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 |
|---|---|---|---|---|---|---|---|---|
exfever_2379
|
The Handmaid's Tale is a futuristic dystopian novel by Canadian author Margaret Atwood, set in a near-future of a region comprising six states in the Western United States including Connecticut, Maine, and Massachusetts.
|
{
"evidence": [
"The Handmaid's Tale",
"New England"
],
"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 Handmaid's Tale is a futuristic dystopian novel by Canadian author Margaret Atwood, set in a near-future of a region comprising six states in the Western United States including Connecticut, Maine, and Massachusetts.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Handmaid's Tale",
"New England"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3284,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4451
|
Whiskey Tango Foxtrot is an American biographical war comedy-drama film, starring an American actress best known for her work on the NBC sketch comedy series Saturday Night Live.
|
{
"evidence": [
"Whiskey Tango Foxtrot (film)",
"Tina Fey"
],
"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:\nWhiskey Tango Foxtrot is an American biographical war comedy-drama film, starring an American actress best known for her work on the NBC sketch comedy series Saturday Night Live.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Whiskey Tango Foxtrot (film)",
"Tina Fey"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3285,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
91797
|
On June 30, 2015, Misia announced she would be hosting a four-date installment of the Misia Candle Night concert series starting in September 2016 where the proceeds after expenses benefited the non-profit Mudef.
|
{
"evidence": [
"Misia Candle Night 2015_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:\nOn June 30, 2015, Misia announced she would be hosting a four-date installment of the Misia Candle Night concert series starting in September 2016 where the proceeds after expenses benefited the non-profit Mudef.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Misia Candle Night 2015_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3286,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.754",
"value": "?search="
},
{
"operation": "search",
"time": "8.198",
"value": "Misia"
},
{
"operation": "Now on",
"time": "9.163",
"value": "Misia"
},
{
"operation": "Page search",
"time": "15.972",
"value": "June 30"
},
{
"operation": "Page search",
"time": "24.779",
"value": "announced"
},
{
"operation": "Page search",
"time": "37.98",
"value": "Candle"
},
{
"operation": "Page search",
"time": "49.66",
"value": "Mudef"
},
{
"operation": "search",
"time": "59.644",
"value": "Misia Candle Night 2015"
},
{
"operation": "Now on",
"time": "60.412",
"value": "Misia Candle Night 2015"
},
{
"operation": "Highlighting",
"time": "64.71",
"value": "Misia Candle Night 2015_sentence_2"
},
{
"operation": "finish",
"time": "77.443",
"value": "finish"
}
]
}
|
exfever_7691
|
The Mughal Empire was the most known Islamic empire in South Asia.
|
{
"evidence": [
"Mughal Empire",
"Early modern period"
],
"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 Mughal Empire was the most known Islamic empire in South Asia.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Mughal Empire",
"Early modern period"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3287,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5620
|
The existence of God is the most controversial topic in human existence.?
|
{
"evidence": [
"Existence of God",
"Popular culture"
],
"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 existence of God is the most controversial topic in human existence.?\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Existence of God",
"Popular culture"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3288,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_24
|
Kingdom Hearts is a video game series that combines elements of action games and the role-playing genre.
|
{
"evidence": [
"Kingdom Hearts",
"Action role-playing game"
],
"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:\nKingdom Hearts is a video game series that combines elements of action games and the role-playing genre.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Kingdom Hearts",
"Action role-playing game"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3289,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
26199
|
Some of the storytelling of Guimba the Tyrant is done through the village griot, a West African historian.
|
{
"evidence": [
"Guimba the Tyrant_sentence_2",
"Griot_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:\nSome of the storytelling of Guimba the Tyrant is done through the village griot, a West African historian.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Guimba the Tyrant_sentence_2",
"Griot_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3290,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.661",
"value": "?search="
},
{
"operation": "search",
"time": "15.667",
"value": "Guimba the Tyrant"
},
{
"operation": "Now on",
"time": "16.394",
"value": "Guimba the Tyrant"
},
{
"operation": "Page search",
"time": "47.266",
"value": "griot"
},
{
"operation": "Highlighting",
"time": "56.524",
"value": "Guimba the Tyrant_sentence_2"
},
{
"operation": "hyperlink",
"time": "58.325",
"value": "griot"
},
{
"operation": "Now on",
"time": "59.09",
"value": "Griot"
},
{
"operation": "Highlighting",
"time": "71.138",
"value": "Griot_sentence_5"
},
{
"operation": "finish",
"time": "79.337",
"value": "finish"
}
]
}
|
70272
|
The 2011–12 UCI Africa Tour began on September 29, 2011, with the Grand Prix Chantal Biya (a professional road bicycle racing stage race held in Cameroon and sponsored by the Fédération Camerounaise de Cyclisme/Ateba Koungou), and in 2015 UCI Africa Tour, ten races had a UCI rating of 1.2 while four races had a UCI rating of 2.2.
|
{
"evidence": [
"2011–12 UCI Africa Tour_sentence_1",
"Grand Prix Chantal Biya_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:\nThe 2011–12 UCI Africa Tour began on September 29, 2011, with the Grand Prix Chantal Biya (a professional road bicycle racing stage race held in Cameroon and sponsored by the Fédération Camerounaise de Cyclisme/Ateba Koungou), and in 2015 UCI Africa Tour, ten races had a UCI rating of 1.2 while four races had a UCI rating of 2.2.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"2011–12 UCI Africa Tour_sentence_1",
"Grand Prix Chantal Biya_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3291,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.808",
"value": "?search="
},
{
"operation": "search",
"time": "21.473",
"value": "2011–12 UCI Africa Tour"
},
{
"operation": "Now on",
"time": "22.592",
"value": "2011–12 UCI Africa Tour"
},
{
"operation": "Highlighting",
"time": "29.602",
"value": "2011–12 UCI Africa Tour_sentence_1"
},
{
"operation": "hyperlink",
"time": "48",
"value": "Grand Prix Chantal Biya"
},
{
"operation": "Now on",
"time": "48.658",
"value": "Grand Prix Chantal Biya"
},
{
"operation": "Highlighting",
"time": "54.812",
"value": "Grand Prix Chantal Biya_sentence_0"
},
{
"operation": "back-button-clicked",
"time": "56.333",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "56.381",
"value": "2011–12 UCI Africa Tour"
},
{
"operation": "search",
"time": "103.368",
"value": "Contains... 2015 UCI Africa Tour"
},
{
"operation": "Now on",
"time": "105.087",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "122.873",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "122.92",
"value": "2011–12 UCI Africa Tour"
},
{
"operation": "hyperlink",
"time": "128.834",
"value": "UCI Africa Tour"
},
{
"operation": "Now on",
"time": "130.301",
"value": "UCI Continental Circuits"
},
{
"operation": "search",
"time": "181.369",
"value": "UCI ratings"
},
{
"operation": "Now on",
"time": "182.82",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "196.743",
"value": "2015 UCI America Tour"
},
{
"operation": "Now on",
"time": "197.538",
"value": "2015 UCI America Tour"
},
{
"operation": "search",
"time": "218.082",
"value": "2015 UCI Africa Tour"
},
{
"operation": "Now on",
"time": "219.478",
"value": "undefined"
},
{
"operation": "search",
"time": "253.849",
"value": "Contains...ten races had a UCI rating of 1.2 while four races had a UCI rating of 2.2"
},
{
"operation": "Now on",
"time": "255.863",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "296.757",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "297.757",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "301.042",
"value": "2015 KZN Autumn Series"
},
{
"operation": "Now on",
"time": "301.779",
"value": "2015 KZN Autumn Series"
},
{
"operation": "back-button-clicked",
"time": "317.935",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "318.974",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "318.983",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "318.987",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "318.99",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "319.032",
"value": "2015 UCI America Tour"
},
{
"operation": "back-button-clicked",
"time": "319.953",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "319.955",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "320.892",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "322.634",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "322.637",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "323.818",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "323.861",
"value": "2011–12 UCI Africa Tour"
},
{
"operation": "search",
"time": "451.001",
"value": "UCI Africa Tour"
},
{
"operation": "Now on",
"time": "452.379",
"value": "UCI Continental Circuits"
},
{
"operation": "Highlighting",
"time": "481.678",
"value": "UCI Continental Circuits_cell_0_11_0"
},
{
"operation": "Highlighting",
"time": "481.68",
"value": "UCI Continental Circuits_title"
},
{
"operation": "Highlighting",
"time": "481.68",
"value": "UCI Continental Circuits_section_0"
},
{
"operation": "Highlighting",
"time": "481.68",
"value": "UCI Continental Circuits_section_1"
},
{
"operation": "Highlighting",
"time": "481.682",
"value": "UCI Continental Circuits_header_cell_0_0_0"
},
{
"operation": "Highlighting deleted",
"time": "484.116",
"value": "UCI Continental Circuits_cell_0_11_0"
},
{
"operation": "Highlighting deleted",
"time": "484.116",
"value": "UCI Continental Circuits_title"
},
{
"operation": "Highlighting deleted",
"time": "484.117",
"value": "UCI Continental Circuits_section_0"
},
{
"operation": "Highlighting deleted",
"time": "484.118",
"value": "UCI Continental Circuits_section_1"
},
{
"operation": "Highlighting deleted",
"time": "484.12",
"value": "UCI Continental Circuits_header_cell_0_0_0"
},
{
"operation": "hyperlink",
"time": "493.389",
"value": "Skydive Dubai Pro Cycling Team–Al Ahli Club"
},
{
"operation": "Now on",
"time": "494.116",
"value": "Skydive Dubai–Al Ahli Pro Cycling Team"
},
{
"operation": "back-button-clicked",
"time": "501.922",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "507.613",
"value": "UCI Continental Circuits"
},
{
"operation": "hyperlink",
"time": "527.835",
"value": "Salah Eddine Mraouni"
},
{
"operation": "Now on",
"time": "528.331",
"value": "Salah Eddine Mraouni"
},
{
"operation": "finish",
"time": "550.635",
"value": "finish"
}
]
}
|
exfever_7868
|
No One Else on Earth" is a song written by Jill Colucci, Stewart Harris and Sam Lorber.
|
{
"evidence": [
"Wynonna Judd",
"I Saw the Light (Wynonna Judd song)"
],
"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:\nNo One Else on Earth\" is a song written by Jill Colucci, Stewart Harris and Sam Lorber.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Wynonna Judd",
"I Saw the Light (Wynonna Judd song)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3292,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8918
|
Las Vegas is the capital city of Clark County and the 26th-most populous city in the United Kingdom.
|
{
"evidence": [
"Las Vegas",
"County seat"
],
"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:\nLas Vegas is the capital city of Clark County and the 26th-most populous city in the United Kingdom.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Las Vegas",
"County seat"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3293,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3931
|
Robert McElhenney III made his debut in the comedy series It's Always Sunny in Philadelphia.
|
{
"evidence": [
"Rob McElhenney",
"FXX"
],
"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:\nRobert McElhenney III made his debut in the comedy series It's Always Sunny in Philadelphia.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Rob McElhenney",
"FXX"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3294,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
59751
|
Anil Gurung, who was born and raised in Nepal, has scored multiple goals for his nation at the SAFF Championship throughout his career.
|
{
"evidence": [
"Anil Gurung_sentence_0",
"Anil Gurung_sentence_3",
"Anil Gurung_sentence_23"
],
"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:\nAnil Gurung, who was born and raised in Nepal, has scored multiple goals for his nation at the SAFF Championship throughout his career.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Anil Gurung_sentence_0",
"Anil Gurung_sentence_3",
"Anil Gurung_sentence_23"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3295,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.732",
"value": "?search="
},
{
"operation": "search",
"time": "34.258",
"value": "Anil Gurung"
},
{
"operation": "Now on",
"time": "35.431",
"value": "Anil Gurung"
},
{
"operation": "Page search",
"time": "43.592",
"value": "raise"
},
{
"operation": "Page search",
"time": "93.793",
"value": "nepal"
},
{
"operation": "Highlighting",
"time": "119.327",
"value": "Anil Gurung_sentence_0"
},
{
"operation": "Highlighting",
"time": "137.48",
"value": "Anil Gurung_sentence_3"
},
{
"operation": "Highlighting deleted",
"time": "160.835",
"value": "Anil Gurung_sentence_3"
},
{
"operation": "search",
"time": "170.596",
"value": "Barahi Secondary School"
},
{
"operation": "Now on",
"time": "171.831",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "179.098",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "179.293",
"value": "Anil Gurung"
},
{
"operation": "Highlighting",
"time": "187.939",
"value": "Anil Gurung_sentence_3"
},
{
"operation": "Page search",
"time": "234.827",
"value": "saff champion"
},
{
"operation": "Highlighting",
"time": "276.019",
"value": "Anil Gurung_sentence_23"
},
{
"operation": "hyperlink",
"time": "288.924",
"value": "2013 SAFF Championship"
},
{
"operation": "Now on",
"time": "289.8",
"value": "2013 SAFF Championship"
},
{
"operation": "Page search",
"time": "295.233",
"value": "gurung"
},
{
"operation": "hyperlink",
"time": "316.296",
"value": "Nepal"
},
{
"operation": "Now on",
"time": "319.161",
"value": "Nepal national football team"
},
{
"operation": "back-button-clicked",
"time": "321.005",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "321.191",
"value": "2013 SAFF Championship"
},
{
"operation": "finish",
"time": "334.455",
"value": "finish"
}
]
}
|
43817
|
Paul Wachs was the son of the French composer Frédéric Wachs, and a student at the Conservatoire de Paris (a college of music and dance founded in 1795 and situated in the avenue Jean Jaurès in the 19th arrondissement of Paris, France).
|
{
"evidence": [
"Paul Wachs_sentence_2",
"Paul Wachs_sentence_3",
"Conservatoire de Paris_sentence_0",
"Conservatoire de Paris_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:\nPaul Wachs was the son of the French composer Frédéric Wachs, and a student at the Conservatoire de Paris (a college of music and dance founded in 1795 and situated in the avenue Jean Jaurès in the 19th arrondissement of Paris, France).\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Paul Wachs_sentence_2",
"Paul Wachs_sentence_3",
"Conservatoire de Paris_sentence_0",
"Conservatoire de Paris_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3296,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.629",
"value": "?search="
},
{
"operation": "search",
"time": "7.433",
"value": "Paul Wachs"
},
{
"operation": "Now on",
"time": "8.527",
"value": "Paul Wachs"
},
{
"operation": "Highlighting",
"time": "15.838",
"value": "Paul Wachs_sentence_2"
},
{
"operation": "Highlighting",
"time": "22.101",
"value": "Paul Wachs_sentence_3"
},
{
"operation": "hyperlink",
"time": "23.163",
"value": "Conservatoire de Paris"
},
{
"operation": "Now on",
"time": "23.96",
"value": "Conservatoire de Paris"
},
{
"operation": "Highlighting",
"time": "30.048",
"value": "Conservatoire de Paris_sentence_0"
},
{
"operation": "Highlighting",
"time": "34.479",
"value": "Conservatoire de Paris_sentence_1"
},
{
"operation": "back-button-clicked",
"time": "36.588",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "36.692",
"value": "Paul Wachs"
},
{
"operation": "finish",
"time": "64.807",
"value": "finish"
}
]
}
|
exfever_7932
|
Sir Ridley Scott was an English film director of the 2000 epic historical drama film written by David Franzoni.
|
{
"evidence": [
"Ridley Scott",
"Gladiator (2000 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:\nSir Ridley Scott was an English film director of the 2000 epic historical drama film written by David Franzoni.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Ridley Scott",
"Gladiator (2000 film)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3297,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
25252
|
Von Ahnska magasinet, a port warehouse located on Storgatan in Umeå, Sweden, was demolished in 1888.
|
{
"evidence": [
"Von Ahnska magasinet_sentence_2",
"Von Ahnska magasinet_sentence_3",
"Von Ahnska magasinet_sentence_4"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nVon Ahnska magasinet, a port warehouse located on Storgatan in Umeå, Sweden, was demolished in 1888.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Von Ahnska magasinet_sentence_2",
"Von Ahnska magasinet_sentence_3",
"Von Ahnska magasinet_sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3298,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.849",
"value": "?search="
},
{
"operation": "search",
"time": "8.296",
"value": "Von Ahnska magasinet"
},
{
"operation": "Now on",
"time": "8.97",
"value": "Von Ahnska magasinet"
},
{
"operation": "Highlighting",
"time": "11.808",
"value": "Von Ahnska magasinet_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "27.668",
"value": "Von Ahnska magasinet_sentence_0"
},
{
"operation": "Highlighting",
"time": "29.957",
"value": "Von Ahnska magasinet_sentence_2"
},
{
"operation": "Highlighting",
"time": "49.937",
"value": "Von Ahnska magasinet_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "54.535",
"value": "Von Ahnska magasinet_sentence_4"
},
{
"operation": "Highlighting",
"time": "56.063",
"value": "Von Ahnska magasinet_sentence_3"
},
{
"operation": "Highlighting",
"time": "56.763",
"value": "Von Ahnska magasinet_sentence_4"
},
{
"operation": "finish",
"time": "99.333",
"value": "finish"
}
]
}
|
exfever_5287
|
The Shield is an American crime drama television series starring an Italian actor, television director, and television producer who played Commissioner Tony Scali in the ABC comedy-drama television series in the United States that aired from September 28, 1991, to January 11, 1996.
|
{
"evidence": [
"The Shield",
"Michael Chiklis",
"The Commish"
],
"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 Shield is an American crime drama television series starring an Italian actor, television director, and television producer who played Commissioner Tony Scali in the ABC comedy-drama television series in the United States that aired from September 28, 1991, to January 11, 1996.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Shield",
"Michael Chiklis",
"The Commish"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3299,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4782
|
Andrew Maxwell Dwyer is a fictional character that appears in the first season as an American actress slacker boyfriend.
|
{
"evidence": [
"Andy Dwyer",
"Rashida Jones"
],
"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:\nAndrew Maxwell Dwyer is a fictional character that appears in the first season as an American actress slacker boyfriend.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Andy Dwyer",
"Rashida Jones"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3300,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7219
|
Hannibal was a Carthaginian statesman whose brother-in-law was a Carthaginian military leader and politician, and the founder of a Spanish city and a major naval station located in the Region of Murcia.
|
{
"evidence": [
"Hannibal",
"Hasdrubal the Fair",
"Cartagena, Spain"
],
"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:\nHannibal was a Carthaginian statesman whose brother-in-law was a Carthaginian military leader and politician, and the founder of a Spanish city and a major naval station located in the Region of Murcia.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Hannibal",
"Hasdrubal the Fair",
"Cartagena, Spain"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3301,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_926
|
Red was released on Blu-ray and DVD on January 25, 2011.
|
{
"evidence": [
"Red (2010 film)",
"Red (WildStorm)"
],
"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:\nRed was released on Blu-ray and DVD on January 25, 2011.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Red (2010 film)",
"Red (WildStorm)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3302,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4342
|
One Direction is an English-Irish pop boy band who signed with Simon Cowell's record label, a defunct division of Syco Entertainment that oversaw releases of the Irish boy band formed in Sligo in 1999.
|
{
"evidence": [
"One Direction",
"Syco Music",
"Westlife"
],
"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:\nOne Direction is an English-Irish pop boy band who signed with Simon Cowell's record label, a defunct division of Syco Entertainment that oversaw releases of the Irish boy band formed in Sligo in 1999.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"One Direction",
"Syco Music",
"Westlife"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3303,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4604
|
Shakira is the female singer who has the most awards including three Grammy Awards, twelve Latin Grammy Awards, four MTV Video Music Awards, seven Billboard Music Awards, and thirty-nine Billboard Latin Music Awards.
|
{
"evidence": [
"Shakira",
"Barranquilla"
],
"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:\nShakira is the female singer who has the most awards including three Grammy Awards, twelve Latin Grammy Awards, four MTV Video Music Awards, seven Billboard Music Awards, and thirty-nine Billboard Latin Music Awards.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Shakira",
"Barranquilla"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3304,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
51437
|
There are three members of the ADAR family, ADARs 1-3, with ADAR 1 and ADAR 2 being the only enzymatically active members and involved in editing at the Q/R site, and ADAR3 is thought to have a regulatory role in the brain and is restricted to the brain only.
|
{
"evidence": [
"5-HT2C receptor_sentence_59",
"5-HT2C receptor_sentence_91",
"5-HT2C receptor_sentence_60",
"5-HT2C receptor_sentence_62"
],
"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:\nThere are three members of the ADAR family, ADARs 1-3, with ADAR 1 and ADAR 2 being the only enzymatically active members and involved in editing at the Q/R site, and ADAR3 is thought to have a regulatory role in the brain and is restricted to the brain only.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"5-HT2C receptor_sentence_59",
"5-HT2C receptor_sentence_91",
"5-HT2C receptor_sentence_60",
"5-HT2C receptor_sentence_62"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3305,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.178",
"value": "?search="
},
{
"operation": "search",
"time": "18.508",
"value": "Contains...ADAR family"
},
{
"operation": "Now on",
"time": "20.025",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "23.738",
"value": "5-HT2C receptor"
},
{
"operation": "Now on",
"time": "25.155",
"value": "5-HT2C receptor"
},
{
"operation": "Highlighting",
"time": "38.176",
"value": "5-HT2C receptor_sentence_59"
},
{
"operation": "Highlighting",
"time": "99.946",
"value": "5-HT2C receptor_sentence_91"
},
{
"operation": "Highlighting",
"time": "112.197",
"value": "5-HT2C receptor_sentence_60"
},
{
"operation": "Highlighting",
"time": "148.77",
"value": "5-HT2C receptor_sentence_62"
},
{
"operation": "finish",
"time": "165.938",
"value": "finish"
}
]
}
|
41284
|
Edward Charles Pearce served on the Shanghai Municipal Council, he was granted the Freedom of Shanghai and settled down in London.
|
{
"evidence": [
"Edward Charles Pearce_sentence_0",
"Edward Charles Pearce_sentence_6",
"Edward Charles Pearce_sentence_8"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nEdward Charles Pearce served on the Shanghai Municipal Council, he was granted the Freedom of Shanghai and settled down in London.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Edward Charles Pearce_sentence_0",
"Edward Charles Pearce_sentence_6",
"Edward Charles Pearce_sentence_8"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3306,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.124",
"value": "undefined"
},
{
"operation": "search",
"time": "13.271",
"value": "Edward Charles Pearce"
},
{
"operation": "Now on",
"time": "14.153",
"value": "Edward Charles Pearce"
},
{
"operation": "Highlighting",
"time": "25.295",
"value": "Edward Charles Pearce_sentence_0"
},
{
"operation": "Highlighting",
"time": "47.996",
"value": "Edward Charles Pearce_sentence_6"
},
{
"operation": "Highlighting",
"time": "55.571",
"value": "Edward Charles Pearce_sentence_8"
},
{
"operation": "finish",
"time": "142.404",
"value": "finish"
}
]
}
|
exfever_8643
|
The Prince of Egypt is a 1995 American animated musical drama film produced by DreamWorks Animation adapted from the second book of the Bible.
|
{
"evidence": [
"The Prince of Egypt",
"Book of Exodus"
],
"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 Prince of Egypt is a 1995 American animated musical drama film produced by DreamWorks Animation adapted from the second book of the Bible.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Prince of Egypt",
"Book of Exodus"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3307,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7232
|
Dwight David Eisenhower was an American military officer and statesman who served as the 34th head of state and head of government of the United States of America.
|
{
"evidence": [
"Dwight D. Eisenhower",
"President of the United States"
],
"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:\nDwight David Eisenhower was an American military officer and statesman who served as the 34th head of state and head of government of the United States of America.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Dwight D. Eisenhower",
"President of the United States"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3308,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
14140
|
Tina Turner has written a total of more than 30 poems from 1969 to 1977.
|
{
"evidence": [
"Tina Turner_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:\nTina Turner has written a total of more than 30 poems from 1969 to 1977.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Tina Turner_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3309,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0",
"value": "?search="
},
{
"operation": "search",
"time": "0.392",
"value": "tina turner"
},
{
"operation": "Now on",
"time": "0.394",
"value": "Tina turner"
},
{
"operation": "Highlighting",
"time": "0.51",
"value": "Tina Turner_sentence_0"
},
{
"operation": "finish",
"time": "0.558",
"value": "finish"
},
{
"operation": "Now on",
"time": "1619479053.323",
"value": "Tina Turner"
},
{
"operation": "finish",
"time": "1619479057.262",
"value": "finish"
}
]
}
|
49731
|
Persion Poet Nezami wrote a legend about Mount Behistun that states that a man named Farhad was told if he cut away the mountain he could marry his lover, but after he got through removing half the mountain he found out his loved died and then he died too.
|
{
"evidence": [
"Mount Behistun_sentence_3",
"Mount Behistun_sentence_4",
"Mount Behistun_sentence_5",
"Mount Behistun_sentence_6"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nPersion Poet Nezami wrote a legend about Mount Behistun that states that a man named Farhad was told if he cut away the mountain he could marry his lover, but after he got through removing half the mountain he found out his loved died and then he died too.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Mount Behistun_sentence_3",
"Mount Behistun_sentence_4",
"Mount Behistun_sentence_5",
"Mount Behistun_sentence_6"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3310,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.788",
"value": "?search="
},
{
"operation": "search",
"time": "23.647",
"value": "Persion Poet Nezami"
},
{
"operation": "Now on",
"time": "25.679",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "29.053",
"value": "Persion Poet Nezami"
},
{
"operation": "Now on",
"time": "29.459",
"value": "undefined"
},
{
"operation": "search",
"time": "35.386",
"value": "nezami"
},
{
"operation": "Now on",
"time": "35.953",
"value": "Nezami"
},
{
"operation": "hyperlink",
"time": "41.81",
"value": "Nizami Ganjavi"
},
{
"operation": "Now on",
"time": "42.43",
"value": "Nizami Ganjavi"
},
{
"operation": "Page search",
"time": "49.942",
"value": "farhad"
},
{
"operation": "Page search",
"time": "70.996",
"value": "Farhahd"
},
{
"operation": "Page search",
"time": "72.687",
"value": "Farhad"
},
{
"operation": "Page search",
"time": "85.889",
"value": "mountain"
},
{
"operation": "hyperlink",
"time": "98.005",
"value": "Nizami Aruzi"
},
{
"operation": "Now on",
"time": "98.66",
"value": "Nizami Aruzi"
},
{
"operation": "Page search",
"time": "104.908",
"value": "Farhad"
},
{
"operation": "hyperlink",
"time": "111.892",
"value": "List of Persian poets and authors"
},
{
"operation": "Now on",
"time": "112.403",
"value": "List of Persian poets and authors"
},
{
"operation": "Page search",
"time": "117.013",
"value": "nezami"
},
{
"operation": "back-button-clicked",
"time": "128.586",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "128.759",
"value": "Nizami Aruzi"
},
{
"operation": "back-button-clicked",
"time": "144.227",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "144.237",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "144.427",
"value": "Nizami Ganjavi"
},
{
"operation": "Page search",
"time": "152.372",
"value": "behistun"
},
{
"operation": "search",
"time": "183.055",
"value": "Mount Behistun"
},
{
"operation": "Now on",
"time": "183.76",
"value": "Mount Behistun"
},
{
"operation": "Page search",
"time": "190.117",
"value": "Farhad"
},
{
"operation": "Highlighting",
"time": "194.482",
"value": "Mount Behistun_sentence_3"
},
{
"operation": "Highlighting",
"time": "198.57",
"value": "Mount Behistun_sentence_4"
},
{
"operation": "Highlighting",
"time": "202.434",
"value": "Mount Behistun_sentence_5"
},
{
"operation": "Highlighting",
"time": "205.772",
"value": "Mount Behistun_sentence_6"
},
{
"operation": "finish",
"time": "228.909",
"value": "finish"
}
]
}
|
3801
|
Chemnitz got its name in 1990 when the city was officially renamed, the city aced difficult tasks after the reunification of Germany (Unification of East and West Germany).
|
{
"evidence": [
"Chemnitz_sentence_17"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nChemnitz got its name in 1990 when the city was officially renamed, the city aced difficult tasks after the reunification of Germany (Unification of East and West Germany).\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Chemnitz_sentence_17"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3311,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "14.11",
"value": "Chemnitz"
},
{
"operation": "Highlighting",
"time": "53.562",
"value": "Chemnitz_sentence_46"
},
{
"operation": "Highlighting",
"time": "53.578",
"value": "Chemnitz_section_5"
},
{
"operation": "Highlighting",
"time": "55.161",
"value": "Chemnitz_sentence_47"
},
{
"operation": "Highlighting",
"time": "72.601",
"value": "Chemnitz_sentence_48"
},
{
"operation": "Highlighting",
"time": "148.825",
"value": "Chemnitz_sentence_17"
},
{
"operation": "Page search",
"time": "191.57",
"value": "first"
},
{
"operation": "Highlighting",
"time": "200.937",
"value": "Chemnitz_sentence_17"
},
{
"operation": "Highlighting",
"time": "213.097",
"value": "Chemnitz_sentence_18"
},
{
"operation": "hyperlink",
"time": "220.025",
"value": "Imperial city"
},
{
"operation": "back-button-clicked",
"time": "222.186",
"value": "back-button-clicked"
},
{
"operation": "Highlighting deleted",
"time": "223.873",
"value": "Chemnitz_sentence_18"
},
{
"operation": "finish",
"time": "240.763",
"value": "finish"
}
]
}
|
12725
|
Each National Olympic Committee (NOC) can enter a maximum of 2 competitors per gender and each of the five continents must be represented in each gender but if no athletes qualify the host nation will also qualify.
|
{
"evidence": [
"Archery at the 2020 Summer Olympics_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:\nEach National Olympic Committee (NOC) can enter a maximum of 2 competitors per gender and each of the five continents must be represented in each gender but if no athletes qualify the host nation will also qualify.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Archery at the 2020 Summer Olympics_sentence_5"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3312,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.886",
"value": "?search="
},
{
"operation": "search",
"time": "96.525",
"value": "National Olympic Committee"
},
{
"operation": "Now on",
"time": "97.402",
"value": "National Olympic Committee"
},
{
"operation": "search",
"time": "161.609",
"value": "Archery at the 2020 Summer Olympics"
},
{
"operation": "Now on",
"time": "162.325",
"value": "Archery at the 2020 Summer Olympics"
},
{
"operation": "Page search",
"time": "166.08",
"value": "per gender"
},
{
"operation": "Highlighting",
"time": "172.127",
"value": "Archery at the 2020 Summer Olympics_sentence_5"
},
{
"operation": "finish",
"time": "190.998",
"value": "finish"
}
]
}
|
exfever_1943
|
Satellite Kid is The Dogs D'Amour's best-selling song.
|
{
"evidence": [
"The Dogs D'Amour",
"The Rolling Stones"
],
"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:\nSatellite Kid is The Dogs D'Amour's best-selling song.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Dogs D'Amour",
"The Rolling Stones"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3313,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1728
|
Ellen Kathleen Pompeo is a Singaporean actress and producer who was born in a city in Middlesex County of most populous state in the New England region of the United States.
|
{
"evidence": [
"Ellen Pompeo",
"Everett, Massachusetts",
"Massachusetts"
],
"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:\nEllen Kathleen Pompeo is a Singaporean actress and producer who was born in a city in Middlesex County of most populous state in the New England region of the United States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Ellen Pompeo",
"Everett, Massachusetts",
"Massachusetts"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3314,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
50683
|
The Alfred M. Landon Lecture Series is a series of speeches on current public affairs, named after Kansas politician Alf Landon, former Governor of Kansas and Democratic presidential candidate.
|
{
"evidence": [
"Landon Lecture Series_sentence_1"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Alfred M. Landon Lecture Series is a series of speeches on current public affairs, named after Kansas politician Alf Landon, former Governor of Kansas and Democratic presidential candidate.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Landon Lecture Series_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3315,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.126",
"value": "?search="
},
{
"operation": "search",
"time": "12.597",
"value": "Alfred M. Landon Lecture Series"
},
{
"operation": "Now on",
"time": "14.636",
"value": "Landon Lecture Series"
},
{
"operation": "Highlighting",
"time": "28.468",
"value": "Landon Lecture Series_sentence_0"
},
{
"operation": "Highlighting",
"time": "30.113",
"value": "Landon Lecture Series_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "39.521",
"value": "Landon Lecture Series_sentence_0"
},
{
"operation": "finish",
"time": "196.36",
"value": "finish"
}
]
}
|
90364
|
Former gun manufacturer, Beistegui Hermanos currently manufactures car parts and fitness equipment, but has not ever manufactured bicycles.
|
{
"evidence": [
"Beistegui Hermanos_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:\nFormer gun manufacturer, Beistegui Hermanos currently manufactures car parts and fitness equipment, but has not ever manufactured bicycles.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Beistegui Hermanos_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3316,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.392",
"value": "undefined"
},
{
"operation": "search",
"time": "17.447",
"value": "Beistegui Hermanos"
},
{
"operation": "Now on",
"time": "19.152",
"value": "Beistegui Hermanos"
},
{
"operation": "Highlighting",
"time": "28.029",
"value": "Beistegui Hermanos_sentence_0"
},
{
"operation": "finish",
"time": "55.547",
"value": "finish"
}
]
}
|
exfever_7038
|
Ocean's Eight was the first comedy film directed by Gary Ross.
|
{
"evidence": [
"Ocean's Eight",
"Heist 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:\nOcean's Eight was the first comedy film directed by Gary Ross.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Ocean's Eight",
"Heist film"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3317,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
24360
|
Sam Sparks (born 1939), is currently employed by a private institution.
|
{
"evidence": [
"United States District Court for the Western District of Texas_sentence_2",
"Sam Sparks_sentence_1"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nSam Sparks (born 1939), is currently employed by a private institution.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"United States District Court for the Western District of Texas_sentence_2",
"Sam Sparks_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3318,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.774",
"value": "?search="
},
{
"operation": "search",
"time": "13.536",
"value": "Sam Sparks"
},
{
"operation": "Now on",
"time": "14.687",
"value": "Sam Sparks"
},
{
"operation": "Highlighting",
"time": "20.066",
"value": "Sam Sparks_sentence_1"
},
{
"operation": "hyperlink",
"time": "49.743",
"value": "United States District Court for the Western District of Texas"
},
{
"operation": "Now on",
"time": "51.526",
"value": "United States District Court for the Western District of Texas"
},
{
"operation": "Page search",
"time": "72.575",
"value": "private"
},
{
"operation": "Page search",
"time": "79.998",
"value": "public"
},
{
"operation": "Page search",
"time": "80.768",
"value": "public"
},
{
"operation": "back-button-clicked",
"time": "88.868",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "88.925",
"value": "Sam Sparks"
},
{
"operation": "hyperlink",
"time": "143.397",
"value": "United States District Court for the Western District of Texas"
},
{
"operation": "Now on",
"time": "143.765",
"value": "United States District Court for the Western District of Texas"
},
{
"operation": "Highlighting",
"time": "160.104",
"value": "United States District Court for the Western District of Texas_sentence_2"
},
{
"operation": "hyperlink",
"time": "161.109",
"value": "federal district court"
},
{
"operation": "Now on",
"time": "162.449",
"value": "United States district court"
},
{
"operation": "Highlighting",
"time": "238.671",
"value": "United States district court_sentence_3"
},
{
"operation": "Highlighting",
"time": "238.672",
"value": "United States district court_title"
},
{
"operation": "hyperlink",
"time": "241.467",
"value": "civil"
},
{
"operation": "Now on",
"time": "242.097",
"value": "Civil law (common law)"
},
{
"operation": "Page search",
"time": "255.891",
"value": "public"
},
{
"operation": "back-button-clicked",
"time": "318.859",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "319.406",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "319.536",
"value": "United States District Court for the Western District of Texas"
},
{
"operation": "back-button-clicked",
"time": "320.112",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "320.117",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "320.17",
"value": "Sam Sparks"
},
{
"operation": "Highlighting",
"time": "323.241",
"value": "Sam Sparks_sentence_1"
},
{
"operation": "finish",
"time": "354.673",
"value": "finish"
}
]
}
|
exfever_2674
|
Kumkum Bhagya is an Indonesian Hindi-language drama television series starring an actress who won the Indian Telly Award for Best Actress in a Leading Role.
|
{
"evidence": [
"Kumkum Bhagya",
"Sriti Jha"
],
"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:\nKumkum Bhagya is an Indonesian Hindi-language drama television series starring an actress who won the Indian Telly Award for Best Actress in a Leading Role.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Kumkum Bhagya",
"Sriti Jha"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3319,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
57702
|
May Lee was in the United States in 1999 as a host for The Oxygen Network and in the Asia-Pacific region in 2004 as an anchor for The Asia Wall Street Journal.
|
{
"evidence": [
"May Lee_sentence_5",
"May Lee_sentence_6"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMay Lee was in the United States in 1999 as a host for The Oxygen Network and in the Asia-Pacific region in 2004 as an anchor for The Asia Wall Street Journal.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"May Lee_sentence_5",
"May Lee_sentence_6"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3320,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.334",
"value": "?search="
},
{
"operation": "search",
"time": "27.269",
"value": "May Lee"
},
{
"operation": "Now on",
"time": "28.276",
"value": "May Lee"
},
{
"operation": "Highlighting",
"time": "47.513",
"value": "May Lee_sentence_5"
},
{
"operation": "Highlighting",
"time": "51.397",
"value": "May Lee_sentence_6"
},
{
"operation": "finish",
"time": "297.241",
"value": "finish"
}
]
}
|
81034
|
The Recorder of Dublin's (a judicial office holder in pre-Independence Ireland) responsibility was to keep the peace, and control the number of pubs in the city and by the 1810s the Recorder was hearing roughly 2,000 cases a year.
|
{
"evidence": [
"Recorder of Dublin_sentence_4"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Recorder of Dublin's (a judicial office holder in pre-Independence Ireland) responsibility was to keep the peace, and control the number of pubs in the city and by the 1810s the Recorder was hearing roughly 2,000 cases a year.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Recorder of Dublin_sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3321,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.981",
"value": "?search="
},
{
"operation": "search",
"time": "18.907",
"value": "Contains...The Recorder of Dublin"
},
{
"operation": "Now on",
"time": "20.116",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "24.922",
"value": "Recorder (judge)"
},
{
"operation": "Now on",
"time": "25.93",
"value": "Recorder (judge)"
},
{
"operation": "hyperlink",
"time": "32.832",
"value": "2.1 Ireland and Northern Ireland Recorder (judge)_section_5"
},
{
"operation": "hyperlink",
"time": "42.487",
"value": "Recorder of Dublin"
},
{
"operation": "Now on",
"time": "43.153",
"value": "Recorder of Dublin"
},
{
"operation": "Highlighting",
"time": "48.187",
"value": "Recorder of Dublin_sentence_0"
},
{
"operation": "Highlighting",
"time": "57.943",
"value": "Recorder of Dublin_sentence_3"
},
{
"operation": "Highlighting",
"time": "71.176",
"value": "Recorder of Dublin_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "72.119",
"value": "Recorder of Dublin_sentence_3"
},
{
"operation": "Highlighting deleted",
"time": "73.039",
"value": "Recorder of Dublin_sentence_0"
},
{
"operation": "finish",
"time": "89.071",
"value": "finish"
}
]
}
|
15637
|
HuniePop designed and programmed by Ryan Koons, was a game released only to macOS in two different versions.
|
{
"evidence": [
"HuniePop_sentence_1"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nHuniePop designed and programmed by Ryan Koons, was a game released only to macOS in two different versions.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"HuniePop_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3322,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.373",
"value": "?search="
},
{
"operation": "search",
"time": "12.025",
"value": "HuniePop"
},
{
"operation": "Now on",
"time": "13.051",
"value": "HuniePop"
},
{
"operation": "Highlighting",
"time": "26.206",
"value": "HuniePop_sentence_0"
},
{
"operation": "Highlighting",
"time": "32.17",
"value": "HuniePop_cell_0_3_1"
},
{
"operation": "Highlighting",
"time": "32.191",
"value": "HuniePop_header_cell_0_3_0"
},
{
"operation": "Highlighting",
"time": "32.192",
"value": "HuniePop_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "32.826",
"value": "HuniePop_cell_0_4_1"
},
{
"operation": "Highlighting",
"time": "32.848",
"value": "HuniePop_header_cell_0_4_0"
},
{
"operation": "Highlighting",
"time": "36.191",
"value": "HuniePop_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "37.13",
"value": "HuniePop_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "38.436",
"value": "HuniePop_cell_0_3_1"
},
{
"operation": "Highlighting deleted",
"time": "38.461",
"value": "HuniePop_header_cell_0_3_0"
},
{
"operation": "Highlighting deleted",
"time": "38.972",
"value": "HuniePop_cell_0_4_1"
},
{
"operation": "Highlighting deleted",
"time": "38.995",
"value": "HuniePop_header_cell_0_4_0"
},
{
"operation": "Highlighting deleted",
"time": "38.999",
"value": "HuniePop_header_cell_0_0_0"
},
{
"operation": "finish",
"time": "61.976",
"value": "finish"
}
]
}
|
27586
|
The Attanagalla Polling Division has never matched the final country result, whereas the Gampaha Electoral District has matched the final country result seven out of seven times,
|
{
"evidence": [
"Attanagalla Polling Division_sentence_1"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Attanagalla Polling Division has never matched the final country result, whereas the Gampaha Electoral District has matched the final country result seven out of seven times,\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Attanagalla Polling Division_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3323,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.47",
"value": "?search="
},
{
"operation": "search",
"time": "13.014",
"value": "Attanagalla Polling Division"
},
{
"operation": "Now on",
"time": "14.923",
"value": "Attanagalla Polling Division"
},
{
"operation": "Highlighting",
"time": "22.432",
"value": "Attanagalla Polling Division_sentence_1"
},
{
"operation": "finish",
"time": "50.11",
"value": "finish"
}
]
}
|
41227
|
In 1789, King Heraclius II of Georgia sent his army into Imereti and helped David-Solomon to expel David II (1756 – 11 January 1795) into Akhaltsikhe, then in 1790, he returned with a Turkish force and deposed Solomon II, but was eventually defeated and fled Imereti.
|
{
"evidence": [
"Kingdom of Georgia_sentence_0"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn 1789, King Heraclius II of Georgia sent his army into Imereti and helped David-Solomon to expel David II (1756 – 11 January 1795) into Akhaltsikhe, then in 1790, he returned with a Turkish force and deposed Solomon II, but was eventually defeated and fled Imereti.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Kingdom of Georgia_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3324,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.635",
"value": "?search="
},
{
"operation": "search",
"time": "53.824",
"value": "King Heraclius II"
},
{
"operation": "Now on",
"time": "54.779",
"value": "Heraclius II of Georgia"
},
{
"operation": "Page search",
"time": "73.284",
"value": "imereti"
},
{
"operation": "Page search",
"time": "107.859",
"value": "1789"
},
{
"operation": "Page search",
"time": "133.348",
"value": "akhaltsikhe"
},
{
"operation": "search",
"time": "256.867",
"value": "Kingdom of Georgia"
},
{
"operation": "Now on",
"time": "257.961",
"value": "Kingdom of Georgia"
},
{
"operation": "Page search",
"time": "265.517",
"value": "imereti"
},
{
"operation": "Highlighting",
"time": "300.764",
"value": "Kingdom of Georgia_sentence_0"
},
{
"operation": "finish",
"time": "315.058",
"value": "finish"
}
]
}
|
exfever_4439
|
The Hui people are East Asian ethnoreligious group who speak a Sinitic language spoken primarily in Kazakhstan and Kyrgyztan.
|
{
"evidence": [
"Hui people",
"Dungan language"
],
"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 Hui people are East Asian ethnoreligious group who speak a Sinitic language spoken primarily in Kazakhstan and Kyrgyztan.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Hui people",
"Dungan language"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3325,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
86769
|
Destroying angels are characterized by having a white stalk and gills where the cap can be pure white, or white at the edge and yellowish, pinkish, or tan at the center.
|
{
"evidence": [
"Destroying angel_sentence_8",
"Destroying angel_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:\nDestroying angels are characterized by having a white stalk and gills where the cap can be pure white, or white at the edge and yellowish, pinkish, or tan at the center.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Destroying angel_sentence_8",
"Destroying angel_sentence_9"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3326,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.564",
"value": "?search="
},
{
"operation": "search",
"time": "12.652",
"value": "Destroying angel (disambiguation)"
},
{
"operation": "Now on",
"time": "13.544",
"value": "Destroying angel (disambiguation)"
},
{
"operation": "hyperlink",
"time": "23.422",
"value": "destroying angel"
},
{
"operation": "Now on",
"time": "24.059",
"value": "Destroying angel"
},
{
"operation": "Highlighting",
"time": "37.751",
"value": "Destroying angel_sentence_8"
},
{
"operation": "Highlighting",
"time": "44.804",
"value": "Destroying angel_sentence_9"
},
{
"operation": "finish",
"time": "72.011",
"value": "finish"
}
]
}
|
50881
|
Rufinatscha is recognised as one of Tyrol's most important artist of the 19th century, his works can be said to form a connection between those of Franz Schubert and some of his works have been recorded.
|
{
"evidence": [
"Johann Rufinatscha_sentence_10",
"Johann Rufinatscha_sentence_11",
"Johann Rufinatscha_sentence_23"
],
"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:\nRufinatscha is recognised as one of Tyrol's most important artist of the 19th century, his works can be said to form a connection between those of Franz Schubert and some of his works have been recorded.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Johann Rufinatscha_sentence_10",
"Johann Rufinatscha_sentence_11",
"Johann Rufinatscha_sentence_23"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3327,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.411",
"value": "?search="
},
{
"operation": "search",
"time": "105.172",
"value": "Rufinatscha"
},
{
"operation": "Now on",
"time": "106.044",
"value": "Johann Rufinatscha"
},
{
"operation": "Page search",
"time": "114.065",
"value": "tyrol"
},
{
"operation": "Highlighting",
"time": "122.879",
"value": "Johann Rufinatscha_sentence_10"
},
{
"operation": "Page search",
"time": "143.33",
"value": "Franz Schubert"
},
{
"operation": "Highlighting",
"time": "151.422",
"value": "Johann Rufinatscha_sentence_11"
},
{
"operation": "Page search",
"time": "163.825",
"value": "record"
},
{
"operation": "Highlighting",
"time": "184.511",
"value": "Johann Rufinatscha_sentence_23"
},
{
"operation": "finish",
"time": "200.735",
"value": "finish"
}
]
}
|
74000
|
The only team to receive tobacco sponsorship for the 2007 season, Ferrari had Philip Morris International as its principal sponsor.
|
{
"evidence": [
"Philip Morris International_sentence_88"
],
"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 only team to receive tobacco sponsorship for the 2007 season, Ferrari had Philip Morris International as its principal sponsor.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Philip Morris International_sentence_88"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3328,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.76",
"value": "?search="
},
{
"operation": "search",
"time": "15.163",
"value": "Ferrari Phillip Morris"
},
{
"operation": "Now on",
"time": "16.842",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "21.98",
"value": "Philip Morris International"
},
{
"operation": "Now on",
"time": "23.322",
"value": "Philip Morris International"
},
{
"operation": "Highlighting",
"time": "30.676",
"value": "Philip Morris International_sentence_88"
},
{
"operation": "search",
"time": "57.98",
"value": "2007 Ferrari"
},
{
"operation": "Now on",
"time": "59.318",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "88.75",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "88.893",
"value": "Philip Morris International"
},
{
"operation": "search",
"time": "107.064",
"value": "Phillip Morris tabacco sponsership"
},
{
"operation": "Now on",
"time": "108.145",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "116.583",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "116.791",
"value": "Philip Morris International"
},
{
"operation": "search",
"time": "169.614",
"value": "Ferrari Formula One"
},
{
"operation": "Now on",
"time": "171.641",
"value": "undefined"
},
{
"operation": "search",
"time": "188.121",
"value": "2007 Ferrari season"
},
{
"operation": "Now on",
"time": "189.886",
"value": "undefined"
},
{
"operation": "search",
"time": "208.674",
"value": "Ferrari Formula One"
},
{
"operation": "Now on",
"time": "209.58",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "212.894",
"value": "Ferrari"
},
{
"operation": "Now on",
"time": "214.474",
"value": "Ferrari"
},
{
"operation": "hyperlink",
"time": "247.994",
"value": "Formula One"
},
{
"operation": "Now on",
"time": "256.59",
"value": "Formula One"
},
{
"operation": "finish",
"time": "282.24",
"value": "finish"
}
]
}
|
exfever_3271
|
Marilyn Manson is an American rock band whose each member's name was created from the combination of the last name of a serial killer and the first name of an American actress and model who became one of the most popular sex symbols of the 1940s.
|
{
"evidence": [
"Marilyn Manson (band)",
"Marilyn Monroe"
],
"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:\nMarilyn Manson is an American rock band whose each member's name was created from the combination of the last name of a serial killer and the first name of an American actress and model who became one of the most popular sex symbols of the 1940s.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Marilyn Manson (band)",
"Marilyn Monroe"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3329,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
64191
|
Aleksandras Stulginskis moved to Vilnius during Word War 1, he was one of the founders of the Lithuanian Christian Democratic Party and he was elected to the Council of Lithuania.
|
{
"evidence": [
"Aleksandras Stulginskis_sentence_9",
"Aleksandras Stulginskis_sentence_10",
"Aleksandras Stulginskis_sentence_14"
],
"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:\nAleksandras Stulginskis moved to Vilnius during Word War 1, he was one of the founders of the Lithuanian Christian Democratic Party and he was elected to the Council of Lithuania.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Aleksandras Stulginskis_sentence_9",
"Aleksandras Stulginskis_sentence_10",
"Aleksandras Stulginskis_sentence_14"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3330,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.8",
"value": "?search="
},
{
"operation": "search",
"time": "11.102",
"value": "Aleksandras Stulginskis"
},
{
"operation": "Now on",
"time": "11.79",
"value": "Aleksandras Stulginskis"
},
{
"operation": "Highlighting",
"time": "22.105",
"value": "Aleksandras Stulginskis_sentence_9"
},
{
"operation": "Highlighting",
"time": "24.836",
"value": "Aleksandras Stulginskis_sentence_10"
},
{
"operation": "Highlighting",
"time": "34.502",
"value": "Aleksandras Stulginskis_sentence_14"
},
{
"operation": "finish",
"time": "53.584",
"value": "finish"
}
]
}
|
26280
|
Byron Bay's own Skegss album Holiday Food was reviewed by three critics.
|
{
"evidence": [
"Skegss_sentence_0",
"Skegss_sentence_6",
"Skegss_sentence_27",
"Skegss_sentence_28",
"Skegss_sentence_37"
],
"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:\nByron Bay's own Skegss album Holiday Food was reviewed by three critics.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Skegss_sentence_0",
"Skegss_sentence_6",
"Skegss_sentence_27",
"Skegss_sentence_28",
"Skegss_sentence_37"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3331,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.358",
"value": "?search="
},
{
"operation": "search",
"time": "20.67",
"value": "Byron Bay Record"
},
{
"operation": "Now on",
"time": "21.681",
"value": "Byron Bay Record"
},
{
"operation": "search",
"time": "34.464",
"value": "Contains...Holiday Food"
},
{
"operation": "Now on",
"time": "36.128",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "38.428",
"value": "Skegss"
},
{
"operation": "Now on",
"time": "39.161",
"value": "Skegss"
},
{
"operation": "Highlighting",
"time": "45.115",
"value": "Skegss_sentence_0"
},
{
"operation": "Page search",
"time": "57.452",
"value": "Holiday Food"
},
{
"operation": "hyperlink",
"time": "71.963",
"value": "4 Discography Skegss_section_3"
},
{
"operation": "Highlighting",
"time": "114.815",
"value": "Skegss_sentence_31"
},
{
"operation": "Highlighting deleted",
"time": "120.523",
"value": "Skegss_sentence_31"
},
{
"operation": "search",
"time": "146.023",
"value": "Contains...\"holiday food\""
},
{
"operation": "Now on",
"time": "147.641",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "161.498",
"value": "Skegss"
},
{
"operation": "Now on",
"time": "162.559",
"value": "Skegss"
},
{
"operation": "Page search",
"time": "180.503",
"value": "holiday food"
},
{
"operation": "Highlighting",
"time": "184.966",
"value": "Skegss_sentence_6"
},
{
"operation": "Highlighting",
"time": "194.122",
"value": "Skegss_sentence_27"
},
{
"operation": "Highlighting",
"time": "198.385",
"value": "Skegss_sentence_28"
},
{
"operation": "Highlighting",
"time": "287.775",
"value": "Skegss_sentence_37"
},
{
"operation": "Now on",
"time": "332.17",
"value": "Skegss"
},
{
"operation": "finish",
"time": "362.032",
"value": "finish"
}
]
}
|
exfever_714
|
Steffi Graf became the first tennis player to win all four major championships,?overseen by the International Tennis Federation rather than the governing body of the men's professional tennis circuit.
|
{
"evidence": [
"Steffi Graf",
"Grand Slam (tennis)",
"Association of Tennis Professionals"
],
"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:\nSteffi Graf became the first tennis player to win all four major championships,?overseen by the International Tennis Federation rather than the governing body of the men's professional tennis circuit.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Steffi Graf",
"Grand Slam (tennis)",
"Association of Tennis Professionals"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3332,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4659
|
"Everybody Needs a Best Friend" is a song from the 2012 feature film Ted, with lyrics by an American screenwriter and actor who starred in a 2014 American Western horror film that features an ensemble cast, including Charlize Theron.
|
{
"evidence": [
"Everybody Needs a Best Friend",
"Seth MacFarlane",
"A Million Ways to Die in the West"
],
"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\"Everybody Needs a Best Friend\" is a song from the 2012 feature film Ted, with lyrics by an American screenwriter and actor who starred in a 2014 American Western horror film that features an ensemble cast, including Charlize Theron.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Everybody Needs a Best Friend",
"Seth MacFarlane",
"A Million Ways to Die in the West"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3333,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_924
|
Leonardo di ser Piero da Vinci was an Irish polymath of the High Renaissance whose collective works composed a contribution to later generations of artists matched only by an Italian sculptor, painter, architect, and poet of the High Renaissance born in a medieval and early modern state that was centered in the Italian city of Florence in Tuscany.
|
{
"evidence": [
"Leonardo da Vinci",
"Michelangelo",
"Republic of Florence"
],
"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:\nLeonardo di ser Piero da Vinci was an Irish polymath of the High Renaissance whose collective works composed a contribution to later generations of artists matched only by an Italian sculptor, painter, architect, and poet of the High Renaissance born in a medieval and early modern state that was centered in the Italian city of Florence in Tuscany.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Leonardo da Vinci",
"Michelangelo",
"Republic of Florence"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3334,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
869
|
The BISF is of a futuristic design, with simple architectural devices of projecting window surrounds encasing Crittall Hope windows, and differing cladding to the upper and lower stories deal with the junction between components in an understated fashion, the framework of which is clad on the lower storey with rendering on metal lath.
|
{
"evidence": [
"BISF house_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:\nThe BISF is of a futuristic design, with simple architectural devices of projecting window surrounds encasing Crittall Hope windows, and differing cladding to the upper and lower stories deal with the junction between components in an understated fashion, the framework of which is clad on the lower storey with rendering on metal lath.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"BISF house_sentence_5"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3335,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "320.255",
"value": "BISF house"
},
{
"operation": "Highlighting",
"time": "412.022",
"value": "BISF house_sentence_5"
},
{
"operation": "finish",
"time": "448.564",
"value": "finish"
}
]
}
|
24165
|
Kenneth Lusaka as fronted by the Jubilee Party but failed to win the speaker seat on second round with 42 votes.
|
{
"evidence": [
"Kenneth Lusaka_sentence_9"
],
"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:\nKenneth Lusaka as fronted by the Jubilee Party but failed to win the speaker seat on second round with 42 votes.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Kenneth Lusaka_sentence_9"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3336,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.551",
"value": "?search="
},
{
"operation": "search",
"time": "15.389",
"value": "Kenneth Lusaka"
},
{
"operation": "Now on",
"time": "16.022",
"value": "Kenneth Lusaka"
},
{
"operation": "Highlighting",
"time": "43.624",
"value": "Kenneth Lusaka_sentence_9"
},
{
"operation": "Highlighting",
"time": "61.533",
"value": "Kenneth Lusaka_sentence_7"
},
{
"operation": "Page search",
"time": "69.25",
"value": "fronted"
},
{
"operation": "Highlighting",
"time": "75.707",
"value": "Kenneth Lusaka_sentence_8"
},
{
"operation": "Highlighting deleted",
"time": "78.208",
"value": "Kenneth Lusaka_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "91.428",
"value": "Kenneth Lusaka_sentence_8"
},
{
"operation": "finish",
"time": "101.971",
"value": "finish"
}
]
}
|
exfever_5278
|
Princess Agents is a 2019 television series starring a Chinese actor who rose to fame for his role in Scarlet Heart.
|
{
"evidence": [
"Princess Agents",
"Lin Gengxin"
],
"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:\nPrincess Agents is a 2019 television series starring a Chinese actor who rose to fame for his role in Scarlet Heart.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Princess Agents",
"Lin Gengxin"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3337,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_219
|
The Boeing 737 is the first narrow-body aircraft produced by Boeing.
|
{
"evidence": [
"Boeing 737",
"Turbofan",
"Bypass ratio"
],
"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 Boeing 737 is the first narrow-body aircraft produced by Boeing.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Boeing 737",
"Turbofan",
"Bypass ratio"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3338,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
41689
|
Carlos Villanueva can play basketball and soccer, but he never excelled in the latter so he pursued the NBA.
|
{
"evidence": [
"Charlie Villanueva_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:\nCarlos Villanueva can play basketball and soccer, but he never excelled in the latter so he pursued the NBA.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Charlie Villanueva_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3339,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.892",
"value": "?search="
},
{
"operation": "search",
"time": "18.285",
"value": "Carlos Villanueva"
},
{
"operation": "Now on",
"time": "19.226",
"value": "Carlos Villanueva"
},
{
"operation": "hyperlink",
"time": "38.079",
"value": "Carlos Villanueva López"
},
{
"operation": "Now on",
"time": "38.507",
"value": "Carlos Villanueva López"
},
{
"operation": "back-button-clicked",
"time": "95.578",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "95.621",
"value": "Carlos Villanueva"
},
{
"operation": "Highlighting",
"time": "101.934",
"value": "Carlos Villanueva_item_0_2"
},
{
"operation": "Highlighting",
"time": "101.935",
"value": "Carlos Villanueva_title"
},
{
"operation": "Highlighting",
"time": "103.376",
"value": "Carlos Villanueva_item_0_3"
},
{
"operation": "Highlighting deleted",
"time": "104.086",
"value": "Carlos Villanueva_item_0_3"
},
{
"operation": "Highlighting deleted",
"time": "104.688",
"value": "Carlos Villanueva_item_0_2"
},
{
"operation": "Highlighting deleted",
"time": "104.689",
"value": "Carlos Villanueva_title"
},
{
"operation": "hyperlink",
"time": "105.898",
"value": "Carlos Villanueva (footballer)"
},
{
"operation": "Now on",
"time": "106.367",
"value": "Carlos Villanueva (footballer)"
},
{
"operation": "Page search",
"time": "135.028",
"value": "basketball"
},
{
"operation": "back-button-clicked",
"time": "139.992",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "140.045",
"value": "Carlos Villanueva"
},
{
"operation": "hyperlink",
"time": "141.899",
"value": "Carlos Villanueva (baseball)"
},
{
"operation": "Now on",
"time": "142.335",
"value": "Carlos Villanueva (baseball)"
},
{
"operation": "Page search",
"time": "150.387",
"value": "basketball"
},
{
"operation": "Page search",
"time": "171.219",
"value": "soccer"
},
{
"operation": "back-button-clicked",
"time": "211.075",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "211.122",
"value": "Carlos Villanueva"
},
{
"operation": "search",
"time": "492.139",
"value": "charlie Villanueva"
},
{
"operation": "Now on",
"time": "494.152",
"value": "Charlie Villanueva"
},
{
"operation": "back-button-clicked",
"time": "520.882",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "520.926",
"value": "Carlos Villanueva"
},
{
"operation": "hyperlink",
"time": "532.439",
"value": "Carlos Villanueva (footballer)"
},
{
"operation": "Now on",
"time": "532.754",
"value": "Carlos Villanueva (footballer)"
},
{
"operation": "search",
"time": "561.285",
"value": "Carlos Villanueva nba"
},
{
"operation": "Now on",
"time": "565.013",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "593.723",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "593.799",
"value": "Carlos Villanueva (footballer)"
},
{
"operation": "back-button-clicked",
"time": "596.179",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "596.182",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "596.234",
"value": "Carlos Villanueva"
},
{
"operation": "hyperlink",
"time": "597.83",
"value": "Carlos Villanueva (baseball)"
},
{
"operation": "Now on",
"time": "598.127",
"value": "Carlos Villanueva (baseball)"
},
{
"operation": "Page search",
"time": "629.735",
"value": "soccer"
},
{
"operation": "Page search",
"time": "629.945",
"value": "soccer"
},
{
"operation": "Page search",
"time": "637.431",
"value": "football"
},
{
"operation": "Page search",
"time": "637.729",
"value": "football"
},
{
"operation": "back-button-clicked",
"time": "641.964",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "642.03",
"value": "Carlos Villanueva"
},
{
"operation": "search",
"time": "650.632",
"value": "Carlos Villanueva Fuentes"
},
{
"operation": "Now on",
"time": "651.642",
"value": "Carlo Villanueva Fuentes"
},
{
"operation": "back-button-clicked",
"time": "659.179",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "659.217",
"value": "Carlos Villanueva"
},
{
"operation": "search",
"time": "674.272",
"value": "charlie Villanueva"
},
{
"operation": "Now on",
"time": "674.714",
"value": "Charlie Villanueva"
},
{
"operation": "Highlighting",
"time": "678.223",
"value": "Charlie Villanueva_sentence_0"
},
{
"operation": "finish",
"time": "687.722",
"value": "finish"
}
]
}
|
93305
|
Sundre, in the Mountain View County, Central Alberta Region, Alberta, Canada, and under the governance of Mayor Terry Leslie and the Sundre Town Council, has a total population of 2,729 in a land area of 11.11 km (4.29 sq mi), based in the 2016 Census of Population conducted neither by Statistics Canada nor by Service Canada.
|
{
"evidence": [
"Sundre_sentence_6"
],
"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:\nSundre, in the Mountain View County, Central Alberta Region, Alberta, Canada, and under the governance of Mayor Terry Leslie and the Sundre Town Council, has a total population of 2,729 in a land area of 11.11 km (4.29 sq mi), based in the 2016 Census of Population conducted neither by Statistics Canada nor by Service Canada.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Sundre_sentence_6"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3340,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.868",
"value": "?search="
},
{
"operation": "search",
"time": "14.072",
"value": "Sundre"
},
{
"operation": "Now on",
"time": "15.31",
"value": "Sundre"
},
{
"operation": "Highlighting",
"time": "22.259",
"value": "Sundre_sentence_2"
},
{
"operation": "Highlighting",
"time": "27.019",
"value": "Sundre_cell_0_3_1"
},
{
"operation": "Highlighting",
"time": "27.041",
"value": "Sundre_header_cell_0_3_0"
},
{
"operation": "Highlighting",
"time": "27.042",
"value": "Sundre_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "31.015",
"value": "Sundre_cell_0_1_1"
},
{
"operation": "Highlighting",
"time": "31.033",
"value": "Sundre_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "46.892",
"value": "Sundre_cell_0_10_1"
},
{
"operation": "Highlighting",
"time": "46.912",
"value": "Sundre_header_cell_0_10_0"
},
{
"operation": "Highlighting",
"time": "46.913",
"value": "Sundre_header_cell_0_9_0"
},
{
"operation": "Highlighting",
"time": "48.435",
"value": "Sundre_cell_0_11_1"
},
{
"operation": "Highlighting",
"time": "48.453",
"value": "Sundre_header_cell_0_11_0"
},
{
"operation": "Highlighting",
"time": "60.695",
"value": "Sundre_cell_0_18_1"
},
{
"operation": "Highlighting",
"time": "60.718",
"value": "Sundre_header_cell_0_18_0"
},
{
"operation": "Highlighting",
"time": "60.718",
"value": "Sundre_header_cell_0_17_0"
},
{
"operation": "Highlighting",
"time": "66.065",
"value": "Sundre_cell_0_15_1"
},
{
"operation": "Highlighting",
"time": "66.085",
"value": "Sundre_header_cell_0_15_0"
},
{
"operation": "Highlighting",
"time": "66.085",
"value": "Sundre_header_cell_0_14_0"
},
{
"operation": "Highlighting",
"time": "85.044",
"value": "Sundre_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "86.804",
"value": "Sundre_cell_0_15_1"
},
{
"operation": "Highlighting deleted",
"time": "86.826",
"value": "Sundre_header_cell_0_15_0"
},
{
"operation": "Highlighting deleted",
"time": "86.827",
"value": "Sundre_header_cell_0_14_0"
},
{
"operation": "Highlighting",
"time": "87.282",
"value": "Sundre_cell_0_15_1"
},
{
"operation": "Highlighting",
"time": "87.302",
"value": "Sundre_header_cell_0_15_0"
},
{
"operation": "Highlighting",
"time": "87.302",
"value": "Sundre_header_cell_0_14_0"
},
{
"operation": "Highlighting deleted",
"time": "88.237",
"value": "Sundre_cell_0_18_1"
},
{
"operation": "Highlighting deleted",
"time": "88.256",
"value": "Sundre_header_cell_0_18_0"
},
{
"operation": "Highlighting deleted",
"time": "88.258",
"value": "Sundre_header_cell_0_17_0"
},
{
"operation": "Highlighting deleted",
"time": "97.016",
"value": "Sundre_cell_0_15_1"
},
{
"operation": "Highlighting deleted",
"time": "97.035",
"value": "Sundre_header_cell_0_15_0"
},
{
"operation": "Highlighting deleted",
"time": "97.038",
"value": "Sundre_header_cell_0_14_0"
},
{
"operation": "Highlighting",
"time": "97.518",
"value": "Sundre_cell_0_12_1"
},
{
"operation": "Highlighting",
"time": "97.542",
"value": "Sundre_header_cell_0_12_0"
},
{
"operation": "Highlighting deleted",
"time": "98.761",
"value": "Sundre_cell_0_12_1"
},
{
"operation": "Highlighting deleted",
"time": "98.784",
"value": "Sundre_header_cell_0_12_0"
},
{
"operation": "Highlighting deleted",
"time": "99.318",
"value": "Sundre_cell_0_11_1"
},
{
"operation": "Highlighting deleted",
"time": "99.339",
"value": "Sundre_header_cell_0_11_0"
},
{
"operation": "Highlighting deleted",
"time": "99.758",
"value": "Sundre_cell_0_10_1"
},
{
"operation": "Highlighting deleted",
"time": "99.778",
"value": "Sundre_header_cell_0_10_0"
},
{
"operation": "Highlighting deleted",
"time": "99.78",
"value": "Sundre_header_cell_0_9_0"
},
{
"operation": "Highlighting deleted",
"time": "101.656",
"value": "Sundre_cell_0_3_1"
},
{
"operation": "Highlighting deleted",
"time": "101.673",
"value": "Sundre_header_cell_0_3_0"
},
{
"operation": "Highlighting deleted",
"time": "102.698",
"value": "Sundre_cell_0_1_1"
},
{
"operation": "Highlighting deleted",
"time": "102.715",
"value": "Sundre_header_cell_0_1_0"
},
{
"operation": "Highlighting deleted",
"time": "102.718",
"value": "Sundre_header_cell_0_0_0"
},
{
"operation": "Highlighting deleted",
"time": "103.43",
"value": "Sundre_sentence_2"
},
{
"operation": "finish",
"time": "115.031",
"value": "finish"
}
]
}
|
64455
|
Greater cane rat can grow to nearly 60 cm, it has rounded ears and is a small family of African hystricognath rodents.
|
{
"evidence": [
"Greater cane rat_sentence_0",
"Greater cane rat_sentence_2",
"Greater cane rat_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:\nGreater cane rat can grow to nearly 60 cm, it has rounded ears and is a small family of African hystricognath rodents.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Greater cane rat_sentence_0",
"Greater cane rat_sentence_2",
"Greater cane rat_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3341,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.725",
"value": "?search="
},
{
"operation": "search",
"time": "11.465",
"value": "Greater cane rat"
},
{
"operation": "Now on",
"time": "12.931",
"value": "Greater cane rat"
},
{
"operation": "Highlighting",
"time": "17.097",
"value": "Greater cane rat_sentence_0"
},
{
"operation": "Highlighting",
"time": "25.065",
"value": "Greater cane rat_sentence_2"
},
{
"operation": "Highlighting",
"time": "29.935",
"value": "Greater cane rat_sentence_3"
},
{
"operation": "finish",
"time": "50.292",
"value": "finish"
}
]
}
|
exfever_913
|
Bill Cosby is an American author who starred in a show with an American director.
|
{
"evidence": [
"Bill Cosby",
"Robert Culp"
],
"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:\nBill Cosby is an American author who starred in a show with an American director.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Bill Cosby",
"Robert Culp"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3342,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7946
|
The Muppet Christmas Carol is a 1995 American Christmas film starring Michael Caine alongside Muppet performer, an American puppeteer who performed in a children's musical fantasy comedy puppet television series about interconnected societies of Muppet creatures.
|
{
"evidence": [
"The Muppet Christmas Carol",
"Jerry Nelson",
"Fraggle Rock"
],
"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 Muppet Christmas Carol is a 1995 American Christmas film starring Michael Caine alongside Muppet performer, an American puppeteer who performed in a children's musical fantasy comedy puppet television series about interconnected societies of Muppet creatures.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Muppet Christmas Carol",
"Jerry Nelson",
"Fraggle Rock"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3343,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2311
|
Winona Ryder is an American actress who rose to prominence in a 1990 American fantasy romance film directed by Tim Burton.
|
{
"evidence": [
"Winona Ryder",
"Edward Scissorhands"
],
"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:\nWinona Ryder is an American actress who rose to prominence in a 1990 American fantasy romance film directed by Tim Burton.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Winona Ryder",
"Edward Scissorhands"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3344,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8364
|
Event management is the application of project management to the creation and development of small and/or large-scale personal or corporate events such as a gathering of individuals who meet at an arranged place and time in order to discuss or engage in some common interest.
|
{
"evidence": [
"Event management",
"Convention (meeting)"
],
"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:\nEvent management is the application of project management to the creation and development of small and/or large-scale personal or corporate events such as a gathering of individuals who meet at an arranged place and time in order to discuss or engage in some common interest.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Event management",
"Convention (meeting)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3345,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6750
|
Iron Man 3 is a 2013 Australian superhero film based on the Marvel Comics character Iron Man and the sequel to a 2010 American superhero film based on the Marvel Comics character Iron Man.
|
{
"evidence": [
"Iron Man 3",
"Iron Man 2"
],
"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:\nIron Man 3 is a 2013 Australian superhero film based on the Marvel Comics character Iron Man and the sequel to a 2010 American superhero film based on the Marvel Comics character Iron Man.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Iron Man 3",
"Iron Man 2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3346,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7022
|
Star Wars is the most prominent American epic space opera film directed by Richard Marquand.
|
{
"evidence": [
"Caroline Blakiston",
"Return of the Jedi",
"Mark Hamill"
],
"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:\nStar Wars is the most prominent American epic space opera film directed by Richard Marquand.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Caroline Blakiston",
"Return of the Jedi",
"Mark Hamill"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3347,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7455
|
Chet Atkins was the youngest musician who helped create the Nashville sound.
|
{
"evidence": [
"Chet Atkins",
"George Barnes (musician)"
],
"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:\nChet Atkins was the youngest musician who helped create the Nashville sound.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Chet Atkins",
"George Barnes (musician)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3348,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
38144
|
1959 British East Indies Championships was the third edition of the track and field competition between British colony nations in the Caribbean which was held in Georgetown in British Guiana and three new men's events were added to the programme: 3000 metres steeplechase, half marathon, and the 3000 metres walk.
|
{
"evidence": [
"1959 British West Indies Championships_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:\n1959 British East Indies Championships was the third edition of the track and field competition between British colony nations in the Caribbean which was held in Georgetown in British Guiana and three new men's events were added to the programme: 3000 metres steeplechase, half marathon, and the 3000 metres walk.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"1959 British West Indies Championships_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3349,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.281",
"value": "?search="
},
{
"operation": "search",
"time": "10.944",
"value": "1959 British East Indies Championships"
},
{
"operation": "Now on",
"time": "13.054",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "19.298",
"value": "1989 british east indies championship"
},
{
"operation": "Now on",
"time": "20.481",
"value": "undefined"
},
{
"operation": "search",
"time": "24.637",
"value": "1959 British East Indies Championships"
},
{
"operation": "Now on",
"time": "25.798",
"value": "undefined"
},
{
"operation": "search",
"time": "73.597",
"value": "1959 British East Indies Championship"
},
{
"operation": "Now on",
"time": "74.587",
"value": "undefined"
},
{
"operation": "search",
"time": "80.901",
"value": "1959 British East Indies"
},
{
"operation": "Now on",
"time": "81.871",
"value": "undefined"
},
{
"operation": "search",
"time": "94.265",
"value": "1959 British East Indies track and field"
},
{
"operation": "Now on",
"time": "95.706",
"value": "undefined"
},
{
"operation": "search",
"time": "108.115",
"value": "British colony nations in the Caribbean which was held in Georgetown in British Guiana"
},
{
"operation": "Now on",
"time": "109.967",
"value": "undefined"
},
{
"operation": "search",
"time": "649.719",
"value": "Contains...1959 british east indies"
},
{
"operation": "Now on",
"time": "650.889",
"value": "undefined"
},
{
"operation": "search",
"time": "659.345",
"value": "track and field competition between British colony nations in the Caribbean"
},
{
"operation": "Now on",
"time": "660.968",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "677.056",
"value": "1959 British West Indies Championships"
},
{
"operation": "Now on",
"time": "678.285",
"value": "1959 British West Indies Championships"
},
{
"operation": "Highlighting",
"time": "684.648",
"value": "1959 British West Indies Championships_sentence_0"
},
{
"operation": "finish",
"time": "695.918",
"value": "finish"
}
]
}
|
80922
|
Giacinto Morera who was awarded the 1878 laurea in engineering and the 1879 laurea in mathematics, attended courses held by Enrico d'Ovidio, Angelo Genocchi and Francesco Siacci (a mathematician, ballistician and officer of the army, who he called his mentor in life).
|
{
"evidence": [
"Giacinto Morera_sentence_8",
"Giacinto Morera_sentence_9",
"Francesco Siacci_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:\nGiacinto Morera who was awarded the 1878 laurea in engineering and the 1879 laurea in mathematics, attended courses held by Enrico d'Ovidio, Angelo Genocchi and Francesco Siacci (a mathematician, ballistician and officer of the army, who he called his mentor in life).\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Giacinto Morera_sentence_8",
"Giacinto Morera_sentence_9",
"Francesco Siacci_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3350,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.98",
"value": "?search="
},
{
"operation": "search",
"time": "42.421",
"value": "Giacinto Morera"
},
{
"operation": "Now on",
"time": "43.222",
"value": "Giacinto Morera"
},
{
"operation": "Page search",
"time": "49.515",
"value": "1878"
},
{
"operation": "Page search",
"time": "56.821",
"value": "1878"
},
{
"operation": "Highlighting",
"time": "79.503",
"value": "Giacinto Morera_sentence_8"
},
{
"operation": "Highlighting",
"time": "125.358",
"value": "Giacinto Morera_sentence_9"
},
{
"operation": "hyperlink",
"time": "131.117",
"value": "Angelo Genocchi"
},
{
"operation": "Now on",
"time": "131.999",
"value": "Angelo Genocchi"
},
{
"operation": "Highlighting",
"time": "143.076",
"value": "Angelo Genocchi_sentence_0"
},
{
"operation": "Highlighting",
"time": "143.077",
"value": "Angelo Genocchi_title"
},
{
"operation": "Highlighting deleted",
"time": "175.67",
"value": "Angelo Genocchi_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "175.671",
"value": "Angelo Genocchi_title"
},
{
"operation": "back-button-clicked",
"time": "181.641",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "181.773",
"value": "Giacinto Morera"
},
{
"operation": "hyperlink",
"time": "184.22",
"value": "Francesco Siacci"
},
{
"operation": "Now on",
"time": "184.694",
"value": "Francesco Siacci"
},
{
"operation": "Highlighting",
"time": "187.269",
"value": "Francesco Siacci_sentence_0"
},
{
"operation": "back-button-clicked",
"time": "189.157",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "189.299",
"value": "Giacinto Morera"
},
{
"operation": "finish",
"time": "314.361",
"value": "finish"
}
]
}
|
exfever_2960
|
For his role in Apocalypse Now, Robert Duvall won his first Academy Award for Best Supporting Actor.
|
{
"evidence": [
"Apocalypse Now",
"Martin Sheen",
"Badlands (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:\nFor his role in Apocalypse Now, Robert Duvall won his first Academy Award for Best Supporting Actor.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Apocalypse Now",
"Martin Sheen",
"Badlands (film)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3351,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7490
|
The Ten Commandments is a 1956 epic religious drama film shot in a higher resolution, widescreen variant of the 35 mm motion picture film format, which was created in 1955.
|
{
"evidence": [
"The Ten Commandments (1956 film)",
"VistaVision"
],
"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 Ten Commandments is a 1956 epic religious drama film shot in a higher resolution, widescreen variant of the 35 mm motion picture film format, which was created in 1955.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Ten Commandments (1956 film)",
"VistaVision"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3352,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
27156
|
During the campaign against Highgate Rabbit Farm by the Animal liberation movement, the activists' efforts never led to any facility closures.
|
{
"evidence": [
"Campaign against Highgate Rabbit Farm_sentence_0",
"Campaign against Highgate Rabbit Farm_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:\nDuring the campaign against Highgate Rabbit Farm by the Animal liberation movement, the activists' efforts never led to any facility closures.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Campaign against Highgate Rabbit Farm_sentence_0",
"Campaign against Highgate Rabbit Farm_sentence_2"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3353,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.209",
"value": "?search="
},
{
"operation": "search",
"time": "159.09",
"value": "Highgate Rabbit Farm"
},
{
"operation": "Now on",
"time": "159.845",
"value": "Campaign against Highgate Rabbit Farm"
},
{
"operation": "Highlighting",
"time": "167.035",
"value": "Campaign against Highgate Rabbit Farm_sentence_0"
},
{
"operation": "Highlighting",
"time": "172.017",
"value": "Campaign against Highgate Rabbit Farm_sentence_2"
},
{
"operation": "Page search",
"time": "183.854",
"value": "facility close"
},
{
"operation": "Page search",
"time": "239.689",
"value": "close"
},
{
"operation": "finish",
"time": "269.498",
"value": "finish"
}
]
}
|
exfever_3779
|
Paul von Hindenburg was a German Field Marshal and statesman who was appointed as Chancellor of Germany by an Austrian-born German politician who was the dictator of Germany from 1933 until his death in 1965.
|
{
"evidence": [
"Paul von Hindenburg",
"Adolf Hitler"
],
"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:\nPaul von Hindenburg was a German Field Marshal and statesman who was appointed as Chancellor of Germany by an Austrian-born German politician who was the dictator of Germany from 1933 until his death in 1965.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Paul von Hindenburg",
"Adolf Hitler"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3354,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4029
|
Monster's Ball is the first feature film directed by Marc Forster.
|
{
"evidence": [
"Stranger than Fiction (2006 film)",
"Marc Forster"
],
"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:\nMonster's Ball is the first feature film directed by Marc Forster.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Stranger than Fiction (2006 film)",
"Marc Forster"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3355,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4887
|
David Judah Simon is an American author who wrote the book that makes the basis for Homicide: Life on the Street, a series by an American English-language commercial broadcast television and radio network which is owned by Comcast through NBCUniversal.
|
{
"evidence": [
"David Simon",
"NBC"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDavid Judah Simon is an American author who wrote the book that makes the basis for Homicide: Life on the Street, a series by an American English-language commercial broadcast television and radio network which is owned by Comcast through NBCUniversal.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"David Simon",
"NBC"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3356,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5095
|
The Crown is a historical drama television series about the reign of Queen Elizabeth II produced by a British film and television production company.
|
{
"evidence": [
"The Crown (TV series)",
"Left Bank Pictures"
],
"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 Crown is a historical drama television series about the reign of Queen Elizabeth II produced by a British film and television production company.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Crown (TV series)",
"Left Bank Pictures"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3357,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
3534
|
Mosa Tegels' (a four-star design hotel located in the Wyck district in Maastricht, Netherlands) launch campaign started in February 2008 where the ingredients stood for: Style (design, architecture and art), Sex (the passion in daily life) and Soup (cuisine, authenticity, and genuine), and the elements of the launch campaign return in the concept of Townhouse by creating an atmosphere of a Maastricht living room.
|
{
"evidence": [
"Townhouse Designhotel Maastricht_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:\nMosa Tegels' (a four-star design hotel located in the Wyck district in Maastricht, Netherlands) launch campaign started in February 2008 where the ingredients stood for: Style (design, architecture and art), Sex (the passion in daily life) and Soup (cuisine, authenticity, and genuine), and the elements of the launch campaign return in the concept of Townhouse by creating an atmosphere of a Maastricht living room.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Townhouse Designhotel Maastricht_sentence_3"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3358,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "19.345",
"value": "mosa tegels"
},
{
"operation": "hyperlink",
"time": "25.314",
"value": "Townhouse Designhotel Maastricht"
},
{
"operation": "Page search",
"time": "29.143",
"value": "mosa te"
},
{
"operation": "Highlighting",
"time": "31.374",
"value": "Townhouse Designhotel Maastricht_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "52.877",
"value": "Townhouse Designhotel Maastricht_sentence_6"
},
{
"operation": "Page search",
"time": "72.719",
"value": "maastri"
},
{
"operation": "page-search-reset",
"time": "76.676",
"value": "page-search-reset"
},
{
"operation": "Page search",
"time": "83.044",
"value": "mosa tege"
},
{
"operation": "Page search",
"time": "97.587",
"value": "2008"
},
{
"operation": "search",
"time": "119.724",
"value": "mosa tegels"
},
{
"operation": "search",
"time": "133.321",
"value": "Maastricht, Netherlands"
},
{
"operation": "Page search",
"time": "143.557",
"value": "mosa teg"
},
{
"operation": "search",
"time": "183.719",
"value": "Townhouse Designhotel Maastricht"
},
{
"operation": "Highlighting",
"time": "216.968",
"value": "Townhouse Designhotel Maastricht_sentence_3"
},
{
"operation": "Highlighting deleted",
"time": "223.843",
"value": "Townhouse Designhotel Maastricht_sentence_3"
},
{
"operation": "Page search",
"time": "227.969",
"value": "mosa teg"
},
{
"operation": "Highlighting",
"time": "237.83",
"value": "Townhouse Designhotel Maastricht_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "238.177",
"value": "Townhouse Designhotel Maastricht_sentence_1"
},
{
"operation": "Highlighting",
"time": "239.418",
"value": "Townhouse Designhotel Maastricht_sentence_3"
},
{
"operation": "finish",
"time": "249.076",
"value": "finish"
}
]
}
|
9151
|
Tharanga Goonetilleke graduated from Converse College in South Carolina in 2005 with a Bachelor of Music degree in Vocal Performance and a minor in Biology; and she entered the Metropolitan Opera National Council Auditions (an annual singing competition sponsored by the Metropolitan Opera) representing South Carolina, and won.
|
{
"evidence": [
"Tharanga Goonetilleke_sentence_5",
"Tharanga Goonetilleke_sentence_6",
"Metropolitan Opera National Council Auditions_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:\nTharanga Goonetilleke graduated from Converse College in South Carolina in 2005 with a Bachelor of Music degree in Vocal Performance and a minor in Biology; and she entered the Metropolitan Opera National Council Auditions (an annual singing competition sponsored by the Metropolitan Opera) representing South Carolina, and won.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Tharanga Goonetilleke_sentence_5",
"Tharanga Goonetilleke_sentence_6",
"Metropolitan Opera National Council Auditions_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3359,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.977",
"value": "?search="
},
{
"operation": "search",
"time": "13.317",
"value": "Tharanga Goonetilleke"
},
{
"operation": "Now on",
"time": "14.006",
"value": "Tharanga Goonetilleke"
},
{
"operation": "Highlighting",
"time": "27.048",
"value": "Tharanga Goonetilleke_sentence_5"
},
{
"operation": "Highlighting",
"time": "54.488",
"value": "Tharanga Goonetilleke_sentence_6"
},
{
"operation": "hyperlink",
"time": "61.664",
"value": "Metropolitan Opera National Council Auditions"
},
{
"operation": "Now on",
"time": "62.393",
"value": "Metropolitan Opera National Council Auditions"
},
{
"operation": "Highlighting",
"time": "65.008",
"value": "Metropolitan Opera National Council Auditions_sentence_0"
},
{
"operation": "finish",
"time": "72.445",
"value": "finish"
}
]
}
|
62440
|
Ireland competed at the 2011 World Championships in Athletics from August 27 to September 4 in Daegu, South Korea, and during this event, only two men out of the six men from Ireland finished the finals, and they were Ciaran O'Lionaird (1500 meters) and Alistair Ian Cragg (5000 meters).
|
{
"evidence": [
"2011 World Championships in Athletics_sentence_1",
"2011 World Championships in Athletics_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:\nIreland competed at the 2011 World Championships in Athletics from August 27 to September 4 in Daegu, South Korea, and during this event, only two men out of the six men from Ireland finished the finals, and they were Ciaran O'Lionaird (1500 meters) and Alistair Ian Cragg (5000 meters).\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"2011 World Championships in Athletics_sentence_1",
"2011 World Championships in Athletics_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3360,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "2.493",
"value": "?search="
},
{
"operation": "Now on",
"time": "2398.039",
"value": "?search="
},
{
"operation": "search",
"time": "2457.412",
"value": "he 2011 World Championships in Athletics"
},
{
"operation": "Now on",
"time": "2461.567",
"value": "undefined"
},
{
"operation": "search",
"time": "2501.063",
"value": "2011 World Championships in Athletics"
},
{
"operation": "Now on",
"time": "2507.485",
"value": "2011 World Championships in Athletics"
},
{
"operation": "Highlighting",
"time": "2522.935",
"value": "2011 World Championships in Athletics_sentence_1"
},
{
"operation": "Highlighting",
"time": "2525.044",
"value": "2011 World Championships in Athletics_sentence_0"
},
{
"operation": "search",
"time": "2620.536",
"value": "Ciaran O'Lionaird"
},
{
"operation": "Now on",
"time": "2622.451",
"value": "Ciarán Ó Lionáird"
},
{
"operation": "hyperlink",
"time": "2634.968",
"value": "2011 IAAF World Championships"
},
{
"operation": "Now on",
"time": "2640.679",
"value": "2011 World Championships in Athletics"
},
{
"operation": "hyperlink",
"time": "2704.655",
"value": "2011 World Championships in Athletics qualification standards"
},
{
"operation": "Now on",
"time": "2706.245",
"value": "2011 World Championships in Athletics qualification standards"
},
{
"operation": "back-button-clicked",
"time": "2710.722",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "2711.045",
"value": "2011 World Championships in Athletics"
},
{
"operation": "search",
"time": "2782.562",
"value": "Alistair Ian Cragg"
},
{
"operation": "Now on",
"time": "2785.081",
"value": "Alistair Cragg"
},
{
"operation": "hyperlink",
"time": "2809.041",
"value": "2011 World Championships in Athletics"
},
{
"operation": "Now on",
"time": "2810.388",
"value": "2011 World Championships in Athletics"
},
{
"operation": "finish",
"time": "2881.233",
"value": "finish"
}
]
}
|
exfever_3052
|
Zachary Galifianakis is an actor who starred in the 2016 American crime comedy film based on the October 1997 Loomis Fargo robbery in South Carolina.
|
{
"evidence": [
"Zach Galifianakis",
"Masterminds (2016 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:\nZachary Galifianakis is an actor who starred in the 2016 American crime comedy film based on the October 1997 Loomis Fargo robbery in South Carolina.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Zach Galifianakis",
"Masterminds (2016 film)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3361,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
43611
|
The story in The Cure for Everything, by writer Severna Park, includes two characters named Maria and N'Lykli, as well as a woman from a North American tribe who could be the source of a hundred new vaccines.
|
{
"evidence": [
"The Cure for Everything_sentence_0",
"The Cure for Everything_sentence_3",
"The Cure for Everything_sentence_5",
"The Cure for Everything_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 story in The Cure for Everything, by writer Severna Park, includes two characters named Maria and N'Lykli, as well as a woman from a North American tribe who could be the source of a hundred new vaccines.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"The Cure for Everything_sentence_0",
"The Cure for Everything_sentence_3",
"The Cure for Everything_sentence_5",
"The Cure for Everything_sentence_8"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3362,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.631",
"value": "?search="
},
{
"operation": "search",
"time": "180.805",
"value": "The Cure for Everything"
},
{
"operation": "Now on",
"time": "181.698",
"value": "The Cure for Everything"
},
{
"operation": "Highlighting",
"time": "392.749",
"value": "The Cure for Everything_sentence_0"
},
{
"operation": "Highlighting",
"time": "408.709",
"value": "The Cure for Everything_sentence_3"
},
{
"operation": "Highlighting",
"time": "417.831",
"value": "The Cure for Everything_sentence_5"
},
{
"operation": "Highlighting",
"time": "443.829",
"value": "The Cure for Everything_sentence_8"
},
{
"operation": "Page search",
"time": "461.361",
"value": "tribe"
},
{
"operation": "search",
"time": "487.772",
"value": "Xingu"
},
{
"operation": "Now on",
"time": "488.729",
"value": "Xingu"
},
{
"operation": "back-button-clicked",
"time": "494.117",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "494.141",
"value": "The Cure for Everything"
},
{
"operation": "Page search",
"time": "510.137",
"value": "america"
},
{
"operation": "finish",
"time": "514.943",
"value": "finish"
}
]
}
|
47964
|
Both First and Sexond degree murders will get you imprisoned for a long time accordjng to the Florida Law.
|
{
"evidence": [
"Florida_sentence_446",
"Florida_sentence_448",
"Florida_sentence_447",
"Florida_sentence_449"
],
"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:\nBoth First and Sexond degree murders will get you imprisoned for a long time accordjng to the Florida Law.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Florida_sentence_446",
"Florida_sentence_448",
"Florida_sentence_447",
"Florida_sentence_449"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3363,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.934",
"value": "?search="
},
{
"operation": "search",
"time": "21.686",
"value": "Florida Law First and Second degree murder"
},
{
"operation": "Now on",
"time": "24.496",
"value": "undefined"
},
{
"operation": "search",
"time": "34.037",
"value": "Florida Law"
},
{
"operation": "Now on",
"time": "34.685",
"value": "Law of Florida"
},
{
"operation": "hyperlink",
"time": "37.884",
"value": "Florida Statutes"
},
{
"operation": "Now on",
"time": "38.177",
"value": "Florida Statutes"
},
{
"operation": "hyperlink",
"time": "43.772",
"value": "Laws of Florida"
},
{
"operation": "Now on",
"time": "44.151",
"value": "Laws of Florida"
},
{
"operation": "search",
"time": "57.828",
"value": "First Degree murder in Florida penalty"
},
{
"operation": "Now on",
"time": "58.788",
"value": "undefined"
},
{
"operation": "search",
"time": "73.123",
"value": "murder penalty in Florida"
},
{
"operation": "Now on",
"time": "73.633",
"value": "undefined"
},
{
"operation": "search",
"time": "114.321",
"value": "imprisonment sentences in Florida"
},
{
"operation": "Now on",
"time": "116.979",
"value": "undefined"
},
{
"operation": "search",
"time": "129.944",
"value": "Florida Law"
},
{
"operation": "Now on",
"time": "130.357",
"value": "Florida law"
},
{
"operation": "hyperlink",
"time": "138.464",
"value": "Florida Statutes"
},
{
"operation": "Now on",
"time": "138.749",
"value": "Florida Statutes"
},
{
"operation": "hyperlink",
"time": "154.671",
"value": "Capital punishment in Florida"
},
{
"operation": "Now on",
"time": "155.253",
"value": "Capital punishment in Florida"
},
{
"operation": "hyperlink",
"time": "159.223",
"value": "2 Capital crimes Capital punishment in Florida_section_1"
},
{
"operation": "back-button-clicked",
"time": "219.47",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "219.518",
"value": "Florida Statutes"
},
{
"operation": "hyperlink",
"time": "221.381",
"value": "Felony murder rule (Florida)"
},
{
"operation": "Now on",
"time": "221.78",
"value": "Felony murder rule (Florida)"
},
{
"operation": "Highlighting",
"time": "269.5",
"value": "Felony murder rule (Florida)_sentence_8"
},
{
"operation": "Highlighting",
"time": "269.501",
"value": "Felony murder rule (Florida)_title"
},
{
"operation": "Highlighting",
"time": "269.501",
"value": "Felony murder rule (Florida)_section_3"
},
{
"operation": "Highlighting deleted",
"time": "278.035",
"value": "Felony murder rule (Florida)_sentence_8"
},
{
"operation": "Highlighting deleted",
"time": "278.035",
"value": "Felony murder rule (Florida)_title"
},
{
"operation": "Highlighting deleted",
"time": "278.037",
"value": "Felony murder rule (Florida)_section_3"
},
{
"operation": "hyperlink",
"time": "296.81",
"value": "first degree murder"
},
{
"operation": "Now on",
"time": "297.41",
"value": "Murder (United States law)"
},
{
"operation": "hyperlink",
"time": "301.282",
"value": "4.3 Florida Murder (United States law)_section_6"
},
{
"operation": "Highlighting",
"time": "332.441",
"value": "Murder (United States law)_sentence_54"
},
{
"operation": "Highlighting",
"time": "332.441",
"value": "Murder (United States law)_title"
},
{
"operation": "Highlighting",
"time": "332.442",
"value": "Murder (United States law)_section_3"
},
{
"operation": "Highlighting",
"time": "332.442",
"value": "Murder (United States law)_section_6"
},
{
"operation": "Highlighting",
"time": "333.025",
"value": "Murder (United States law)_sentence_55"
},
{
"operation": "Highlighting",
"time": "333.745",
"value": "Murder (United States law)_sentence_56"
},
{
"operation": "Highlighting",
"time": "334.833",
"value": "Murder (United States law)_sentence_57"
},
{
"operation": "search",
"time": "355.729",
"value": "First degree murder in Florida"
},
{
"operation": "Now on",
"time": "356.892",
"value": "undefined"
},
{
"operation": "search",
"time": "369.528",
"value": "Florida penal code"
},
{
"operation": "Now on",
"time": "370.336",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "405.455",
"value": "next 20"
},
{
"operation": "Now on",
"time": "406.22",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "412.758",
"value": "Florida Murder (United States law)_section_6"
},
{
"operation": "Now on",
"time": "413.065",
"value": "Murder (United States law)#Florida Murder .28United States law.29 section 6"
},
{
"operation": "hyperlink",
"time": "417.254",
"value": "Florida"
},
{
"operation": "Now on",
"time": "418.222",
"value": "Florida"
},
{
"operation": "Page search",
"time": "422.961",
"value": "murder"
},
{
"operation": "Highlighting",
"time": "453.966",
"value": "Florida_sentence_446"
},
{
"operation": "Highlighting",
"time": "459.14",
"value": "Florida_sentence_448"
},
{
"operation": "Highlighting",
"time": "472.004",
"value": "Florida_sentence_447"
},
{
"operation": "Highlighting",
"time": "474.812",
"value": "Florida_sentence_449"
},
{
"operation": "finish",
"time": "586.448",
"value": "finish"
}
]
}
|
exfever_7419
|
In 2012, The Economic Times was the world's second-most widely read English-language business newspaper and is published simultaneously in 19 cities, including the capital city of the Indian state of Tamil Nadu.
|
{
"evidence": [
"The Economic Times",
"Chennai"
],
"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:\nIn 2012, The Economic Times was the world's second-most widely read English-language business newspaper and is published simultaneously in 19 cities, including the capital city of the Indian state of Tamil Nadu.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Economic Times",
"Chennai"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3364,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
71751
|
The Westville Dam's entrance is near the bridge over the Quinebaug river, a river originating in East Brimfield Lake.
|
{
"evidence": [
"Westville Dam_sentence_13",
"Quinebaug River_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:\nThe Westville Dam's entrance is near the bridge over the Quinebaug river, a river originating in East Brimfield Lake.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Westville Dam_sentence_13",
"Quinebaug River_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3365,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.241",
"value": "?search="
},
{
"operation": "search",
"time": "8.795",
"value": "Westville Dam"
},
{
"operation": "Now on",
"time": "9.656",
"value": "Westville Dam"
},
{
"operation": "Highlighting",
"time": "76.612",
"value": "Westville Dam_sentence_13"
},
{
"operation": "hyperlink",
"time": "99.927",
"value": "Quinebaug River"
},
{
"operation": "Now on",
"time": "101.905",
"value": "Quinebaug River"
},
{
"operation": "Highlighting",
"time": "121.472",
"value": "Quinebaug River_sentence_5"
},
{
"operation": "finish",
"time": "133.849",
"value": "finish"
}
]
}
|
32666
|
English Actor Sope Dirisu, who studied Economics and played football for the University of Texas, has had roles on ten TV shows during his career.
|
{
"evidence": [
"Sope Dirisu_sentence_4"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nEnglish Actor Sope Dirisu, who studied Economics and played football for the University of Texas, has had roles on ten TV shows during his career.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Sope Dirisu_sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3366,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.971",
"value": "?search="
},
{
"operation": "search",
"time": "11.097",
"value": "Sope Dirisu"
},
{
"operation": "Now on",
"time": "14.149",
"value": "Sope Dirisu"
},
{
"operation": "Highlighting",
"time": "18.187",
"value": "Sope Dirisu_sentence_0"
},
{
"operation": "Highlighting",
"time": "22.012",
"value": "Sope Dirisu_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "34.32",
"value": "Sope Dirisu_sentence_0"
},
{
"operation": "finish",
"time": "48.641",
"value": "finish"
}
]
}
|
exfever_850
|
Mike Shinoda is the founder of the rock band Linkin Park.
|
{
"evidence": [
"A Thousand Suns",
"Rick Rubin"
],
"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:\nMike Shinoda is the founder of the rock band Linkin Park.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"A Thousand Suns",
"Rick Rubin"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3367,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5281
|
Andrew Stanton is an American animator whose work includes co-directing Pixar's American computer-animated science-fiction film that stars the voice of an American stand-up comedian and actor born on June 5, 1962.
|
{
"evidence": [
"Andrew Stanton",
"WALL-E",
"Jeff Garlin"
],
"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:\nAndrew Stanton is an American animator whose work includes co-directing Pixar's American computer-animated science-fiction film that stars the voice of an American stand-up comedian and actor born on June 5, 1962.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Andrew Stanton",
"WALL-E",
"Jeff Garlin"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3368,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_598
|
Helen Elizabeth Hunt rose to fame portraying Jamie Buchman in the British television sitcom, that initially aired on NBC.
|
{
"evidence": [
"Helen Hunt",
"Mad About You"
],
"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:\nHelen Elizabeth Hunt rose to fame portraying Jamie Buchman in the British television sitcom, that initially aired on NBC.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Helen Hunt",
"Mad About You"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3369,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
66503
|
Elden Henson is an American actor who starred in The Battle of Shaker Heights(2003) and ran the film production company, Paramount Pictures.
|
{
"evidence": [
"The Battle of Shaker Heights_sentence_1",
"Elden Henson_sentence_0",
"Paramount Pictures_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:\nElden Henson is an American actor who starred in The Battle of Shaker Heights(2003) and ran the film production company, Paramount Pictures.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"The Battle of Shaker Heights_sentence_1",
"Elden Henson_sentence_0",
"Paramount Pictures_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3370,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.215",
"value": "?search="
},
{
"operation": "search",
"time": "8.611",
"value": "Elden Henson"
},
{
"operation": "Now on",
"time": "12.32",
"value": "Elden Henson"
},
{
"operation": "search",
"time": "23.691",
"value": "The Battle of Shaker Heights"
},
{
"operation": "Now on",
"time": "24.606",
"value": "The Battle of Shaker Heights"
},
{
"operation": "Highlighting",
"time": "27.039",
"value": "The Battle of Shaker Heights_cell_0_4_1"
},
{
"operation": "Highlighting",
"time": "27.053",
"value": "The Battle of Shaker Heights_header_cell_0_4_0"
},
{
"operation": "Highlighting",
"time": "27.053",
"value": "The Battle of Shaker Heights_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "33.198",
"value": "The Battle of Shaker Heights_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "34.849",
"value": "The Battle of Shaker Heights_cell_0_4_1"
},
{
"operation": "Highlighting deleted",
"time": "34.862",
"value": "The Battle of Shaker Heights_header_cell_0_4_0"
},
{
"operation": "Highlighting deleted",
"time": "34.864",
"value": "The Battle of Shaker Heights_header_cell_0_0_0"
},
{
"operation": "hyperlink",
"time": "36.228",
"value": "Elden Henson"
},
{
"operation": "Now on",
"time": "36.635",
"value": "Elden Henson"
},
{
"operation": "Highlighting",
"time": "38.85",
"value": "Elden Henson_sentence_0"
},
{
"operation": "back-button-clicked",
"time": "41.038",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "41.157",
"value": "The Battle of Shaker Heights"
},
{
"operation": "hyperlink",
"time": "50.44",
"value": "Elden Henson"
},
{
"operation": "Now on",
"time": "50.87",
"value": "Elden Henson"
},
{
"operation": "search",
"time": "74.866",
"value": "paramount pictures"
},
{
"operation": "Now on",
"time": "77.149",
"value": "Paramount Pictures"
},
{
"operation": "hyperlink",
"time": "157.986",
"value": "Charles Bluhdorn"
},
{
"operation": "Now on",
"time": "158.678",
"value": "Charles Bluhdorn"
},
{
"operation": "search",
"time": "193.661",
"value": "Paramount Pictures"
},
{
"operation": "Now on",
"time": "195.649",
"value": "Paramount Pictures"
},
{
"operation": "Highlighting",
"time": "212.904",
"value": "Paramount Pictures_sentence_0"
},
{
"operation": "finish",
"time": "228.753",
"value": "finish"
}
]
}
|
exfever_5568
|
Christina María Aguilera is a singer who rose to stardom with her album that incorporates dance-pop and a subgenre of pop music that is oriented towards preteens and teenagers.
|
{
"evidence": [
"Christina Aguilera",
"Christina Aguilera (album)",
"Teen pop"
],
"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:\nChristina María Aguilera is a singer who rose to stardom with her album that incorporates dance-pop and a subgenre of pop music that is oriented towards preteens and teenagers.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Christina Aguilera",
"Christina Aguilera (album)",
"Teen pop"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3371,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
72004
|
The 1985 German Formula Three Championship started with Circuit Zolder and ended at the Salzburgring in Austria; with a German team winning by over 30 points.
|
{
"evidence": [
"1985 German Formula Three Championship_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 1985 German Formula Three Championship started with Circuit Zolder and ended at the Salzburgring in Austria; with a German team winning by over 30 points.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"1985 German Formula Three Championship_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3372,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.752",
"value": "?search="
},
{
"operation": "search",
"time": "24.461",
"value": "1985 German Formula Three Championship"
},
{
"operation": "Now on",
"time": "25.814",
"value": "1985 German Formula Three Championship"
},
{
"operation": "Highlighting",
"time": "41.032",
"value": "1985 German Formula Three Championship_sentence_2"
},
{
"operation": "finish",
"time": "61.528",
"value": "finish"
}
]
}
|
exfever_2837
|
Nicholas Jerry Jonas is an American singer who formed a band with his older brothers that appeared in the widely successful 2010 American musical television film directed by Matthew Diamond and starring Demi Lovato, Meaghan Martin, Maria Canals-Barrera, Daniel Fathers, and Alyson Stoner.
|
{
"evidence": [
"Nick Jonas",
"Camp Rock"
],
"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:\nNicholas Jerry Jonas is an American singer who formed a band with his older brothers that appeared in the widely successful 2010 American musical television film directed by Matthew Diamond and starring Demi Lovato, Meaghan Martin, Maria Canals-Barrera, Daniel Fathers, and Alyson Stoner.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Nick Jonas",
"Camp Rock"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3373,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7179
|
After studying acting at the Drama Centre London, Edward Thomas Hardy made his film debut in Black Hawk Down (2001) by an English film director and producer.
|
{
"evidence": [
"Tom Hardy",
"Ridley Scott"
],
"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:\nAfter studying acting at the Drama Centre London, Edward Thomas Hardy made his film debut in Black Hawk Down (2001) by an English film director and producer.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Tom Hardy",
"Ridley Scott"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3374,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
5915
|
Michiel Bothma won four playoffs in the Sunshine Tour held in 2002 and 2010 against runner-ups Mark Murless and Jaco van Zyl.
|
{
"evidence": [
"Michiel Bothma_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:\nMichiel Bothma won four playoffs in the Sunshine Tour held in 2002 and 2010 against runner-ups Mark Murless and Jaco van Zyl.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Michiel Bothma_sentence_5"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3375,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.935",
"value": "?search="
},
{
"operation": "search",
"time": "7.209",
"value": "Michiel Bothma"
},
{
"operation": "Now on",
"time": "8.078",
"value": "Michiel Bothma"
},
{
"operation": "Page search",
"time": "11.796",
"value": "Sunshine"
},
{
"operation": "page-search-reset",
"time": "15.216",
"value": "page-search-reset"
},
{
"operation": "Highlighting",
"time": "30.516",
"value": "Michiel Bothma_sentence_5"
},
{
"operation": "Highlighting",
"time": "35.544",
"value": "Michiel Bothma_cell_2_1_0"
},
{
"operation": "Highlighting",
"time": "35.552",
"value": "Michiel Bothma_header_cell_2_0_0"
},
{
"operation": "Highlighting",
"time": "36.858",
"value": "Michiel Bothma_cell_2_1_1"
},
{
"operation": "Highlighting",
"time": "36.869",
"value": "Michiel Bothma_header_cell_2_0_1"
},
{
"operation": "Highlighting deleted",
"time": "37.936",
"value": "Michiel Bothma_cell_2_1_0"
},
{
"operation": "Highlighting deleted",
"time": "37.947",
"value": "Michiel Bothma_header_cell_2_0_0"
},
{
"operation": "Highlighting deleted",
"time": "92.191",
"value": "Michiel Bothma_cell_2_1_1"
},
{
"operation": "Highlighting deleted",
"time": "92.199",
"value": "Michiel Bothma_header_cell_2_0_1"
},
{
"operation": "Highlighting",
"time": "94.521",
"value": "Michiel Bothma_cell_1_1_1"
},
{
"operation": "Highlighting",
"time": "94.532",
"value": "Michiel Bothma_header_cell_1_0_1"
},
{
"operation": "Highlighting",
"time": "100.917",
"value": "Michiel Bothma_header_cell_1_0_3"
},
{
"operation": "Highlighting",
"time": "103.722",
"value": "Michiel Bothma_cell_1_1_3"
},
{
"operation": "Highlighting",
"time": "103.736",
"value": "Michiel Bothma_header_cell_1_0_3"
},
{
"operation": "Highlighting deleted",
"time": "104.664",
"value": "Michiel Bothma_cell_1_1_3"
},
{
"operation": "Highlighting deleted",
"time": "104.675",
"value": "Michiel Bothma_header_cell_1_0_3"
},
{
"operation": "Highlighting deleted",
"time": "106.182",
"value": "Michiel Bothma_cell_1_1_1"
},
{
"operation": "Highlighting deleted",
"time": "106.19",
"value": "Michiel Bothma_header_cell_1_0_1"
},
{
"operation": "Highlighting deleted",
"time": "107.29",
"value": "Michiel Bothma_sentence_5"
},
{
"operation": "Highlighting",
"time": "107.733",
"value": "Michiel Bothma_sentence_5"
},
{
"operation": "finish",
"time": "132.735",
"value": "finish"
}
]
}
|
exfever_7926
|
Black Panther is a 2018 American superhero film produced by Marvel Studios and distributed by an American film distribution studio within the Disney Media and Entertainment Distribution division of the Walt Disney Company.
|
{
"evidence": [
"Black Panther (film)",
"Walt Disney Studios Motion Pictures"
],
"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:\nBlack Panther is a 2018 American superhero film produced by Marvel Studios and distributed by an American film distribution studio within the Disney Media and Entertainment Distribution division of the Walt Disney Company.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Black Panther (film)",
"Walt Disney Studios Motion Pictures"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3376,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5231
|
Sansa Stark is a fictional character in the A Song of Ice and Fire series of epic fantasy novels by American author George R. R. Martin and its television adaptation of an American fantasy drama series created by David Benioff and D. B. Weiss for Netflix.
|
{
"evidence": [
"Sansa Stark",
"Game of Thrones"
],
"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:\nSansa Stark is a fictional character in the A Song of Ice and Fire series of epic fantasy novels by American author George R. R. Martin and its television adaptation of an American fantasy drama series created by David Benioff and D. B. Weiss for Netflix.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Sansa Stark",
"Game of Thrones"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3377,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_318
|
Greyson Chance is an American singer-songwriter and musician who rose to national attention in 2010 with his performance of a song by an American singer, songwriter, and actress best known for her image reinventions and musical versatility, from her debut studio album, The Fame which was released as the album's fifth and final single by Interscope Records.
|
{
"evidence": [
"Greyson Chance",
"Paparazzi (Lady Gaga song)",
"Lady Gaga"
],
"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:\nGreyson Chance is an American singer-songwriter and musician who rose to national attention in 2010 with his performance of a song by an American singer, songwriter, and actress best known for her image reinventions and musical versatility, from her debut studio album, The Fame which was released as the album's fifth and final single by Interscope Records.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Greyson Chance",
"Paparazzi (Lady Gaga song)",
"Lady Gaga"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3378,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
91540
|
The Machine Question is only a fairy tale without chapters.
|
{
"evidence": [
"The Machine Question_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 Machine Question is only a fairy tale without chapters.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"The Machine Question_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3379,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.045",
"value": "?search="
},
{
"operation": "search",
"time": "12.578",
"value": "The Machine Question"
},
{
"operation": "Now on",
"time": "13.383",
"value": "The Machine Question"
},
{
"operation": "Page search",
"time": "31.408",
"value": "fairy"
},
{
"operation": "Page search",
"time": "38.672",
"value": "chapt"
},
{
"operation": "Highlighting",
"time": "41.477",
"value": "The Machine Question_sentence_2"
},
{
"operation": "finish",
"time": "50.593",
"value": "finish"
}
]
}
|
exfever_3027
|
Coldplay is a British rock band whose members include an English singer-songwriter who was a co-founder of the rock band Linkin Park.
|
{
"evidence": [
"Coldplay",
"Chris Martin"
],
"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:\nColdplay is a British rock band whose members include an English singer-songwriter who was a co-founder of the rock band Linkin Park.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Coldplay",
"Chris Martin"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3380,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3865
|
Vincent Cassel is an Italian actor who garnered wide recognition with English-speaking audiences for his performances in a 2004 American heist comedy film directed by Steven Soderbergh.
|
{
"evidence": [
"Vincent Cassel",
"Ocean's Twelve"
],
"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:\nVincent Cassel is an Italian actor who garnered wide recognition with English-speaking audiences for his performances in a 2004 American heist comedy film directed by Steven Soderbergh.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Vincent Cassel",
"Ocean's Twelve"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3381,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5549
|
Rick and Morty is an American adult animated science fiction sitcom created by Justin Roiland and an American writer, producer, and actor best known as the creator and producer of the NBC/Yahoo! Screen sitcom Community.
|
{
"evidence": [
"Rick and Morty",
"Dan Harmon"
],
"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:\nRick and Morty is an American adult animated science fiction sitcom created by Justin Roiland and an American writer, producer, and actor best known as the creator and producer of the NBC/Yahoo! Screen sitcom Community.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Rick and Morty",
"Dan Harmon"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3382,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4779
|
Duane Hopwood is a 2008 American comedy-drama film starring an American actor and comedian known for playing the role of writer Frank Rossitano on 30 Rock.
|
{
"evidence": [
"Duane Hopwood",
"Judah Friedlander"
],
"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:\nDuane Hopwood is a 2008 American comedy-drama film starring an American actor and comedian known for playing the role of writer Frank Rossitano on 30 Rock.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Duane Hopwood",
"Judah Friedlander"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3383,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.