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 |
|---|---|---|---|---|---|---|---|---|
66128
|
Sergio Santín made his official debut on July 18, 1980, he represented his native country and played club football for Danubio which was founded by European brothers.
|
{
"evidence": [
"Sergio Santín_sentence_2",
"Sergio Santín_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:\nSergio Santín made his official debut on July 18, 1980, he represented his native country and played club football for Danubio which was founded by European brothers.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Sergio Santín_sentence_2",
"Sergio Santín_sentence_4"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 188,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.249",
"value": "?search="
},
{
"operation": "search",
"time": "16.801",
"value": "Sergio Santín"
},
{
"operation": "Now on",
"time": "17.547",
"value": "Sergio Santín"
},
{
"operation": "Highlighting",
"time": "74.19",
"value": "Sergio Santín_sentence_2"
},
{
"operation": "Highlighting",
"time": "99.122",
"value": "Sergio Santín_sentence_4"
},
{
"operation": "hyperlink",
"time": "101.786",
"value": "Danubio"
},
{
"operation": "Now on",
"time": "102.303",
"value": "Danubio F.C."
},
{
"operation": "Page search",
"time": "129.273",
"value": "European brothers"
},
{
"operation": "finish",
"time": "148.922",
"value": "finish"
}
]
}
|
39396
|
Federico Plummer made his professional debut, a sixth-round knockout of Sam Langford on May 23, 1947; then he won his first six fights before losing to Mike Ross in April 1948.
|
{
"evidence": [
"Federico Plummer_sentence_2",
"Federico Plummer_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:\nFederico Plummer made his professional debut, a sixth-round knockout of Sam Langford on May 23, 1947; then he won his first six fights before losing to Mike Ross in April 1948.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Federico Plummer_sentence_2",
"Federico Plummer_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 189,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.74",
"value": "?search="
},
{
"operation": "search",
"time": "10.95",
"value": "Federico Plummer"
},
{
"operation": "Now on",
"time": "11.897",
"value": "Federico Plummer"
},
{
"operation": "Highlighting",
"time": "19.51",
"value": "Federico Plummer_sentence_2"
},
{
"operation": "Highlighting",
"time": "22.981",
"value": "Federico Plummer_sentence_3"
},
{
"operation": "finish",
"time": "31.988",
"value": "finish"
}
]
}
|
79665
|
Christian made his senior debut for Denmark in the 1994 ICC Trophy (formally the ABN AMRO ICC Trophy), then his next appearance for Denmark came against the Somerset Second XI in 1997.
|
{
"evidence": [
"1994 ICC Trophy squads_sentence_15",
"Denmark national cricket team_sentence_24",
"Denmark national cricket team_sentence_0",
"1994 ICC Trophy_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:\nChristian made his senior debut for Denmark in the 1994 ICC Trophy (formally the ABN AMRO ICC Trophy), then his next appearance for Denmark came against the Somerset Second XI in 1997.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"1994 ICC Trophy squads_sentence_15",
"Denmark national cricket team_sentence_24",
"Denmark national cricket team_sentence_0",
"1994 ICC Trophy_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 190,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.688",
"value": "undefined"
},
{
"operation": "search",
"time": "26.764",
"value": "1994 ICC Trophy squads"
},
{
"operation": "Now on",
"time": "27.983",
"value": "1994 ICC Trophy squads"
},
{
"operation": "hyperlink",
"time": "34.524",
"value": "5 Denmark 1994 ICC Trophy squads_section_4"
},
{
"operation": "Highlighting",
"time": "39.655",
"value": "1994 ICC Trophy squads_sentence_15"
},
{
"operation": "search",
"time": "66.143",
"value": "1994 ICC Trophy"
},
{
"operation": "Now on",
"time": "67.479",
"value": "1994 ICC Trophy"
},
{
"operation": "Highlighting",
"time": "85.083",
"value": "1994 ICC Trophy_cell_4_3_0"
},
{
"operation": "Highlighting",
"time": "85.084",
"value": "1994 ICC Trophy_section_1"
},
{
"operation": "Highlighting",
"time": "85.084",
"value": "1994 ICC Trophy_section_5"
},
{
"operation": "Highlighting",
"time": "85.086",
"value": "1994 ICC Trophy_header_cell_4_0_0"
},
{
"operation": "Highlighting",
"time": "87.511",
"value": "1994 ICC Trophy_cell_4_3_1"
},
{
"operation": "Highlighting",
"time": "87.513",
"value": "1994 ICC Trophy_header_cell_4_0_1"
},
{
"operation": "Highlighting",
"time": "89.908",
"value": "1994 ICC Trophy_cell_4_3_6"
},
{
"operation": "Highlighting",
"time": "89.91",
"value": "1994 ICC Trophy_header_cell_4_0_6"
},
{
"operation": "Highlighting deleted",
"time": "98.21",
"value": "1994 ICC Trophy_cell_4_3_0"
},
{
"operation": "Highlighting deleted",
"time": "98.211",
"value": "1994 ICC Trophy_header_cell_4_0_0"
},
{
"operation": "Highlighting deleted",
"time": "99.234",
"value": "1994 ICC Trophy_cell_4_3_1"
},
{
"operation": "Highlighting deleted",
"time": "99.236",
"value": "1994 ICC Trophy_header_cell_4_0_1"
},
{
"operation": "Highlighting deleted",
"time": "100.217",
"value": "1994 ICC Trophy_cell_4_3_6"
},
{
"operation": "Highlighting deleted",
"time": "100.219",
"value": "1994 ICC Trophy_section_1"
},
{
"operation": "Highlighting deleted",
"time": "100.219",
"value": "1994 ICC Trophy_section_5"
},
{
"operation": "Highlighting deleted",
"time": "100.221",
"value": "1994 ICC Trophy_header_cell_4_0_6"
},
{
"operation": "search",
"time": "126.267",
"value": "Christian cricket"
},
{
"operation": "Now on",
"time": "127.674",
"value": "undefined"
},
{
"operation": "search",
"time": "140.537",
"value": "Christian Louboutin"
},
{
"operation": "Now on",
"time": "141.786",
"value": "Christian Louboutin"
},
{
"operation": "back-button-clicked",
"time": "151.032",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "151.805",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "174.624",
"value": "Dan Christian"
},
{
"operation": "Now on",
"time": "175.656",
"value": "Dan Christian"
},
{
"operation": "Highlighting",
"time": "183.204",
"value": "Dan Christian_sentence_0"
},
{
"operation": "Highlighting",
"time": "183.204",
"value": "Dan Christian_title"
},
{
"operation": "back-button-clicked",
"time": "208.547",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "209.362",
"value": "undefined"
},
{
"operation": "search",
"time": "268.636",
"value": "1994 ICC Trophy squads"
},
{
"operation": "Now on",
"time": "269.287",
"value": "1994 ICC Trophy squads"
},
{
"operation": "search",
"time": "284.183",
"value": "1994 ICC Trophy"
},
{
"operation": "Now on",
"time": "285.114",
"value": "1994 ICC Trophy"
},
{
"operation": "search",
"time": "508.056",
"value": "*christian"
},
{
"operation": "Now on",
"time": "549.009",
"value": "undefined"
},
{
"operation": "search",
"time": "584.843",
"value": "Christian cricket"
},
{
"operation": "Now on",
"time": "585.581",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "598.599",
"value": "Prospect Cricket Club"
},
{
"operation": "Now on",
"time": "599.607",
"value": "Prospect Cricket Club"
},
{
"operation": "hyperlink",
"time": "687.481",
"value": "3 Notable players Prospect Cricket Club_section_2"
},
{
"operation": "back-button-clicked",
"time": "704.232",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "705.011",
"value": "undefined"
},
{
"operation": "search",
"time": "729.802",
"value": "Christian"
},
{
"operation": "Now on",
"time": "730.817",
"value": "Christian"
},
{
"operation": "hyperlink",
"time": "775.778",
"value": "1 Names Christian_section_0"
},
{
"operation": "back-button-clicked",
"time": "801.333",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "802.106",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "811.524",
"value": "Cricket at the 1900 Summer Olympics"
},
{
"operation": "Now on",
"time": "812.388",
"value": "Cricket at the 1900 Summer Olympics"
},
{
"operation": "back-button-clicked",
"time": "826.962",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "827.746",
"value": "undefined"
},
{
"operation": "search",
"time": "890.376",
"value": "Contains...Somerset Second XI in 1997"
},
{
"operation": "Now on",
"time": "892.338",
"value": "undefined"
},
{
"operation": "search",
"time": "952.206",
"value": "Denmark Cricket"
},
{
"operation": "Now on",
"time": "953.44",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "957.498",
"value": "List of Denmark Twenty20 International cricketers"
},
{
"operation": "Now on",
"time": "958.797",
"value": "List of Denmark Twenty20 International cricketers"
},
{
"operation": "back-button-clicked",
"time": "1012.949",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "1013.758",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "1020.28",
"value": "Denmark national cricket team"
},
{
"operation": "Now on",
"time": "1021.55",
"value": "Denmark national cricket team"
},
{
"operation": "search",
"time": "1045.601",
"value": "Contains...Denmark national cricket team 1994"
},
{
"operation": "Now on",
"time": "1047.375",
"value": "undefined"
},
{
"operation": "search",
"time": "1131.921",
"value": "Denmark national cricket team"
},
{
"operation": "Now on",
"time": "1132.809",
"value": "Denmark national cricket team"
},
{
"operation": "Highlighting",
"time": "1159.057",
"value": "Denmark national cricket team_sentence_26"
},
{
"operation": "Highlighting deleted",
"time": "1168.485",
"value": "Denmark national cricket team_sentence_26"
},
{
"operation": "Highlighting",
"time": "1178.342",
"value": "Denmark national cricket team_sentence_24"
},
{
"operation": "Highlighting deleted",
"time": "1184.016",
"value": "Denmark national cricket team_sentence_24"
},
{
"operation": "Highlighting",
"time": "1186.725",
"value": "Denmark national cricket team_sentence_24"
},
{
"operation": "hyperlink",
"time": "1195.519",
"value": "ICC Trophy"
},
{
"operation": "Now on",
"time": "1197.021",
"value": "ICC World Cup Qualifier"
},
{
"operation": "hyperlink",
"time": "1211.585",
"value": "1994"
},
{
"operation": "Now on",
"time": "1212.147",
"value": "1994 ICC Trophy"
},
{
"operation": "hyperlink",
"time": "1232.979",
"value": "Denmark"
},
{
"operation": "Now on",
"time": "1233.468",
"value": "Denmark national cricket team"
},
{
"operation": "Highlighting",
"time": "1236.587",
"value": "Denmark national cricket team_sentence_0"
},
{
"operation": "back-button-clicked",
"time": "1279.664",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "1279.789",
"value": "1994 ICC Trophy"
},
{
"operation": "Highlighting",
"time": "1287.296",
"value": "1994 ICC Trophy_sentence_0"
},
{
"operation": "finish",
"time": "1326.838",
"value": "finish"
}
]
}
|
exfever_1095
|
Led Zeppelin was the most prominent band of their generation.
|
{
"evidence": [
"Led Zeppelin",
"Hard rock"
],
"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:\nLed Zeppelin was the most prominent band of their generation.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Led Zeppelin",
"Hard rock"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 191,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
exfever_8472
|
Justin Trudeau is the second-youngest prime minister in Canadian history after a statesman, businessman, writer, and politician who won the leadership of a centre-right federal political party in Canada.
|
{
"evidence": [
"Justin Trudeau",
"Joe Clark",
"Progressive Conservative Party of Canada"
],
"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:\nJustin Trudeau is the second-youngest prime minister in Canadian history after a statesman, businessman, writer, and politician who won the leadership of a centre-right federal political party in Canada.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Justin Trudeau",
"Joe Clark",
"Progressive Conservative Party of Canada"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 192,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
41948
|
Johnson and Clark batted well before Clark got out leaving Australia with a healthy score of 343, then the Indian team were sent into bat and Dravid and Jaffer batted very slowly before losing early wickets.
|
{
"evidence": [
"Indian cricket team in Australia in 2007–08_sentence_8",
"Indian cricket team in Australia in 2007–08_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:\nJohnson and Clark batted well before Clark got out leaving Australia with a healthy score of 343, then the Indian team were sent into bat and Dravid and Jaffer batted very slowly before losing early wickets.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Indian cricket team in Australia in 2007–08_sentence_8",
"Indian cricket team in Australia in 2007–08_sentence_9"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 193,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.606",
"value": "?search="
},
{
"operation": "search",
"time": "22.897",
"value": "Contains...Johnson and Clark batted well before Clark got out leaving Australia with a healthy score of 343"
},
{
"operation": "Now on",
"time": "25.634",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "28.498",
"value": "Indian cricket team in Australia in 2007–08"
},
{
"operation": "Now on",
"time": "29.442",
"value": "Indian cricket team in Australia in 2007–08"
},
{
"operation": "Page search",
"time": "33.121",
"value": "Johnson and Clark batted well before Clark got out leaving Australia with a healthy score of 343"
},
{
"operation": "Highlighting",
"time": "37.581",
"value": "Indian cricket team in Australia in 2007–08_sentence_8"
},
{
"operation": "Highlighting",
"time": "44.364",
"value": "Indian cricket team in Australia in 2007–08_sentence_9"
},
{
"operation": "finish",
"time": "57.591",
"value": "finish"
}
]
}
|
exfever_4892
|
The New York Giants compete in the National Football League as a member club of one of the two conferences of the National Football League and its counterpart is organized into 6 groups of teams who compete against each other for a championship.
|
{
"evidence": [
"New York Giants",
"National Football Conference",
"Division (sport)"
],
"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 New York Giants compete in the National Football League as a member club of one of the two conferences of the National Football League and its counterpart is organized into 6 groups of teams who compete against each other for a championship.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"New York Giants",
"National Football Conference",
"Division (sport)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 194,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
20086
|
To make traditional ginger ale, yeast which contains zymase is added with sugar which is the generic name for sweet-tasting soluble carbohydrates.
|
{
"evidence": [
"Ginger ale_sentence_25",
"Ginger ale_sentence_28",
"Sugar_sentence_5"
],
"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:\nTo make traditional ginger ale, yeast which contains zymase is added with sugar which is the generic name for sweet-tasting soluble carbohydrates.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Ginger ale_sentence_25",
"Ginger ale_sentence_28",
"Sugar_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 195,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.022",
"value": "?search="
},
{
"operation": "search",
"time": "20.547",
"value": "Ginger ale"
},
{
"operation": "Now on",
"time": "21.391",
"value": "Ginger ale"
},
{
"operation": "hyperlink",
"time": "26.622",
"value": "1 History Ginger ale_section_0"
},
{
"operation": "Highlighting",
"time": "42.007",
"value": "Ginger ale_sentence_25"
},
{
"operation": "Highlighting",
"time": "47.739",
"value": "Ginger ale_sentence_28"
},
{
"operation": "hyperlink",
"time": "58.509",
"value": "sugar"
},
{
"operation": "Now on",
"time": "59.219",
"value": "Sugar"
},
{
"operation": "Highlighting",
"time": "65.399",
"value": "Sugar_sentence_5"
},
{
"operation": "finish",
"time": "97.463",
"value": "finish"
}
]
}
|
32352
|
Before the auspicious time, Karadaiyan Nonbu (a festival performed by married women from the Tamil Brahmin community) is done women draw kolam infront of their pooja room, take bath and officiate the ceremony with the pooja; wherein an offering made out of rice powder and jaggery is prepared and offered to the goddess.
|
{
"evidence": [
"Karadaiyan Nonbu_sentence_0",
"Karadaiyan Nonbu_sentence_6",
"Karadaiyan Nonbu_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:\nBefore the auspicious time, Karadaiyan Nonbu (a festival performed by married women from the Tamil Brahmin community) is done women draw kolam infront of their pooja room, take bath and officiate the ceremony with the pooja; wherein an offering made out of rice powder and jaggery is prepared and offered to the goddess.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Karadaiyan Nonbu_sentence_0",
"Karadaiyan Nonbu_sentence_6",
"Karadaiyan Nonbu_sentence_7"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 196,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.103",
"value": "?search="
},
{
"operation": "search",
"time": "9.162",
"value": "Karadaiyan Nonbu"
},
{
"operation": "Now on",
"time": "10.02",
"value": "Karadaiyan Nonbu"
},
{
"operation": "Highlighting",
"time": "13.693",
"value": "Karadaiyan Nonbu_sentence_0"
},
{
"operation": "Highlighting",
"time": "22.286",
"value": "Karadaiyan Nonbu_sentence_6"
},
{
"operation": "Highlighting",
"time": "58.942",
"value": "Karadaiyan Nonbu_sentence_7"
},
{
"operation": "finish",
"time": "73.934",
"value": "finish"
}
]
}
|
76453
|
Fort Kalamata was under the British forces in 1625, before being conquered by Spain when they invaded the Spice Islands.
|
{
"evidence": [
"Fort Kalamata_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:\nFort Kalamata was under the British forces in 1625, before being conquered by Spain when they invaded the Spice Islands.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Fort Kalamata_sentence_7"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 197,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.226",
"value": "?search="
},
{
"operation": "search",
"time": "13.089",
"value": "Fort Kalamata"
},
{
"operation": "Now on",
"time": "14.272",
"value": "Fort Kalamata"
},
{
"operation": "Highlighting",
"time": "94.871",
"value": "Fort Kalamata_sentence_7"
},
{
"operation": "Highlighting",
"time": "105.791",
"value": "Fort Kalamata_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "109.951",
"value": "Fort Kalamata_sentence_7"
},
{
"operation": "Highlighting",
"time": "117.343",
"value": "Fort Kalamata_sentence_6"
},
{
"operation": "Highlighting",
"time": "127.053",
"value": "Fort Kalamata_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "160.945",
"value": "Fort Kalamata_sentence_7"
},
{
"operation": "search",
"time": "166.847",
"value": "Geen Huigen Schapen"
},
{
"operation": "Now on",
"time": "173.222",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "178.973",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "179.208",
"value": "Fort Kalamata"
},
{
"operation": "Highlighting",
"time": "185.081",
"value": "Fort Kalamata_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "186.095",
"value": "Fort Kalamata_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "186.677",
"value": "Fort Kalamata_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "188.807",
"value": "Fort Kalamata_sentence_7"
},
{
"operation": "Highlighting",
"time": "190.009",
"value": "Fort Kalamata_sentence_5"
},
{
"operation": "Highlighting",
"time": "191.973",
"value": "Fort Kalamata_sentence_6"
},
{
"operation": "Highlighting",
"time": "192.588",
"value": "Fort Kalamata_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "207.616",
"value": "Fort Kalamata_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "208.637",
"value": "Fort Kalamata_sentence_5"
},
{
"operation": "finish",
"time": "215.131",
"value": "finish"
}
]
}
|
79603
|
Richard Bloomingdale is a labor union activist who has served as President of the Pennsylvania AFL-CIO.
|
{
"evidence": [
"Richard Bloomingdale_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:\nRichard Bloomingdale is a labor union activist who has served as President of the Pennsylvania AFL-CIO.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Richard Bloomingdale_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 198,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.437",
"value": "?search="
},
{
"operation": "search",
"time": "13.737",
"value": "Richard Bloomingdale"
},
{
"operation": "Now on",
"time": "14.886",
"value": "Richard Bloomingdale"
},
{
"operation": "Highlighting",
"time": "20.829",
"value": "Richard Bloomingdale_sentence_0"
},
{
"operation": "finish",
"time": "69.968",
"value": "finish"
}
]
}
|
31560
|
DEVS (Discrete Event System Specification) formalism for modeling and analyzing general systems was invented by Bernard P. Zeigler published in his first book released in 1976.
|
{
"evidence": [
"DEVS_sentence_1",
"DEVS_sentence_4",
"DEVS_sentence_5"
],
"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:\nDEVS (Discrete Event System Specification) formalism for modeling and analyzing general systems was invented by Bernard P. Zeigler published in his first book released in 1976.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"DEVS_sentence_1",
"DEVS_sentence_4",
"DEVS_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 199,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.705",
"value": "?search="
},
{
"operation": "search",
"time": "14.449",
"value": "DEVS"
},
{
"operation": "Now on",
"time": "15.361",
"value": "DEVS"
},
{
"operation": "Highlighting",
"time": "17.709",
"value": "DEVS_sentence_1"
},
{
"operation": "Highlighting",
"time": "64.879",
"value": "DEVS_sentence_4"
},
{
"operation": "Highlighting",
"time": "73.403",
"value": "DEVS_sentence_5"
},
{
"operation": "finish",
"time": "132.974",
"value": "finish"
}
]
}
|
exfever_2966
|
Django Unchained is a 2011 revisionist Western film starring an American actor who has appeared in television series, including The Shield.
|
{
"evidence": [
"Django Unchained",
"Walton Goggins"
],
"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:\nDjango Unchained is a 2011 revisionist Western film starring an American actor who has appeared in television series, including The Shield.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Django Unchained",
"Walton Goggins"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 200,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
exfever_3938
|
Saving Private Ryan is a 1998 American war film whose genre of filmmaking is with large-scale, sweeping scope and spectacle, directed by Steven Spielberg.
|
{
"evidence": [
"Saving Private Ryan",
"Epic 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:\nSaving Private Ryan is a 1998 American war film whose genre of filmmaking is with large-scale, sweeping scope and spectacle, directed by Steven Spielberg.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Saving Private Ryan",
"Epic film"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 201,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
7690
|
Since 1978 the Calgary Flames Hockey Team has been winning awards along with its staff and up until the 2018-2019 season have acquired a total of 16 but have never won the Calder Memorial Trophy for Rookie of the Year.
|
{
"evidence": [
"Calgary Flames_sentence_78",
"Calgary Flames_sentence_96",
"Calgary Flames_sentence_336"
],
"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:\nSince 1978 the Calgary Flames Hockey Team has been winning awards along with its staff and up until the 2018-2019 season have acquired a total of 16 but have never won the Calder Memorial Trophy for Rookie of the Year.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Calgary Flames_sentence_78",
"Calgary Flames_sentence_96",
"Calgary Flames_sentence_336"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 202,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.534",
"value": "?search="
},
{
"operation": "search",
"time": "24.305",
"value": "Calgary Flames awards"
},
{
"operation": "Now on",
"time": "26.542",
"value": "undefined"
},
{
"operation": "search",
"time": "43.076",
"value": "Calgary Flames"
},
{
"operation": "Now on",
"time": "44.208",
"value": "Calgary Flames"
},
{
"operation": "Highlighting",
"time": "137.355",
"value": "Calgary Flames_sentence_78"
},
{
"operation": "Highlighting",
"time": "158.101",
"value": "Calgary Flames_sentence_96"
},
{
"operation": "Highlighting",
"time": "164.093",
"value": "Calgary Flames_sentence_336"
},
{
"operation": "finish",
"time": "174.203",
"value": "finish"
}
]
}
|
exfever_1870
|
Aluminium is the preferred material for construction.
|
{
"evidence": [
"Aluminium",
"Metal"
],
"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:\nAluminium is the preferred material for construction.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Aluminium",
"Metal"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 203,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
10877
|
Linda Tripp was a colleague of Monica Lewinsky who had a physical relationship with a US president.
|
{
"evidence": [
"Linda Tripp_sentence_14",
"Monica Lewinsky_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:\nLinda Tripp was a colleague of Monica Lewinsky who had a physical relationship with a US president.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Linda Tripp_sentence_14",
"Monica Lewinsky_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 204,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.534",
"value": "?search="
},
{
"operation": "search",
"time": "4.659",
"value": "Linda Tripp"
},
{
"operation": "Now on",
"time": "5.364",
"value": "Linda Tripp"
},
{
"operation": "Page search",
"time": "12.537",
"value": "Monica Lewinsky"
},
{
"operation": "Highlighting",
"time": "39.205",
"value": "Linda Tripp_sentence_14"
},
{
"operation": "hyperlink",
"time": "48.238",
"value": "Monica Lewinsky"
},
{
"operation": "Now on",
"time": "48.911",
"value": "Monica Lewinsky"
},
{
"operation": "Highlighting",
"time": "55.386",
"value": "Monica Lewinsky_sentence_3"
},
{
"operation": "finish",
"time": "115.686",
"value": "finish"
}
]
}
|
exfever_185
|
Peter Jackson directed King Kong and The Lord of The Rings.
|
{
"evidence": [
"King Kong (2005 film)",
"Epic film"
],
"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:\nPeter Jackson directed King Kong and The Lord of The Rings.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"King Kong (2005 film)",
"Epic film"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 205,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
exfever_6512
|
Conan the Destroyer was the first film appearance of Wilt Chamberlain.
|
{
"evidence": [
"Wilt Chamberlain",
"Philadelphia 76ers"
],
"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:\nConan the Destroyer was the first film appearance of Wilt Chamberlain.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Wilt Chamberlain",
"Philadelphia 76ers"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 206,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
exfever_7458
|
Paul Labile Pogba is a French professional footballer who plays mostly as a central midfielder but can also be used as a deep-lying player who directs the team's play and is frequently involved in offensive and defensive passing sequences that result in goals.
|
{
"evidence": [
"Paul Pogba",
"Playmaker"
],
"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:\nPaul Labile Pogba is a French professional footballer who plays mostly as a central midfielder but can also be used as a deep-lying player who directs the team's play and is frequently involved in offensive and defensive passing sequences that result in goals.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Paul Pogba",
"Playmaker"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 207,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
94227
|
Bad Sülze is located in the Vorpommern-Rügen district of Germany.
|
{
"evidence": [
"Bad Sülze_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:\nBad Sülze is located in the Vorpommern-Rügen district of Germany.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Bad Sülze_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 208,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.29",
"value": "?search="
},
{
"operation": "search",
"time": "12.682",
"value": "Bad Sülze"
},
{
"operation": "Now on",
"time": "13.891",
"value": "Bad Sülze"
},
{
"operation": "Highlighting",
"time": "20.769",
"value": "Bad Sülze_sentence_0"
},
{
"operation": "finish",
"time": "35.17",
"value": "finish"
}
]
}
|
20022
|
In the Dominican Republic, a plate, which shows the period number, is issued to a vehicle biannually.
|
{
"evidence": [
"Vehicle registration plates of the Dominican Republic_sentence_2",
"Vehicle registration plates of the Dominican Republic_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:\nIn the Dominican Republic, a plate, which shows the period number, is issued to a vehicle biannually.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Vehicle registration plates of the Dominican Republic_sentence_2",
"Vehicle registration plates of the Dominican Republic_sentence_4"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 209,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.128",
"value": "?search="
},
{
"operation": "search",
"time": "22.536",
"value": "Dominican Republic"
},
{
"operation": "Now on",
"time": "29.882",
"value": "Dominican Republic"
},
{
"operation": "search",
"time": "89.143",
"value": "dominican republic driving"
},
{
"operation": "Now on",
"time": "92.253",
"value": "undefined"
},
{
"operation": "search",
"time": "119.612",
"value": "dominican republic license plates"
},
{
"operation": "Now on",
"time": "121.669",
"value": "undefined"
},
{
"operation": "search",
"time": "155.334",
"value": "Vehicle registration plates of dominican republic"
},
{
"operation": "Now on",
"time": "157.927",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "159.96",
"value": "Vehicle registration plates of the Dominican Republic"
},
{
"operation": "Now on",
"time": "162.98",
"value": "Vehicle registration plates of the Dominican Republic"
},
{
"operation": "Highlighting",
"time": "167.844",
"value": "Vehicle registration plates of the Dominican Republic_sentence_2"
},
{
"operation": "Highlighting",
"time": "185.178",
"value": "Vehicle registration plates of the Dominican Republic_sentence_4"
},
{
"operation": "finish",
"time": "205.057",
"value": "finish"
}
]
}
|
13187
|
The 1905 Clemson Tigers football team won three out of six games under coach Eddie Cochems.
|
{
"evidence": [
"1905 Clemson Tigers football team_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 1905 Clemson Tigers football team won three out of six games under coach Eddie Cochems.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"1905 Clemson Tigers football team_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 210,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.053",
"value": "?search="
},
{
"operation": "search",
"time": "7.692",
"value": "1905 Clemson Tigers football team"
},
{
"operation": "Now on",
"time": "8.67",
"value": "1905 Clemson Tigers football team"
},
{
"operation": "Highlighting",
"time": "13.238",
"value": "1905 Clemson Tigers football team_sentence_1"
},
{
"operation": "finish",
"time": "23.424",
"value": "finish"
}
]
}
|
80927
|
A 336 km metre gauge line of the North Bengal Railway linked Saraghat on the northern bank of the Padma to Siliguri and it was during this period that Abdulpur (a railway junction in Natore District of Rajshahi Division in Bangladesh) came up as a railway station.
|
{
"evidence": [
"Abdulpur Junction railway station_sentence_3",
"Abdulpur Junction railway station_sentence_4",
"Abdulpur Junction railway station_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:\nA 336 km metre gauge line of the North Bengal Railway linked Saraghat on the northern bank of the Padma to Siliguri and it was during this period that Abdulpur (a railway junction in Natore District of Rajshahi Division in Bangladesh) came up as a railway station.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Abdulpur Junction railway station_sentence_3",
"Abdulpur Junction railway station_sentence_4",
"Abdulpur Junction railway station_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 211,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.048",
"value": "?search="
},
{
"operation": "search",
"time": "25.023",
"value": "Contains...Saraghat on the northern bank of the Padma to Siliguri"
},
{
"operation": "Now on",
"time": "26.495",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "69.38",
"value": "Parbatipur Junction railway station"
},
{
"operation": "Now on",
"time": "70.06",
"value": "Parbatipur Junction railway station"
},
{
"operation": "Highlighting",
"time": "79.855",
"value": "Parbatipur Junction railway station_sentence_3"
},
{
"operation": "Highlighting",
"time": "79.856",
"value": "Parbatipur Junction railway station_title"
},
{
"operation": "Highlighting",
"time": "79.857",
"value": "Parbatipur Junction railway station_section_0"
},
{
"operation": "Highlighting deleted",
"time": "95.676",
"value": "Parbatipur Junction railway station_sentence_3"
},
{
"operation": "Highlighting deleted",
"time": "95.679",
"value": "Parbatipur Junction railway station_title"
},
{
"operation": "Highlighting deleted",
"time": "95.684",
"value": "Parbatipur Junction railway station_section_0"
},
{
"operation": "Page search",
"time": "106.591",
"value": "Abdulpur"
},
{
"operation": "search",
"time": "117.298",
"value": "Abdulpur Junction railway station"
},
{
"operation": "Now on",
"time": "118.326",
"value": "Abdulpur Junction railway station"
},
{
"operation": "Highlighting",
"time": "121.302",
"value": "Abdulpur Junction railway station_sentence_3"
},
{
"operation": "Highlighting",
"time": "128.397",
"value": "Abdulpur Junction railway station_sentence_4"
},
{
"operation": "Highlighting",
"time": "203.782",
"value": "Abdulpur Junction railway station_sentence_0"
},
{
"operation": "finish",
"time": "212.017",
"value": "finish"
}
]
}
|
exfever_5529
|
Michigan, with a population of nearly 10.1 million is a state in the bi-national Canadian–American region that includes portions of the eight constituent political entities, of which there are currently 50.
|
{
"evidence": [
"Michigan",
"Great Lakes region",
"U.S. state"
],
"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:\nMichigan, with a population of nearly 10.1 million is a state in the bi-national Canadian–American region that includes portions of the eight constituent political entities, of which there are currently 50.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Michigan",
"Great Lakes region",
"U.S. state"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 212,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
exfever_1858
|
iO was an improv theater and training center with a former branch in the capital of the state of North Carolina and the seat of Wake County in the United Kingdom.
|
{
"evidence": [
"IO Theater",
"Raleigh, North Carolina"
],
"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:\niO was an improv theater and training center with a former branch in the capital of the state of North Carolina and the seat of Wake County in the United Kingdom.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"IO Theater",
"Raleigh, North Carolina"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 213,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
52994
|
Born on 2 April 1910, Peter Beatty was the oldest son of David Beatty, a poor peasant, and his wife Ethel, and as a baby, he was born without any infectious disease or any injuries.
|
{
"evidence": [
"Peter Beatty_sentence_2",
"Peter Beatty_sentence_5",
"Neonatal conjunctivitis_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:\nBorn on 2 April 1910, Peter Beatty was the oldest son of David Beatty, a poor peasant, and his wife Ethel, and as a baby, he was born without any infectious disease or any injuries.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Peter Beatty_sentence_2",
"Peter Beatty_sentence_5",
"Neonatal conjunctivitis_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 214,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.659",
"value": "?search="
},
{
"operation": "search",
"time": "16.36",
"value": "Peter Beatty"
},
{
"operation": "Now on",
"time": "17.727",
"value": "Peter Beatty"
},
{
"operation": "Highlighting",
"time": "77.387",
"value": "Peter Beatty_sentence_1"
},
{
"operation": "Highlighting",
"time": "88.821",
"value": "Peter Beatty_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "90.044",
"value": "Peter Beatty_sentence_1"
},
{
"operation": "Highlighting",
"time": "121.838",
"value": "Peter Beatty_sentence_5"
},
{
"operation": "hyperlink",
"time": "131.606",
"value": "ophthalmia neonatorum"
},
{
"operation": "Now on",
"time": "132.931",
"value": "Neonatal conjunctivitis"
},
{
"operation": "Highlighting",
"time": "136.578",
"value": "Neonatal conjunctivitis_sentence_0"
},
{
"operation": "back-button-clicked",
"time": "159.366",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "159.456",
"value": "Peter Beatty"
},
{
"operation": "hyperlink",
"time": "165.183",
"value": "David Beatty, 1st Earl Beatty"
},
{
"operation": "Now on",
"time": "167.632",
"value": "David Beatty, 1st Earl Beatty"
},
{
"operation": "finish",
"time": "186.638",
"value": "finish"
}
]
}
|
exfever_8897
|
Babe Ruth was an American professional baseball player whose name was elected into the Hall of Fame, a historical museum operated by private interests.
|
{
"evidence": [
"Babe Ruth",
"National Baseball Hall of Fame and Museum"
],
"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:\nBabe Ruth was an American professional baseball player whose name was elected into the Hall of Fame, a historical museum operated by private interests.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Babe Ruth",
"National Baseball Hall of Fame and Museum"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 215,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
65068
|
Known as Gramzow Provincial School, Old Carbrook State School served as a location for German church services and language classes for German children.
|
{
"evidence": [
"Old Carbrook State School_sentence_28",
"Old Carbrook State School_sentence_27",
"Old Carbrook State School_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:\nKnown as Gramzow Provincial School, Old Carbrook State School served as a location for German church services and language classes for German children.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Old Carbrook State School_sentence_28",
"Old Carbrook State School_sentence_27",
"Old Carbrook State School_sentence_2"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 216,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0",
"value": "?search="
},
{
"operation": "search",
"time": "0.545",
"value": "Old Carbrook"
},
{
"operation": "Now on",
"time": "0.547",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "0.55",
"value": "Old Carbrook State School"
},
{
"operation": "Now on",
"time": "0.56",
"value": "Old Carbrook State School"
},
{
"operation": "Highlighting",
"time": "0.579",
"value": "Old Carbrook State School_sentence_2"
},
{
"operation": "Page search",
"time": "0.593",
"value": "church"
},
{
"operation": "Highlighting",
"time": "0.753",
"value": "Old Carbrook State School_sentence_28"
},
{
"operation": "Highlighting deleted",
"time": "0.764",
"value": "Old Carbrook State School_sentence_2"
},
{
"operation": "Highlighting",
"time": "0.812",
"value": "Old Carbrook State School_sentence_27"
},
{
"operation": "Highlighting",
"time": "0.885",
"value": "Old Carbrook State School_sentence_2"
},
{
"operation": "Highlighting",
"time": "1.053",
"value": "Old Carbrook State School_sentence_28"
},
{
"operation": "finish",
"time": "1.09",
"value": "finish"
},
{
"operation": "Now on",
"time": "1621670748.248",
"value": "Old Carbrook State School"
},
{
"operation": "Highlighting",
"time": "1621670762.839",
"value": "Old Carbrook State School_sentence_27"
},
{
"operation": "Highlighting",
"time": "1621670767.331",
"value": "Old Carbrook State School_sentence_2"
},
{
"operation": "finish",
"time": "1621670794.23",
"value": "finish"
}
]
}
|
exfever_7504
|
Indian is the first film produced by A. M. Rathnam.
|
{
"evidence": [
"Indian (1996 film)",
"Manisha Koirala"
],
"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:\nIndian is the first film produced by A. M. Rathnam.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Indian (1996 film)",
"Manisha Koirala"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 217,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
44037
|
Vidame de Chartres was never put in the Bastille according to the novel La Princesse de Clèves.
|
{
"evidence": [
"Vidame de Chartres_sentence_22"
],
"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:\nVidame de Chartres was never put in the Bastille according to the novel La Princesse de Clèves.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Vidame de Chartres_sentence_22"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 218,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.063",
"value": "?search="
},
{
"operation": "search",
"time": "7",
"value": "Vidame de Chartres"
},
{
"operation": "Now on",
"time": "8.207",
"value": "Vidame de Chartres"
},
{
"operation": "Page search",
"time": "18.021",
"value": "bastille"
},
{
"operation": "Highlighting",
"time": "32.146",
"value": "Vidame de Chartres_sentence_21"
},
{
"operation": "Page search",
"time": "47.211",
"value": "cleves"
},
{
"operation": "Page search",
"time": "60.158",
"value": "cl"
},
{
"operation": "hyperlink",
"time": "101.247",
"value": "La Princesse de Clèves"
},
{
"operation": "Now on",
"time": "104.667",
"value": "La Princesse de Clèves"
},
{
"operation": "Page search",
"time": "116.24",
"value": "bastille"
},
{
"operation": "back-button-clicked",
"time": "128.287",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "128.432",
"value": "Vidame de Chartres"
},
{
"operation": "Highlighting",
"time": "171.407",
"value": "Vidame de Chartres_sentence_22"
},
{
"operation": "Highlighting deleted",
"time": "201.84",
"value": "Vidame de Chartres_sentence_22"
},
{
"operation": "Highlighting deleted",
"time": "202.852",
"value": "Vidame de Chartres_sentence_21"
},
{
"operation": "Highlighting",
"time": "207.336",
"value": "Vidame de Chartres_sentence_22"
},
{
"operation": "hyperlink",
"time": "209.395",
"value": "La Princesse de Clèves"
},
{
"operation": "Now on",
"time": "210.384",
"value": "La Princesse de Clèves"
},
{
"operation": "finish",
"time": "249.19",
"value": "finish"
}
]
}
|
27616
|
Kula Kei Uluivuya was founded by Viliame Cokanauto at the age of 19 and started with birthday gigs before playing at the University of the South Pacific when introduced through a battle of the bands.
|
{
"evidence": [
"Kula Kei Uluivuya_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:\nKula Kei Uluivuya was founded by Viliame Cokanauto at the age of 19 and started with birthday gigs before playing at the University of the South Pacific when introduced through a battle of the bands.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Kula Kei Uluivuya_sentence_6"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 219,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.943",
"value": "?search="
},
{
"operation": "search",
"time": "8.455",
"value": "Kula Kei Uluivuya"
},
{
"operation": "Now on",
"time": "9.601",
"value": "Kula Kei Uluivuya"
},
{
"operation": "Now on",
"time": "360.192",
"value": "Kula Kei Uluivuya"
},
{
"operation": "Now on",
"time": "396.844",
"value": "Kula Kei Uluivuya"
},
{
"operation": "Page search",
"time": "434.985",
"value": "Viliame Cokanauto"
},
{
"operation": "Highlighting",
"time": "437.407",
"value": "Kula Kei Uluivuya_sentence_4"
},
{
"operation": "Highlighting",
"time": "440.532",
"value": "Kula Kei Uluivuya_sentence_5"
},
{
"operation": "Highlighting",
"time": "447.206",
"value": "Kula Kei Uluivuya_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "448.609",
"value": "Kula Kei Uluivuya_sentence_6"
},
{
"operation": "Highlighting",
"time": "459.421",
"value": "Kula Kei Uluivuya_sentence_6"
},
{
"operation": "Page search",
"time": "478.456",
"value": "battle"
},
{
"operation": "Page search",
"time": "513.29",
"value": "battle"
},
{
"operation": "Highlighting",
"time": "517.001",
"value": "Kula Kei Uluivuya_sentence_8"
},
{
"operation": "Highlighting deleted",
"time": "524.674",
"value": "Kula Kei Uluivuya_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "525.613",
"value": "Kula Kei Uluivuya_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "526.329",
"value": "Kula Kei Uluivuya_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "541.846",
"value": "Kula Kei Uluivuya_sentence_8"
},
{
"operation": "Highlighting",
"time": "546.072",
"value": "Kula Kei Uluivuya_sentence_6"
},
{
"operation": "finish",
"time": "556.134",
"value": "finish"
}
]
}
|
exfever_7463
|
School 2013 is a 2020 South Korean television series starring a South Korean actor and model.
|
{
"evidence": [
"School 2013",
"Lee Jong-suk"
],
"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:\nSchool 2013 is a 2020 South Korean television series starring a South Korean actor and model.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"School 2013",
"Lee Jong-suk"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 220,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
29513
|
Carmen de Burgos, a Spanish journalist, writer, translator and women's rights activist, enrolled at the local college to obtain a teaching certificate after her infant son died and was able to teach primary within a year.
|
{
"evidence": [
"Carmen de Burgos_sentence_0",
"Carmen de Burgos_sentence_9",
"Carmen de Burgos_sentence_10"
],
"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:\nCarmen de Burgos, a Spanish journalist, writer, translator and women's rights activist, enrolled at the local college to obtain a teaching certificate after her infant son died and was able to teach primary within a year.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Carmen de Burgos_sentence_0",
"Carmen de Burgos_sentence_9",
"Carmen de Burgos_sentence_10"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 221,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.628",
"value": "?search="
},
{
"operation": "search",
"time": "4.233",
"value": "Carmen de Burgos"
},
{
"operation": "Now on",
"time": "4.903",
"value": "Carmen de Burgos"
},
{
"operation": "Highlighting",
"time": "6.288",
"value": "Carmen de Burgos_sentence_0"
},
{
"operation": "Highlighting",
"time": "12.435",
"value": "Carmen de Burgos_sentence_9"
},
{
"operation": "Highlighting",
"time": "21.901",
"value": "Carmen de Burgos_sentence_10"
},
{
"operation": "finish",
"time": "27.834",
"value": "finish"
}
]
}
|
exfever_6596
|
Seth Woodbury MacFarlane is an American actor who starred in a 2014 American comedy film whose genre of fiction and film is set primarily in the latter half of the 19th century and the early 20th century in the Western United States.
|
{
"evidence": [
"Seth MacFarlane",
"A Million Ways to Die in the West",
"Western (genre)"
],
"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:\nSeth Woodbury MacFarlane is an American actor who starred in a 2014 American comedy film whose genre of fiction and film is set primarily in the latter half of the 19th century and the early 20th century in the Western United States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Seth MacFarlane",
"A Million Ways to Die in the West",
"Western (genre)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 222,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
exfever_2232
|
Django Unchained is a 2012 Mexican revisionist Western film, starring an American actor, comedian, and singer who is the recipient of various accolades, including an Academy Award.
|
{
"evidence": [
"Django Unchained",
"Jamie Foxx"
],
"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:\nDjango Unchained is a 2012 Mexican revisionist Western film, starring an American actor, comedian, and singer who is the recipient of various accolades, including an Academy Award.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Django Unchained",
"Jamie Foxx"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 223,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
46594
|
If I Could Only Remember My Name is the debut solo album produced and released on 22 February 1971 by David Crosby which received a rating of D– from The Village Voice and 8.7/10 from Pitchfork.
|
{
"evidence": [
"If I Could Only Remember My Name_sentence_0",
"If I Could Only Remember My Name_sentence_13"
],
"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:\nIf I Could Only Remember My Name is the debut solo album produced and released on 22 February 1971 by David Crosby which received a rating of D– from The Village Voice and 8.7/10 from Pitchfork.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"If I Could Only Remember My Name_sentence_0",
"If I Could Only Remember My Name_sentence_13"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 224,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.841",
"value": "?search="
},
{
"operation": "Now on",
"time": "10.002",
"value": "?search="
},
{
"operation": "search",
"time": "16.437",
"value": "If I Could Only Remember My Name"
},
{
"operation": "Now on",
"time": "17.247",
"value": "If I Could Only Remember My Name"
},
{
"operation": "Highlighting",
"time": "48.31",
"value": "If I Could Only Remember My Name_cell_0_5_1"
},
{
"operation": "Highlighting",
"time": "48.317",
"value": "If I Could Only Remember My Name_section_2"
},
{
"operation": "Highlighting",
"time": "48.324",
"value": "If I Could Only Remember My Name_header_cell_0_1_1"
},
{
"operation": "Highlighting",
"time": "48.326",
"value": "If I Could Only Remember My Name_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "49.73",
"value": "If I Could Only Remember My Name_cell_0_5_0"
},
{
"operation": "Highlighting",
"time": "49.737",
"value": "If I Could Only Remember My Name_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "51.982",
"value": "If I Could Only Remember My Name_cell_0_3_1"
},
{
"operation": "Highlighting",
"time": "52.999",
"value": "If I Could Only Remember My Name_cell_0_3_0"
},
{
"operation": "Highlighting",
"time": "74.349",
"value": "If I Could Only Remember My Name_sentence_0"
},
{
"operation": "Highlighting",
"time": "99.642",
"value": "If I Could Only Remember My Name_cell_0_6_1"
},
{
"operation": "Highlighting deleted",
"time": "101.947",
"value": "If I Could Only Remember My Name_cell_0_6_1"
},
{
"operation": "Highlighting deleted",
"time": "102.5",
"value": "If I Could Only Remember My Name_cell_0_5_1"
},
{
"operation": "Highlighting deleted",
"time": "103.11",
"value": "If I Could Only Remember My Name_cell_0_3_1"
},
{
"operation": "Highlighting deleted",
"time": "103.117",
"value": "If I Could Only Remember My Name_header_cell_0_1_1"
},
{
"operation": "Highlighting deleted",
"time": "104.069",
"value": "If I Could Only Remember My Name_cell_0_5_0"
},
{
"operation": "Highlighting deleted",
"time": "104.617",
"value": "If I Could Only Remember My Name_cell_0_3_0"
},
{
"operation": "Highlighting deleted",
"time": "104.62",
"value": "If I Could Only Remember My Name_section_2"
},
{
"operation": "Highlighting deleted",
"time": "104.624",
"value": "If I Could Only Remember My Name_header_cell_0_1_0"
},
{
"operation": "Highlighting deleted",
"time": "104.627",
"value": "If I Could Only Remember My Name_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "115.375",
"value": "If I Could Only Remember My Name_sentence_13"
},
{
"operation": "finish",
"time": "155.272",
"value": "finish"
}
]
}
|
49325
|
In the film Smile Before Death, Marco has retired to live with Gianna in a luxurious lakeshore villa and his daughter Nancy notifies him of his plan to take over her mother's estate very soon.
|
{
"evidence": [
"Smile Before Death_sentence_4",
"Smile Before Death_sentence_5"
],
"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 the film Smile Before Death, Marco has retired to live with Gianna in a luxurious lakeshore villa and his daughter Nancy notifies him of his plan to take over her mother's estate very soon.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Smile Before Death_sentence_4",
"Smile Before Death_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 225,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.403",
"value": "?search="
},
{
"operation": "search",
"time": "11.947",
"value": "Smile Before Death"
},
{
"operation": "Now on",
"time": "13.219",
"value": "Smile Before Death"
},
{
"operation": "Highlighting",
"time": "196.74",
"value": "Smile Before Death_sentence_4"
},
{
"operation": "Highlighting",
"time": "205.069",
"value": "Smile Before Death_sentence_5"
},
{
"operation": "finish",
"time": "245.588",
"value": "finish"
}
]
}
|
exfever_6123
|
Batman was the last American comic book published by DC Comics.
|
{
"evidence": [
"Kim Basinger",
"Vicki Vale",
"Superhero"
],
"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:\nBatman was the last American comic book published by DC Comics.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Kim Basinger",
"Vicki Vale",
"Superhero"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 226,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
exfever_550
|
Holly Patricia Hunter earned three additional Academy Award nominations for a 1987 Canadian romantic comedy-drama film written, produced, and directed by James L. Brooks.
|
{
"evidence": [
"Holly Hunter",
"Broadcast News (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:\nHolly Patricia Hunter earned three additional Academy Award nominations for a 1987 Canadian romantic comedy-drama film written, produced, and directed by James L. Brooks.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Holly Hunter",
"Broadcast News (film)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 227,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
49204
|
Emydochampsa, a non-mammalian synapsid described in paleontology in 1912, is 300 million years old.
|
{
"evidence": [
"Emydochampsa_sentence_0",
"Synapsid_sentence_8",
"Vertebrate paleontology_sentence_19",
"Vertebrate paleontology_sentence_20"
],
"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:\nEmydochampsa, a non-mammalian synapsid described in paleontology in 1912, is 300 million years old.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Emydochampsa_sentence_0",
"Synapsid_sentence_8",
"Vertebrate paleontology_sentence_19",
"Vertebrate paleontology_sentence_20"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 228,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.075",
"value": "?search="
},
{
"operation": "search",
"time": "12.599",
"value": "Emydochampsa"
},
{
"operation": "Now on",
"time": "13.856",
"value": "Emydochampsa"
},
{
"operation": "Highlighting",
"time": "20.353",
"value": "Emydochampsa_sentence_0"
},
{
"operation": "hyperlink",
"time": "45.734",
"value": "List of therapsids"
},
{
"operation": "Now on",
"time": "46.191",
"value": "List of therapsids"
},
{
"operation": "Page search",
"time": "54.636",
"value": "emdochamp"
},
{
"operation": "search",
"time": "87.772",
"value": "Contains...Emydochampsa, a non-mammalian synapsid described in paleontology in 1912, is 300 million years old."
},
{
"operation": "Now on",
"time": "88.967",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "94.561",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "94.719",
"value": "List of therapsids"
},
{
"operation": "back-button-clicked",
"time": "113.032",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "113.047",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "113.22",
"value": "Emydochampsa"
},
{
"operation": "hyperlink",
"time": "128.754",
"value": "List of therapsids"
},
{
"operation": "Now on",
"time": "129.139",
"value": "List of therapsids"
},
{
"operation": "hyperlink",
"time": "139.453",
"value": "Synapsid"
},
{
"operation": "Now on",
"time": "140.288",
"value": "Synapsid"
},
{
"operation": "Highlighting",
"time": "149.479",
"value": "Synapsid_sentence_8"
},
{
"operation": "Page search",
"time": "154.917",
"value": "1912"
},
{
"operation": "Page search",
"time": "169.914",
"value": "paleo"
},
{
"operation": "hyperlink",
"time": "199.884",
"value": "Vertebrate paleontology"
},
{
"operation": "Now on",
"time": "200.583",
"value": "Vertebrate paleontology"
},
{
"operation": "Page search",
"time": "210.544",
"value": "synap"
},
{
"operation": "Page search",
"time": "237.347",
"value": "describ"
},
{
"operation": "Highlighting",
"time": "253.54",
"value": "Vertebrate paleontology_sentence_19"
},
{
"operation": "Highlighting",
"time": "257.638",
"value": "Vertebrate paleontology_sentence_20"
},
{
"operation": "finish",
"time": "276.181",
"value": "finish"
}
]
}
|
exfever_8689
|
Quartet was the first comedy-drama film by Ronald Harwood.
|
{
"evidence": [
"Quartet (2012 film)",
"Buckinghamshire"
],
"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:\nQuartet was the first comedy-drama film by Ronald Harwood.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Quartet (2012 film)",
"Buckinghamshire"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 229,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
exfever_40
|
Amy Poehler co-founded the improvisational-comedy troupe Upright Citizens Brigade that moved to the most populous city in the United States.
|
{
"evidence": [
"Amy Poehler",
"New York City"
],
"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:\nAmy Poehler co-founded the improvisational-comedy troupe Upright Citizens Brigade that moved to the most populous city in the United States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Amy Poehler",
"New York City"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 230,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
16604
|
In the plot of Fantasy Life characters Yuelia and Noelia must gather people's wishes to save the world, which they fail to do when there aren't enough wishes to complete their task.
|
{
"evidence": [
"Fantasy Life_sentence_10",
"Fantasy Life_sentence_11",
"Fantasy Life_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:\nIn the plot of Fantasy Life characters Yuelia and Noelia must gather people's wishes to save the world, which they fail to do when there aren't enough wishes to complete their task.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Fantasy Life_sentence_10",
"Fantasy Life_sentence_11",
"Fantasy Life_sentence_13"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 231,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.818",
"value": "?search="
},
{
"operation": "search",
"time": "10.458",
"value": "Fantasy Life"
},
{
"operation": "Now on",
"time": "11.344",
"value": "Fantasy Life"
},
{
"operation": "Highlighting",
"time": "70.035",
"value": "Fantasy Life_sentence_10"
},
{
"operation": "Highlighting",
"time": "85.983",
"value": "Fantasy Life_sentence_11"
},
{
"operation": "Highlighting",
"time": "150.987",
"value": "Fantasy Life_sentence_13"
},
{
"operation": "finish",
"time": "184.154",
"value": "finish"
}
]
}
|
39761
|
Mohd Amirul Izwan Yaacob was awarded his FIFA badge in 2012 and refereed some matches in AFC Champions League only.
|
{
"evidence": [
"Mohd Amirul Izwan Yaacob_sentence_5"
],
"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:\nMohd Amirul Izwan Yaacob was awarded his FIFA badge in 2012 and refereed some matches in AFC Champions League only.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Mohd Amirul Izwan Yaacob_sentence_5"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 232,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.334",
"value": "?search="
},
{
"operation": "search",
"time": "5.83",
"value": "Mohd Amirul Izwan Yaacob"
},
{
"operation": "Now on",
"time": "8.148",
"value": "Mohd Amirul Izwan Yaacob"
},
{
"operation": "Highlighting",
"time": "14.001",
"value": "Mohd Amirul Izwan Yaacob_sentence_2"
},
{
"operation": "Highlighting",
"time": "52.886",
"value": "Mohd Amirul Izwan Yaacob_sentence_3"
},
{
"operation": "Highlighting",
"time": "111.447",
"value": "Mohd Amirul Izwan Yaacob_sentence_5"
},
{
"operation": "finish",
"time": "117.284",
"value": "finish"
}
]
}
|
exfever_8045
|
The Atlantic City stars include an American actor.
|
{
"evidence": [
"Atlantic City (1980 film)",
"Burt Lancaster"
],
"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 Atlantic City stars include an American actor.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Atlantic City (1980 film)",
"Burt Lancaster"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 233,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
26001
|
Bob Dylan's Never Ending Tour 2006 had a big show in Kilkenny.
|
{
"evidence": [
"Never Ending Tour_sentence_35",
"Bob Dylan_sentence_591",
"Bob Dylan_sentence_594"
],
"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:\nBob Dylan's Never Ending Tour 2006 had a big show in Kilkenny.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Never Ending Tour_sentence_35",
"Bob Dylan_sentence_591",
"Bob Dylan_sentence_594"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 234,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.608",
"value": "?search="
},
{
"operation": "search",
"time": "15.156",
"value": "Contains...Bob Dylan's Never Ending Tour 2006"
},
{
"operation": "Now on",
"time": "16.558",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "20.411",
"value": "Never Ending Tour Bob Dylan_section_21"
},
{
"operation": "Now on",
"time": "21.762",
"value": "Bob Dylan#Never Ending Tour Bob Dylan section 21"
},
{
"operation": "Page search",
"time": "50.952",
"value": "2006"
},
{
"operation": "hyperlink",
"time": "71.564",
"value": "2 Never Ending Tour Bob Dylan_section_21"
},
{
"operation": "Highlighting",
"time": "114.592",
"value": "Bob Dylan_sentence_591"
},
{
"operation": "Highlighting",
"time": "115.933",
"value": "Bob Dylan_sentence_594"
},
{
"operation": "Highlighting deleted",
"time": "129.71",
"value": "Bob Dylan_sentence_591"
},
{
"operation": "hyperlink",
"time": "131.854",
"value": "Never Ending Tour"
},
{
"operation": "Now on",
"time": "132.386",
"value": "Never Ending Tour"
},
{
"operation": "hyperlink",
"time": "140.572",
"value": "4 Tours Never Ending Tour_section_4"
},
{
"operation": "Page search",
"time": "151.764",
"value": "kilkenny"
},
{
"operation": "Highlighting",
"time": "221.26",
"value": "Never Ending Tour_sentence_35"
},
{
"operation": "back-button-clicked",
"time": "245.755",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "245.979",
"value": "Bob Dylan#Never Ending Tour Bob Dylan section 21"
},
{
"operation": "Highlighting",
"time": "250.984",
"value": "Bob Dylan_sentence_591"
},
{
"operation": "Highlighting deleted",
"time": "254.505",
"value": "Bob Dylan_sentence_591"
},
{
"operation": "Highlighting",
"time": "255.177",
"value": "Bob Dylan_sentence_591"
},
{
"operation": "Highlighting deleted",
"time": "256.364",
"value": "Bob Dylan_sentence_594"
},
{
"operation": "Highlighting",
"time": "257.032",
"value": "Bob Dylan_sentence_594"
},
{
"operation": "finish",
"time": "266.884",
"value": "finish"
}
]
}
|
exfever_7536
|
Plutonium is a radioactive chemical element that normally exhibits six allotropes and four hypothetical charges of an atom if all of its bonds to different atoms are fully ionic.
|
{
"evidence": [
"Plutonium",
"Oxidation 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:\nPlutonium is a radioactive chemical element that normally exhibits six allotropes and four hypothetical charges of an atom if all of its bonds to different atoms are fully ionic.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Plutonium",
"Oxidation state"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 235,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
36072
|
Translations for John 3:16 into Kurdish, which belongs to the Iranian language family, have yet to be done.
|
{
"evidence": [
"Kurdish languages_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:\nTranslations for John 3:16 into Kurdish, which belongs to the Iranian language family, have yet to be done.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Kurdish languages_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 236,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.765",
"value": "?search="
},
{
"operation": "search",
"time": "17.088",
"value": "john 3"
},
{
"operation": "Now on",
"time": "20.764",
"value": "undefined"
},
{
"operation": "search",
"time": "35.568",
"value": "john 3"
},
{
"operation": "Now on",
"time": "36.735",
"value": "undefined"
},
{
"operation": "search",
"time": "97.025",
"value": "john 3"
},
{
"operation": "Now on",
"time": "98.196",
"value": "undefined"
},
{
"operation": "search",
"time": "124.429",
"value": "kurdish"
},
{
"operation": "Now on",
"time": "124.876",
"value": "Kurdish"
},
{
"operation": "hyperlink",
"time": "129.193",
"value": "Kurdish languages"
},
{
"operation": "Now on",
"time": "130.067",
"value": "Kurdish languages"
},
{
"operation": "Highlighting",
"time": "135.939",
"value": "Kurdish languages_sentence_0"
},
{
"operation": "search",
"time": "152.472",
"value": "john 3"
},
{
"operation": "Now on",
"time": "153.452",
"value": "undefined"
},
{
"operation": "search",
"time": "199.283",
"value": "Translations for John 3"
},
{
"operation": "Now on",
"time": "200.392",
"value": "undefined"
},
{
"operation": "finish",
"time": "228.564",
"value": "finish"
}
]
}
|
72192
|
Centrocoris spiniger are green-colored bugs that can reach a length of 30-40 millimeters (1.18–1.57 in).
|
{
"evidence": [
"Centrocoris spiniger_sentence_4",
"Centrocoris spiniger_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:\nCentrocoris spiniger are green-colored bugs that can reach a length of 30-40 millimeters (1.18–1.57 in).\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Centrocoris spiniger_sentence_4",
"Centrocoris spiniger_sentence_3"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 237,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.926",
"value": "?search="
},
{
"operation": "search",
"time": "21.097",
"value": "Centrocoris spiniger"
},
{
"operation": "Now on",
"time": "27.27",
"value": "Centrocoris spiniger"
},
{
"operation": "Highlighting",
"time": "86.384",
"value": "Centrocoris spiniger_sentence_4"
},
{
"operation": "Highlighting",
"time": "109.904",
"value": "Centrocoris spiniger_sentence_3"
},
{
"operation": "finish",
"time": "186.487",
"value": "finish"
}
]
}
|
exfever_1857
|
British Airways is the third largest UK based carrier, behind a British multinational low-cost airline group.
|
{
"evidence": [
"British Airways",
"EasyJet"
],
"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:\nBritish Airways is the third largest UK based carrier, behind a British multinational low-cost airline group.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"British Airways",
"EasyJet"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 238,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
86464
|
Darts fell from popularity in the 1990s after a boom in the late 1980s.
|
{
"evidence": [
"Darts_sentence_1"
],
"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:\nDarts fell from popularity in the 1990s after a boom in the late 1980s.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Darts_sentence_1"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 239,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.746",
"value": "?search="
},
{
"operation": "search",
"time": "42.189",
"value": "Darts"
},
{
"operation": "Now on",
"time": "43.937",
"value": "Darts"
},
{
"operation": "search",
"time": "83.495",
"value": "Contains...Darts fell from popularity in the 1990s"
},
{
"operation": "Now on",
"time": "85.861",
"value": "undefined"
},
{
"operation": "search",
"time": "102.356",
"value": "Darts"
},
{
"operation": "Now on",
"time": "103.345",
"value": "Darts"
},
{
"operation": "search",
"time": "282.757",
"value": "Contains...after a boom in the late 1980s"
},
{
"operation": "Now on",
"time": "287.147",
"value": "undefined"
},
{
"operation": "search",
"time": "380.602",
"value": "Contains...fell from popularity in the 1990s"
},
{
"operation": "Now on",
"time": "382.86",
"value": "undefined"
},
{
"operation": "search",
"time": "431.94",
"value": "darts popularity"
},
{
"operation": "Now on",
"time": "433.07",
"value": "undefined"
},
{
"operation": "search",
"time": "469.134",
"value": "Contains...darts"
},
{
"operation": "Now on",
"time": "469.988",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "479.739",
"value": "BDO World Darts Championship"
},
{
"operation": "Now on",
"time": "481.996",
"value": "BDO World Darts Championship"
},
{
"operation": "search",
"time": "498.567",
"value": "Darts"
},
{
"operation": "Now on",
"time": "499.415",
"value": "Darts"
},
{
"operation": "Highlighting",
"time": "508.434",
"value": "Darts_sentence_1"
},
{
"operation": "finish",
"time": "516.827",
"value": "finish"
}
]
}
|
exfever_5676
|
Aluminium is a chemical element with the symbol Al and has nuclear charges 13, the charge number of an atomic nucleus.
|
{
"evidence": [
"Aluminium",
"Atomic number"
],
"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:\nAluminium is a chemical element with the symbol Al and has nuclear charges 13, the charge number of an atomic nucleus.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Aluminium",
"Atomic number"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 240,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
exfever_2149
|
The Walking Dead is the longest American post-apocalyptic comic book series.
|
{
"evidence": [
"The Walking Dead (comic book)",
"Image Comics"
],
"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 Walking Dead is the longest American post-apocalyptic comic book series.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Walking Dead (comic book)",
"Image Comics"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 241,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
19637
|
Boeung Kak was leased to Shukaku Inc for 99 years by the Cambodian government in 2007.
|
{
"evidence": [
"Boeung Kak_sentence_5"
],
"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:\nBoeung Kak was leased to Shukaku Inc for 99 years by the Cambodian government in 2007.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Boeung Kak_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 242,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.595",
"value": "?search="
},
{
"operation": "search",
"time": "11.472",
"value": "Boeung Kak"
},
{
"operation": "Now on",
"time": "12.414",
"value": "Boeung Kak"
},
{
"operation": "Page search",
"time": "151.048",
"value": "Shukaku"
},
{
"operation": "Highlighting",
"time": "167.524",
"value": "Boeung Kak_sentence_5"
},
{
"operation": "finish",
"time": "255.649",
"value": "finish"
}
]
}
|
72886
|
Anna Epps first worked as a technologist in the clinical laboratories of Our Lady of Mercy Hospital and acting head of the medical technology department at Arkansas University.
|
{
"evidence": [
"Anna Epps_sentence_15"
],
"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:\nAnna Epps first worked as a technologist in the clinical laboratories of Our Lady of Mercy Hospital and acting head of the medical technology department at Arkansas University.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Anna Epps_sentence_15"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 243,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.385",
"value": "?search="
},
{
"operation": "search",
"time": "66.76",
"value": "Anna Epps"
},
{
"operation": "Now on",
"time": "67.738",
"value": "Anna Epps"
},
{
"operation": "Highlighting",
"time": "81.766",
"value": "Anna Epps_sentence_14"
},
{
"operation": "Highlighting",
"time": "90.721",
"value": "Anna Epps_sentence_15"
},
{
"operation": "Highlighting deleted",
"time": "95.004",
"value": "Anna Epps_sentence_14"
},
{
"operation": "finish",
"time": "120.468",
"value": "finish"
}
]
}
|
70115
|
The census of the literacy in CD Blocks of Bardhaman district shows that the highest literacy percentage is in Purbasthali I – (Kalna subdivision), while the lowest literacy percentage in Baraban (Asansol subdivision).
|
{
"evidence": [
"List of West Bengal districts ranked by literacy rate_sentence_1"
],
"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 census of the literacy in CD Blocks of Bardhaman district shows that the highest literacy percentage is in Purbasthali I – (Kalna subdivision), while the lowest literacy percentage in Baraban (Asansol subdivision).\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"List of West Bengal districts ranked by literacy rate_sentence_1"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 244,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.849",
"value": "?search="
},
{
"operation": "search",
"time": "94.76",
"value": "Contains...census of the literacy in CD"
},
{
"operation": "Now on",
"time": "97.015",
"value": "undefined"
},
{
"operation": "search",
"time": "126.614",
"value": "Contains...census of the literacy in CD Blocks"
},
{
"operation": "Now on",
"time": "127.907",
"value": "undefined"
},
{
"operation": "search",
"time": "164.477",
"value": "Bardhaman district"
},
{
"operation": "Now on",
"time": "165.701",
"value": "Bardhaman district"
},
{
"operation": "hyperlink",
"time": "219.776",
"value": "2011 census"
},
{
"operation": "Now on",
"time": "225.24",
"value": "2011 Census of India"
},
{
"operation": "hyperlink",
"time": "228.322",
"value": "7 Literacy 2011 Census of India_section_9"
},
{
"operation": "Highlighting",
"time": "259.643",
"value": "2011 Census of India_cell_3_5_1"
},
{
"operation": "Highlighting",
"time": "259.644",
"value": "2011 Census of India_title"
},
{
"operation": "Highlighting",
"time": "259.644",
"value": "2011 Census of India_section_9"
},
{
"operation": "Highlighting",
"time": "259.645",
"value": "2011 Census of India_header_cell_3_0_1"
},
{
"operation": "Highlighting deleted",
"time": "260.179",
"value": "2011 Census of India_cell_3_5_1"
},
{
"operation": "Highlighting deleted",
"time": "260.18",
"value": "2011 Census of India_title"
},
{
"operation": "Highlighting deleted",
"time": "260.182",
"value": "2011 Census of India_section_9"
},
{
"operation": "Highlighting deleted",
"time": "260.183",
"value": "2011 Census of India_header_cell_3_0_1"
},
{
"operation": "Highlighting",
"time": "352.961",
"value": "2011 Census of India_item_1_0"
},
{
"operation": "Highlighting",
"time": "352.961",
"value": "2011 Census of India_title"
},
{
"operation": "Highlighting",
"time": "352.962",
"value": "2011 Census of India_section_9"
},
{
"operation": "Highlighting deleted",
"time": "353.808",
"value": "2011 Census of India_item_1_0"
},
{
"operation": "Highlighting deleted",
"time": "353.809",
"value": "2011 Census of India_title"
},
{
"operation": "Highlighting deleted",
"time": "353.811",
"value": "2011 Census of India_section_9"
},
{
"operation": "search",
"time": "370.944",
"value": "Bardhaman district literacy"
},
{
"operation": "Now on",
"time": "373.023",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "377.824",
"value": "Literacy and education Birbhum district_section_15"
},
{
"operation": "Now on",
"time": "384.634",
"value": "Birbhum district#Literacy and education Birbhum district section 15"
},
{
"operation": "back-button-clicked",
"time": "407.076",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "407.887",
"value": "undefined"
},
{
"operation": "search",
"time": "470.172",
"value": "Contains... literacy in CD Blocks of Bardhaman"
},
{
"operation": "Now on",
"time": "471.926",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "535.58",
"value": "Literacy Purbasthali II_section_3"
},
{
"operation": "Now on",
"time": "538.434",
"value": "Purbasthali II#Literacy Purbasthali II section 3"
},
{
"operation": "hyperlink",
"time": "543.046",
"value": "Purba Bardhaman district"
},
{
"operation": "Now on",
"time": "544.691",
"value": "Purba Bardhaman district"
},
{
"operation": "hyperlink",
"time": "609.818",
"value": "Kalna"
},
{
"operation": "Now on",
"time": "611.58",
"value": "Kalna subdivision"
},
{
"operation": "hyperlink",
"time": "645.424",
"value": "List of West Bengal districts ranked by literacy rate"
},
{
"operation": "Now on",
"time": "647.585",
"value": "List of West Bengal districts ranked by literacy rate"
},
{
"operation": "hyperlink",
"time": "651.929",
"value": "Indian states ranking by literacy rate"
},
{
"operation": "Now on",
"time": "658.658",
"value": "List of Indian states and union territories by literacy rate"
},
{
"operation": "Highlighting",
"time": "669.229",
"value": "List of Indian states and union territories by literacy rate_cell_0_35_0"
},
{
"operation": "Highlighting",
"time": "669.231",
"value": "List of Indian states and union territories by literacy rate_title"
},
{
"operation": "Highlighting",
"time": "669.232",
"value": "List of Indian states and union territories by literacy rate_section_1"
},
{
"operation": "Highlighting",
"time": "669.27",
"value": "List of Indian states and union territories by literacy rate_header_cell_0_0_0"
},
{
"operation": "hyperlink",
"time": "670.419",
"value": "West Bengal"
},
{
"operation": "Now on",
"time": "673.445",
"value": "West Bengal"
},
{
"operation": "hyperlink",
"time": "703.42",
"value": "Paschim Bardhaman"
},
{
"operation": "Now on",
"time": "708.038",
"value": "Paschim Bardhaman district"
},
{
"operation": "back-button-clicked",
"time": "729.632",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "729.84",
"value": "West Bengal"
},
{
"operation": "back-button-clicked",
"time": "730.539",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "730.545",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "733.717",
"value": "List of Indian states and union territories by literacy rate"
},
{
"operation": "back-button-clicked",
"time": "741.401",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "741.406",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "741.488",
"value": "List of West Bengal districts ranked by literacy rate"
},
{
"operation": "search",
"time": "778.006",
"value": "Contains... census of the literacy in CD Blocks of Bardhaman district"
},
{
"operation": "Now on",
"time": "781.173",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "838.442",
"value": "next 20"
},
{
"operation": "Now on",
"time": "839.551",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "856.094",
"value": "Alphabetical listing of West Bengal Districts List of districts of West Bengal_section_3"
},
{
"operation": "Now on",
"time": "862.779",
"value": "List of districts of West Bengal#Alphabetical listing of West Bengal Districts List of districts of West Bengal section 3"
},
{
"operation": "hyperlink",
"time": "880.533",
"value": "Bardhaman"
},
{
"operation": "Now on",
"time": "880.85",
"value": "Bardhaman district"
},
{
"operation": "hyperlink",
"time": "910.653",
"value": "List of West Bengal districts ranked by literacy rate"
},
{
"operation": "Now on",
"time": "911.162",
"value": "List of West Bengal districts ranked by literacy rate"
},
{
"operation": "Highlighting",
"time": "915.271",
"value": "List of West Bengal districts ranked by literacy rate_sentence_1"
},
{
"operation": "finish",
"time": "942.84",
"value": "finish"
}
]
}
|
exfever_3481
|
Istanbul is the city that has the most Heritage Sites listed by UNESCO.
|
{
"evidence": [
"Istanbul",
"Strait"
],
"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:\nIstanbul is the city that has the most Heritage Sites listed by UNESCO.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Istanbul",
"Strait"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 245,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
13823
|
Kolla Ashok Kumar, whose first film was the political thriller Bharat Bandh, appeared in print ads between 1991 and 2012.
|
{
"evidence": [
"Kolla Ashok Kumar_sentence_3"
],
"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:\nKolla Ashok Kumar, whose first film was the political thriller Bharat Bandh, appeared in print ads between 1991 and 2012.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Kolla Ashok Kumar_sentence_3"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 246,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.948",
"value": "?search="
},
{
"operation": "search",
"time": "10.978",
"value": "Kolla Ashok Kumar"
},
{
"operation": "Now on",
"time": "11.856",
"value": "Kolla Ashok Kumar"
},
{
"operation": "Highlighting",
"time": "17.447",
"value": "Kolla Ashok Kumar_sentence_3"
},
{
"operation": "search",
"time": "66.092",
"value": "Kolla Ashok Kumar print ads"
},
{
"operation": "Now on",
"time": "67.416",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "75.307",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "75.418",
"value": "Kolla Ashok Kumar"
},
{
"operation": "hyperlink",
"time": "89.965",
"value": "Bharat Bandh"
},
{
"operation": "Now on",
"time": "90.495",
"value": "Bharat Bandh"
},
{
"operation": "back-button-clicked",
"time": "117.753",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "117.804",
"value": "Kolla Ashok Kumar"
},
{
"operation": "finish",
"time": "142.199",
"value": "finish"
}
]
}
|
72576
|
Thomas Muntzer opposed Martin Luther, distributing pamphlets in 1523 despite not being officially appointed within the church; he may be partially responsible for the creation of the first document regarding human rights after the Roman Empire.
|
{
"evidence": [
"Thomas Müntzer_sentence_0",
"Thomas Müntzer_sentence_95"
],
"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:\nThomas Muntzer opposed Martin Luther, distributing pamphlets in 1523 despite not being officially appointed within the church; he may be partially responsible for the creation of the first document regarding human rights after the Roman Empire.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Thomas Müntzer_sentence_0",
"Thomas Müntzer_sentence_95"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 247,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.13",
"value": "?search="
},
{
"operation": "search",
"time": "12.123",
"value": "Thomas Müntzer"
},
{
"operation": "Now on",
"time": "16.239",
"value": "Thomas Müntzer"
},
{
"operation": "Highlighting",
"time": "21.279",
"value": "Thomas Müntzer_sentence_0"
},
{
"operation": "Highlighting",
"time": "102.334",
"value": "Thomas Müntzer_sentence_95"
},
{
"operation": "Page search",
"time": "110.308",
"value": "human rights"
},
{
"operation": "hyperlink",
"time": "138.017",
"value": "Martin Luther"
},
{
"operation": "Now on",
"time": "140.365",
"value": "Martin Luther"
},
{
"operation": "back-button-clicked",
"time": "160.868",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "161.03",
"value": "Thomas Müntzer"
},
{
"operation": "search",
"time": "206.876",
"value": "distributing pamphlets in 1523"
},
{
"operation": "Now on",
"time": "208.219",
"value": "undefined"
},
{
"operation": "search",
"time": "231.664",
"value": "creation of the first document regarding human rights"
},
{
"operation": "Now on",
"time": "235.029",
"value": "undefined"
},
{
"operation": "finish",
"time": "263.055",
"value": "finish"
}
]
}
|
exfever_8012
|
Stanley Tucci is an actor who starred in The Terminal by an American director who received an?annual honor given to those in the performing arts for their lifetime of contributions to American culture.
|
{
"evidence": [
"Stanley Tucci",
"Steven Spielberg",
"Kennedy Center Honors"
],
"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:\nStanley Tucci is an actor who starred in The Terminal by an American director who received an?annual honor given to those in the performing arts for their lifetime of contributions to American culture.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Stanley Tucci",
"Steven Spielberg",
"Kennedy Center Honors"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 248,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
66127
|
Lars Fletre was the youngest of seven children born to Ola Oddson Fletre and Kristine Olsdatter ytre Kvårmo and married Helen Svensson Fletre, a Swedish-born, American author and journalist.
|
{
"evidence": [
"Lars Fletre_sentence_2",
"Lars Fletre_sentence_4",
"Helen Svensson Fletre_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:\nLars Fletre was the youngest of seven children born to Ola Oddson Fletre and Kristine Olsdatter ytre Kvårmo and married Helen Svensson Fletre, a Swedish-born, American author and journalist.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Lars Fletre_sentence_2",
"Lars Fletre_sentence_4",
"Helen Svensson Fletre_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 249,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "10.651",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "12.711",
"value": "Lars Fletre"
},
{
"operation": "Now on",
"time": "16.011",
"value": "Lars Fletre"
},
{
"operation": "Highlighting",
"time": "22.724",
"value": "Lars Fletre_sentence_2"
},
{
"operation": "Highlighting",
"time": "27.859",
"value": "Lars Fletre_sentence_4"
},
{
"operation": "hyperlink",
"time": "30.628",
"value": "Helen Svensson Fletre"
},
{
"operation": "Now on",
"time": "33.967",
"value": "Helen Svensson Fletre"
},
{
"operation": "Highlighting",
"time": "36.564",
"value": "Helen Svensson Fletre_sentence_0"
},
{
"operation": "finish",
"time": "40.323",
"value": "finish"
}
]
}
|
exfever_5031
|
Tennis is an Olympic racket sport that can be played against a single or doubles opponent on a firm square surface with a low net stretched across the center.
|
{
"evidence": [
"Tennis",
"Tennis court"
],
"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:\nTennis is an Olympic racket sport that can be played against a single or doubles opponent on a firm square surface with a low net stretched across the center.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Tennis",
"Tennis court"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 250,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
exfever_8725
|
Wilt Chamberlain was an American professional basketball player who joined the NBA after playing for the intercollegiate men's basketball program of 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 joined the NBA after playing for the intercollegiate men's basketball program of the University of Kansas.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Wilt Chamberlain",
"Kansas Jayhawks men's basketball"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 251,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
72190
|
Umpire Harry Wendelstedt waved off the game after a 90-minute rain delay after the game was stopped by a steady rain with one out in the top of the sixth, however, due to a statistical rule change in 1991, no-hitters must last at least nine innings to count; thus, this game and twenty-five others are no longer considered no-hitters.
|
{
"evidence": [
"1988 Philadelphia Phillies season_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:\nUmpire Harry Wendelstedt waved off the game after a 90-minute rain delay after the game was stopped by a steady rain with one out in the top of the sixth, however, due to a statistical rule change in 1991, no-hitters must last at least nine innings to count; thus, this game and twenty-five others are no longer considered no-hitters.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"1988 Philadelphia Phillies season_sentence_12"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 252,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.17",
"value": "?search="
},
{
"operation": "search",
"time": "125.34",
"value": "1988 Philadelphia Phillies season"
},
{
"operation": "Now on",
"time": "128.751",
"value": "1988 Philadelphia Phillies season"
},
{
"operation": "Highlighting",
"time": "143.735",
"value": "1988 Philadelphia Phillies season_sentence_10"
},
{
"operation": "Highlighting",
"time": "156.799",
"value": "1988 Philadelphia Phillies season_sentence_11"
},
{
"operation": "Highlighting",
"time": "164.047",
"value": "1988 Philadelphia Phillies season_sentence_12"
},
{
"operation": "Highlighting deleted",
"time": "168.615",
"value": "1988 Philadelphia Phillies season_sentence_10"
},
{
"operation": "Highlighting deleted",
"time": "169.247",
"value": "1988 Philadelphia Phillies season_sentence_11"
},
{
"operation": "finish",
"time": "175.925",
"value": "finish"
}
]
}
|
53741
|
Elegast exposes Eggeric as a traitor and kills him, then marries Eggeric's wife, in Karel ende Elegast, a lengthy narrative poem.
|
{
"evidence": [
"Elegast_sentence_16",
"Elegast_sentence_17",
"Elegast_sentence_0",
"Epic poetry_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:\nElegast exposes Eggeric as a traitor and kills him, then marries Eggeric's wife, in Karel ende Elegast, a lengthy narrative poem.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Elegast_sentence_16",
"Elegast_sentence_17",
"Elegast_sentence_0",
"Epic poetry_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 253,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.035",
"value": "?search="
},
{
"operation": "search",
"time": "10.5",
"value": "Elegast"
},
{
"operation": "Now on",
"time": "11.058",
"value": "Elegast"
},
{
"operation": "search",
"time": "40.901",
"value": "Karel"
},
{
"operation": "Now on",
"time": "41.422",
"value": "Karel"
},
{
"operation": "back-button-clicked",
"time": "42.047",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "42.111",
"value": "Elegast"
},
{
"operation": "Highlighting",
"time": "43.536",
"value": "Elegast_sentence_0"
},
{
"operation": "Page search",
"time": "58.845",
"value": "narrative"
},
{
"operation": "Page search",
"time": "68.069",
"value": "length"
},
{
"operation": "search",
"time": "76.117",
"value": "Karel ende Elegast"
},
{
"operation": "Now on",
"time": "76.645",
"value": "Elegast"
},
{
"operation": "hyperlink",
"time": "89.18",
"value": "epic poem"
},
{
"operation": "Now on",
"time": "89.73",
"value": "Epic poetry"
},
{
"operation": "Highlighting",
"time": "91.247",
"value": "Epic poetry_sentence_1"
},
{
"operation": "back-button-clicked",
"time": "97.675",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "97.735",
"value": "Elegast"
},
{
"operation": "Page search",
"time": "109.752",
"value": "Eggeric"
},
{
"operation": "Highlighting",
"time": "131.227",
"value": "Elegast_sentence_13"
},
{
"operation": "Highlighting",
"time": "135.302",
"value": "Elegast_sentence_14"
},
{
"operation": "Highlighting",
"time": "144.065",
"value": "Elegast_sentence_16"
},
{
"operation": "Highlighting deleted",
"time": "146.263",
"value": "Elegast_sentence_14"
},
{
"operation": "Highlighting deleted",
"time": "146.888",
"value": "Elegast_sentence_13"
},
{
"operation": "Highlighting",
"time": "153.963",
"value": "Elegast_sentence_17"
},
{
"operation": "Highlighting",
"time": "371.96",
"value": "Elegast_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "373.898",
"value": "Elegast_sentence_0"
},
{
"operation": "Highlighting",
"time": "374.87",
"value": "Elegast_sentence_0"
},
{
"operation": "hyperlink",
"time": "375.662",
"value": "epic poem"
},
{
"operation": "Now on",
"time": "376.396",
"value": "Epic poetry"
},
{
"operation": "Highlighting",
"time": "377.927",
"value": "Epic poetry_sentence_1"
},
{
"operation": "finish",
"time": "394.824",
"value": "finish"
}
]
}
|
exfever_2219
|
Jon Watts directed and co-wrote the 2014 American horror film, which was produced by Mac Cappuccino, Eli Roth, and Cody Ryder.
|
{
"evidence": [
"Jon Watts",
"Clown (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:\nJon Watts directed and co-wrote the 2014 American horror film, which was produced by Mac Cappuccino, Eli Roth, and Cody Ryder.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jon Watts",
"Clown (film)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 254,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
21963
|
PP1 is involved in regulation of various cellular processes and cardiac function but has never been visible in complex health conditions.
|
{
"evidence": [
"Protein phosphatase 1_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:\nPP1 is involved in regulation of various cellular processes and cardiac function but has never been visible in complex health conditions.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Protein phosphatase 1_sentence_4"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 255,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.987",
"value": "?search="
},
{
"operation": "search",
"time": "9.361",
"value": "PP1"
},
{
"operation": "Now on",
"time": "10.318",
"value": "Protein phosphatase 1"
},
{
"operation": "Page search",
"time": "27.921",
"value": "cardiac"
},
{
"operation": "Page search",
"time": "40.304",
"value": "cellular process"
},
{
"operation": "Page search",
"time": "54.392",
"value": "regulation"
},
{
"operation": "Highlighting",
"time": "141.75",
"value": "Protein phosphatase 1_sentence_4"
},
{
"operation": "search",
"time": "151.355",
"value": "Cellular process"
},
{
"operation": "Now on",
"time": "152.599",
"value": "Cell (biology)"
},
{
"operation": "search",
"time": "165.426",
"value": "pp1"
},
{
"operation": "Now on",
"time": "165.958",
"value": "Pp1"
},
{
"operation": "back-button-clicked",
"time": "172.087",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "173.449",
"value": "Cell (biology)"
},
{
"operation": "Page search",
"time": "175.891",
"value": "pp1"
},
{
"operation": "Page search",
"time": "177.053",
"value": "pp1"
},
{
"operation": "hyperlink",
"time": "243.753",
"value": "Prokaryote"
},
{
"operation": "Now on",
"time": "244.524",
"value": "Prokaryote"
},
{
"operation": "back-button-clicked",
"time": "246.309",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "246.423",
"value": "Cell (biology)"
},
{
"operation": "back-button-clicked",
"time": "247.654",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "247.658",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "247.662",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "248.492",
"value": "Protein phosphatase 1"
},
{
"operation": "Highlighting deleted",
"time": "253.972",
"value": "Protein phosphatase 1_sentence_4"
},
{
"operation": "Page search",
"time": "262.21",
"value": "complex health conditions"
},
{
"operation": "Page search",
"time": "263.418",
"value": "complex health conditions"
},
{
"operation": "Page search",
"time": "287.16",
"value": "function"
},
{
"operation": "hyperlink",
"time": "289.188",
"value": "3 Biological function Protein phosphatase 1_section_3"
},
{
"operation": "hyperlink",
"time": "352.958",
"value": "Pp1"
},
{
"operation": "Now on",
"time": "353.591",
"value": "Pp1"
},
{
"operation": "hyperlink",
"time": "361.788",
"value": "Ribonuclease PP1"
},
{
"operation": "Now on",
"time": "362.353",
"value": "Ribonuclease T1"
},
{
"operation": "back-button-clicked",
"time": "378.851",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "378.895",
"value": "Pp1"
},
{
"operation": "back-button-clicked",
"time": "392.412",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "392.418",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "392.489",
"value": "Protein phosphatase 1#Biological function Protein phosphatase 1 section 3"
},
{
"operation": "Page search",
"time": "436.494",
"value": "card"
},
{
"operation": "hyperlink",
"time": "466.71",
"value": "Pp1"
},
{
"operation": "Now on",
"time": "467.137",
"value": "Pp1 (disambiguation)"
},
{
"operation": "hyperlink",
"time": "468.693",
"value": "1PP"
},
{
"operation": "Now on",
"time": "469.137",
"value": "1 Police Plaza"
},
{
"operation": "back-button-clicked",
"time": "472.587",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "472.633",
"value": "Pp1"
},
{
"operation": "hyperlink",
"time": "473.846",
"value": "1-PP"
},
{
"operation": "Now on",
"time": "474.287",
"value": "Pyrimidinylpiperazine"
},
{
"operation": "back-button-clicked",
"time": "487.611",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "487.659",
"value": "Pp1"
},
{
"operation": "back-button-clicked",
"time": "492.822",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "492.827",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "492.832",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "492.903",
"value": "Protein phosphatase 1#Biological function Protein phosphatase 1 section 3"
},
{
"operation": "hyperlink",
"time": "517.297",
"value": "muscle contraction"
},
{
"operation": "Now on",
"time": "518.46",
"value": "Muscle contraction"
},
{
"operation": "back-button-clicked",
"time": "546.391",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "546.463",
"value": "Protein phosphatase 1#Biological function Protein phosphatase 1 section 3"
},
{
"operation": "hyperlink",
"time": "561.411",
"value": "protein serine/threonine phosphatases"
},
{
"operation": "Now on",
"time": "561.833",
"value": "Threonine phosphatase"
},
{
"operation": "back-button-clicked",
"time": "576.648",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "576.718",
"value": "Protein phosphatase 1#Biological function Protein phosphatase 1 section 3"
},
{
"operation": "search",
"time": "624.718",
"value": "visible in complex health conditions."
},
{
"operation": "Now on",
"time": "626.989",
"value": "undefined"
},
{
"operation": "search",
"time": "634.656",
"value": "visible in complex health conditions"
},
{
"operation": "Now on",
"time": "635.669",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "664.996",
"value": "next 20"
},
{
"operation": "Now on",
"time": "666.2",
"value": "undefined"
},
{
"operation": "search",
"time": "678.006",
"value": "regulation of various cellular processes"
},
{
"operation": "Now on",
"time": "680.22",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "709.603",
"value": "Adenosine triphosphate"
},
{
"operation": "Now on",
"time": "710.744",
"value": "Adenosine triphosphate"
},
{
"operation": "Page search",
"time": "717.141",
"value": "cardi"
},
{
"operation": "Page search",
"time": "731.524",
"value": "cardiac function"
},
{
"operation": "search",
"time": "783.266",
"value": "cellular proces"
},
{
"operation": "Now on",
"time": "783.638",
"value": "undefined"
},
{
"operation": "search",
"time": "789.481",
"value": "Cellular process"
},
{
"operation": "Now on",
"time": "789.957",
"value": "Cellular process"
},
{
"operation": "Page search",
"time": "800.679",
"value": "various"
},
{
"operation": "Page search",
"time": "801.264",
"value": "various"
},
{
"operation": "Page search",
"time": "822.558",
"value": "regulation"
},
{
"operation": "Page search",
"time": "842.294",
"value": "regulate"
},
{
"operation": "hyperlink",
"time": "895.798",
"value": "Prokaryotic"
},
{
"operation": "Now on",
"time": "896.613",
"value": "Prokaryote"
},
{
"operation": "search",
"time": "1007.112",
"value": "PP1"
},
{
"operation": "Now on",
"time": "1008.104",
"value": "PP1"
},
{
"operation": "hyperlink",
"time": "1014.173",
"value": "protein serine/threonine phosphatases"
},
{
"operation": "Now on",
"time": "1014.688",
"value": "Threonine phosphatase"
},
{
"operation": "Page search",
"time": "1026.622",
"value": "card"
},
{
"operation": "Page search",
"time": "1026.96",
"value": "card"
},
{
"operation": "hyperlink",
"time": "1030.618",
"value": "PPP1"
},
{
"operation": "Now on",
"time": "1031.554",
"value": "Protein phosphatase 1"
},
{
"operation": "back-button-clicked",
"time": "1044.979",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "1045.04",
"value": "Threonine phosphatase"
},
{
"operation": "hyperlink",
"time": "1047.471",
"value": "PPP2"
},
{
"operation": "Now on",
"time": "1048.021",
"value": "Protein phosphatase 2"
},
{
"operation": "back-button-clicked",
"time": "1057.617",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "1057.661",
"value": "Threonine phosphatase"
},
{
"operation": "hyperlink",
"time": "1059.831",
"value": "PPP3"
},
{
"operation": "Now on",
"time": "1060.264",
"value": "Calcineurin"
},
{
"operation": "Page search",
"time": "1074.524",
"value": "cardiac"
},
{
"operation": "Page search",
"time": "1098.811",
"value": "complex"
},
{
"operation": "back-button-clicked",
"time": "1102.001",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "1102.056",
"value": "Threonine phosphatase"
},
{
"operation": "hyperlink",
"time": "1104.178",
"value": "PPP4"
},
{
"operation": "Now on",
"time": "1104.487",
"value": "PPP4C"
},
{
"operation": "back-button-clicked",
"time": "1108.072",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "1108.118",
"value": "Threonine phosphatase"
},
{
"operation": "hyperlink",
"time": "1109.315",
"value": "PPP5"
},
{
"operation": "Now on",
"time": "1109.731",
"value": "PPP5C"
},
{
"operation": "back-button-clicked",
"time": "1123.448",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "1123.496",
"value": "Threonine phosphatase"
},
{
"operation": "hyperlink",
"time": "1125",
"value": "PPP6"
},
{
"operation": "Now on",
"time": "1125.327",
"value": "PPP6C"
},
{
"operation": "back-button-clicked",
"time": "1129.044",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "1129.093",
"value": "Threonine phosphatase"
},
{
"operation": "hyperlink",
"time": "1135.975",
"value": "PPP1"
},
{
"operation": "Now on",
"time": "1136.297",
"value": "Protein phosphatase 1"
},
{
"operation": "Page search",
"time": "1142.227",
"value": "complex"
},
{
"operation": "Page search",
"time": "1172.345",
"value": "visible"
},
{
"operation": "Highlighting",
"time": "1256.629",
"value": "Protein phosphatase 1_sentence_4"
},
{
"operation": "finish",
"time": "1264.429",
"value": "finish"
}
]
}
|
23011
|
Look Communications was an energy supplier company founded in Ontario in 1997.
|
{
"evidence": [
"Look Communications_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:\nLook Communications was an energy supplier company founded in Ontario in 1997.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Look Communications_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 256,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.627",
"value": "?search="
},
{
"operation": "search",
"time": "6.624",
"value": "Look Communications"
},
{
"operation": "Now on",
"time": "7.531",
"value": "Look Communications"
},
{
"operation": "Highlighting",
"time": "14.486",
"value": "Look Communications_cell_0_2_1"
},
{
"operation": "Highlighting",
"time": "14.492",
"value": "Look Communications_header_cell_0_2_0"
},
{
"operation": "Highlighting",
"time": "26.715",
"value": "Look Communications_cell_0_3_1"
},
{
"operation": "Highlighting",
"time": "26.722",
"value": "Look Communications_header_cell_0_3_0"
},
{
"operation": "Highlighting",
"time": "38.797",
"value": "Look Communications_sentence_0"
},
{
"operation": "finish",
"time": "61.852",
"value": "finish"
}
]
}
|
57452
|
Aeritalia and Aermacchi produced a joint proposal for the Italian Air Force's requirement for 187 new strike fighters to replace old G91s rather than compete for the contract, and in April 1978 work on the joint venture formally commenced.
|
{
"evidence": [
"AMX International AMX_sentence_8",
"AMX International AMX_sentence_4",
"AMX International AMX_sentence_10"
],
"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:\nAeritalia and Aermacchi produced a joint proposal for the Italian Air Force's requirement for 187 new strike fighters to replace old G91s rather than compete for the contract, and in April 1978 work on the joint venture formally commenced.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"AMX International AMX_sentence_8",
"AMX International AMX_sentence_4",
"AMX International AMX_sentence_10"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 257,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.889",
"value": "undefined"
},
{
"operation": "search",
"time": "51.845",
"value": "Aeritalia and Aermacchi"
},
{
"operation": "Now on",
"time": "52.806",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "58.505",
"value": "AMX International AMX"
},
{
"operation": "Now on",
"time": "59.947",
"value": "AMX International AMX"
},
{
"operation": "Page search",
"time": "71.443",
"value": "aeritalia"
},
{
"operation": "Highlighting",
"time": "93.389",
"value": "AMX International AMX_sentence_8"
},
{
"operation": "Page search",
"time": "119.517",
"value": "187"
},
{
"operation": "Highlighting",
"time": "127.211",
"value": "AMX International AMX_sentence_4"
},
{
"operation": "Page search",
"time": "187.531",
"value": "compete"
},
{
"operation": "Page search",
"time": "209.51",
"value": "contract"
},
{
"operation": "Page search",
"time": "248.296",
"value": "1978"
},
{
"operation": "Page search",
"time": "248.747",
"value": "1978"
},
{
"operation": "Page search",
"time": "249.521",
"value": "1978"
},
{
"operation": "Highlighting",
"time": "258.156",
"value": "AMX International AMX_sentence_10"
},
{
"operation": "finish",
"time": "280.751",
"value": "finish"
}
]
}
|
exfever_5385
|
The Ink Master series was produced by an American production company founded in 2002.
|
{
"evidence": [
"Ink Master",
"Original Media"
],
"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 Ink Master series was produced by an American production company founded in 2002.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Ink Master",
"Original Media"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 258,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
exfever_5990
|
Borussia Dortmund is best known for its men's professional football team, which plays in a professional association football league in Germany.?
|
{
"evidence": [
"Borussia Dortmund",
"Bundesliga"
],
"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:\nBorussia Dortmund is best known for its men's professional football team, which plays in a professional association football league in Germany.?\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Borussia Dortmund",
"Bundesliga"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 259,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
exfever_6186
|
Snowden is a biographical film directed by Oliver Stone also classified as a suspense film that evokes excitement and suspense in the audience.
|
{
"evidence": [
"Snowden (film)",
"Thriller 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:\nSnowden is a biographical film directed by Oliver Stone also classified as a suspense film that evokes excitement and suspense in the audience.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Snowden (film)",
"Thriller film"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 260,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
76536
|
Between 2006 and 2017, Kenya repaid its debts, which was used to build a railway, to China, whose debt accounts of 72% of overall foreign debt.
|
{
"evidence": [
"Debt-trap diplomacy_sentence_64",
"Debt-trap diplomacy_sentence_65",
"Debt-trap diplomacy_sentence_67"
],
"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:\nBetween 2006 and 2017, Kenya repaid its debts, which was used to build a railway, to China, whose debt accounts of 72% of overall foreign debt.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Debt-trap diplomacy_sentence_64",
"Debt-trap diplomacy_sentence_65",
"Debt-trap diplomacy_sentence_67"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 261,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.723",
"value": "?search="
},
{
"operation": "search",
"time": "60.495",
"value": "Debt-trap diplomacy"
},
{
"operation": "Now on",
"time": "61.71",
"value": "Debt-trap diplomacy"
},
{
"operation": "Page search",
"time": "65.738",
"value": "kenya"
},
{
"operation": "page-search-reset",
"time": "76.185",
"value": "page-search-reset"
},
{
"operation": "hyperlink",
"time": "78.731",
"value": "3.3 Kenya Debt-trap diplomacy_section_7"
},
{
"operation": "Highlighting",
"time": "85.144",
"value": "Debt-trap diplomacy_sentence_64"
},
{
"operation": "Highlighting",
"time": "102.44",
"value": "Debt-trap diplomacy_sentence_67"
},
{
"operation": "Highlighting deleted",
"time": "105.275",
"value": "Debt-trap diplomacy_sentence_67"
},
{
"operation": "Highlighting",
"time": "105.959",
"value": "Debt-trap diplomacy_sentence_65"
},
{
"operation": "Highlighting",
"time": "106.504",
"value": "Debt-trap diplomacy_sentence_67"
},
{
"operation": "hyperlink",
"time": "113.785",
"value": "China–Kenya relations"
},
{
"operation": "Now on",
"time": "114.486",
"value": "China–Kenya relations"
},
{
"operation": "Page search",
"time": "129.522",
"value": "repaid"
},
{
"operation": "Page search",
"time": "140.778",
"value": "debt"
},
{
"operation": "finish",
"time": "163.726",
"value": "finish"
}
]
}
|
exfever_5135
|
AC/DC are an Australian rock band formed in Sydney in 1972 by Scottish-born brothers Malcolm and an Australian musician known for his energetic performances, schoolboy-uniform stage outfits, and his version of Chuck Berry's duckwalk.
|
{
"evidence": [
"AC/DC",
"Angus Young"
],
"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:\nAC/DC are an Australian rock band formed in Sydney in 1972 by Scottish-born brothers Malcolm and an Australian musician known for his energetic performances, schoolboy-uniform stage outfits, and his version of Chuck Berry's duckwalk.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"AC/DC",
"Angus Young"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 262,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
50406
|
Nucleoporin 153, a protein which in humans is encoded by the NUP153 gene, is an essential component of the basket of nuclear pore complexes (NPCs- which span the nuclear envelope, which is the double membrane surrounding the eukaryotic cell nucleus) in vertebrates, and acts as the docking site of Karyopherin.
|
{
"evidence": [
"Nucleoporin 153_sentence_0",
"Nuclear pore_sentence_0",
"Nucleoporin 153_sentence_1",
"Nucleoporin 153_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:\nNucleoporin 153, a protein which in humans is encoded by the NUP153 gene, is an essential component of the basket of nuclear pore complexes (NPCs- which span the nuclear envelope, which is the double membrane surrounding the eukaryotic cell nucleus) in vertebrates, and acts as the docking site of Karyopherin.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Nucleoporin 153_sentence_0",
"Nuclear pore_sentence_0",
"Nucleoporin 153_sentence_1",
"Nucleoporin 153_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 263,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.048",
"value": "?search="
},
{
"operation": "search",
"time": "25.512",
"value": "Nucleoporin 153"
},
{
"operation": "Now on",
"time": "26.744",
"value": "Nucleoporin 153"
},
{
"operation": "Highlighting",
"time": "68.233",
"value": "Nucleoporin 153_sentence_0"
},
{
"operation": "hyperlink",
"time": "97.261",
"value": "nuclear pore complexes"
},
{
"operation": "Now on",
"time": "98.29",
"value": "Nuclear pore"
},
{
"operation": "Highlighting",
"time": "104.167",
"value": "Nuclear pore_sentence_0"
},
{
"operation": "back-button-clicked",
"time": "107.405",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "107.525",
"value": "Nucleoporin 153"
},
{
"operation": "Highlighting",
"time": "127.161",
"value": "Nucleoporin 153_sentence_1"
},
{
"operation": "Highlighting",
"time": "129.027",
"value": "Nucleoporin 153_sentence_2"
},
{
"operation": "finish",
"time": "156.678",
"value": "finish"
}
]
}
|
16816
|
The Universal Postal Union Collection has over 93,000 stamps from 1908 and up, this is thanks to Wilmot Corfield who proposed a National Collection Committee to work with the British Museum to create a modern stamp collection.
|
{
"evidence": [
"Universal Postal Union Collection_sentence_0",
"Universal Postal Union Collection_sentence_4",
"Universal Postal Union Collection_sentence_2",
"Universal Postal Union Collection_sentence_5",
"Universal Postal Union Collection_sentence_6",
"Universal Postal Union Collection_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:\nThe Universal Postal Union Collection has over 93,000 stamps from 1908 and up, this is thanks to Wilmot Corfield who proposed a National Collection Committee to work with the British Museum to create a modern stamp collection.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Universal Postal Union Collection_sentence_0",
"Universal Postal Union Collection_sentence_4",
"Universal Postal Union Collection_sentence_2",
"Universal Postal Union Collection_sentence_5",
"Universal Postal Union Collection_sentence_6",
"Universal Postal Union Collection_sentence_8"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 264,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.698",
"value": "?search="
},
{
"operation": "search",
"time": "21.062",
"value": "Universal Postal Union Collection"
},
{
"operation": "Now on",
"time": "22.356",
"value": "Universal Postal Union Collection"
},
{
"operation": "Highlighting",
"time": "30.258",
"value": "Universal Postal Union Collection_sentence_0"
},
{
"operation": "Highlighting",
"time": "59.556",
"value": "Universal Postal Union Collection_sentence_4"
},
{
"operation": "Highlighting",
"time": "78.889",
"value": "Universal Postal Union Collection_sentence_2"
},
{
"operation": "Highlighting",
"time": "122.795",
"value": "Universal Postal Union Collection_sentence_5"
},
{
"operation": "Highlighting",
"time": "139.232",
"value": "Universal Postal Union Collection_sentence_6"
},
{
"operation": "Highlighting",
"time": "193.217",
"value": "Universal Postal Union Collection_sentence_8"
},
{
"operation": "finish",
"time": "224.127",
"value": "finish"
}
]
}
|
exfever_1361
|
Hybrid Theory was the best-selling album by Linkin Park.
|
{
"evidence": [
"Chester Bennington",
"Recording Industry Association of America"
],
"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:\nHybrid Theory was the best-selling album by Linkin Park.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Chester Bennington",
"Recording Industry Association of America"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 265,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
45390
|
Emil Roy (May 26, 1907 – January 5, 1997) was a native of Brighton, Massachusetts who played summer baseball in the Cape Cod Baseball League and later was signed by the Philadelphia Athletics.
|
{
"evidence": [
"Emil Roy_sentence_0",
"Emil Roy_sentence_3",
"Emil Roy_sentence_4",
"Emil Roy_sentence_5"
],
"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:\nEmil Roy (May 26, 1907 – January 5, 1997) was a native of Brighton, Massachusetts who played summer baseball in the Cape Cod Baseball League and later was signed by the Philadelphia Athletics.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Emil Roy_sentence_0",
"Emil Roy_sentence_3",
"Emil Roy_sentence_4",
"Emil Roy_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 266,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.816",
"value": "?search="
},
{
"operation": "search",
"time": "6.386",
"value": "Emil Roy"
},
{
"operation": "Now on",
"time": "7.653",
"value": "Emil Roy"
},
{
"operation": "Highlighting",
"time": "12.189",
"value": "Emil Roy_sentence_0"
},
{
"operation": "Highlighting",
"time": "36.517",
"value": "Emil Roy_sentence_3"
},
{
"operation": "Highlighting",
"time": "39.62",
"value": "Emil Roy_sentence_4"
},
{
"operation": "Highlighting",
"time": "44.596",
"value": "Emil Roy_sentence_5"
},
{
"operation": "finish",
"time": "56.135",
"value": "finish"
}
]
}
|
exfever_3749
|
Zach Galifianakis is an American actor who has starred in films, including the 2012 American political satire comedy film.
|
{
"evidence": [
"Zach Galifianakis",
"The Campaign (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:\nZach Galifianakis is an American actor who has starred in films, including the 2012 American political satire comedy film.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Zach Galifianakis",
"The Campaign (film)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 267,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
exfever_516
|
Chris Terrio's first film was Heights in 2005.
|
{
"evidence": [
"Chris Terrio",
"Man of Steel (film)",
"Zack Snyder"
],
"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:\nChris Terrio's first film was Heights in 2005.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Chris Terrio",
"Man of Steel (film)",
"Zack Snyder"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 268,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
91782
|
Rachid Ammar was the chief of staff of the Tunisian Armed Forces who was sacked by Zine El Abidine Ben Ali when he refused to give the order to shoot the Tunisian protesters.
|
{
"evidence": [
"Rachid Ammar_sentence_0",
"Rachid Ammar_sentence_8",
"Rachid Ammar_sentence_10"
],
"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:\nRachid Ammar was the chief of staff of the Tunisian Armed Forces who was sacked by Zine El Abidine Ben Ali when he refused to give the order to shoot the Tunisian protesters.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Rachid Ammar_sentence_0",
"Rachid Ammar_sentence_8",
"Rachid Ammar_sentence_10"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 269,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.227",
"value": "?search="
},
{
"operation": "search",
"time": "11.795",
"value": "Rachid Ammar"
},
{
"operation": "Now on",
"time": "12.627",
"value": "Rachid Ammar"
},
{
"operation": "Highlighting",
"time": "19.489",
"value": "Rachid Ammar_sentence_2"
},
{
"operation": "Highlighting",
"time": "19.492",
"value": "Rachid Ammar_section_0"
},
{
"operation": "Highlighting deleted",
"time": "48.614",
"value": "Rachid Ammar_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "48.622",
"value": "Rachid Ammar_section_0"
},
{
"operation": "Highlighting",
"time": "50.19",
"value": "Rachid Ammar_sentence_0"
},
{
"operation": "Highlighting",
"time": "79.768",
"value": "Rachid Ammar_sentence_8"
},
{
"operation": "Highlighting",
"time": "91.719",
"value": "Rachid Ammar_sentence_10"
},
{
"operation": "finish",
"time": "118.696",
"value": "finish"
}
]
}
|
exfever_8852
|
Fiji is an island country in Melanesia, part of a geographic region that span the Southern and the half of Earth which lies west of the prime meridian.
|
{
"evidence": [
"Fiji",
"Oceania",
"Western Hemisphere"
],
"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:\nFiji is an island country in Melanesia, part of a geographic region that span the Southern and the half of Earth which lies west of the prime meridian.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Fiji",
"Oceania",
"Western Hemisphere"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 270,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
74990
|
Ben Swift (born 5 November 1987), an English professional track and road racing cyclist, who began cycling with Mossley Cycle Racing Team at the age of 12, came second to fellow Olympian Steven Burke in the under 16 national scratch race championship in 2003.
|
{
"evidence": [
"Ben Swift_sentence_0",
"Ben Swift_sentence_3",
"Ben Swift_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:\nBen Swift (born 5 November 1987), an English professional track and road racing cyclist, who began cycling with Mossley Cycle Racing Team at the age of 12, came second to fellow Olympian Steven Burke in the under 16 national scratch race championship in 2003.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Ben Swift_sentence_0",
"Ben Swift_sentence_3",
"Ben Swift_sentence_4"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 271,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.821",
"value": "?search="
},
{
"operation": "search",
"time": "8.216",
"value": "Ben Swift"
},
{
"operation": "Now on",
"time": "9.139",
"value": "Ben Swift"
},
{
"operation": "Highlighting",
"time": "13.777",
"value": "Ben Swift_sentence_0"
},
{
"operation": "Highlighting",
"time": "44.198",
"value": "Ben Swift_sentence_3"
},
{
"operation": "Highlighting",
"time": "55.78",
"value": "Ben Swift_sentence_4"
},
{
"operation": "finish",
"time": "64.772",
"value": "finish"
}
]
}
|
90114
|
Declan Quigley, the chief motorsport writer for the Irish Independent, began commentating on cycle racing for Eurosport in 2013 but was replaced by David Harman.
|
{
"evidence": [
"Declan Quigley_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:\nDeclan Quigley, the chief motorsport writer for the Irish Independent, began commentating on cycle racing for Eurosport in 2013 but was replaced by David Harman.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Declan Quigley_sentence_3"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 272,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "2.73",
"value": "?search="
},
{
"operation": "search",
"time": "14.924",
"value": "Declan Quigley"
},
{
"operation": "Now on",
"time": "16.685",
"value": "Declan Quigley"
},
{
"operation": "Highlighting",
"time": "20.548",
"value": "Declan Quigley_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "28.373",
"value": "Declan Quigley_sentence_1"
},
{
"operation": "Highlighting",
"time": "29.42",
"value": "Declan Quigley_sentence_3"
},
{
"operation": "finish",
"time": "43.948",
"value": "finish"
}
]
}
|
84184
|
Military Regions of Laos, roughly corresponding to the areas of the country's 15 provinces, had air transportations that could not accommodate anything larger than twin-engine aircraft and defense was entrusted to local troops.
|
{
"evidence": [
"Military Regions of Laos_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:\nMilitary Regions of Laos, roughly corresponding to the areas of the country's 15 provinces, had air transportations that could not accommodate anything larger than twin-engine aircraft and defense was entrusted to local troops.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Military Regions of Laos_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 273,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.139",
"value": "?search="
},
{
"operation": "search",
"time": "23.807",
"value": "Military Regions of Laos"
},
{
"operation": "Now on",
"time": "24.747",
"value": "Military Regions of Laos"
},
{
"operation": "Highlighting",
"time": "33.316",
"value": "Military Regions of Laos_sentence_0"
},
{
"operation": "finish",
"time": "40.045",
"value": "finish"
}
]
}
|
exfever_2969
|
The Formula is Marlon Brando's first mystery film.
|
{
"evidence": [
"The Formula (1980 film)",
"John G. Avildsen"
],
"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 Formula is Marlon Brando's first mystery film.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Formula (1980 film)",
"John G. Avildsen"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 274,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
exfever_334
|
X-Men is a 2000 American film that depicts a world where an unknown proportion focuses on the actions of superheroes dedicated to protecting the enemy.
|
{
"evidence": [
"X-Men (film)",
"Superhero 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:\nX-Men is a 2000 American film that depicts a world where an unknown proportion focuses on the actions of superheroes dedicated to protecting the enemy.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"X-Men (film)",
"Superhero film"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 275,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
84755
|
Gilbert Benson (discography consist of 36 studio albums) with the Hammock Underground Band "Erotic Moods" (1978) are from the same session and he wasn't the front man and it was not his scheduled release.
|
{
"evidence": [
"George Benson discography_sentence_0",
"George Benson discography_sentence_11"
],
"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:\nGilbert Benson (discography consist of 36 studio albums) with the Hammock Underground Band \"Erotic Moods\" (1978) are from the same session and he wasn't the front man and it was not his scheduled release.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"George Benson discography_sentence_0",
"George Benson discography_sentence_11"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 276,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.906",
"value": "?search="
},
{
"operation": "search",
"time": "151.69",
"value": "Contains...Erotic Moods"
},
{
"operation": "Now on",
"time": "153.512",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "186.779",
"value": "George Benson discography"
},
{
"operation": "Now on",
"time": "190.497",
"value": "George Benson discography"
},
{
"operation": "Highlighting",
"time": "212.525",
"value": "George Benson discography_sentence_0"
},
{
"operation": "Highlighting",
"time": "320.097",
"value": "George Benson discography_sentence_11"
},
{
"operation": "finish",
"time": "335.449",
"value": "finish"
}
]
}
|
exfever_6921
|
Marvel's Iron Fist is an American television series created by a television writer who has written several television series, including a sitcom that aired on CBS from September 13, 1996, to May 16, 2005.
|
{
"evidence": [
"Iron Fist (TV series)",
"Scott Buck",
"Everybody Loves Raymond"
],
"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:\nMarvel's Iron Fist is an American television series created by a television writer who has written several television series, including a sitcom that aired on CBS from September 13, 1996, to May 16, 2005.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Iron Fist (TV series)",
"Scott Buck",
"Everybody Loves Raymond"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 277,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
exfever_5496
|
"Life Is Peachy" is an album by Korn, released on an Australian record label owned by a subsidiary of an American subsidiary of the Japanese conglomerate Sony Group Corporation.
|
{
"evidence": [
"Life Is Peachy",
"Epic Records",
"Sony Corporation of America"
],
"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\"Life Is Peachy\" is an album by Korn, released on an Australian record label owned by a subsidiary of an American subsidiary of the Japanese conglomerate Sony Group Corporation.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Life Is Peachy",
"Epic Records",
"Sony Corporation of America"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 278,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
74388
|
Shane Bond, a New Zealand cricket coach and former cricketer and present bowling coach of Mumbai Indians, was the only player in the 2010 Indian Premier League that was sold at $750,000.
|
{
"evidence": [
"Shane Bond_sentence_1",
"Shane Bond_sentence_11",
"Shane Bond_sentence_42"
],
"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:\nShane Bond, a New Zealand cricket coach and former cricketer and present bowling coach of Mumbai Indians, was the only player in the 2010 Indian Premier League that was sold at $750,000.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Shane Bond_sentence_1",
"Shane Bond_sentence_11",
"Shane Bond_sentence_42"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 279,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.646",
"value": "?search="
},
{
"operation": "search",
"time": "13.978",
"value": "Shane Bond"
},
{
"operation": "Now on",
"time": "17.156",
"value": "Shane Bond"
},
{
"operation": "Highlighting",
"time": "20.13",
"value": "Shane Bond_sentence_1"
},
{
"operation": "Highlighting",
"time": "44.479",
"value": "Shane Bond_sentence_11"
},
{
"operation": "Highlighting",
"time": "76.486",
"value": "Shane Bond_sentence_42"
},
{
"operation": "hyperlink",
"time": "99.379",
"value": "IPL"
},
{
"operation": "Now on",
"time": "102.72",
"value": "Indian Premier League"
},
{
"operation": "Page search",
"time": "122.403",
"value": "2010"
},
{
"operation": "hyperlink",
"time": "173.362",
"value": "Controversies involving the Indian Premier League"
},
{
"operation": "Now on",
"time": "174.674",
"value": "Controversies involving the Indian Premier League"
},
{
"operation": "hyperlink",
"time": "189.017",
"value": "2 Financial controversies Controversies involving the Indian Premier League_section_6"
},
{
"operation": "hyperlink",
"time": "219.633",
"value": "2 Financial controversies Controversies involving the Indian Premier League_section_6"
},
{
"operation": "back-button-clicked",
"time": "225.211",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "225.524",
"value": "Indian Premier League"
},
{
"operation": "back-button-clicked",
"time": "227.153",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "227.185",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "227.34",
"value": "Shane Bond"
},
{
"operation": "hyperlink",
"time": "256.081",
"value": "Kolkata Knight Riders"
},
{
"operation": "Now on",
"time": "259.184",
"value": "Kolkata Knight Riders"
},
{
"operation": "Page search",
"time": "261.976",
"value": "2010"
},
{
"operation": "finish",
"time": "303.873",
"value": "finish"
}
]
}
|
43117
|
Hollye Moss, professor at the Western Carolina University which is part of the University of North Carolina system, commented that Vitasek's book is "thought-provoking."
|
{
"evidence": [
"Kate Vitasek_sentence_13",
"Kate Vitasek_sentence_14",
"Western Carolina University_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:\nHollye Moss, professor at the Western Carolina University which is part of the University of North Carolina system, commented that Vitasek's book is \"thought-provoking.\"\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Kate Vitasek_sentence_13",
"Kate Vitasek_sentence_14",
"Western Carolina University_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 280,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.895",
"value": "?search="
},
{
"operation": "search",
"time": "9.794",
"value": "Contains...Hollye Moss"
},
{
"operation": "Now on",
"time": "10.538",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "13.984",
"value": "Kate Vitasek"
},
{
"operation": "Now on",
"time": "14.698",
"value": "Kate Vitasek"
},
{
"operation": "Page search",
"time": "18.451",
"value": "Hollye Moss"
},
{
"operation": "Highlighting",
"time": "22.879",
"value": "Kate Vitasek_sentence_13"
},
{
"operation": "Highlighting",
"time": "31.774",
"value": "Kate Vitasek_sentence_14"
},
{
"operation": "hyperlink",
"time": "35.554",
"value": "Western Carolina University"
},
{
"operation": "Now on",
"time": "36.948",
"value": "Western Carolina University"
},
{
"operation": "Highlighting",
"time": "38.115",
"value": "Western Carolina University_sentence_1"
},
{
"operation": "back-button-clicked",
"time": "40.115",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "40.212",
"value": "Kate Vitasek"
},
{
"operation": "finish",
"time": "70.579",
"value": "finish"
}
]
}
|
exfever_946
|
The Mummy is a 2017 American fantasy action-adventure film directed by an American film and television writer who co-write the scripts to a 2013 American science fiction action film with Roberto Orci.
|
{
"evidence": [
"The Mummy (2017 film)",
"Alex Kurtzman",
"Star Trek Into Darkness"
],
"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 Mummy is a 2017 American fantasy action-adventure film directed by an American film and television writer who co-write the scripts to a 2013 American science fiction action film with Roberto Orci.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Mummy (2017 film)",
"Alex Kurtzman",
"Star Trek Into Darkness"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 281,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
exfever_8455
|
Francis "Frank" Arthur Sedgman is a retired world No. 1 Australian player of a racket sport that can be played either individually or by two players.
|
{
"evidence": [
"Frank Sedgman",
"Tennis"
],
"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:\nFrancis \"Frank\" Arthur Sedgman is a retired world No. 1 Australian player of a racket sport that can be played either individually or by two players.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Frank Sedgman",
"Tennis"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 282,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
91966
|
The Wolayita Zone in Ghana includes the Paracho administrative town in the Abala Abaya woreda.
|
{
"evidence": [
"Wolayita Zone_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 Wolayita Zone in Ghana includes the Paracho administrative town in the Abala Abaya woreda.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Wolayita Zone_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 283,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.675",
"value": "?search="
},
{
"operation": "search",
"time": "12.953",
"value": "Wolayita Zone"
},
{
"operation": "Now on",
"time": "14.285",
"value": "Wolayita Zone"
},
{
"operation": "search",
"time": "313.868",
"value": "The Wolayita Zone in Ghana"
},
{
"operation": "Now on",
"time": "315.764",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "319.364",
"value": "Solanum nigrum"
},
{
"operation": "Now on",
"time": "321.001",
"value": "Solanum nigrum"
},
{
"operation": "search",
"time": "344.418",
"value": "Abala Abaya woreda"
},
{
"operation": "Now on",
"time": "345.511",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "347.5",
"value": "Wolayita Zone"
},
{
"operation": "Now on",
"time": "348.014",
"value": "Wolayita Zone"
},
{
"operation": "Highlighting",
"time": "351.942",
"value": "Wolayita Zone_sentence_2"
},
{
"operation": "finish",
"time": "357.802",
"value": "finish"
}
]
}
|
exfever_693
|
Grease is the third musical romantic comedy film written by Bronte Woodard.
|
{
"evidence": [
"Grease (film)",
"Warren Casey",
"Writer"
],
"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:\nGrease is the third musical romantic comedy film written by Bronte Woodard.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Grease (film)",
"Warren Casey",
"Writer"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 284,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
81600
|
Treffpunkt Flughafen is an East German television series produced by the DEFA-Studio fur Spielfilme on behalf of Fernsehen der DDR in 1985 and 1988 where the show follows the life and adventures of the crew of an IL-62 operated by the East German airline Interflug.
|
{
"evidence": [
"Treffpunkt Flughafen_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:\nTreffpunkt Flughafen is an East German television series produced by the DEFA-Studio fur Spielfilme on behalf of Fernsehen der DDR in 1985 and 1988 where the show follows the life and adventures of the crew of an IL-62 operated by the East German airline Interflug.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Treffpunkt Flughafen_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 285,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.104",
"value": "?search="
},
{
"operation": "search",
"time": "16.025",
"value": "Treffpunkt Flughafen"
},
{
"operation": "Now on",
"time": "17.123",
"value": "Treffpunkt Flughafen"
},
{
"operation": "Highlighting",
"time": "24.694",
"value": "Treffpunkt Flughafen_sentence_0"
},
{
"operation": "finish",
"time": "53.972",
"value": "finish"
}
]
}
|
exfever_6157
|
Fargo's first season was set primarily in Minnesota and in a U.S. state in the Upper Midwest that is bordered to the east by the U.S. state in the upper Midwestern United States and the 12th largest U.S. state in the area.
|
{
"evidence": [
"Fargo (TV series)",
"North Dakota",
"Minnesota"
],
"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:\nFargo's first season was set primarily in Minnesota and in a U.S. state in the Upper Midwest that is bordered to the east by the U.S. state in the upper Midwestern United States and the 12th largest U.S. state in the area.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Fargo (TV series)",
"North Dakota",
"Minnesota"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 286,
"split": "test_on"
}
|
{
"annotator_operations": []
}
|
20845
|
J. Earl Danieley, born on May 28, 1924 in Burlington, North Carolina, was the President of Elon College preceded by Leon Edgar Smith.
|
{
"evidence": [
"James Danieley_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:\nJ. Earl Danieley, born on May 28, 1924 in Burlington, North Carolina, was the President of Elon College preceded by Leon Edgar Smith.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"James Danieley_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 287,
"split": "test_on"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.632",
"value": "?search="
},
{
"operation": "search",
"time": "6.185",
"value": "Contains...J. Earl Danieley"
},
{
"operation": "Now on",
"time": "6.976",
"value": "undefined"
},
{
"operation": "search",
"time": "18.458",
"value": "James Danieley"
},
{
"operation": "Now on",
"time": "19.313",
"value": "James Danieley"
},
{
"operation": "Highlighting",
"time": "23.499",
"value": "James Danieley_sentence_0"
},
{
"operation": "finish",
"time": "46.372",
"value": "finish"
}
]
}
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.