
"Hamed Abderrahaman Ahmad_sentence_6"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nHamed Abderrahaman Ahmad is a Spanish terrorist who was arrested in Pakistan for fighting for al-Qaeda in Afghanistan.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Hamed Abderrahaman Ahmad_sentence_0",
"Hamed Abderrahaman Ahmad_sentence_3",
"Hamed Abderrahaman Ahmad_sentence_4",
"Hamed Abderrahaman Ahmad_sentence_31",
"Hamed Abderrahaman Ahmad_sentence_32",
"Hamed Abderrahaman Ahmad_sentence_33",
"Hamed Abderrahaman Ahmad_sentence_34",
"Hamed Abderrahaman Ahmad_sentence_5",
"Hamed Abderrahaman Ahmad_sentence_6"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2429,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.166",
"value": "?search="
},
{
"operation": "search",
"time": "11.334",
"value": "Hamed Abderrahaman Ahmad"
},
{
"operation": "Now on",
"time": "12.215",
"value": "Hamed Abderrahaman Ahmad"
},
{
"operation": "Highlighting",
"time": "36.966",
"value": "Hamed Abderrahaman Ahmad_sentence_0"
},
{
"operation": "Highlighting",
"time": "39.002",
"value": "Hamed Abderrahaman Ahmad_sentence_3"
},
{
"operation": "Highlighting",
"time": "41.314",
"value": "Hamed Abderrahaman Ahmad_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "59.882",
"value": "Hamed Abderrahaman Ahmad_sentence_5"
},
{
"operation": "Highlighting",
"time": "60.714",
"value": "Hamed Abderrahaman Ahmad_sentence_4"
},
{
"operation": "Highlighting",
"time": "62.011",
"value": "Hamed Abderrahaman Ahmad_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "132.998",
"value": "Hamed Abderrahaman Ahmad_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "135.862",
"value": "Hamed Abderrahaman Ahmad_sentence_4"
},
{
"operation": "Highlighting",
"time": "138.813",
"value": "Hamed Abderrahaman Ahmad_sentence_4"
},
{
"operation": "Highlighting",
"time": "202.495",
"value": "Hamed Abderrahaman Ahmad_sentence_31"
},
{
"operation": "Highlighting",
"time": "203.815",
"value": "Hamed Abderrahaman Ahmad_sentence_32"
},
{
"operation": "Highlighting",
"time": "215.248",
"value": "Hamed Abderrahaman Ahmad_sentence_33"
},
{
"operation": "Highlighting",
"time": "216.144",
"value": "Hamed Abderrahaman Ahmad_sentence_34"
},
{
"operation": "Highlighting",
"time": "245.122",
"value": "Hamed Abderrahaman Ahmad_sentence_5"
},
{
"operation": "Highlighting",
"time": "247.009",
"value": "Hamed Abderrahaman Ahmad_sentence_6"
},
{
"operation": "finish",
"time": "309.043",
"value": "finish"
}
]
}
|
28777
|
Bethania Lutheran Church s a heritage-listed church built in 1872 in Queensland, Australia with gable roofs and a 1972 concrete block entry vestibule.
|
{
"evidence": [
"Bethania Lutheran Church_sentence_0",
"Bethania Lutheran Church_sentence_1",
"Bethania Lutheran Church_sentence_11",
"Bethania Lutheran Church_sentence_20"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBethania Lutheran Church s a heritage-listed church built in 1872 in Queensland, Australia with gable roofs and a 1972 concrete block entry vestibule.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Bethania Lutheran Church_sentence_0",
"Bethania Lutheran Church_sentence_1",
"Bethania Lutheran Church_sentence_11",
"Bethania Lutheran Church_sentence_20"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2430,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.859",
"value": "?search="
},
{
"operation": "search",
"time": "8.974",
"value": "Bethania Lutheran Church"
},
{
"operation": "Now on",
"time": "9.546",
"value": "Bethania Lutheran Church"
},
{
"operation": "Highlighting",
"time": "12.359",
"value": "Bethania Lutheran Church_sentence_0"
},
{
"operation": "Highlighting",
"time": "17.274",
"value": "Bethania Lutheran Church_sentence_1"
},
{
"operation": "Page search",
"time": "37.289",
"value": "gable"
},
{
"operation": "Highlighting",
"time": "45.313",
"value": "Bethania Lutheran Church_sentence_11"
},
{
"operation": "Highlighting",
"time": "52.648",
"value": "Bethania Lutheran Church_sentence_20"
},
{
"operation": "finish",
"time": "82.404",
"value": "finish"
}
]
}
|
85660
|
Babylon Mystery Religion did not sell well.
|
{
"evidence": [
"Babylon Mystery Religion_sentence_0",
"Babylon Mystery Religion_sentence_3",
"Babylon Mystery Religion_sentence_4"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBabylon Mystery Religion did not sell well.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Babylon Mystery Religion_sentence_0",
"Babylon Mystery Religion_sentence_3",
"Babylon Mystery Religion_sentence_4"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2431,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.817",
"value": "?search="
},
{
"operation": "search",
"time": "58.171",
"value": "Babylon Mystery Religion"
},
{
"operation": "Now on",
"time": "59.062",
"value": "Babylon Mystery Religion"
},
{
"operation": "search",
"time": "445.861",
"value": "Babylon Mystery Religion did not sell well"
},
{
"operation": "Now on",
"time": "448.885",
"value": "undefined"
},
{
"operation": "search",
"time": "463.61",
"value": "Contains...Babylon Mystery Religion"
},
{
"operation": "Now on",
"time": "464.762",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "481.813",
"value": "Cremation of Care"
},
{
"operation": "Now on",
"time": "482.562",
"value": "Cremation of Care"
},
{
"operation": "search",
"time": "544.943",
"value": "Contains...\"Babylon Mystery Religion\""
},
{
"operation": "Now on",
"time": "545.307",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "549.026",
"value": "Babylon Mystery Religion"
},
{
"operation": "Now on",
"time": "549.654",
"value": "Babylon Mystery Religion"
},
{
"operation": "Highlighting",
"time": "552.876",
"value": "Babylon Mystery Religion_sentence_0"
},
{
"operation": "Highlighting",
"time": "557.804",
"value": "Babylon Mystery Religion_sentence_3"
},
{
"operation": "Highlighting",
"time": "559.569",
"value": "Babylon Mystery Religion_sentence_4"
},
{
"operation": "finish",
"time": "577.167",
"value": "finish"
}
]
}
|
50320
|
San Francisco Rush 2049, a racing video game, was developed by Atari Games for arcades in single-player mode only, serving as the final game for the Atari Games label.
|
{
"evidence": [
"San Francisco Rush 2049_sentence_12"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nSan Francisco Rush 2049, a racing video game, was developed by Atari Games for arcades in single-player mode only, serving as the final game for the Atari Games label.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"San Francisco Rush 2049_sentence_12"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2432,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.141",
"value": "?search="
},
{
"operation": "search",
"time": "12.667",
"value": "San Francisco Rush 2049"
},
{
"operation": "Now on",
"time": "13.856",
"value": "San Francisco Rush 2049"
},
{
"operation": "Highlighting",
"time": "19.354",
"value": "San Francisco Rush 2049_sentence_0"
},
{
"operation": "Highlighting",
"time": "37.661",
"value": "San Francisco Rush 2049_sentence_12"
},
{
"operation": "Highlighting deleted",
"time": "43.029",
"value": "San Francisco Rush 2049_sentence_0"
},
{
"operation": "finish",
"time": "56.428",
"value": "finish"
}
]
}
|
9830
|
Emmy award-winning producer Steve Benen served as publisher of the Political Animal blog and published The Carpetbagger Report.
|
{
"evidence": [
"Steve Benen_sentence_0",
"Steve Benen_sentence_7",
"Steve Benen_sentence_6"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nEmmy award-winning producer Steve Benen served as publisher of the Political Animal blog and published The Carpetbagger Report.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Steve Benen_sentence_0",
"Steve Benen_sentence_7",
"Steve Benen_sentence_6"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2433,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.003",
"value": "undefined"
},
{
"operation": "search",
"time": "0.026",
"value": "Steve Benen"
},
{
"operation": "Now on",
"time": "0.028",
"value": "Steve Benen"
},
{
"operation": "Highlighting",
"time": "0.03",
"value": "Steve Benen_sentence_0"
},
{
"operation": "Page search",
"time": "0.045",
"value": "publisher"
},
{
"operation": "Highlighting",
"time": "0.05",
"value": "Steve Benen_sentence_7"
},
{
"operation": "Highlighting",
"time": "0.067",
"value": "Steve Benen_sentence_6"
},
{
"operation": "finish",
"time": "0.093",
"value": "finish"
},
{
"operation": "Now on",
"time": "1619143615.214",
"value": "Steve Benen"
},
{
"operation": "finish",
"time": "1619143617.523",
"value": "finish"
}
]
}
|
exfever_6020
|
Drax the Destroyer is a fictional character who often battled a stereotypical fictional character that possesses superpowers and abilities beyond those of ordinary people, whom audiences recognize from frequent recurrences in a particular literary tradition.
|
{
"evidence": [
"Drax the Destroyer",
"Superhero",
"Stock character"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDrax the Destroyer is a fictional character who often battled a stereotypical fictional character that possesses superpowers and abilities beyond those of ordinary people, whom audiences recognize from frequent recurrences in a particular literary tradition.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Drax the Destroyer",
"Superhero",
"Stock character"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2434,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
41296
|
Racial democracy was believed to obscure very real forms of racial oppression as argued by Skidmore and that the ideology of racial democracy aids in the effective action to combat racial discrimination according to Michael Hanchard.
|
{
"evidence": [
"Racial democracy_sentence_13"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nRacial democracy was believed to obscure very real forms of racial oppression as argued by Skidmore and that the ideology of racial democracy aids in the effective action to combat racial discrimination according to Michael Hanchard.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Racial democracy_sentence_13"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2435,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.897",
"value": "?search="
},
{
"operation": "search",
"time": "11.507",
"value": "Racial democracy"
},
{
"operation": "Now on",
"time": "13.477",
"value": "Racial democracy"
},
{
"operation": "Page search",
"time": "21.716",
"value": "obscure"
},
{
"operation": "Highlighting",
"time": "27.909",
"value": "Racial democracy_sentence_12"
},
{
"operation": "Highlighting",
"time": "42.78",
"value": "Racial democracy_sentence_13"
},
{
"operation": "Highlighting deleted",
"time": "73.075",
"value": "Racial democracy_sentence_12"
},
{
"operation": "finish",
"time": "107.969",
"value": "finish"
}
]
}
|
exfever_1765
|
Penny Dreadful is a horror drama television series co-produced by a Polish pay television channel owned by Sky Group Limited broadcast.
|
{
"evidence": [
"Penny Dreadful (TV series)",
"Sky Atlantic"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nPenny Dreadful is a horror drama television series co-produced by a Polish pay television channel owned by Sky Group Limited broadcast.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Penny Dreadful (TV series)",
"Sky Atlantic"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2436,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
68421
|
Yésica Bopp is an Argentine professional boxer which is regulated, sanctioned boxing.
|
{
"evidence": [
"Yésica Bopp_sentence_0",
"Professional boxing_sentence_2"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nYésica Bopp is an Argentine professional boxer which is regulated, sanctioned boxing.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Yésica Bopp_sentence_0",
"Professional boxing_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2437,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.178",
"value": "?search="
},
{
"operation": "search",
"time": "12.51",
"value": "Yésica Bopp"
},
{
"operation": "Now on",
"time": "26.491",
"value": "Yésica Bopp"
},
{
"operation": "Highlighting",
"time": "29.609",
"value": "Yésica Bopp_sentence_0"
},
{
"operation": "hyperlink",
"time": "32.52",
"value": "professional boxer"
},
{
"operation": "Now on",
"time": "41.451",
"value": "Professional boxing"
},
{
"operation": "Highlighting",
"time": "43.07",
"value": "Professional boxing_sentence_2"
},
{
"operation": "finish",
"time": "52.953",
"value": "finish"
}
]
}
|
exfever_2826
|
Poland is a country in Central Europe with its capital as the largest city in Poland and the 7th most populous capital city in a political and economic union that is located in Asia.
|
{
"evidence": [
"Poland",
"Warsaw",
"European Union"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nPoland is a country in Central Europe with its capital as the largest city in Poland and the 7th most populous capital city in a political and economic union that is located in Asia.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Poland",
"Warsaw",
"European Union"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2438,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2654
|
Leighton Marissa Meester is an American actress who has also appeared in films, including the 2011 American adventure-romantic comedy film based on Headhunters.
|
{
"evidence": [
"Leighton Meester",
"Monte Carlo (2011 film)"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLeighton Marissa Meester is an American actress who has also appeared in films, including the 2011 American adventure-romantic comedy film based on Headhunters.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Leighton Meester",
"Monte Carlo (2011 film)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2439,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
59253
|
James Alexander Hamilton Irwin, born in Feeny, Ireland, was ordained as a Presbyterian minister; but, he later on toured America with his lifelong friend Éamon de Valera wherein they decided to move in the country after 15 years.
|
{
"evidence": [
"James Alexander Hamilton Irwin_sentence_6",
"James Alexander Hamilton Irwin_sentence_4",
"James Alexander Hamilton Irwin_sentence_8"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJames Alexander Hamilton Irwin, born in Feeny, Ireland, was ordained as a Presbyterian minister; but, he later on toured America with his lifelong friend Éamon de Valera wherein they decided to move in the country after 15 years.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"James Alexander Hamilton Irwin_sentence_6",
"James Alexander Hamilton Irwin_sentence_4",
"James Alexander Hamilton Irwin_sentence_8"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2440,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.698",
"value": "?search="
},
{
"operation": "search",
"time": "14.704",
"value": "James Alexander Hamilton Irwin"
},
{
"operation": "Now on",
"time": "15.405",
"value": "James Alexander Hamilton Irwin"
},
{
"operation": "Highlighting",
"time": "50.37",
"value": "James Alexander Hamilton Irwin_sentence_6"
},
{
"operation": "Highlighting",
"time": "56.206",
"value": "James Alexander Hamilton Irwin_sentence_4"
},
{
"operation": "Highlighting",
"time": "67.041",
"value": "James Alexander Hamilton Irwin_sentence_8"
},
{
"operation": "finish",
"time": "74.85",
"value": "finish"
}
]
}
|
74257
|
Tun Perpatih Putih, the 6th bendahara of the Sultanate of Oman, was described as an ineffective bendahara who was then subsequently succeeded by Bendahara Sri Maharaja Tun Mutahir.
|
{
"evidence": [
"Tun Perpatih Putih_sentence_0"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nTun Perpatih Putih, the 6th bendahara of the Sultanate of Oman, was described as an ineffective bendahara who was then subsequently succeeded by Bendahara Sri Maharaja Tun Mutahir.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Tun Perpatih Putih_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2441,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.131",
"value": "undefined"
},
{
"operation": "search",
"time": "16.103",
"value": "Tun Perpatih Putih"
},
{
"operation": "Now on",
"time": "18.607",
"value": "Tun Perpatih Putih"
},
{
"operation": "Highlighting",
"time": "21.201",
"value": "Tun Perpatih Putih_sentence_0"
},
{
"operation": "finish",
"time": "32.328",
"value": "finish"
}
]
}
|
47940
|
Aonyōbō is a blue-skinned ogre-like spirit of poverty and misfortune who takes the appearance of an ancient court noblewoman, and legend has it that they are court ladies who serve noble families until they are married off to a worthy suitor.
|
{
"evidence": [
"Aonyōbō_sentence_1",
"Aonyōbō_sentence_3"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAonyōbō is a blue-skinned ogre-like spirit of poverty and misfortune who takes the appearance of an ancient court noblewoman, and legend has it that they are court ladies who serve noble families until they are married off to a worthy suitor.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Aonyōbō_sentence_1",
"Aonyōbō_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2442,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.846",
"value": "?search="
},
{
"operation": "search",
"time": "23.797",
"value": "Aonyōbō"
},
{
"operation": "Now on",
"time": "24.996",
"value": "Aonyōbō"
},
{
"operation": "Highlighting",
"time": "27.897",
"value": "Aonyōbō_sentence_1"
},
{
"operation": "Highlighting",
"time": "39.387",
"value": "Aonyōbō_sentence_3"
},
{
"operation": "finish",
"time": "63.827",
"value": "finish"
}
]
}
|
exfever_1788
|
John Lennon was an English singer, songwriter, and musician who is known as the co-lead vocalist and rhythm guitarist of an English rock band, formed in Liverpool in 1960, that comprised four members including an English musician, singer-songwriter, and music and film producer who was known for the song Taxman.
|
{
"evidence": [
"John Lennon",
"The Beatles",
"George Harrison"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJohn Lennon was an English singer, songwriter, and musician who is known as the co-lead vocalist and rhythm guitarist of an English rock band, formed in Liverpool in 1960, that comprised four members including an English musician, singer-songwriter, and music and film producer who was known for the song Taxman.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"John Lennon",
"The Beatles",
"George Harrison"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2443,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
48575
|
Carol Deene is an English former folk singer who recorded the songs "Sad Movies (Make Me Cry)" and "Norman" in 1961.
|
{
"evidence": [
"Carol Deene_sentence_0"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nCarol Deene is an English former folk singer who recorded the songs \"Sad Movies (Make Me Cry)\"\tand \"Norman\" in 1961.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Carol Deene_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2444,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.806",
"value": "?search="
},
{
"operation": "search",
"time": "54.101",
"value": "Carol Deene"
},
{
"operation": "Now on",
"time": "55.045",
"value": "Carol Deene"
},
{
"operation": "Highlighting",
"time": "69.423",
"value": "Carol Deene_sentence_0"
},
{
"operation": "Page search",
"time": "75.872",
"value": "folk"
},
{
"operation": "finish",
"time": "101.71",
"value": "finish"
}
]
}
|
12539
|
Theodoor Boeyermans was influenced by Rubens and van Dyck but failed to champion the more emotional style of van Dyck.
|
{
"evidence": [
"Theodoor Boeyermans_sentence_27"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nTheodoor Boeyermans was influenced by Rubens and van Dyck but failed to champion the more emotional style of van Dyck.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Theodoor Boeyermans_sentence_27"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2445,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.634",
"value": "?search="
},
{
"operation": "search",
"time": "13.464",
"value": "Theodoor Boeyermans"
},
{
"operation": "Now on",
"time": "14.36",
"value": "Theodoor Boeyermans"
},
{
"operation": "Highlighting",
"time": "33.107",
"value": "Theodoor Boeyermans_sentence_0"
},
{
"operation": "Page search",
"time": "37.079",
"value": "emotion"
},
{
"operation": "Highlighting",
"time": "44.288",
"value": "Theodoor Boeyermans_sentence_27"
},
{
"operation": "Highlighting deleted",
"time": "46.796",
"value": "Theodoor Boeyermans_sentence_0"
},
{
"operation": "finish",
"time": "64.285",
"value": "finish"
}
]
}
|
85796
|
Ronna and Beverly are the 50-something characters created and embodied by actor/comedians Jessica Chaffin and Jamie Denbo who never hosted a podcast on the Earwolf network from 2011 to 2017.
|
{
"evidence": [
"Ronna and Beverly_sentence_1"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nRonna and Beverly are the 50-something characters created and embodied by actor/comedians Jessica Chaffin and Jamie Denbo who never hosted a podcast on the Earwolf network from 2011 to 2017.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Ronna and Beverly_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2446,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.338",
"value": "?search="
},
{
"operation": "search",
"time": "15.636",
"value": "Ronna and Beverly"
},
{
"operation": "Now on",
"time": "17.538",
"value": "Ronna and Beverly"
},
{
"operation": "Highlighting",
"time": "19.719",
"value": "Ronna and Beverly_sentence_0"
},
{
"operation": "Highlighting",
"time": "24.907",
"value": "Ronna and Beverly_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "25.468",
"value": "Ronna and Beverly_sentence_0"
},
{
"operation": "finish",
"time": "33.003",
"value": "finish"
}
]
}
|
exfever_1135
|
"Wild Ones" is a song from the fourth studio album of the same name by a Mexican rapper who has sold over 80 million records worldwide, making him one of the best-selling music artists.
|
{
"evidence": [
"Wild Ones (song)",
"Flo Rida"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\n\"Wild Ones\" is a song from the fourth studio album of the same name by a Mexican rapper who has sold over 80 million records worldwide, making him one of the best-selling music artists.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Wild Ones (song)",
"Flo Rida"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2447,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
35812
|
Sorensen Parkway, which travels east from Blair High Road and North 90th Street, and has been seen as the northern boundary that defines North Omaha (a community area located north of Downtown Omaha, Nebraska).
|
{
"evidence": [
"Sorensen Parkway_sentence_3",
"Sorensen Parkway_sentence_5",
"North Omaha, Nebraska_sentence_0"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nSorensen Parkway, which travels east from Blair High Road and North 90th Street, and has been seen as the northern boundary that defines North Omaha (a community area located north of Downtown Omaha, Nebraska).\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Sorensen Parkway_sentence_3",
"Sorensen Parkway_sentence_5",
"North Omaha, Nebraska_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2448,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.69",
"value": "?search="
},
{
"operation": "search",
"time": "11.182",
"value": "Sorensen Parkway"
},
{
"operation": "Now on",
"time": "12.1",
"value": "Sorensen Parkway"
},
{
"operation": "Highlighting",
"time": "26.691",
"value": "Sorensen Parkway_sentence_3"
},
{
"operation": "Highlighting",
"time": "35.474",
"value": "Sorensen Parkway_sentence_5"
},
{
"operation": "hyperlink",
"time": "37.546",
"value": "North Omaha"
},
{
"operation": "Now on",
"time": "38.445",
"value": "North Omaha, Nebraska"
},
{
"operation": "Highlighting",
"time": "42.474",
"value": "North Omaha, Nebraska_sentence_0"
},
{
"operation": "finish",
"time": "52.61",
"value": "finish"
}
]
}
|
77124
|
In her career as a researcher, Sharon Anderson has expanded understanding of the pathophysiology of diabetic nephropathy, and has uncovered the important effect of ACE inhibitors in modifying the progression of diabetic nephropathy.
|
{
"evidence": [
"Sharon Anderson_sentence_14",
"Sharon Anderson_sentence_15"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn her career as a researcher, Sharon Anderson has expanded understanding of the pathophysiology of diabetic nephropathy, and has uncovered the important effect of ACE inhibitors in modifying the progression of diabetic nephropathy.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Sharon Anderson_sentence_14",
"Sharon Anderson_sentence_15"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2449,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.754",
"value": "?search="
},
{
"operation": "search",
"time": "17.948",
"value": "Sharon Anderson"
},
{
"operation": "Now on",
"time": "18.863",
"value": "Sharon Anderson"
},
{
"operation": "Highlighting",
"time": "32.22",
"value": "Sharon Anderson_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "37.369",
"value": "Sharon Anderson_sentence_1"
},
{
"operation": "Highlighting",
"time": "42.358",
"value": "Sharon Anderson_sentence_3"
},
{
"operation": "Highlighting deleted",
"time": "51.209",
"value": "Sharon Anderson_sentence_3"
},
{
"operation": "Highlighting",
"time": "61.442",
"value": "Sharon Anderson_sentence_14"
},
{
"operation": "Highlighting",
"time": "70.217",
"value": "Sharon Anderson_sentence_15"
},
{
"operation": "finish",
"time": "82.96",
"value": "finish"
}
]
}
|
exfever_5769
|
Leicester City Football Club is a professional football club based in Leicester in one of nine official regions of England at the first level of ITL for statistical purposes.
|
{
"evidence": [
"Leicester City F.C.",
"East Midlands"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLeicester City Football Club is a professional football club based in Leicester in one of nine official regions of England at the first level of ITL for statistical purposes.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Leicester City F.C.",
"East Midlands"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2450,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4038
|
Jeffrey Lionel Dahmer committed his first murder in June 1978.
|
{
"evidence": [
"Jeffrey Dahmer",
"Borderline personality disorder"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJeffrey Lionel Dahmer committed his first murder in June 1978.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jeffrey Dahmer",
"Borderline personality disorder"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2451,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2360
|
Beijing is the richest capital city in the world.
|
{
"evidence": [
"Beijing",
"China"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBeijing is the richest capital city in the world.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Beijing",
"China"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2452,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
39218
|
Beijing has more municipal (public university administered by the municipality) than any other type of universities and colleges.
|
{
"evidence": [
"Peking University_sentence_2"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBeijing has more municipal (public university administered by the municipality) than any other type of universities and colleges.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Peking University_sentence_2"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2453,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "2.866",
"value": "?search="
},
{
"operation": "search",
"time": "16.29",
"value": "Contains...beijing universities"
},
{
"operation": "Now on",
"time": "18.839",
"value": "undefined"
},
{
"operation": "search",
"time": "61.115",
"value": "beijing"
},
{
"operation": "Now on",
"time": "66.339",
"value": "Beijing"
},
{
"operation": "hyperlink",
"time": "71.514",
"value": "Beijing (disambiguation)"
},
{
"operation": "Now on",
"time": "73.031",
"value": "Beijing (disambiguation)"
},
{
"operation": "Page search",
"time": "78.556",
"value": "universi"
},
{
"operation": "Page search",
"time": "78.722",
"value": "universi"
},
{
"operation": "search",
"time": "140.07",
"value": "Beijing has more municipal than any other type of universities"
},
{
"operation": "Now on",
"time": "142.204",
"value": "undefined"
},
{
"operation": "search",
"time": "161.303",
"value": "Beijing University"
},
{
"operation": "Now on",
"time": "163.051",
"value": "Beijing University"
},
{
"operation": "hyperlink",
"time": "168.157",
"value": "Beida (disambiguation)"
},
{
"operation": "Now on",
"time": "169.243",
"value": "Beida (disambiguation)"
},
{
"operation": "hyperlink",
"time": "176.355",
"value": "Peking University"
},
{
"operation": "Now on",
"time": "177.366",
"value": "Peking University"
},
{
"operation": "Page search",
"time": "194.179",
"value": "municipal"
},
{
"operation": "Page search",
"time": "194.36",
"value": "municipal"
},
{
"operation": "Page search",
"time": "194.502",
"value": "municipal"
},
{
"operation": "search",
"time": "247.016",
"value": "Contains...beijing municipal"
},
{
"operation": "Now on",
"time": "248.367",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "259.606",
"value": "Beijing College Student Film Festival"
},
{
"operation": "Now on",
"time": "260.722",
"value": "Beijing College Student Film Festival"
},
{
"operation": "search",
"time": "281.765",
"value": "beijing municipal colleges"
},
{
"operation": "Now on",
"time": "282.906",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "293.43",
"value": "Peking University"
},
{
"operation": "Now on",
"time": "294.35",
"value": "Peking University"
},
{
"operation": "Highlighting",
"time": "304.92",
"value": "Peking University_sentence_2"
},
{
"operation": "finish",
"time": "310.63",
"value": "finish"
}
]
}
|
exfever_5292
|
Norovirus is the most common cause of gastroenteritis, usually spread by the particular route of transmission of a disease wherein pathogens in fecal particles pass from one person's mouth to another person.
|
{
"evidence": [
"Norovirus",
"Fecal–oral route"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nNorovirus is the most common cause of gastroenteritis, usually spread by the particular route of transmission of a disease wherein pathogens in fecal particles pass from one person's mouth to another person.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Norovirus",
"Fecal–oral route"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2454,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
42688
|
Canada was known to produce the most diamonds in 2016 with 11.1 million carats produced.
|
{
"evidence": [
"Diamond_sentence_350",
"Diamond_sentence_369"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nCanada was known to produce the most diamonds in 2016 with 11.1 million carats produced.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Diamond_sentence_350",
"Diamond_sentence_369"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2455,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.618",
"value": "?search="
},
{
"operation": "search",
"time": "13.796",
"value": "diamond production"
},
{
"operation": "Now on",
"time": "16.274",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "22.071",
"value": "Diamond"
},
{
"operation": "Now on",
"time": "22.995",
"value": "Diamond"
},
{
"operation": "search",
"time": "165.026",
"value": "Canada diamonds"
},
{
"operation": "Now on",
"time": "165.745",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "175.824",
"value": "Canadian diamonds"
},
{
"operation": "Now on",
"time": "176.437",
"value": "Canadian diamonds"
},
{
"operation": "search",
"time": "251.478",
"value": "diamond producers"
},
{
"operation": "Now on",
"time": "252.49",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "256.368",
"value": "diamond producer"
},
{
"operation": "Now on",
"time": "256.815",
"value": "undefined"
},
{
"operation": "search",
"time": "262.475",
"value": "diamond mines"
},
{
"operation": "Now on",
"time": "263.208",
"value": "List of diamond mines"
},
{
"operation": "hyperlink",
"time": "272.017",
"value": "4.1 Canada List of diamond mines_section_11"
},
{
"operation": "hyperlink",
"time": "296.224",
"value": "4.1 Canada List of diamond mines_section_11"
},
{
"operation": "hyperlink",
"time": "299.518",
"value": "Ekati Diamond Mine"
},
{
"operation": "Now on",
"time": "299.972",
"value": "Ekati Diamond Mine"
},
{
"operation": "back-button-clicked",
"time": "320.974",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "321.428",
"value": "List of diamond mines#Canada List of diamond mines section 11"
},
{
"operation": "hyperlink",
"time": "323.125",
"value": "Jericho Diamond Mine"
},
{
"operation": "Now on",
"time": "323.819",
"value": "Jericho Diamond Mine"
},
{
"operation": "back-button-clicked",
"time": "337.628",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "337.702",
"value": "List of diamond mines#Canada List of diamond mines section 11"
},
{
"operation": "hyperlink",
"time": "339.078",
"value": "Snap Lake Diamond Mine"
},
{
"operation": "Now on",
"time": "339.523",
"value": "Snap Lake Diamond Mine"
},
{
"operation": "back-button-clicked",
"time": "350.088",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "350.17",
"value": "List of diamond mines#Canada List of diamond mines section 11"
},
{
"operation": "hyperlink",
"time": "351.934",
"value": "Victor Diamond Mine"
},
{
"operation": "Now on",
"time": "352.588",
"value": "Victor Diamond Mine"
},
{
"operation": "back-button-clicked",
"time": "372.76",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "372.84",
"value": "List of diamond mines#Canada List of diamond mines section 11"
},
{
"operation": "hyperlink",
"time": "374.45",
"value": "Gahcho Kue Diamond Mine Project"
},
{
"operation": "Now on",
"time": "374.923",
"value": "Gahcho Kue Diamond Mine"
},
{
"operation": "back-button-clicked",
"time": "387.543",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "387.624",
"value": "List of diamond mines#Canada List of diamond mines section 11"
},
{
"operation": "hyperlink",
"time": "389.323",
"value": "Renard Diamond Mine"
},
{
"operation": "Now on",
"time": "389.778",
"value": "Renard diamond mine"
},
{
"operation": "search",
"time": "404.774",
"value": "greatest Diamond mines"
},
{
"operation": "Now on",
"time": "406.058",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "409.419",
"value": "Diamond"
},
{
"operation": "Now on",
"time": "409.829",
"value": "Diamond"
},
{
"operation": "Page search",
"time": "414.156",
"value": "Canada"
},
{
"operation": "Highlighting",
"time": "475.072",
"value": "Diamond_sentence_350"
},
{
"operation": "search",
"time": "540.804",
"value": "Biggest produce of diamonds"
},
{
"operation": "Now on",
"time": "541.656",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "548.58",
"value": "Gem-grade diamonds Diamond_section_24"
},
{
"operation": "Now on",
"time": "548.885",
"value": "Diamond#Gem-grade diamonds Diamond section 24"
},
{
"operation": "Highlighting",
"time": "589.99",
"value": "Diamond_sentence_369"
},
{
"operation": "finish",
"time": "621.991",
"value": "finish"
}
]
}
|
52955
|
Samuel Rhoads (1711 – April 7, 1784) was an architect from US who first worked as a carpenter and builder before becoming a merchant and later served as the 59th mayor of Philadelphia, Pennsylvania and master of the Carpenters' Company of Philadelphia.
|
{
"evidence": [
"Samuel Rhoads_sentence_6",
"Samuel Rhoads_sentence_13",
"Samuel Rhoads_sentence_16",
"Samuel Rhoads_sentence_0"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nSamuel Rhoads (1711 – April 7, 1784) was an architect from US who first worked as a carpenter and builder before becoming a merchant and later served as the 59th mayor of Philadelphia, Pennsylvania and master of the Carpenters' Company of Philadelphia.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Samuel Rhoads_sentence_6",
"Samuel Rhoads_sentence_13",
"Samuel Rhoads_sentence_16",
"Samuel Rhoads_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2456,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "2.375",
"value": "?search="
},
{
"operation": "search",
"time": "14.827",
"value": "Samuel Rhoads"
},
{
"operation": "Now on",
"time": "15.874",
"value": "Samuel Rhoads"
},
{
"operation": "Highlighting",
"time": "20.865",
"value": "Samuel Rhoads_cell_0_3_1"
},
{
"operation": "Highlighting",
"time": "20.869",
"value": "Samuel Rhoads_header_cell_0_3_0"
},
{
"operation": "Highlighting",
"time": "20.869",
"value": "Samuel Rhoads_header_cell_0_2_0"
},
{
"operation": "Highlighting",
"time": "20.869",
"value": "Samuel Rhoads_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "20.869",
"value": "Samuel Rhoads_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "21.556",
"value": "Samuel Rhoads_cell_0_4_1"
},
{
"operation": "Highlighting",
"time": "21.559",
"value": "Samuel Rhoads_header_cell_0_4_0"
},
{
"operation": "Highlighting",
"time": "45.554",
"value": "Samuel Rhoads_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "54.389",
"value": "Samuel Rhoads_sentence_0"
},
{
"operation": "Highlighting",
"time": "66.594",
"value": "Samuel Rhoads_sentence_6"
},
{
"operation": "Highlighting",
"time": "86.611",
"value": "Samuel Rhoads_sentence_13"
},
{
"operation": "Highlighting",
"time": "120.126",
"value": "Samuel Rhoads_sentence_16"
},
{
"operation": "Highlighting",
"time": "155.606",
"value": "Samuel Rhoads_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "157.946",
"value": "Samuel Rhoads_cell_0_3_1"
},
{
"operation": "Highlighting deleted",
"time": "157.95",
"value": "Samuel Rhoads_header_cell_0_3_0"
},
{
"operation": "Highlighting deleted",
"time": "158.784",
"value": "Samuel Rhoads_cell_0_4_1"
},
{
"operation": "Highlighting deleted",
"time": "158.788",
"value": "Samuel Rhoads_header_cell_0_4_0"
},
{
"operation": "Highlighting deleted",
"time": "158.789",
"value": "Samuel Rhoads_header_cell_0_2_0"
},
{
"operation": "Highlighting deleted",
"time": "158.79",
"value": "Samuel Rhoads_header_cell_0_1_0"
},
{
"operation": "Highlighting deleted",
"time": "158.791",
"value": "Samuel Rhoads_header_cell_0_0_0"
},
{
"operation": "finish",
"time": "218.917",
"value": "finish"
}
]
}
|
exfever_2296
|
iO was a central Chicago improv theater and training center with notable alumni, such as the American comedian, political commentator, and television host, best known for hosting The Colbert Report.
|
{
"evidence": [
"IO Theater",
"Stephen Colbert"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\niO was a central Chicago improv theater and training center with notable alumni, such as the American comedian, political commentator, and television host, best known for hosting The Colbert Report.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"IO Theater",
"Stephen Colbert"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2457,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
35139
|
Deacon Chiu rose to fame when he bought the Rediffusion Television (the first television station in Hong Kong) and later renamed it as Asia Television which he eventually sold in 1989,
|
{
"evidence": [
"Deacon Chiu_sentence_9",
"Rediffusion Television_sentence_3"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDeacon Chiu rose to fame when he bought the Rediffusion Television (the first television station in Hong Kong) and later renamed it as Asia Television which he eventually sold in 1989,\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Deacon Chiu_sentence_9",
"Rediffusion Television_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2458,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.681",
"value": "?search="
},
{
"operation": "search",
"time": "9.79",
"value": "Deacon Chiu"
},
{
"operation": "Now on",
"time": "10.689",
"value": "Deacon Chiu"
},
{
"operation": "Highlighting",
"time": "65.798",
"value": "Deacon Chiu_sentence_9"
},
{
"operation": "hyperlink",
"time": "70.234",
"value": "Rediffusion Television"
},
{
"operation": "Now on",
"time": "71.126",
"value": "Rediffusion Television"
},
{
"operation": "Highlighting",
"time": "73.993",
"value": "Rediffusion Television_sentence_3"
},
{
"operation": "finish",
"time": "139.075",
"value": "finish"
}
]
}
|
exfever_7188
|
Tom Brady is a baseball player who spent the first 20 seasons with the professional American football team based in Greater Boston.
|
{
"evidence": [
"Tom Brady",
"New England Patriots"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nTom Brady is a baseball player who spent the first 20 seasons with the professional American football team based in Greater Boston.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Tom Brady",
"New England Patriots"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2459,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_826
|
Little Miss Sunshine is a 2006 American film that starred a multi-awarded American actor, director, and screenwriter.
|
{
"evidence": [
"Little Miss Sunshine",
"Alan Arkin"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLittle Miss Sunshine is a 2006 American film that starred a multi-awarded American actor, director, and screenwriter.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Little Miss Sunshine",
"Alan Arkin"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2460,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6093
|
Cary Grant was a British-American actor who was one of classic Hollywood's definitive actors, which is the primary male protagonist in a film, television, or play, such as an American actor who was known for his natural acting style and versatility.
|
{
"evidence": [
"Cary Grant",
"Leading man",
"Spencer Tracy"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nCary Grant was a British-American actor who was one of classic Hollywood's definitive actors, which is the primary male protagonist in a film, television, or play, such as an American actor who was known for his natural acting style and versatility.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Cary Grant",
"Leading man",
"Spencer Tracy"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2461,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7993
|
"Dirt" is an American television series starring an actress best known for her role as an ambitious, strong-willed journalist and writer in the horror film franchise Scream.
|
{
"evidence": [
"Dirt (TV series)",
"Courteney Cox",
"Gale Weathers"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\n\"Dirt\" is an American television series starring an actress best known for her role as an ambitious, strong-willed journalist and writer in the horror film franchise Scream.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Dirt (TV series)",
"Courteney Cox",
"Gale Weathers"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2462,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
63471
|
In the 1819 United States Senate elections in Maryland, the two top vote-getters (both Democratic-Republicans) were declared the winners, with the first place winner, Edward Lloyd taking the Class 3 seat, and the second place winner, William Pinkney, taking the Class 2 seat.
|
{
"evidence": [
"1819 United States Senate elections in Maryland_sentence_4"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn the 1819 United States Senate elections in Maryland, the two top vote-getters (both Democratic-Republicans) were declared the winners, with the first place winner, Edward Lloyd taking the Class 3 seat, and the second place winner, William Pinkney, taking the Class 2 seat.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"1819 United States Senate elections in Maryland_sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2463,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.051",
"value": "?search="
},
{
"operation": "search",
"time": "14.681",
"value": "Contains...1819 United States Senate elections in Maryland,"
},
{
"operation": "Now on",
"time": "17.114",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "22.211",
"value": "1819 United States Senate elections in Maryland"
},
{
"operation": "Now on",
"time": "23.056",
"value": "1819 United States Senate elections in Maryland"
},
{
"operation": "Highlighting",
"time": "29.652",
"value": "1819 United States Senate elections in Maryland_sentence_3"
},
{
"operation": "Highlighting",
"time": "55.457",
"value": "1819 United States Senate elections in Maryland_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "57.977",
"value": "1819 United States Senate elections in Maryland_sentence_3"
},
{
"operation": "finish",
"time": "74.819",
"value": "finish"
}
]
}
|
exfever_5349
|
Patch Adams is a film starring an Nigerian actor, director, screenwriter, author and narrator of films, theatre, television, and audiobooks.
|
{
"evidence": [
"Patch Adams (film)",
"Peter Coyote"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nPatch Adams is a film starring an Nigerian actor, director, screenwriter, author and narrator of films, theatre, television, and audiobooks.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Patch Adams (film)",
"Peter Coyote"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2464,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
58753
|
Pentraxin-related protein PTX3, also known as TNF-inducible gene 14 protein, is a member of the pentraxin superfamily or phylogenetically conserved humoral mediators of innate immunity, a protein that in humans is encoded by the PTX3 gene.
|
{
"evidence": [
"PTX3_sentence_2",
"PTX3_sentence_3",
"PTX3_sentence_17"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nPentraxin-related protein PTX3, also known as TNF-inducible gene 14 protein, is a member of the pentraxin superfamily or phylogenetically conserved humoral mediators of innate immunity, a protein that in humans is encoded by the PTX3 gene.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"PTX3_sentence_2",
"PTX3_sentence_3",
"PTX3_sentence_17"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2465,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.612",
"value": "?search="
},
{
"operation": "search",
"time": "17.099",
"value": "PTX3"
},
{
"operation": "Now on",
"time": "18.022",
"value": "PTX3"
},
{
"operation": "Highlighting",
"time": "45.327",
"value": "PTX3_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "45.703",
"value": "PTX3_sentence_1"
},
{
"operation": "Highlighting",
"time": "46.167",
"value": "PTX3_sentence_2"
},
{
"operation": "Highlighting",
"time": "53.377",
"value": "PTX3_sentence_3"
},
{
"operation": "hyperlink",
"time": "64.314",
"value": "pentraxin"
},
{
"operation": "Now on",
"time": "65.043",
"value": "Pentraxins"
},
{
"operation": "back-button-clicked",
"time": "71.491",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "71.574",
"value": "PTX3"
},
{
"operation": "Highlighting",
"time": "99.616",
"value": "PTX3_sentence_17"
},
{
"operation": "search",
"time": "154.634",
"value": "Phylogenetically conserved"
},
{
"operation": "Now on",
"time": "155.913",
"value": "Conserved sequence"
},
{
"operation": "finish",
"time": "177.513",
"value": "finish"
}
]
}
|
exfever_3887
|
Timothy David Olyphant is an American actor who appeared in the 2001 American action heist film starring Nicolas Cage.
|
{
"evidence": [
"Timothy Olyphant",
"Gone in 60 Seconds (2000 film)"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nTimothy David Olyphant is an American actor who appeared in the 2001 American action heist film starring Nicolas Cage.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Timothy Olyphant",
"Gone in 60 Seconds (2000 film)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2466,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6228
|
"Inhumans" is an American television series that is set in a shared universe centered on a series of superhero films.
|
{
"evidence": [
"Inhumans (TV series)",
"Marvel Cinematic Universe"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\n\"Inhumans\" is an American television series that is set in a shared universe centered on a series of superhero films.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Inhumans (TV series)",
"Marvel Cinematic Universe"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2467,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1301
|
Joe Chappelle is the youngest film producer.
|
{
"evidence": [
"Phantoms (film)",
"Peter O'Toole"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJoe Chappelle is the youngest film producer.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Phantoms (film)",
"Peter O'Toole"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2468,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
62692
|
Kohei Hirate came twelfth for Team Rosberg, he was a Toyota test driver and competed in the Super GT.
|
{
"evidence": [
"Kohei Hirate_sentence_6",
"Kohei Hirate_sentence_8"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nKohei Hirate came twelfth for Team Rosberg, he was a Toyota test driver and competed in the Super GT.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Kohei Hirate_sentence_6",
"Kohei Hirate_sentence_8"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2469,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.695",
"value": "?search="
},
{
"operation": "search",
"time": "8.757",
"value": "Kohei Hirate"
},
{
"operation": "Now on",
"time": "10.066",
"value": "Kohei Hirate"
},
{
"operation": "Page search",
"time": "13.995",
"value": "twle"
},
{
"operation": "Page search",
"time": "15.009",
"value": "twel"
},
{
"operation": "Highlighting",
"time": "21.551",
"value": "Kohei Hirate_sentence_6"
},
{
"operation": "Highlighting",
"time": "75.436",
"value": "Kohei Hirate_sentence_8"
},
{
"operation": "finish",
"time": "153.591",
"value": "finish"
}
]
}
|
exfever_6861
|
Bridesmaids is Paul Feig's third directorial film.
|
{
"evidence": [
"Paul Feig",
"The Office (U.S. TV series)"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBridesmaids is Paul Feig's third directorial film.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Paul Feig",
"The Office (U.S. TV series)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2470,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
50857
|
Musicogenic epilepsy is found in one out of 10,000,000 people, and is not more prominent in one sex over the other.
|
{
"evidence": [
"Musicogenic epilepsy_sentence_7"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMusicogenic epilepsy is found in one out of 10,000,000 people, and is not more prominent in one sex over the other.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Musicogenic epilepsy_sentence_7"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2471,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.85",
"value": "?search="
},
{
"operation": "search",
"time": "13.626",
"value": "Musicogenic epilepsy"
},
{
"operation": "Now on",
"time": "14.207",
"value": "Musicogenic epilepsy"
},
{
"operation": "Page search",
"time": "26.59",
"value": "one"
},
{
"operation": "Page search",
"time": "43.756",
"value": "prominent"
},
{
"operation": "Highlighting",
"time": "58.628",
"value": "Musicogenic epilepsy_sentence_7"
},
{
"operation": "finish",
"time": "100.724",
"value": "finish"
}
]
}
|
exfever_827
|
Peggy Sue Got Married is a 1986 British fantasy comedy-drama film directed by Francis Ford Coppola and starring an American actress known for her distinctive voice.
|
{
"evidence": [
"Peggy Sue Got Married",
"Kathleen Turner"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nPeggy Sue Got Married is a 1986 British fantasy comedy-drama film directed by Francis Ford Coppola and starring an American actress known for her distinctive voice.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Peggy Sue Got Married",
"Kathleen Turner"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2472,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
85487
|
Although the Spark Arena hosted many entertainment events in 2014, none of these acts were female musical artists.
|
{
"evidence": [
"Spark Arena_sentence_12",
"Spark Arena_sentence_40"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAlthough the Spark Arena hosted many entertainment events in 2014, none of these acts were female musical artists.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Spark Arena_sentence_12",
"Spark Arena_sentence_40"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2473,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.845",
"value": "?search="
},
{
"operation": "search",
"time": "20.438",
"value": "Spark Arena"
},
{
"operation": "Now on",
"time": "21.435",
"value": "Spark Arena"
},
{
"operation": "hyperlink",
"time": "56.821",
"value": "5 Other events Spark Arena_section_4"
},
{
"operation": "search",
"time": "77.457",
"value": "2014 Spark Arena event"
},
{
"operation": "Now on",
"time": "80.19",
"value": "undefined"
},
{
"operation": "search",
"time": "94.082",
"value": "Spark Arena"
},
{
"operation": "Now on",
"time": "95.174",
"value": "Spark Arena"
},
{
"operation": "Highlighting",
"time": "99.901",
"value": "Spark Arena_sentence_12"
},
{
"operation": "Highlighting",
"time": "108.092",
"value": "Spark Arena_sentence_40"
},
{
"operation": "finish",
"time": "122.146",
"value": "finish"
}
]
}
|
23630
|
Langham Place was built on the site of the former Bird Street; construction of the replacement garden was completed in March 1997 on a more peripheral site.
|
{
"evidence": [
"Yuen Po Street Bird Garden_sentence_3",
"Yuen Po Street Bird Garden_sentence_4"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLangham Place was built on the site of the former Bird Street; construction of the replacement garden was completed in March 1997 on a more peripheral site.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Yuen Po Street Bird Garden_sentence_3",
"Yuen Po Street Bird Garden_sentence_4"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2474,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.639",
"value": "?search="
},
{
"operation": "search",
"time": "7.76",
"value": "Langham Place"
},
{
"operation": "Now on",
"time": "8.463",
"value": "Langham Place"
},
{
"operation": "search",
"time": "21.581",
"value": "Yuen Po Street Bird Garden"
},
{
"operation": "Now on",
"time": "22.254",
"value": "Yuen Po Street Bird Garden"
},
{
"operation": "Page search",
"time": "31.698",
"value": "Bird Street"
},
{
"operation": "page-search-reset",
"time": "34.775",
"value": "page-search-reset"
},
{
"operation": "Highlighting",
"time": "36.192",
"value": "Yuen Po Street Bird Garden_sentence_3"
},
{
"operation": "Highlighting",
"time": "69.949",
"value": "Yuen Po Street Bird Garden_sentence_4"
},
{
"operation": "finish",
"time": "76.165",
"value": "finish"
}
]
}
|
exfever_2660
|
Your Highness is a 2018 American stoner comedy fantasy film starring an American actress known for her deadpan roles in comedy films.
|
{
"evidence": [
"Your Highness",
"Zooey Deschanel"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nYour Highness is a 2018 American stoner comedy fantasy film starring an American actress known for her deadpan roles in comedy films.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Your Highness",
"Zooey Deschanel"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2475,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
2344
|
Backbone Flute, written by a Russian and Soviet poet, playwright, artist, and actor in the autumn of 1915 and completed in November, was originally called Verses for Her.
|
{
"evidence": [
"Backbone Flute_sentence_0",
"Vladimir Mayakovsky_sentence_5",
"Backbone Flute_sentence_16"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBackbone Flute, written by a Russian and Soviet poet, playwright, artist, and actor in the autumn of 1915 and completed in November, was originally called Verses for Her.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Backbone Flute_sentence_0",
"Vladimir Mayakovsky_sentence_5",
"Backbone Flute_sentence_16"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2476,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "22.511",
"value": "Backbone Flute"
},
{
"operation": "Highlighting",
"time": "85.388",
"value": "Backbone Flute_sentence_0"
},
{
"operation": "hyperlink",
"time": "94.339",
"value": "Vladimir Mayakovsky"
},
{
"operation": "Highlighting",
"time": "99.292",
"value": "Vladimir Mayakovsky_sentence_5"
},
{
"operation": "back-button-clicked",
"time": "103.848",
"value": "back-button-clicked"
},
{
"operation": "Highlighting",
"time": "122.174",
"value": "Backbone Flute_sentence_16"
},
{
"operation": "finish",
"time": "136.447",
"value": "finish"
}
]
}
|
24172
|
Agreed upon by Prime Minister Winston Churchill and President Roosevelt at the Second Quebec Conference, the Morgenthau Plan was formulated by Henry Morgenthau and his associate Harry White.
|
{
"evidence": [
"Morgenthau Plan_sentence_11",
"Morgenthau Plan_sentence_2",
"Morgenthau Plan_sentence_9"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAgreed upon by Prime Minister Winston Churchill and President Roosevelt at the Second Quebec Conference, the Morgenthau Plan was formulated by Henry Morgenthau and his associate Harry White.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Morgenthau Plan_sentence_11",
"Morgenthau Plan_sentence_2",
"Morgenthau Plan_sentence_9"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2477,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.829",
"value": "?search="
},
{
"operation": "search",
"time": "28.72",
"value": "Morgenthau Plan"
},
{
"operation": "Now on",
"time": "29.459",
"value": "Morgenthau Plan"
},
{
"operation": "Highlighting",
"time": "101.117",
"value": "Morgenthau Plan_sentence_11"
},
{
"operation": "Page search",
"time": "143.756",
"value": "Harry White"
},
{
"operation": "Highlighting",
"time": "210.273",
"value": "Morgenthau Plan_sentence_2"
},
{
"operation": "Highlighting",
"time": "214.777",
"value": "Morgenthau Plan_sentence_9"
},
{
"operation": "finish",
"time": "235.385",
"value": "finish"
}
]
}
|
74021
|
To recruit skilled staff for the project, a newspaper advertisement was posted; but most of the applicants had no records of prior education because those were lost neither in the Holocaust nor during immigration.
|
{
"evidence": [
"WEIZAC_sentence_20"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nTo recruit skilled staff for the project, a newspaper advertisement was posted; but most of the applicants had no records of prior education because those were lost neither in the Holocaust nor during immigration.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"WEIZAC_sentence_20"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2478,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.587",
"value": "?search="
},
{
"operation": "search",
"time": "151.296",
"value": "\"To recruit skilled staff for the project, a newspaper advertisement was posted\""
},
{
"operation": "Now on",
"time": "152.75",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "155.817",
"value": "WEIZAC"
},
{
"operation": "Now on",
"time": "156.706",
"value": "WEIZAC"
},
{
"operation": "Page search",
"time": "159.592",
"value": "To recruit skilled staff for the project, a newspaper advertisement was posted"
},
{
"operation": "Highlighting",
"time": "162.703",
"value": "WEIZAC_sentence_19"
},
{
"operation": "Highlighting",
"time": "165.421",
"value": "WEIZAC_sentence_20"
},
{
"operation": "Highlighting deleted",
"time": "172.524",
"value": "WEIZAC_sentence_19"
},
{
"operation": "finish",
"time": "185.34",
"value": "finish"
}
]
}
|
75866
|
The "Who's more popular now?" in Only Visiting This Planet was referenced to Albert Einstein and the Beatles, an English rock band formed in Liverpool in 1960.
|
{
"evidence": [
"Only Visiting This Planet_sentence_16",
"Only Visiting This Planet_sentence_17"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe \"Who's more popular now?\" in Only Visiting This Planet was referenced to Albert Einstein and the Beatles, an English rock band formed in Liverpool in 1960.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Only Visiting This Planet_sentence_16",
"Only Visiting This Planet_sentence_17"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2479,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.805",
"value": "?search="
},
{
"operation": "search",
"time": "34.771",
"value": "Only Visiting This Planet"
},
{
"operation": "Now on",
"time": "35.492",
"value": "Only Visiting This Planet"
},
{
"operation": "Highlighting",
"time": "109.695",
"value": "Only Visiting This Planet_sentence_16"
},
{
"operation": "Highlighting",
"time": "112.403",
"value": "Only Visiting This Planet_sentence_17"
},
{
"operation": "finish",
"time": "169.307",
"value": "finish"
}
]
}
|
exfever_6246
|
The Simpsons is a satirical depiction of American life, epitomized by the Simpson family, which consists of Homer, Marge, Lisa, Maggie and the eldest child, whose most prominent traits include his mischievousness.
|
{
"evidence": [
"The Simpsons",
"Bart Simpson"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Simpsons is a satirical depiction of American life, epitomized by the Simpson family, which consists of Homer, Marge, Lisa, Maggie and the eldest child, whose most prominent traits include his mischievousness.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Simpsons",
"Bart Simpson"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2480,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
77303
|
Ioannis Vilaras studied medicine in Padova, Italy where he remained practicing medicine until his death in 1823.
|
{
"evidence": [
"Ioannis Vilaras_sentence_4"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIoannis Vilaras studied medicine in Padova, Italy where he remained practicing medicine until his death in 1823.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Ioannis Vilaras_sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2481,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.942",
"value": "?search="
},
{
"operation": "search",
"time": "11.222",
"value": "Ioannis Vilaras"
},
{
"operation": "Now on",
"time": "12.017",
"value": "Ioannis Vilaras"
},
{
"operation": "Page search",
"time": "15.749",
"value": "Padova"
},
{
"operation": "Highlighting",
"time": "18.546",
"value": "Ioannis Vilaras_sentence_1"
},
{
"operation": "Highlighting",
"time": "34.205",
"value": "Ioannis Vilaras_sentence_2"
},
{
"operation": "Highlighting",
"time": "40.102",
"value": "Ioannis Vilaras_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "60.152",
"value": "Ioannis Vilaras_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "66.398",
"value": "Ioannis Vilaras_sentence_2"
},
{
"operation": "finish",
"time": "79.972",
"value": "finish"
}
]
}
|
35744
|
The Opening Ceremony of the 2015 European Games was the official opening ceremony held on June 12, 2015 in the Baku Olympic Stadium in Baku, Azerbaijan, which was attended by more than 100,000 spectators gathered at the Olympic Stadium, as well as hundreds of millions of online viewers.
|
{
"evidence": [
"2015 European Games opening ceremony_sentence_1"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Opening Ceremony of the 2015 European Games was the official opening ceremony held on June 12, 2015 in the Baku Olympic Stadium in Baku, Azerbaijan, which was attended by more than 100,000 spectators gathered at the Olympic Stadium, as well as hundreds of millions of online viewers.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"2015 European Games opening ceremony_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2482,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.769",
"value": "?search="
},
{
"operation": "search",
"time": "12.303",
"value": "Contains...Opening Ceremony of the 2015 European Games"
},
{
"operation": "Now on",
"time": "14.319",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "20.658",
"value": "2015 European Games opening ceremony"
},
{
"operation": "Now on",
"time": "21.162",
"value": "2015 European Games opening ceremony"
},
{
"operation": "Highlighting",
"time": "24.202",
"value": "2015 European Games opening ceremony_sentence_0"
},
{
"operation": "Highlighting",
"time": "43.993",
"value": "2015 European Games opening ceremony_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "44.969",
"value": "2015 European Games opening ceremony_sentence_0"
},
{
"operation": "finish",
"time": "50.774",
"value": "finish"
}
]
}
|
exfever_2662
|
The Belko Experiment is an Australian horror-thriller film starring an American actor and musician known for originating the role of Moritz Stiefel in Duncan Sheik and Steven Sater's rock musical Spring Awakening.
|
{
"evidence": [
"The Belko Experiment",
"John Gallagher Jr."
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Belko Experiment is an Australian horror-thriller film starring an American actor and musician known for originating the role of Moritz Stiefel in Duncan Sheik and Steven Sater's rock musical Spring Awakening.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Belko Experiment",
"John Gallagher Jr."
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2483,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.