data_source
stringclasses
2 values
prompt
listlengths
2
2
ability
stringclasses
1 value
reward_model
dict
extra_info
dict
metadata
dict
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who laughs on dark side of the moon?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Peter Watts" ] }, "style": "rule" }
{ "index": 600, "need_tools_kwargs": true, "question": "who laughs on dark side of the moon?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Peter Watts" ] }, "question": "who laughs on dark side of the moon?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who sang the original version of earth angel?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "the Penguins" ] }, "style": "rule" }
{ "index": 601, "need_tools_kwargs": true, "question": "who sang the original version of earth angel?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "the Penguins" ] }, "question": "who sang the original version of earth angel?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when did world war 2 star and end?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "lasted from 1939 to 1945" ] }, "style": "rule" }
{ "index": 602, "need_tools_kwargs": true, "question": "when did world war 2 star and end?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "lasted from 1939 to 1945" ] }, "question": "when did world war 2 star and end?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: whats the record for most wins in a row mlb?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "26 games" ] }, "style": "rule" }
{ "index": 603, "need_tools_kwargs": true, "question": "whats the record for most wins in a row mlb?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "26 games" ] }, "question": "whats the record for most wins in a row mlb?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: the lieutenant governor of delhi new delhi delhi?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Anil Baijal" ] }, "style": "rule" }
{ "index": 604, "need_tools_kwargs": true, "question": "the lieutenant governor of delhi new delhi delhi?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Anil Baijal" ] }, "question": "the lieutenant governor of delhi new delhi delhi?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when was a nightmare on elm street made?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "November 9, 1984" ] }, "style": "rule" }
{ "index": 605, "need_tools_kwargs": true, "question": "when was a nightmare on elm street made?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "November 9, 1984" ] }, "question": "when was a nightmare on elm street made?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: where was the tv series yellowstone filmed at?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "near Park City, Utah", "various locations in Montana" ] }, "style": "rule" }
{ "index": 606, "need_tools_kwargs": true, "question": "where was the tv series yellowstone filmed at?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "near Park City, Utah", "various locations in Montana" ] }, "question": "where was the tv series yellowstone filmed at?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: which river is important in the development of chinese civilization?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Wei River" ] }, "style": "rule" }
{ "index": 607, "need_tools_kwargs": true, "question": "which river is important in the development of chinese civilization?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Wei River" ] }, "question": "which river is important in the development of chinese civilization?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when did the lyrics to o canada change?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "1980" ] }, "style": "rule" }
{ "index": 608, "need_tools_kwargs": true, "question": "when did the lyrics to o canada change?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "1980" ] }, "question": "when did the lyrics to o canada change?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when was the catechism of the catholic church written?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "1986" ] }, "style": "rule" }
{ "index": 609, "need_tools_kwargs": true, "question": "when was the catechism of the catholic church written?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "1986" ] }, "question": "when was the catechism of the catholic church written?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who sang beauty school dropout in the film grease?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Frankie Avalon" ] }, "style": "rule" }
{ "index": 610, "need_tools_kwargs": true, "question": "who sang beauty school dropout in the film grease?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Frankie Avalon" ] }, "question": "who sang beauty school dropout in the film grease?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when did the movie little monsters come out?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "August 25, 1989" ] }, "style": "rule" }
{ "index": 611, "need_tools_kwargs": true, "question": "when did the movie little monsters come out?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "August 25, 1989" ] }, "question": "when did the movie little monsters come out?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: which of the seven wonders of the ancient world can you still see today?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Great Pyramid of Giza" ] }, "style": "rule" }
{ "index": 612, "need_tools_kwargs": true, "question": "which of the seven wonders of the ancient world can you still see today?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Great Pyramid of Giza" ] }, "question": "which of the seven wonders of the ancient world can you still see today?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who did rashad jennings play for in the nfl?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "the Oakland Raiders", "the Jacksonville Jaguars", "the New York Giants" ] }, "style": "rule" }
{ "index": 613, "need_tools_kwargs": true, "question": "who did rashad jennings play for in the nfl?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "the Oakland Raiders", "the Jacksonville Jaguars", "the New York Giants" ] }, "question": "who did rashad jennings play for in the nfl?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when was the best animated feature film category first introduced?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "2001" ] }, "style": "rule" }
{ "index": 614, "need_tools_kwargs": true, "question": "when was the best animated feature film category first introduced?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "2001" ] }, "question": "when was the best animated feature film category first introduced?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when is the new modern family episode coming out?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "February 28, 2018" ] }, "style": "rule" }
{ "index": 615, "need_tools_kwargs": true, "question": "when is the new modern family episode coming out?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "February 28, 2018" ] }, "question": "when is the new modern family episode coming out?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who won the indy car race in phoenix?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Mario Andretti" ] }, "style": "rule" }
{ "index": 616, "need_tools_kwargs": true, "question": "who won the indy car race in phoenix?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Mario Andretti" ] }, "question": "who won the indy car race in phoenix?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who sang the song it's my life?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Talk Talk" ] }, "style": "rule" }
{ "index": 617, "need_tools_kwargs": true, "question": "who sang the song it's my life?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Talk Talk" ] }, "question": "who sang the song it's my life?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when did twizzlers pull and peel come out?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "1994" ] }, "style": "rule" }
{ "index": 618, "need_tools_kwargs": true, "question": "when did twizzlers pull and peel come out?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "1994" ] }, "question": "when did twizzlers pull and peel come out?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when does season 7 episode 8 of scandal air?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "January 18, 2018" ] }, "style": "rule" }
{ "index": 619, "need_tools_kwargs": true, "question": "when does season 7 episode 8 of scandal air?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "January 18, 2018" ] }, "question": "when does season 7 episode 8 of scandal air?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who hit the shot heard round the world golf?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Gene Sarazen" ] }, "style": "rule" }
{ "index": 620, "need_tools_kwargs": true, "question": "who hit the shot heard round the world golf?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Gene Sarazen" ] }, "question": "who hit the shot heard round the world golf?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who holds the world record for 100 meters?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Usain Bolt" ] }, "style": "rule" }
{ "index": 621, "need_tools_kwargs": true, "question": "who holds the world record for 100 meters?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Usain Bolt" ] }, "question": "who holds the world record for 100 meters?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: where does the columbia river meet the pacific ocean?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Columbia Bar" ] }, "style": "rule" }
{ "index": 622, "need_tools_kwargs": true, "question": "where does the columbia river meet the pacific ocean?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Columbia Bar" ] }, "question": "where does the columbia river meet the pacific ocean?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: what is ralph angels real name on queen sugar?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Nana-Kofi Siriboe" ] }, "style": "rule" }
{ "index": 623, "need_tools_kwargs": true, "question": "what is ralph angels real name on queen sugar?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Nana-Kofi Siriboe" ] }, "question": "what is ralph angels real name on queen sugar?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: what does aron hakodesh refer to in english?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Holy Ark" ] }, "style": "rule" }
{ "index": 624, "need_tools_kwargs": true, "question": "what does aron hakodesh refer to in english?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Holy Ark" ] }, "question": "what does aron hakodesh refer to in english?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who plays charlie in kinky boots on broadway?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Stark Sands" ] }, "style": "rule" }
{ "index": 625, "need_tools_kwargs": true, "question": "who plays charlie in kinky boots on broadway?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Stark Sands" ] }, "question": "who plays charlie in kinky boots on broadway?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when was the first snowfall by james russell lowell written?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "1847" ] }, "style": "rule" }
{ "index": 626, "need_tools_kwargs": true, "question": "when was the first snowfall by james russell lowell written?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "1847" ] }, "question": "when was the first snowfall by james russell lowell written?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: voice of obi wan in the clone wars?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "James Arnold Taylor" ] }, "style": "rule" }
{ "index": 627, "need_tools_kwargs": true, "question": "voice of obi wan in the clone wars?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "James Arnold Taylor" ] }, "question": "voice of obi wan in the clone wars?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who does the voice of the cat in the hat?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Martin Short" ] }, "style": "rule" }
{ "index": 628, "need_tools_kwargs": true, "question": "who does the voice of the cat in the hat?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Martin Short" ] }, "question": "who does the voice of the cat in the hat?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: what is the line in the middle of a time zone that in theory determines the time of that zone?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "mean solar time" ] }, "style": "rule" }
{ "index": 629, "need_tools_kwargs": true, "question": "what is the line in the middle of a time zone that in theory determines the time of that zone?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "mean solar time" ] }, "question": "what is the line in the middle of a time zone that in theory determines the time of that zone?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who do you play as in assassin's creed syndicate?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Evie", "Jacob" ] }, "style": "rule" }
{ "index": 630, "need_tools_kwargs": true, "question": "who do you play as in assassin's creed syndicate?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Evie", "Jacob" ] }, "question": "who do you play as in assassin's creed syndicate?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when was the last time north america won ms universe?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "2012" ] }, "style": "rule" }
{ "index": 631, "need_tools_kwargs": true, "question": "when was the last time north america won ms universe?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "2012" ] }, "question": "when was the last time north america won ms universe?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who did snow white fall in love with?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "prince" ] }, "style": "rule" }
{ "index": 632, "need_tools_kwargs": true, "question": "who did snow white fall in love with?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "prince" ] }, "question": "who did snow white fall in love with?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: which disney princess did not get married at the end of the movie?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Aurora" ] }, "style": "rule" }
{ "index": 633, "need_tools_kwargs": true, "question": "which disney princess did not get married at the end of the movie?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Aurora" ] }, "question": "which disney princess did not get married at the end of the movie?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who wrote the song these boots are made for walking?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Lee Hazlewood" ] }, "style": "rule" }
{ "index": 634, "need_tools_kwargs": true, "question": "who wrote the song these boots are made for walking?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Lee Hazlewood" ] }, "question": "who wrote the song these boots are made for walking?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when does the truth or dare movie come out?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "April 13, 2018" ] }, "style": "rule" }
{ "index": 635, "need_tools_kwargs": true, "question": "when does the truth or dare movie come out?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "April 13, 2018" ] }, "question": "when does the truth or dare movie come out?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: how many swear words in south park movie?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "399" ] }, "style": "rule" }
{ "index": 636, "need_tools_kwargs": true, "question": "how many swear words in south park movie?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "399" ] }, "question": "how many swear words in south park movie?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when did starcraft 2 go free to play?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "April 19, 2017" ] }, "style": "rule" }
{ "index": 637, "need_tools_kwargs": true, "question": "when did starcraft 2 go free to play?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "April 19, 2017" ] }, "question": "when did starcraft 2 go free to play?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when did the first episode of columbo air?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "February 20, 1968" ] }, "style": "rule" }
{ "index": 638, "need_tools_kwargs": true, "question": "when did the first episode of columbo air?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "February 20, 1968" ] }, "question": "when did the first episode of columbo air?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: the highest officials in the government of the roman republic were?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "consuls" ] }, "style": "rule" }
{ "index": 639, "need_tools_kwargs": true, "question": "the highest officials in the government of the roman republic were?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "consuls" ] }, "question": "the highest officials in the government of the roman republic were?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: what are the dates of the second new deal?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "1935–1938" ] }, "style": "rule" }
{ "index": 640, "need_tools_kwargs": true, "question": "what are the dates of the second new deal?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "1935–1938" ] }, "question": "what are the dates of the second new deal?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when do the arrow and the flash crossover?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "December 2, 2014" ] }, "style": "rule" }
{ "index": 641, "need_tools_kwargs": true, "question": "when do the arrow and the flash crossover?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "December 2, 2014" ] }, "question": "when do the arrow and the flash crossover?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when did the first rocket launched into space?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "1944" ] }, "style": "rule" }
{ "index": 642, "need_tools_kwargs": true, "question": "when did the first rocket launched into space?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "1944" ] }, "question": "when did the first rocket launched into space?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: what is the gold by manchester orchestra about?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "learning to communicate", "issues of trust" ] }, "style": "rule" }
{ "index": 643, "need_tools_kwargs": true, "question": "what is the gold by manchester orchestra about?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "learning to communicate", "issues of trust" ] }, "question": "what is the gold by manchester orchestra about?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who is directing the new star wars movie?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Rian Johnson" ] }, "style": "rule" }
{ "index": 644, "need_tools_kwargs": true, "question": "who is directing the new star wars movie?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Rian Johnson" ] }, "question": "who is directing the new star wars movie?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when did the buccaneers win the super bowl?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "January 26, 2003" ] }, "style": "rule" }
{ "index": 645, "need_tools_kwargs": true, "question": "when did the buccaneers win the super bowl?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "January 26, 2003" ] }, "question": "when did the buccaneers win the super bowl?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when is the season 5 of the originals coming out?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "April 20, 2018" ] }, "style": "rule" }
{ "index": 646, "need_tools_kwargs": true, "question": "when is the season 5 of the originals coming out?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "April 20, 2018" ] }, "question": "when is the season 5 of the originals coming out?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when was the passion of the christ made?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "2004" ] }, "style": "rule" }
{ "index": 647, "need_tools_kwargs": true, "question": "when was the passion of the christ made?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "2004" ] }, "question": "when was the passion of the christ made?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who won the academy award for baby jane?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Norma Koch" ] }, "style": "rule" }
{ "index": 648, "need_tools_kwargs": true, "question": "who won the academy award for baby jane?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Norma Koch" ] }, "question": "who won the academy award for baby jane?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who holds the record for rushing yards in a season college?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Barry Sanders" ] }, "style": "rule" }
{ "index": 649, "need_tools_kwargs": true, "question": "who holds the record for rushing yards in a season college?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Barry Sanders" ] }, "question": "who holds the record for rushing yards in a season college?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: what size engine does a subaru brz have?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "2.0-litre" ] }, "style": "rule" }
{ "index": 650, "need_tools_kwargs": true, "question": "what size engine does a subaru brz have?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "2.0-litre" ] }, "question": "what size engine does a subaru brz have?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who plays young fiona in american horror story coven?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Riley Voelkel" ] }, "style": "rule" }
{ "index": 651, "need_tools_kwargs": true, "question": "who plays young fiona in american horror story coven?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Riley Voelkel" ] }, "question": "who plays young fiona in american horror story coven?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who played the fugitive in the original tv series?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "David Janssen" ] }, "style": "rule" }
{ "index": 652, "need_tools_kwargs": true, "question": "who played the fugitive in the original tv series?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "David Janssen" ] }, "question": "who played the fugitive in the original tv series?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who won the uefa champions league 2018 final?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Real Madrid" ] }, "style": "rule" }
{ "index": 653, "need_tools_kwargs": true, "question": "who won the uefa champions league 2018 final?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Real Madrid" ] }, "question": "who won the uefa champions league 2018 final?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when was the drinking age raised to 21?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "By mid-1988" ] }, "style": "rule" }
{ "index": 654, "need_tools_kwargs": true, "question": "when was the drinking age raised to 21?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "By mid-1988" ] }, "question": "when was the drinking age raised to 21?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who was cast into the lion's den?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Daniel" ] }, "style": "rule" }
{ "index": 655, "need_tools_kwargs": true, "question": "who was cast into the lion's den?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Daniel" ] }, "question": "who was cast into the lion's den?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who has the most consecutive wins on jeopardy?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Ken Jennings" ] }, "style": "rule" }
{ "index": 656, "need_tools_kwargs": true, "question": "who has the most consecutive wins on jeopardy?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Ken Jennings" ] }, "question": "who has the most consecutive wins on jeopardy?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: what type of battery does a wii remote use?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "AA Battery" ] }, "style": "rule" }
{ "index": 657, "need_tools_kwargs": true, "question": "what type of battery does a wii remote use?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "AA Battery" ] }, "question": "what type of battery does a wii remote use?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: which level of u.s. government pays for the most healthcare?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "federal government" ] }, "style": "rule" }
{ "index": 658, "need_tools_kwargs": true, "question": "which level of u.s. government pays for the most healthcare?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "federal government" ] }, "question": "which level of u.s. government pays for the most healthcare?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who is the 1st governor general of india?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Warren Hastings" ] }, "style": "rule" }
{ "index": 659, "need_tools_kwargs": true, "question": "who is the 1st governor general of india?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Warren Hastings" ] }, "question": "who is the 1st governor general of india?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when was the legal age of consent introduced uk?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "In 1275" ] }, "style": "rule" }
{ "index": 660, "need_tools_kwargs": true, "question": "when was the legal age of consent introduced uk?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "In 1275" ] }, "question": "when was the legal age of consent introduced uk?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who decides the outcome in civil and criminal cases?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "judge" ] }, "style": "rule" }
{ "index": 661, "need_tools_kwargs": true, "question": "who decides the outcome in civil and criminal cases?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "judge" ] }, "question": "who decides the outcome in civil and criminal cases?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: what was the dog's name on the jetsons?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Astro" ] }, "style": "rule" }
{ "index": 662, "need_tools_kwargs": true, "question": "what was the dog's name on the jetsons?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Astro" ] }, "question": "what was the dog's name on the jetsons?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: what was the number 1 hit single in the us on may 26 1990?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "\"Vogue\"" ] }, "style": "rule" }
{ "index": 663, "need_tools_kwargs": true, "question": "what was the number 1 hit single in the us on may 26 1990?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "\"Vogue\"" ] }, "question": "what was the number 1 hit single in the us on may 26 1990?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: what are the different sizes of macbook pro?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "17-inch", "13-inch", "15-inch" ] }, "style": "rule" }
{ "index": 664, "need_tools_kwargs": true, "question": "what are the different sizes of macbook pro?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "17-inch", "13-inch", "15-inch" ] }, "question": "what are the different sizes of macbook pro?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: from the french meaning millers wife a cookery method?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Meunière" ] }, "style": "rule" }
{ "index": 665, "need_tools_kwargs": true, "question": "from the french meaning millers wife a cookery method?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Meunière" ] }, "question": "from the french meaning millers wife a cookery method?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who played xander on days of our lives?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Paul Telfer" ] }, "style": "rule" }
{ "index": 666, "need_tools_kwargs": true, "question": "who played xander on days of our lives?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Paul Telfer" ] }, "question": "who played xander on days of our lives?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: where did lorelai gilmore go to high school?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Chilton Preparatory School" ] }, "style": "rule" }
{ "index": 667, "need_tools_kwargs": true, "question": "where did lorelai gilmore go to high school?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Chilton Preparatory School" ] }, "question": "where did lorelai gilmore go to high school?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when do they award the nobel peace prize?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "10 December" ] }, "style": "rule" }
{ "index": 668, "need_tools_kwargs": true, "question": "when do they award the nobel peace prize?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "10 December" ] }, "question": "when do they award the nobel peace prize?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: where did the black panther party take place?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "United", "United Kingdom", "United States" ] }, "style": "rule" }
{ "index": 669, "need_tools_kwargs": true, "question": "where did the black panther party take place?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "United", "United Kingdom", "United States" ] }, "question": "where did the black panther party take place?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who wrote the song it's the climb?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Alexander" ] }, "style": "rule" }
{ "index": 670, "need_tools_kwargs": true, "question": "who wrote the song it's the climb?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Alexander" ] }, "question": "who wrote the song it's the climb?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: how many scheduled tribes are there in india?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "645" ] }, "style": "rule" }
{ "index": 671, "need_tools_kwargs": true, "question": "how many scheduled tribes are there in india?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "645" ] }, "question": "how many scheduled tribes are there in india?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: total number of nuclear power plants in india?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "7" ] }, "style": "rule" }
{ "index": 672, "need_tools_kwargs": true, "question": "total number of nuclear power plants in india?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "7" ] }, "question": "total number of nuclear power plants in india?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: what is the prize for winning europa league?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "€8,500,000" ] }, "style": "rule" }
{ "index": 673, "need_tools_kwargs": true, "question": "what is the prize for winning europa league?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "€8,500,000" ] }, "question": "what is the prize for winning europa league?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when did the black death get to europe?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "1347" ] }, "style": "rule" }
{ "index": 674, "need_tools_kwargs": true, "question": "when did the black death get to europe?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "1347" ] }, "question": "when did the black death get to europe?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when did the philippines become an independent country?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "July 4, 1946" ] }, "style": "rule" }
{ "index": 675, "need_tools_kwargs": true, "question": "when did the philippines become an independent country?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "July 4, 1946" ] }, "question": "when did the philippines become an independent country?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: what year did the cadillac srx come out?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "2004" ] }, "style": "rule" }
{ "index": 676, "need_tools_kwargs": true, "question": "what year did the cadillac srx come out?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "2004" ] }, "question": "what year did the cadillac srx come out?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who was the captain of pakistan cricket team in 2011 cricket world cup?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Shahid Afridi" ] }, "style": "rule" }
{ "index": 677, "need_tools_kwargs": true, "question": "who was the captain of pakistan cricket team in 2011 cricket world cup?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Shahid Afridi" ] }, "question": "who was the captain of pakistan cricket team in 2011 cricket world cup?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who helped israel in the yom kippur war?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "United States" ] }, "style": "rule" }
{ "index": 678, "need_tools_kwargs": true, "question": "who helped israel in the yom kippur war?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "United States" ] }, "question": "who helped israel in the yom kippur war?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who sings i've got the moves like jagger?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Maroon 5" ] }, "style": "rule" }
{ "index": 679, "need_tools_kwargs": true, "question": "who sings i've got the moves like jagger?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Maroon 5" ] }, "question": "who sings i've got the moves like jagger?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: how many panes of glass in the louvre pyramid?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "603 rhombus-shaped", "70 triangular glass segments" ] }, "style": "rule" }
{ "index": 680, "need_tools_kwargs": true, "question": "how many panes of glass in the louvre pyramid?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "603 rhombus-shaped", "70 triangular glass segments" ] }, "question": "how many panes of glass in the louvre pyramid?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who played lily on the drew carey show?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Tammy Lauren" ] }, "style": "rule" }
{ "index": 681, "need_tools_kwargs": true, "question": "who played lily on the drew carey show?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Tammy Lauren" ] }, "question": "who played lily on the drew carey show?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who plays mayor in blue bloods season 8?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Lorraine Bracco" ] }, "style": "rule" }
{ "index": 682, "need_tools_kwargs": true, "question": "who plays mayor in blue bloods season 8?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Lorraine Bracco" ] }, "question": "who plays mayor in blue bloods season 8?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who is the longest serving coach of the nz warriors?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Ivan Cleary" ] }, "style": "rule" }
{ "index": 683, "need_tools_kwargs": true, "question": "who is the longest serving coach of the nz warriors?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Ivan Cleary" ] }, "question": "who is the longest serving coach of the nz warriors?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who has won the afc east the most?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "the Patriots" ] }, "style": "rule" }
{ "index": 684, "need_tools_kwargs": true, "question": "who has won the afc east the most?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "the Patriots" ] }, "question": "who has won the afc east the most?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: prokaryotic cells may also contain small rings of dna known as?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "replicon" ] }, "style": "rule" }
{ "index": 685, "need_tools_kwargs": true, "question": "prokaryotic cells may also contain small rings of dna known as?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "replicon" ] }, "question": "prokaryotic cells may also contain small rings of dna known as?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who got the most rings in the nfl?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Bill Belichick", "Neal Dahlen" ] }, "style": "rule" }
{ "index": 686, "need_tools_kwargs": true, "question": "who got the most rings in the nfl?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Bill Belichick", "Neal Dahlen" ] }, "question": "who got the most rings in the nfl?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: what's the video with the most views on youtube?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "\"Despacito\"" ] }, "style": "rule" }
{ "index": 687, "need_tools_kwargs": true, "question": "what's the video with the most views on youtube?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "\"Despacito\"" ] }, "question": "what's the video with the most views on youtube?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: when is the 100 book 2 coming out?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "September 25, 2014" ] }, "style": "rule" }
{ "index": 688, "need_tools_kwargs": true, "question": "when is the 100 book 2 coming out?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "September 25, 2014" ] }, "question": "when is the 100 book 2 coming out?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: how many soldiers in a company in india?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "80–150" ] }, "style": "rule" }
{ "index": 689, "need_tools_kwargs": true, "question": "how many soldiers in a company in india?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "80–150" ] }, "question": "how many soldiers in a company in india?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: how many languages in new 500 rupee note?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "17" ] }, "style": "rule" }
{ "index": 690, "need_tools_kwargs": true, "question": "how many languages in new 500 rupee note?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "17" ] }, "question": "how many languages in new 500 rupee note?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who has longest field goal in nfl history?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Matt Prater" ] }, "style": "rule" }
{ "index": 691, "need_tools_kwargs": true, "question": "who has longest field goal in nfl history?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Matt Prater" ] }, "question": "who has longest field goal in nfl history?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who plays brenda's mom on six feet under?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Joanna Cassidy" ] }, "style": "rule" }
{ "index": 692, "need_tools_kwargs": true, "question": "who plays brenda's mom on six feet under?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Joanna Cassidy" ] }, "question": "who plays brenda's mom on six feet under?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who does elizabeth banks play in hunger games?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Effie Trinket" ] }, "style": "rule" }
{ "index": 693, "need_tools_kwargs": true, "question": "who does elizabeth banks play in hunger games?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Effie Trinket" ] }, "question": "who does elizabeth banks play in hunger games?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: where was casino royale filmed in the bahamas?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "on Paradise Island", "Several locations around New Providence" ] }, "style": "rule" }
{ "index": 694, "need_tools_kwargs": true, "question": "where was casino royale filmed in the bahamas?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "on Paradise Island", "Several locations around New Providence" ] }, "question": "where was casino royale filmed in the bahamas?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: what are the odds of getting a royal flush in poker?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "649,739 : 1" ] }, "style": "rule" }
{ "index": 695, "need_tools_kwargs": true, "question": "what are the odds of getting a royal flush in poker?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "649,739 : 1" ] }, "question": "what are the odds of getting a royal flush in poker?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: who was the double agent in tinker tailor soldier spy?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Haydon" ] }, "style": "rule" }
{ "index": 696, "need_tools_kwargs": true, "question": "who was the double agent in tinker tailor soldier spy?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Haydon" ] }, "question": "who was the double agent in tinker tailor soldier spy?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: what is the deepest canyon in north america?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Hells Canyon" ] }, "style": "rule" }
{ "index": 697, "need_tools_kwargs": true, "question": "what is the deepest canyon in north america?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Hells Canyon" ] }, "question": "what is the deepest canyon in north america?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: part of mitosis wherein the chromosomes align at the equator?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "Metaphase" ] }, "style": "rule" }
{ "index": 698, "need_tools_kwargs": true, "question": "part of mitosis wherein the chromosomes align at the equator?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "Metaphase" ] }, "question": "part of mitosis wherein the chromosomes align at the equator?" } } } }
null
nq
[ { "content": "You are a helpful and harmless assistant.", "role": "system" }, { "content": "Answer the given question step by step.\n Instructions:\n 1. First, conduct reasoning inside <think> and </think> tags whenever you receive new information.\n 2. If you need external knowledge, you can search using <search>query</search>.\n 3. When you receive search results, evaluate their usefulness and put your judgment inside <judge>Yes</judge> or <judge>No</judge> tags.\n 4. Based on your judgment, follow these strict rules:\n a. If the information is useful AND you now have sufficient information to provide a complete final answer, proceed directly to step 5.\n b. If the information is useful BUT you still need more details, you MUST search again with <search>...</search>.\n c. If the information is not useful, you MUST search again with <search>...</search>. You MUST NOT provide an answer in this case.\n 5. Provide your final answer in <answer>...</answer> tags. The <answer> tag marks the end of the task. After providing the <answer>, you MUST stop and generate no further text.\n\n Question: is costa rica north or south of the equator?", "role": "user" } ]
fact-reasoning
{ "ground_truth": { "target": [ "north of the Equator" ] }, "style": "rule" }
{ "index": 699, "need_tools_kwargs": true, "question": "is costa rica north or south of the equator?", "split": "test", "tools_kwargs": { "search": { "create_kwargs": { "data_source": "nq", "ground_truth": { "target": [ "north of the Equator" ] }, "question": "is costa rica north or south of the equator?" } } } }
null