test/00059.json

[{"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: This morning I attended what is known as \" Committee Day \" for my state 's teacher union where we come together and put together a general direction and more of a focus of priorities for the office for the upcoming year . Well I have been to a couple of these in the past and it was always a joke how I am one of the few Republican union members but I found that today , not to many people were laughing at the joke . It never dawned on me in what this meeting would be like a couple of months before a presidential election . The board room was plastered with \" NEA for Obama \" posters and various placards for other ONLY Dem candidates .\n", "labels": "What type of politics does the narrator have ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-88fe9a72c81b49df984674b91f80b171"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: Not long after they moved in , someone dumped a pregnant stray cat out in the neighborhood and she had her kittens . One got road - killed , and the other two now live with us outdoors . They ' ve been neutered and vaccinated .\n", "labels": "Why do the cats not live indoors with you guys ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-9e791b3686654f65b8bf60beff633c51"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: so last night when i went clubbing this girl said to me in the washroom wow you look super skinny ! and i was like i m sorry what ? and made her repeat because i was in such disbelief !.\n", "labels": "Why was the writer in such disbelief ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-52bac497816b4e8b9e82d1baa5f0c70d"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: El dropped me off at B. 's house - not too long of a drive from Mann 's surprisingly . She welcomed El . and me into her home .\n", "labels": "Why did she welcome us into the house ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-b9363c7c38fb4e3ca83bd5b3f8ceff0b"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: I wondered for a bit if we were going to make it to Otttawa . In fact , I was n't sure we were going to make it out of the city . We were about twenty minutes late starting out ( nothing unusual there ) and everything was fine until we turned on to the 401 to go east . Nothing but a sea of parked cars , appearing not to be going anywhere .\n", "labels": "What country is the person presently located in ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-8125ec294d1b4d27976fb3cfd8daf2e7"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: They are SO gorgeous , and I had a smile on my face all day once I had seen my mail . The beautiful apron is a perfect fit and even makes me want to cook - that 's an achievement ! I thought it warranted a photo shoot , though trust me , it 's better that most of my head is chopped out of this shot ! I 'll show the rejects some time ! !.\n", "labels": "What does the speaker think is gorgeous ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-67faffc3ce5e461e969260873c9a7b89"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: The warm golden scenery betraying just how biting cold it was outside . Darren clutched himself and shivered slightly despite the cozy warmth that fulfilled the house . He considered returning to bed and into the arms of his husband but decided to go downstairs to be alone and in silence while he tried to remember his dream .\n", "labels": "What do you know about the narrator 's Husband ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-e2d70e937cf24d1e9d7c1248eda0796a"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: Well . I feel like crap right now . Dunno why . Can you imagine my monotone , bored - to - death , could - care - less voice ?.\n", "labels": "Why might I feel like crap even though I do n't know why ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-3aa01ff55e5f4f3582ef2b24dc21add8"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: I could nt hold on to the tears any longer . I really hated saying goodbye my fiancee , my family , my friends , my clients ( which are also my friends ) . I got to Jyvaskyla with no sleep and not enough warm clothes ! I was shocked to get off the plane and still be cold in a sweater and Jacket .\n", "labels": "Why was I still cold in a sweater and jacket ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-8ddef0ed3a1c427885921ec83b17a9dc"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: Today church was the most disappointing experience I ' ve had all summer . THREE people 's cell phones rang . Why do you even have to bring it in with you ? You ca n't live an hour without it ?.\n", "labels": "Why did the narrator not feel moved at Church ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-1d0410af2e094f299e6252f87dbf82b5"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: I mentioned that I thought a guy was cute ( did n't really know , just found him attractive ) . My friend had had a class with him , so she took it upon herself to ask if he was single . It was a nice gesture , but now I feel like I ' m in High School . \" OMG ASK BRIAN IF HE LIKES ME ! ! \".\n", "labels": "What may be a fact about the speaker ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-9603bb46bf1b413e962da3d84c4c8097"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: I was sitting at my terminal taking calls , when I noticed that something was scratching at my navel . Thinking thaty my belt buckle was out of place or somesuch , I reached to adjust , only to deeply scratch my finger . Ow !.\n", "labels": "How did I feel at the end ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-147d329867414aca83dbc88198f7f038"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: Did I win ? No , the woman was already beginning her workout . As irrelavent to my day that was , I ended up going to work with her . Well , she took me to my school for that envelope service work I had to do .\n", "labels": "What 's a possible reason they went to work together ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-55b673ed331047a8b5a1a3d81fdfe97b"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: Since it seems to be the difficult one for me . I did my first eight hours in a long while . I spent literally no time at the library because I went work at 6 pm and then stayed until 2 am . That is good for my paycheck and I do not mind because today I will be adding to my work schedule and I will be getting thirty hours exactly .\n", "labels": "What may be the reason for getting out of work so late ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-69e5e156f20f43d6ba459c9779e600c6"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: I ca n't stand girl who think that they are the hottest ones at a party and therefore have to be all up on every guy . We get it . You like to flirt ... and make out ... and wear massive amounts of makeup .\n", "labels": "What am I bothered by when I go to parties ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-665afc2e09a542399752701741a3ac4f"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: I ca n't stand girl who think that they are the hottest ones at a party and therefore have to be all up on every guy . We get it . You like to flirt ... and make out ... and wear massive amounts of makeup .\n", "labels": "What usually happens when I go to parties ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-665afc2e09a542399752701741a3ac4f"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: I feel like I am now ministering out of the overflow of the Spirit of God , not out of the leftovers of an overly busy and stressed out life . I have truly learned much about \" being with God \" and then \" doing for God \" out of that time together . So , while I have a few new ideas , no , I do n't have a new ministry strategy . I do n't have some brilliant answer to age old problems .\n", "labels": "What is your plan for the ministry of God ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-27d46a1b92144f209470c81312f0d9dd"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: I made the best of it and made my first ever snowman ! Take that , eternal summer of Singapore ! We went on a couple of hikes , but nowhere near as much as last year . The crowd this year did n't like hiking that much I guess .\n", "labels": "Why do you think both you and other people did n't hike much this year ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-3a1d6ef57fd347019da63c14526d36da"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: Slaton has two injury - prone guys ahead of him on the depth chart , and could be a sleeper if he sees a lot of action . I failed to get a legitimate number one WR along the lines of Terrell Owens or Reggie Wayne , but Boldin could be a good value and Cotchery seems like a solid sleeper ( though I may have slightly overpaid for him ) . It seemed to me that Lee Evans went fairly cheaply , but he has proven to be inconsistent in his career .\n", "labels": "What sport is the narrator watching ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-294ea8eb9975468ab80ff8af1ed17055"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: We both have this \" lighter \" feeling about our home and ourselves . It makes me feel good in preparation for a new family member . Our home is n't the only thing rolling along . The clocks rolled ahead , as did my age and the baby is rolling along , literally , it seems .\n", "labels": "What does it mean that we are preparing for a \" new family \" ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-d6f55b73262e49069dd95cdb0f578e97"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: My dad has already told my late landlord 's son ( who was willed his father 's share of the park ) that under no circumstances will we give our cats up . We own our own house and are only renting the land . Can the \" new \" landlord demand we get rid of our cats ? Our three cats never go outside except to be carried to thet car in pet carriers for vet visits .\n", "labels": "What may be the reason for them worrying about their cats ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-ed9fa2f460bd4577bc418f1d2c9a4fd6"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: Man , it is all raining and horrifyingly dank and gothic outside , and the whole situation just makes me want to take a nap , preferably with something cuddly , like a cat or George Clooney . I have gotten nothing done for the last hour , and I am giving the heck up . I do n't know if I ' ve mentioned this before , but I work on the 23rd floor of my building , and so whenever it rains , it seems like the weather is just getting all up in my business , and I have to stop whatever I am doing and watch to make sure that lightning does n't come in through the windows and strangle me at my desk . Which , I am sure it ... would , and all .\n", "labels": "Why is the narrator distracted at work ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-6a14f7b900734402a0d4c5d69a0fec67"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: This was all going on to the soundtrack of the \" English speaking tour guide \" we 'd all paid 300 bhat for ( roughly a fiver ) who did n't so much speak English as slur some vaguely European sounding syllables together . I was sitting about 6 feet away from him and I still could n't make it out . But he was very jovial and he got his important points across ( be back on the boat by THREE PM or I leave you ! ) and he was also dressed like he got out of bed about 3 seconds before the boat left , so it was hard to be cheesed at him .\n", "labels": "What may have been the type of tour we were paying 300 bhat for ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-6f0d9a65a53c49db929ab3dae8986ad9"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: everywhere in the street children as well as adults sneak up on each other and throw water and brightly coloured powder at each other until everyone in the city is soaking wet and neon all over . unfortunately , my passport , which i thought was safe in my money belt , also participated in holi and is now covered in pink dye . i have to spend a few days in stinky delhi , applying for a temporary one .\n", "labels": "What might have happened if my passport was n't covered in pink dye ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-eb85bd302de54afb83aaf4d6c0e0d7a9"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: I guess that ... lack ... bothered me more than I was willing to admit . I guess admitting it would have hurt more than pretending I was okay with how things were . Or were not , in this case . It was a transformational weekend .\n", "labels": "How would you feel afterward ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-4ef9e3a809774b089e8a31a4f5a27a06"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: After all the discussion about the Omnivore 's Hundred , much of which revolved around how much curry is too much , I suddenly had a craving for Indian food . This curry - esque recipe was simple , made my kitchen smell yummy , and was pretty tasty to boot . Unfortunately , I had to omit the fresh basil , as the bunch that I bought earlier in the week was no longer edible .\n", "labels": "What made the kitchen smell good .?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-030aa63098554e9b9ec78d571c12b445"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: 3rd august . where i officially become 18!although the fact that my parents sucks the starting of it , overall i had a blasting birthday!all thanks to my dearestsss .. i was drenched with coke by the weirdos exactly at 12midnight.looking damn messy with the make - up i went home.then my mum spoilt the mood totally.i do nt think she remembers its my birthday at all . but who cares .\n", "labels": "What may have happened during their birthday celebration ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-75f2c00556cd484fb033ed7f933e2a14"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: I got the above picture with the 4x optical zoom on my camera while shooting into the sun . I took the picture and got the heck out of there before mom showed up . Click the picture to enlarge . I rode another 20 miles before getting back to Hwy 2 .\n", "labels": "Why am I suggesting to click the picture ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-3f219da854aa4eb0a6ced9899bd3851c"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: I realised today I hate the weekends because I am stalked by people who I really do n't want in my life . For some reason , they do n't stopped texting after I have ignored them for a week . Tired and pissed off that every weekend I have to fend off the donkeys .\n", "labels": "Why does the narrator not seem to be happy with the weekend ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-a62a9ce00cf149eda7932e7da4365226"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: sorry it took me this long to write about our second game . we lost 9 - 2 but the score does not describe the game at all . we are very even with the other team , they are not better then us in any way but there were a few factors that lead to such a big defeat . first we had our back up goalie in net .\n", "labels": "How would you describe the narrator ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-10b2ba3d286748deb458af7884281e54"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: I did drool for a bit over the Heritage Stripe Blue diaper bag . It is a perfect size and fit tote . Of course , I would n't need the diaper pad , but otherwise , it fit comfortably over my shoulder , plus had gobs of pockets inside and out , and had a long strap that could be attached or not . If I was working full time , I would have purchased it for sure .\n", "labels": "What may have been your reason for not getting the Heritage Stripe Blue diaper bag ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-52c32baf344e452a9d2de47a03f62494"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: I could never get the hang of mondays . I got up late this morning , because I could n't fall asleep until about three or so ... Anyway , I got to class , took a history test and had a full 45 minutes before my next class ... I thought it was time for some coffee but I realized that I had a presentation to give in my next course , so I ran to the library , did the prep work and ran to class . I did n't have to give it until Wednesday , or maybe even Friday .\n", "labels": "How may I be feeling right now ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-b169896c7f6043bca45e669248cb9da5"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: Mama Anna provided shawls to wear , for the chilly weather . Her husband then led us up from the house through coffee and banana fields . Pointing out trees and crops , prompting us to give continually incorrect guesses as to the products . Though socked - in , it was still quite beautiful .\n", "labels": "What type of environment does the narrator 's friend live ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-2ad5a21bf59d4fac86bf73482b01ac37"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: Not good . So of course we keep the protein bars with chocolate in them in the cupboard . But we keep the non - chocolate ones in the snack dish on the counter .\n", "labels": "Which food is in the cupboard ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-23e1ba15614546d2852bb0ccd7a7c79f"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: He 's always requesting custom videos from Me and Monique and Ceara ... and I ' m sure others , too . He 's actually really fucking annoying . Well apparently little miss jamie is a pain - slut too .\n", "labels": "Why is the narrator annoyed with \" him \" ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-b45662f75e504633bc2235a3c7f2d1b8"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: and yeah , we do have a lot of fun . but I would n't even dare to not look happy around them . everybody I know thinks that I am a happy person without any problems . The thing is , I m buried in their problems so I never have the time to complain about my own problems .\n", "labels": "Why do n't I complain about my problems .?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-80d8a42f899543b0a61657aa9e188cb3"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: Ate two candied almonds , so 25 calories , and that was that untill we headed home . Lots of walking all day . Then when we fueled up on the way home , I grabbed a 110 calorie protein bar .\n", "labels": "Why is the narrator focused on calories ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-dcf327476f1a4837b3ad8fd3518e2e61"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: One would think it 'd be warm . But noooo , I was freezing . Next to me stood a guy with a breath as smelly as Jack Sparrows after his drinking fest on the island .\n", "labels": "What was the reason the pirate had halitosis ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-163880fa61a24c1b8b1a9956a3fd7263"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: So Angela and I were rotting in Late Show today when we got a call from the gay bar ; the place had been broken into and their sound mixer had been stolen . Now both the gay bar and Late Show are owned by Crae , whose father died a few days ago . He was at the memorial service today when he was called and informed of the burglary .\n", "labels": "Why did what happened to the place happened ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-a8663eb583584ce0a09247d5b883b12d"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: I almost do n't want to blog about this - but on the chance that you might pray for them , I ca n't not do it . I got a prayer request last night for the Drews family . I have been online friends with Marsha for at least 3 years since I launched into the blogosphere . We have always kept in touch and commented each other .\n", "labels": "What motivation does the speaker have to blog about the Drews family ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-f27ae1850b05455ca389c66e295b6cea"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: I turned 33 yesterday . I never imagined I 'd be 33 and be pregnant with my first baby . I thought I 'd either have kids young or not have them at all . We ended up spending my birthday trying to fix a computer that got a virus at work . In baby news : Everything is still going well .\n", "labels": "What may happen after the birthday ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-3cb1375d014146619e40657a8c723375"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: Because I remember mine , and the best part of the whole thing for the pilgrim is that there are people from hours away who come , who do n't even know you , and they come just to sing to you and to encourage you and just to demonstrate the love of God . Anyways , last night we went and we got there about 7:30 PM and for about an hour and half all we did was sing , pray for the pilgrims , and just enjoy God . It 's great getting outside of your usual church home and experiencing praise and worship with total strangers .\n", "labels": "What can be said about the person who wrote this post .?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-e5c0fbe46d7b4c12a600df5b62e88a92"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: Also , we put in another door right behind their nesting boxes so we can collect eggs without going inside the coop . And of course there is a door for the chickens to enter the coop from their enclosed yard up a ramp , along with door for us to enter the coop , mostly just to clean it out . We 're hoping to get the roof and siding done this week along with the posts for the yard and chicken wire for the fencing .\n", "labels": "Why is the speaker putting this much effort into being able to collect the eggs efficiently ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-fff6fdcd06c84f1fa27442b0ef669641"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: First of all , I hated this version of Sleeping Beauty . I did nt like the fact that she had to sleep for a hundred years , and I did nt like how there were so many fairies and lastly I did nt like the fact that the prince did nt kiss her in order for her to wake up . I grew up with the Disney version of Sleeping Beauty and that movie is one of my favorite princess movies , going into this story I thought it would be more like that version than what it actually was .\n", "labels": "What will the narrator recommend to friends related to movies in the future ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-ef6e9524458043c991de55c6c13b5816"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: We ran into her while she was signing up and Stephen was taking his ASVAB test . She tried talking to him then and he ignored her . She later on that night sent him a myspace message saying congrats to us getting married but she 'd always love him no matter what .\n", "labels": "What may happen after taking the ASVAB test ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-1716a23dbb71490b96de0bf5416c322d"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: I decided that I want some extra cuddles with Mom . She 's always so busy and never sits down for long . So I pretended to be uncooperative so that she would carry me down the street to the park . She cuddled me the whole way .\n", "labels": "Which of the following is true of the mom ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-523cb64cfb304fff93544e62eb923d28"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: Man those boys like the Jaeger Bombs and showtunes . After that it was time for a few tacos and one last shot @ Hydrate before heading for home . There might be a video , if I get clearance .\n", "labels": "Why was it time for a few tacos ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-cc74195f7d7842859c4a511056a01f5a"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: The brilliance of the stars is not noticed until the darkness comes . I denied his request for a final kiss and went to the kitchen bin and helped the flower find its final resting place . I continue on with the same idea of living my life with no regrets . I do not choose to leave the outcome to chance .\n", "labels": "What did I do with the flower ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-3b2b942639c64a3e86dce8045c17a6eb"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: Currently , I ' ve been off fast food for a couple of months , probably for the first time in my life . Partly because I do n't have the money . And also because it would just make me feel horrible , in my head and in my stomach afterwards . I ' m not considering taking on the Vegan lifestyle full on , or protesting against cattle ranches , or only consuming \" organic \" foods ; right now it 's just sort of in the back of my mind .\n", "labels": "Why has the narrator quit on the fast food ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-5b2cb4b169474e569e44865de488ffd7"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: \" Are you planning on staring at me all evening ? \" he asked . Her eyes glanced away immediately . \" Why have n't you hurt me ? \" she asked , voicing her biggest fear .\n", "labels": "Why is she afraid of being hurt by me ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-23884f0d6df74ee7b204850c0d8ba6a1"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: Sweet girl was so happy to see us ! She looks a little , pieced together , right now ( poor tummy shaved for the ultrasound ! ) , but she 's as happy as can be . We 're trying to make do without the cone collar , and instead are trying a tee - shirt to keep her from messing with her stitches .\n", "labels": "What is the sweet girl ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-ccc9cb1710ab4ed5b98641841da32780"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: Okay kids - so one thing that you may want to know about me is that your dad and I really enjoy watching T.V. together . After you are in bed that is what we like to do .\n", "labels": "Why am I telling the kids that their dad and I like to watch TV ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-ae242f7257d84626b17dd9f2a339cb00"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: Got cancelled on around 4 times for the same jogging appointment with someone . Although this time round i told that someone that he / she need nt jog with me since he / she is tired and waking up early will only add to his / her fustration , i kinda felt irritated and sad . Afterall who likes being cancelled upon 4 times ? I really thought we 'd finally run this sunday morning .. Guess not .\n", "labels": "Why is the narrator growing suspicious of their friend ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-472eeec20492446a9516dfc89bf981fd"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: Then Ciara and I decided to go see Tropic Thunder at the movie theater while Nicole and Monica were going to stay behind and do karoake . Ciara and I get to the movies and the box office was closed , the dude said it closed 10 minutes before we got there . : ( So we decided to go get milkshakes at McDonald 's only to realize it was the opposite way we were going . So we settled for Burger King , who as we waited at the drive thru said they were n't taking any more orders .\n", "labels": "What did the narrator grab to eat ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-8e35dbca1a484336af0bc2b36a0cef44"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: They would n't give me drugs . They kept suggesting warm baths . I wanted to kill them all . Finally after 30 hours of that torture I begged someone to \" check \" me and when she did she said \" Oh , wow , you are at 6 !.\n", "labels": "Why is the narrator being so demanding ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-a3be06b4d7384c6c9bbcf36ca9625c3f"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: It happened 2 months ago . She went WAY over her texting and minutes and then they got in a big fight about it and she gave him attitude and I guess he was drunk and that made him furious . And he spanked her on the bare bottom for a long time and hit her with a switch a bunch of times and then with a wooden spoon and a metal soup ladel too and then he made her do corner time for a while before he let her get dressed again . That has to be seual abuse , or physical maybe ?.\n", "labels": "What has led the speaker to question this relationship and consider labeling it abusive ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-a8a6d4e981384618b7cc4bff05215a48"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: In college , I learned all about media law , Southern literature and even ballroom dance ( hey , I had to fulfill that P.E. requirement somehow ) . I did not , though , learn anything about managing my money .\n", "labels": "What may be a fact about this person ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-a0131fe14775457c8582d3db0351c272"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: One of the more interesting discoveries in Waukesha this weekend was a group set up and playing decent tunes outside Martha Merrill 's . The idea of Guitars For Vets started just a year ago when Dan Van Buskirk , a Vietnam Vet with PTSD , related his relief and the healing he felt from music to Patrick , his guitar teacher . In a rather short period of time , through donations and sponsors , this incredible organization now provides six guitar lessons and a guitar to U.S. Military Vets who ' ve suffered trauma and are receiving treatment through the VA system . They are currently working with VA centers in Milwaukee , Madison and Tomah but hope to expand to the Midwest and around the country by next year .\n", "labels": "What may happen after the music prograqm is set up with the VA ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-bad975fcfea74925b59d7f9659dea1f0"}, {"text": "Definition: Based on the given context, craft a common-sense question, especially those that are LONG, INTERESTING, and COMPLEX. The goal is to write questions that are easy for humans and hard for AI machines! To create such questions, here are some suggestions: A. What may (or may not) be the plausible reason for an event? B. What may (or may not) happen before (or after, or during) an event? C. What may (or may not) be a plausible fact about someone (or something)? D. What may (or may not) happen if an event happens (or did not happen)? You can also create other types of questions. DO NOT make your question answerable without looking at the context, or question of which the correct answer can be directly extracted from the context. DO NOT ask a question that requires very specialized knowledge that is not common sense. DO NOT ask too simple or too short questions. Your question must be related to the context and answerable with common sense. Try to add more variations and complexity to the questions.\nInput: Context: I ' m glad I made a reservation - there was an hour and a half wait for anyone else at 5:30PM . We were seated next to the Volcano Bar which has a working volcano overtop of it . We both had cheeseburgers and yummy drinks . The food was really quite good and fast , I was surprised .\n", "labels": "Why would someone actually want to sit near a working Volcano ?", "task_name": "task023_cosmosqa_question_generation", "task_category": "question_generation", "id": "task023-093734a27ff347ea9195aced2bf336d5"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: Hoping to rebound from their loss to the Patriots, the Raiders stayed at home for a Week 16 duel with the Houston Texans.  Oakland would get the early lead in the first quarter as quarterback JaMarcus Russell completed a 20-yard touchdown pass to rookie wide receiver Chaz Schilens.  The Texans would respond with fullback Vonta Leach getting a 1-yard touchdown run, yet the Raiders would answer with kicker Sebastian Janikowski getting a 33-yard and a 30-yard field goal.  Houston would tie the game in the second quarter with kicker Kris Brown getting a 53-yard and a 24-yard field goal. Oakland would take the lead in the third quarter with wide receiver Johnnie Lee Higgins catching a 29-yard touchdown pass from Russell, followed up by an 80-yard punt return for a touchdown.  The Texans tried to rally in the fourth quarter as Brown nailed a 40-yard field goal, yet the Raiders' defense would shut down any possible attempt.\n", "labels": "How many yards longer was the longest field goal than the shortest?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-2f2ea72410884c348d567d2eb2523022"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: Hoping to rebound from their loss to the Patriots, the Raiders stayed at home for a Week 16 duel with the Houston Texans.  Oakland would get the early lead in the first quarter as quarterback JaMarcus Russell completed a 20-yard touchdown pass to rookie wide receiver Chaz Schilens.  The Texans would respond with fullback Vonta Leach getting a 1-yard touchdown run, yet the Raiders would answer with kicker Sebastian Janikowski getting a 33-yard and a 30-yard field goal.  Houston would tie the game in the second quarter with kicker Kris Brown getting a 53-yard and a 24-yard field goal. Oakland would take the lead in the third quarter with wide receiver Johnnie Lee Higgins catching a 29-yard touchdown pass from Russell, followed up by an 80-yard punt return for a touchdown.  The Texans tried to rally in the fourth quarter as Brown nailed a 40-yard field goal, yet the Raiders' defense would shut down any possible attempt.\n", "labels": "How many yards was the longest field goal in the game?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-2f2ea72410884c348d567d2eb2523022"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: Hoping to rebound from their loss to the Patriots, the Raiders stayed at home for a Week 16 duel with the Houston Texans.  Oakland would get the early lead in the first quarter as quarterback JaMarcus Russell completed a 20-yard touchdown pass to rookie wide receiver Chaz Schilens.  The Texans would respond with fullback Vonta Leach getting a 1-yard touchdown run, yet the Raiders would answer with kicker Sebastian Janikowski getting a 33-yard and a 30-yard field goal.  Houston would tie the game in the second quarter with kicker Kris Brown getting a 53-yard and a 24-yard field goal. Oakland would take the lead in the third quarter with wide receiver Johnnie Lee Higgins catching a 29-yard touchdown pass from Russell, followed up by an 80-yard punt return for a touchdown.  The Texans tried to rally in the fourth quarter as Brown nailed a 40-yard field goal, yet the Raiders' defense would shut down any possible attempt.\n", "labels": "How many yards was the longest touchdown of the game?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-2f2ea72410884c348d567d2eb2523022"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: Still searching for their first win, the Bengals flew to Texas Stadium for a Week 5 interconference duel with the Dallas Cowboys.  In the first quarter, Cincinnati trailed early as Cowboys kicker Nick Folk got a 30-yard field goal, along with RB Felix Jones getting a 33-yard TD run.  In the second quarter, Dallas increased its lead as QB Tony Romo completed a 4-yard TD pass to TE Jason Witten.  The Bengals would end the half with kicker Shayne Graham getting a 41-yard and a 31-yard field goal. In the third quarter, Cincinnati tried to rally as QB Carson Palmer completed an 18-yard TD pass to WR T. J. Houshmandzadeh.  In the fourth quarter, the Bengals got closer as Graham got a 40-yard field goal, yet the Cowboys answered with Romo completing a 57-yard TD pass to WR Terrell Owens.  Cincinnati tried to come back as Palmer completed a 10-yard TD pass to Houshmandzadeh (with a failed 2-point conversion), but Dallas pulled away with Romo completing a 15-yard TD pass to WR Patrick Crayton.\n", "labels": "How many field goals did Shayne Graham kick in the first half?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-4c609ed0fb724a88bc95c8dd38e2c3f5"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: Still searching for their first win, the Bengals flew to Texas Stadium for a Week 5 interconference duel with the Dallas Cowboys.  In the first quarter, Cincinnati trailed early as Cowboys kicker Nick Folk got a 30-yard field goal, along with RB Felix Jones getting a 33-yard TD run.  In the second quarter, Dallas increased its lead as QB Tony Romo completed a 4-yard TD pass to TE Jason Witten.  The Bengals would end the half with kicker Shayne Graham getting a 41-yard and a 31-yard field goal. In the third quarter, Cincinnati tried to rally as QB Carson Palmer completed an 18-yard TD pass to WR T. J. Houshmandzadeh.  In the fourth quarter, the Bengals got closer as Graham got a 40-yard field goal, yet the Cowboys answered with Romo completing a 57-yard TD pass to WR Terrell Owens.  Cincinnati tried to come back as Palmer completed a 10-yard TD pass to Houshmandzadeh (with a failed 2-point conversion), but Dallas pulled away with Romo completing a 15-yard TD pass to WR Patrick Crayton.\n", "labels": "How many points did Dallas score in the first quarter?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-4c609ed0fb724a88bc95c8dd38e2c3f5"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: Still searching for their first win, the Bengals flew to Texas Stadium for a Week 5 interconference duel with the Dallas Cowboys.  In the first quarter, Cincinnati trailed early as Cowboys kicker Nick Folk got a 30-yard field goal, along with RB Felix Jones getting a 33-yard TD run.  In the second quarter, Dallas increased its lead as QB Tony Romo completed a 4-yard TD pass to TE Jason Witten.  The Bengals would end the half with kicker Shayne Graham getting a 41-yard and a 31-yard field goal. In the third quarter, Cincinnati tried to rally as QB Carson Palmer completed an 18-yard TD pass to WR T. J. Houshmandzadeh.  In the fourth quarter, the Bengals got closer as Graham got a 40-yard field goal, yet the Cowboys answered with Romo completing a 57-yard TD pass to WR Terrell Owens.  Cincinnati tried to come back as Palmer completed a 10-yard TD pass to Houshmandzadeh (with a failed 2-point conversion), but Dallas pulled away with Romo completing a 15-yard TD pass to WR Patrick Crayton.\n", "labels": "How many times did the Cowboys score in the first half?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-4c609ed0fb724a88bc95c8dd38e2c3f5"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: Still searching for their first win, the Bengals flew to Texas Stadium for a Week 5 interconference duel with the Dallas Cowboys.  In the first quarter, Cincinnati trailed early as Cowboys kicker Nick Folk got a 30-yard field goal, along with RB Felix Jones getting a 33-yard TD run.  In the second quarter, Dallas increased its lead as QB Tony Romo completed a 4-yard TD pass to TE Jason Witten.  The Bengals would end the half with kicker Shayne Graham getting a 41-yard and a 31-yard field goal. In the third quarter, Cincinnati tried to rally as QB Carson Palmer completed an 18-yard TD pass to WR T. J. Houshmandzadeh.  In the fourth quarter, the Bengals got closer as Graham got a 40-yard field goal, yet the Cowboys answered with Romo completing a 57-yard TD pass to WR Terrell Owens.  Cincinnati tried to come back as Palmer completed a 10-yard TD pass to Houshmandzadeh (with a failed 2-point conversion), but Dallas pulled away with Romo completing a 15-yard TD pass to WR Patrick Crayton.\n", "labels": "How many touchdowns did Jones have?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-4c609ed0fb724a88bc95c8dd38e2c3f5"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: Coming off their road loss to Green Bay, Washington returned home for a duel at FedExField with the Detroit Lions, matching up with them for the first time since 2010, but the first time in DC since 2007. After teams exchanged punts in the first, Washington struck first when Lions quarterback Matthew Stafford was intercepted by Redskins cornerback DeAngelo Hall who returned the interception 9 yards for a 7-0 Washington lead.Detroit responded right away, however, as running back Joique Bell broke numerous tackles on his way to a 12-yard touchdown, evening the score at 7-7. The first quarter would conclude without any further scoring. Early in the second, Detroit finished off a drive with Stafford finding tight end Joseph Fauria for a 5-yard score to give Detroit the 14-7 lead. Later in the second, after a few missed opportunities by both teams, Redskins running back Alfred Morris took off for a 30-yard score, evening the score at 14-14 late in the second. Washington's much-maligned defense, however, could not hold it and Detroit retook the lead before halftime thanks to a David Akers 32-yard field goal, giving Detroit a 17-14 edge and halfway to its first win ever in Washington. Early in the third, Washington was able to equalize with John Potter making his first career field goal from 43 yards out.Both teams failed to score again, and the teams went into the final quarter tied at 17. Early in the fourth, Detroit took the lead again, this time by way of Akers hitting a 28-yard field goal, giving Detroit a 20-17 lead. On Washington's ensuing possession, it appeared as if Washington took the lead, as Robert Griffin III found wide receiver Aldrick Robinson for a 57-yard score and an apparent 24-20 lead. However, upon video replay, it showed Robinson was bobbling the ball as he fell to the ground, thus the apparent pass was ruled incomplete, and Washington could not score again, punting. Detroit marched down the field and all but sealed the deal, as Stafford connected with superstar Calvin Johnson with just under four minutes remaining in the game. Washington didn't give up, however, and John Potter connected from 21 yards out to cut the lead to 27-20. However, after not recovering the onside kick, Washington needed to stop Detroit. They did this, and got the ball back with just under 50 seconds remaining. Washington got to midfield, but Griffin's hail mary pass was off the mark, and Washington lost once again. With the loss. Washington fell to 0-3 for the first time since 2001, and also lost to the Detroit Lions for the third straight time, having not beaten them since 2008 when they didn't even manage to win a single game. However, Washington lost to Detroit at home for the first time since 1935, when they were still the Boston Redskins. They had never, prior to this day, lost to Detroit in Washington or Landover.\n", "labels": "What was the score at the end of the first quarter?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-45addc474af7412da21f923dade9ae63"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: In 1085, Guadalajara was retaken by the Christian forces of Alfonso VI . The chronicles say that the Christian army was led by Alvar Fanez de Minaya, one of the lieutenants  of El Cid. From 1085 until the Battle of Las Navas de Tolosa in 1212, the city suffered wars against the Almoravid and the Almohad Empires. In spite of the wars, the Christian population could definitely settle down in the area thanks to the repopulation with people from the North  who received their first fuero in 1133 from Alfonso VII.In 1219, the king Fernando III gave a new fuero to the city .During the reign of Alfonso X of Castile, the protection of the king allowed the city to develop its economy by protecting merchants and allowing markets.\n", "labels": "How many years passed from when Guadalajara was retaken by the Christian forces of Alfonso VI  until the Battle of Las Navas de Tolosa?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-f84eb6f3d3c3440d8719ee49d4900b6b"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: In 1085, Guadalajara was retaken by the Christian forces of Alfonso VI . The chronicles say that the Christian army was led by Alvar Fanez de Minaya, one of the lieutenants  of El Cid. From 1085 until the Battle of Las Navas de Tolosa in 1212, the city suffered wars against the Almoravid and the Almohad Empires. In spite of the wars, the Christian population could definitely settle down in the area thanks to the repopulation with people from the North  who received their first fuero in 1133 from Alfonso VII.In 1219, the king Fernando III gave a new fuero to the city .During the reign of Alfonso X of Castile, the protection of the king allowed the city to develop its economy by protecting merchants and allowing markets.\n", "labels": "When did the first battle against Guadalajara take place?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-f84eb6f3d3c3440d8719ee49d4900b6b"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: Trying to snap a six-game losing skid, the Lions returned home for an NFC North rematch the-now 2-time NFC North champion Chicago Bears.  In the first quarter, the Bears struck first with kicker Robbie Gould nailing a 36-yard field goal.  Afterwards, the Lions took the lead with QB Jon Kitna completing a 23-yard TD pass to TE Dan Campbell.  In the second quarter, Chicago bounced back with QB Rex Grossman completing a 13-yard TD pass to WR Bernard Berrian.  Afterwards, RB Adrian Peterson got a 2-yard TD run.  In the third quarter, Detroit retook the lead with Kitna completing a 20-yard TD pass to WR Mike Furrey and a 2-yard TD pass to WR Roy Williams.  However, in the fourth quarter, the inconsistency that continues to plague the Lions showed as the Bears won with Gould getting a 36-yard field goal, a 39-yard field goal, and a 44-yard field goal and on a dropped pass by Mike Williams in the endzone on the last play of the game.  With their seventh-straight loss, the Lions fell to 2-13 as they were swept by their division rivals.\n", "labels": "How many more touchdown passes did Jon Kitna throw compared to everyone else?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-80d80797efe34c6c9ffca69afa3d0371"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: In 1905, 1,003 Korean immigrants, which included 802 men and 231 women and children, departed from the port of Chemulpo, Incheon aboard the ship Ilford to Salina Cruz, Oaxaca, Mexico. The journey took 45 days, after which they took a train to Coatzacoalcos, Veracruz. In the Veracruz port, another boat was taken to the port of Progreso with the final destination being the capital city of M\u00e9rida, Yucatan. They arrived in May 1905, with previously signed contracts for four years' work as indentured laborers on the Yucat\u00e1n henequen haciendas. Many of these Koreans were distributed throughout the Yucat\u00e1n in 32 henequen haciendas. The town of Motul, Yucatan, located in the heart of the henequen zone, was a destination for many of the Korean immigrants. Subsequently, in 1909, at the end of their contracts, they began a new stage in which they scattered even further  Thus, the majority of those who came were single men who made or remade their family lives with Yucatecan especially Maya women. While Korean girls were much more subject to marriages arranged by Korean parents, males had greater freedom when it came to making a family. This rapid intermarriage by Koreans, coupled with geographic dispersal, prevented the establishment of close social networks among these migrants and therefore provided the basis for Korean descendants among the Yucatan Peninsula. After that 1905 ship, no further entries of Koreans into Mexico were recorded, until many years later, leading to a new community of Koreans with completely different characteristics from those who entered in 1905. These descendants have started the Museo Conmemorativo de la Inmigraci\u00f3n Coreana a Yucat\u00e1n, a museum for the remembrance of their ancestors journey.\n", "labels": "How many Korean immigrants were more men than woman and children?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-7d524b81791f4ea9954d6b9fa3cf7641"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: In 1905, 1,003 Korean immigrants, which included 802 men and 231 women and children, departed from the port of Chemulpo, Incheon aboard the ship Ilford to Salina Cruz, Oaxaca, Mexico. The journey took 45 days, after which they took a train to Coatzacoalcos, Veracruz. In the Veracruz port, another boat was taken to the port of Progreso with the final destination being the capital city of M\u00e9rida, Yucatan. They arrived in May 1905, with previously signed contracts for four years' work as indentured laborers on the Yucat\u00e1n henequen haciendas. Many of these Koreans were distributed throughout the Yucat\u00e1n in 32 henequen haciendas. The town of Motul, Yucatan, located in the heart of the henequen zone, was a destination for many of the Korean immigrants. Subsequently, in 1909, at the end of their contracts, they began a new stage in which they scattered even further  Thus, the majority of those who came were single men who made or remade their family lives with Yucatecan especially Maya women. While Korean girls were much more subject to marriages arranged by Korean parents, males had greater freedom when it came to making a family. This rapid intermarriage by Koreans, coupled with geographic dispersal, prevented the establishment of close social networks among these migrants and therefore provided the basis for Korean descendants among the Yucatan Peninsula. After that 1905 ship, no further entries of Koreans into Mexico were recorded, until many years later, leading to a new community of Koreans with completely different characteristics from those who entered in 1905. These descendants have started the Museo Conmemorativo de la Inmigraci\u00f3n Coreana a Yucat\u00e1n, a museum for the remembrance of their ancestors journey.\n", "labels": "How many more Korean immigrants in 1905 that departed from Chemulpo were men than women or children?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-7d524b81791f4ea9954d6b9fa3cf7641"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: In 1905, 1,003 Korean immigrants, which included 802 men and 231 women and children, departed from the port of Chemulpo, Incheon aboard the ship Ilford to Salina Cruz, Oaxaca, Mexico. The journey took 45 days, after which they took a train to Coatzacoalcos, Veracruz. In the Veracruz port, another boat was taken to the port of Progreso with the final destination being the capital city of M\u00e9rida, Yucatan. They arrived in May 1905, with previously signed contracts for four years' work as indentured laborers on the Yucat\u00e1n henequen haciendas. Many of these Koreans were distributed throughout the Yucat\u00e1n in 32 henequen haciendas. The town of Motul, Yucatan, located in the heart of the henequen zone, was a destination for many of the Korean immigrants. Subsequently, in 1909, at the end of their contracts, they began a new stage in which they scattered even further  Thus, the majority of those who came were single men who made or remade their family lives with Yucatecan especially Maya women. While Korean girls were much more subject to marriages arranged by Korean parents, males had greater freedom when it came to making a family. This rapid intermarriage by Koreans, coupled with geographic dispersal, prevented the establishment of close social networks among these migrants and therefore provided the basis for Korean descendants among the Yucatan Peninsula. After that 1905 ship, no further entries of Koreans into Mexico were recorded, until many years later, leading to a new community of Koreans with completely different characteristics from those who entered in 1905. These descendants have started the Museo Conmemorativo de la Inmigraci\u00f3n Coreana a Yucat\u00e1n, a museum for the remembrance of their ancestors journey.\n", "labels": "How many years did the immigrants have to work as indentured laborers?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-7d524b81791f4ea9954d6b9fa3cf7641"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: In 1905, 1,003 Korean immigrants, which included 802 men and 231 women and children, departed from the port of Chemulpo, Incheon aboard the ship Ilford to Salina Cruz, Oaxaca, Mexico. The journey took 45 days, after which they took a train to Coatzacoalcos, Veracruz. In the Veracruz port, another boat was taken to the port of Progreso with the final destination being the capital city of M\u00e9rida, Yucatan. They arrived in May 1905, with previously signed contracts for four years' work as indentured laborers on the Yucat\u00e1n henequen haciendas. Many of these Koreans were distributed throughout the Yucat\u00e1n in 32 henequen haciendas. The town of Motul, Yucatan, located in the heart of the henequen zone, was a destination for many of the Korean immigrants. Subsequently, in 1909, at the end of their contracts, they began a new stage in which they scattered even further  Thus, the majority of those who came were single men who made or remade their family lives with Yucatecan especially Maya women. While Korean girls were much more subject to marriages arranged by Korean parents, males had greater freedom when it came to making a family. This rapid intermarriage by Koreans, coupled with geographic dispersal, prevented the establishment of close social networks among these migrants and therefore provided the basis for Korean descendants among the Yucatan Peninsula. After that 1905 ship, no further entries of Koreans into Mexico were recorded, until many years later, leading to a new community of Koreans with completely different characteristics from those who entered in 1905. These descendants have started the Museo Conmemorativo de la Inmigraci\u00f3n Coreana a Yucat\u00e1n, a museum for the remembrance of their ancestors journey.\n", "labels": "Where did the immigrants arrive first, Salina Cruz or Coatzacoalcos?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-7d524b81791f4ea9954d6b9fa3cf7641"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: The institutional framework of Navarre was preserved following the 1512 invasion. Once Ferdinand II of Aragon died in January, the Parliament of Navarre gathered in Pamplona, urging Charles V  to attend a coronation ceremony in the town following tradition, but the envoys of the Parliament were met with the Emperor's utter indifference if not contempt. He refused to attend any ceremony and responded with a brief \"let's say I am happy and  pleases me.\" Eventually the Parliament met in 1517 without Charles V, represented instead by the Duke of Najera pronouncing an array of promises of little certitude, while the acting Parliament kept piling up grievances and demands for damages due to the Emperor, totalling 67\u2014the 2nd Viceroy of Navarre Fadrique de Acu\u00f1a was deposed in 1515 probably for acceding to send grievances.:39-40 Contradictions inherent to the documents accounting for the Emperor's non-existent oath pledge in 1516 point to a contemporary manipulation of the records.\n", "labels": "Who represented the Emperor at Parliament?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-5464c6a2b0694ab689b89d50bd4fe504"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: The first issue in 1942 consisted of denominations of 1, 5, 10 and 50 centavos and 1, 5, and 10 Pesos. The next year brought \"replacement notes\" of the 1, 5 and 10 Pesos while 1944 ushered in a 100 Peso note and soon after an inflationary 500 Pesos note. In 1945, the Japanese issued a 1,000 Pesos note. This set of new money, which was printed even before the war, became known in the Philippines as Mickey Mouse money due to its very low value caused by severe inflation. Anti-Japanese newspapers portrayed stories of going to the market laden with suitcases or \"bayong\" (native bags made of woven coconut or Corypha leaf strips) overflowing with the Japanese-issued bills. In 1944, a box of matches cost more than 100 Mickey Mouse pesos. In 1945, a kilogram of camote cost around 1000 Mickey Mouse pesos. Inflation plagued the country with the devaluation of the Japanese money, evidenced by a 60% inflation experienced in January 1944.\n", "labels": "How many denominations were there in 1942?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-67f2725d0ae544628b46999edd217a4d"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: The first issue in 1942 consisted of denominations of 1, 5, 10 and 50 centavos and 1, 5, and 10 Pesos. The next year brought \"replacement notes\" of the 1, 5 and 10 Pesos while 1944 ushered in a 100 Peso note and soon after an inflationary 500 Pesos note. In 1945, the Japanese issued a 1,000 Pesos note. This set of new money, which was printed even before the war, became known in the Philippines as Mickey Mouse money due to its very low value caused by severe inflation. Anti-Japanese newspapers portrayed stories of going to the market laden with suitcases or \"bayong\" (native bags made of woven coconut or Corypha leaf strips) overflowing with the Japanese-issued bills. In 1944, a box of matches cost more than 100 Mickey Mouse pesos. In 1945, a kilogram of camote cost around 1000 Mickey Mouse pesos. Inflation plagued the country with the devaluation of the Japanese money, evidenced by a 60% inflation experienced in January 1944.\n", "labels": "How many years after the first issue of 1, 5, 10 and 50 centavos did the Japanese issue a 1,000 Pesos note?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-67f2725d0ae544628b46999edd217a4d"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: The first issue in 1942 consisted of denominations of 1, 5, 10 and 50 centavos and 1, 5, and 10 Pesos. The next year brought \"replacement notes\" of the 1, 5 and 10 Pesos while 1944 ushered in a 100 Peso note and soon after an inflationary 500 Pesos note. In 1945, the Japanese issued a 1,000 Pesos note. This set of new money, which was printed even before the war, became known in the Philippines as Mickey Mouse money due to its very low value caused by severe inflation. Anti-Japanese newspapers portrayed stories of going to the market laden with suitcases or \"bayong\" (native bags made of woven coconut or Corypha leaf strips) overflowing with the Japanese-issued bills. In 1944, a box of matches cost more than 100 Mickey Mouse pesos. In 1945, a kilogram of camote cost around 1000 Mickey Mouse pesos. Inflation plagued the country with the devaluation of the Japanese money, evidenced by a 60% inflation experienced in January 1944.\n", "labels": "Was the largest peso note printed in 1944 or 1945?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-67f2725d0ae544628b46999edd217a4d"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: The first issue in 1942 consisted of denominations of 1, 5, 10 and 50 centavos and 1, 5, and 10 Pesos. The next year brought \"replacement notes\" of the 1, 5 and 10 Pesos while 1944 ushered in a 100 Peso note and soon after an inflationary 500 Pesos note. In 1945, the Japanese issued a 1,000 Pesos note. This set of new money, which was printed even before the war, became known in the Philippines as Mickey Mouse money due to its very low value caused by severe inflation. Anti-Japanese newspapers portrayed stories of going to the market laden with suitcases or \"bayong\" (native bags made of woven coconut or Corypha leaf strips) overflowing with the Japanese-issued bills. In 1944, a box of matches cost more than 100 Mickey Mouse pesos. In 1945, a kilogram of camote cost around 1000 Mickey Mouse pesos. Inflation plagued the country with the devaluation of the Japanese money, evidenced by a 60% inflation experienced in January 1944.\n", "labels": "What was more expensive, matches in 1944 or a kilogram of camote in 1945?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-67f2725d0ae544628b46999edd217a4d"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: The first issue in 1942 consisted of denominations of 1, 5, 10 and 50 centavos and 1, 5, and 10 Pesos. The next year brought \"replacement notes\" of the 1, 5 and 10 Pesos while 1944 ushered in a 100 Peso note and soon after an inflationary 500 Pesos note. In 1945, the Japanese issued a 1,000 Pesos note. This set of new money, which was printed even before the war, became known in the Philippines as Mickey Mouse money due to its very low value caused by severe inflation. Anti-Japanese newspapers portrayed stories of going to the market laden with suitcases or \"bayong\" (native bags made of woven coconut or Corypha leaf strips) overflowing with the Japanese-issued bills. In 1944, a box of matches cost more than 100 Mickey Mouse pesos. In 1945, a kilogram of camote cost around 1000 Mickey Mouse pesos. Inflation plagued the country with the devaluation of the Japanese money, evidenced by a 60% inflation experienced in January 1944.\n", "labels": "Which item cost the highest mickey mouse pesos in the 1940s, a kilogram of Camote or matches?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-67f2725d0ae544628b46999edd217a4d"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: Trying to stop a two-game skid, the Bengals went home for their second Monday Night game of the year.  They would wear their alternate uniforms as they played against the New England Patriots.  In the first quarter, Cincinnati trailed early as Patriots kicker Stephen Gostkowski got a 31-yard field goal, while QB Tom Brady completed a 1-yard TD pass to LB Mike Vrabel.  In the second quarter, the Bengals managed to get on the board with QB Carson Palmer completing a 1-yard TD pass to WR T. J. Houshmandzadeh.  However, New England would respond with Brady completing a 7-yard TD pass to WR Randy Moss. Near the end of the 2nd quarter, the Bengals had a chance to get closer to the Pats, but a miscommunication between Palmer and Chad Johnson resulted in a New England interception. Palmer and Johnson argued on the sidelines and all the way back to the locker room. In the third quarter, Cincinnati continued to struggle as Patriots RB Sammy Morris getting a 7-yard TD run.  The Bengals' only response would be kicker Shayne Graham nailing a 40-yard field goal.  In the fourth quarter, New England increased its lead with Gostkowski kicking a 36-yard field goal.  Cincinnati's final response was Graham kicking a 48-yard field goal.  Afterwards, the Patriots sealed the win with Brady and Moss hooking up with each other again on a 14-yard TD pass. With their third-straight loss, the Bengals entered their bye week at 1-3.  It also marked Cincinnati's 5th loss to New England in the last 6 meetings. After the game, Marvin Lewis was heard yelling at his players for their 1-3 record. He called a team meeting a few days later and said that \"anyone who doesn't want to play shouldn't show up\".\n", "labels": "How many field goals were there in the second half?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-830990b49cb44267bb63625b8a07b7f3"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: Trying to stop a two-game skid, the Bengals went home for their second Monday Night game of the year.  They would wear their alternate uniforms as they played against the New England Patriots.  In the first quarter, Cincinnati trailed early as Patriots kicker Stephen Gostkowski got a 31-yard field goal, while QB Tom Brady completed a 1-yard TD pass to LB Mike Vrabel.  In the second quarter, the Bengals managed to get on the board with QB Carson Palmer completing a 1-yard TD pass to WR T. J. Houshmandzadeh.  However, New England would respond with Brady completing a 7-yard TD pass to WR Randy Moss. Near the end of the 2nd quarter, the Bengals had a chance to get closer to the Pats, but a miscommunication between Palmer and Chad Johnson resulted in a New England interception. Palmer and Johnson argued on the sidelines and all the way back to the locker room. In the third quarter, Cincinnati continued to struggle as Patriots RB Sammy Morris getting a 7-yard TD run.  The Bengals' only response would be kicker Shayne Graham nailing a 40-yard field goal.  In the fourth quarter, New England increased its lead with Gostkowski kicking a 36-yard field goal.  Cincinnati's final response was Graham kicking a 48-yard field goal.  Afterwards, the Patriots sealed the win with Brady and Moss hooking up with each other again on a 14-yard TD pass. With their third-straight loss, the Bengals entered their bye week at 1-3.  It also marked Cincinnati's 5th loss to New England in the last 6 meetings. After the game, Marvin Lewis was heard yelling at his players for their 1-3 record. He called a team meeting a few days later and said that \"anyone who doesn't want to play shouldn't show up\".\n", "labels": "How many games in a row did the Bengals lose?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-830990b49cb44267bb63625b8a07b7f3"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: Trying to stop a two-game skid, the Bengals went home for their second Monday Night game of the year.  They would wear their alternate uniforms as they played against the New England Patriots.  In the first quarter, Cincinnati trailed early as Patriots kicker Stephen Gostkowski got a 31-yard field goal, while QB Tom Brady completed a 1-yard TD pass to LB Mike Vrabel.  In the second quarter, the Bengals managed to get on the board with QB Carson Palmer completing a 1-yard TD pass to WR T. J. Houshmandzadeh.  However, New England would respond with Brady completing a 7-yard TD pass to WR Randy Moss. Near the end of the 2nd quarter, the Bengals had a chance to get closer to the Pats, but a miscommunication between Palmer and Chad Johnson resulted in a New England interception. Palmer and Johnson argued on the sidelines and all the way back to the locker room. In the third quarter, Cincinnati continued to struggle as Patriots RB Sammy Morris getting a 7-yard TD run.  The Bengals' only response would be kicker Shayne Graham nailing a 40-yard field goal.  In the fourth quarter, New England increased its lead with Gostkowski kicking a 36-yard field goal.  Cincinnati's final response was Graham kicking a 48-yard field goal.  Afterwards, the Patriots sealed the win with Brady and Moss hooking up with each other again on a 14-yard TD pass. With their third-straight loss, the Bengals entered their bye week at 1-3.  It also marked Cincinnati's 5th loss to New England in the last 6 meetings. After the game, Marvin Lewis was heard yelling at his players for their 1-3 record. He called a team meeting a few days later and said that \"anyone who doesn't want to play shouldn't show up\".\n", "labels": "How many points did the Patriots have at the start of the second quarter?7?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-830990b49cb44267bb63625b8a07b7f3"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: Trying to stop a two-game skid, the Bengals went home for their second Monday Night game of the year.  They would wear their alternate uniforms as they played against the New England Patriots.  In the first quarter, Cincinnati trailed early as Patriots kicker Stephen Gostkowski got a 31-yard field goal, while QB Tom Brady completed a 1-yard TD pass to LB Mike Vrabel.  In the second quarter, the Bengals managed to get on the board with QB Carson Palmer completing a 1-yard TD pass to WR T. J. Houshmandzadeh.  However, New England would respond with Brady completing a 7-yard TD pass to WR Randy Moss. Near the end of the 2nd quarter, the Bengals had a chance to get closer to the Pats, but a miscommunication between Palmer and Chad Johnson resulted in a New England interception. Palmer and Johnson argued on the sidelines and all the way back to the locker room. In the third quarter, Cincinnati continued to struggle as Patriots RB Sammy Morris getting a 7-yard TD run.  The Bengals' only response would be kicker Shayne Graham nailing a 40-yard field goal.  In the fourth quarter, New England increased its lead with Gostkowski kicking a 36-yard field goal.  Cincinnati's final response was Graham kicking a 48-yard field goal.  Afterwards, the Patriots sealed the win with Brady and Moss hooking up with each other again on a 14-yard TD pass. With their third-straight loss, the Bengals entered their bye week at 1-3.  It also marked Cincinnati's 5th loss to New England in the last 6 meetings. After the game, Marvin Lewis was heard yelling at his players for their 1-3 record. He called a team meeting a few days later and said that \"anyone who doesn't want to play shouldn't show up\".\n", "labels": "Which kicker scored the most field goals in the second half?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-830990b49cb44267bb63625b8a07b7f3"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: The total number of active military personnel in the Croatian Armed Forces stands at 14,506 and 6,000 reserves working in various service branches of the armed forces. In May 2016, Armed Forces had 16,019 members, of which 14,506 were active military personnel and 1,513 civil servants. Of the 14,506 active military personnel, 3,183 were officers, 5,389 non-commissioned officers, 5,393 soldiers, 520 military specialists, 337 civil servants and 1,176 other employees.\n", "labels": "How many active military personnel and reserve are in the Croatian Armed Forces?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-21f32f941e034d6f91f52e6e8b3883bc"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: The total number of active military personnel in the Croatian Armed Forces stands at 14,506 and 6,000 reserves working in various service branches of the armed forces. In May 2016, Armed Forces had 16,019 members, of which 14,506 were active military personnel and 1,513 civil servants. Of the 14,506 active military personnel, 3,183 were officers, 5,389 non-commissioned officers, 5,393 soldiers, 520 military specialists, 337 civil servants and 1,176 other employees.\n", "labels": "How many more active military personnel were in the Armed Forces than civil servants?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-21f32f941e034d6f91f52e6e8b3883bc"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: The total number of active military personnel in the Croatian Armed Forces stands at 14,506 and 6,000 reserves working in various service branches of the armed forces. In May 2016, Armed Forces had 16,019 members, of which 14,506 were active military personnel and 1,513 civil servants. Of the 14,506 active military personnel, 3,183 were officers, 5,389 non-commissioned officers, 5,393 soldiers, 520 military specialists, 337 civil servants and 1,176 other employees.\n", "labels": "How many more non commissioned officers than military specialists were in the Croatian Armed Forces in 2016?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-21f32f941e034d6f91f52e6e8b3883bc"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: The total number of active military personnel in the Croatian Armed Forces stands at 14,506 and 6,000 reserves working in various service branches of the armed forces. In May 2016, Armed Forces had 16,019 members, of which 14,506 were active military personnel and 1,513 civil servants. Of the 14,506 active military personnel, 3,183 were officers, 5,389 non-commissioned officers, 5,393 soldiers, 520 military specialists, 337 civil servants and 1,176 other employees.\n", "labels": "How many of the personnel were not civil servants?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-21f32f941e034d6f91f52e6e8b3883bc"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: The total number of active military personnel in the Croatian Armed Forces stands at 14,506 and 6,000 reserves working in various service branches of the armed forces. In May 2016, Armed Forces had 16,019 members, of which 14,506 were active military personnel and 1,513 civil servants. Of the 14,506 active military personnel, 3,183 were officers, 5,389 non-commissioned officers, 5,393 soldiers, 520 military specialists, 337 civil servants and 1,176 other employees.\n", "labels": "How many of the personnel were not soldiers?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-21f32f941e034d6f91f52e6e8b3883bc"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: Coming off a bye week, the Bears battled rival Detroit Lions on Monday Night Football. The team kept the Lions from scoring until the fourth quarter, and forced four takeaways. The first turnover forced was in the first half, when Lance Briggs stripped the ball from Mikel Leshoure, which was recovered by Julius Peppers. The second and third turnovers were forced in the third quarter on Zack Bowman's muffed punt recovery, and Brian Urlacher recovering Joique Bell's fumble. The final turnover occurred when Lions quarterback Matthew Stafford's pass was intercepted by D. J. Moore. The Bears struck first on Jay Cutler's touchdown pass to Brandon Marshall, and Robbie Gould kicked a field goal to increase the first half score to 10-0. During the game, Lions defensive tackle Ndamukong Suh threw Cutler to the ground, injuring his ribs. Cutler was eventually replaced by Jason Campbell for a play, before returning to the game. In the second half, Gould kicked another field goal, and prevented the Lions from scoring until the final 36 seconds of the game, when Stafford threw a 12-yard touchdown pass to Ryan Broyles to narrow the margin to six points, but the Bears sealed the victory by recovering the ensuring onside kick.\n", "labels": "How many quarters did the Lions play before scoring any points?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-57c1e88916734d7f9e5e3ebedd28c56b"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: Coming off a bye week, the Bears battled rival Detroit Lions on Monday Night Football. The team kept the Lions from scoring until the fourth quarter, and forced four takeaways. The first turnover forced was in the first half, when Lance Briggs stripped the ball from Mikel Leshoure, which was recovered by Julius Peppers. The second and third turnovers were forced in the third quarter on Zack Bowman's muffed punt recovery, and Brian Urlacher recovering Joique Bell's fumble. The final turnover occurred when Lions quarterback Matthew Stafford's pass was intercepted by D. J. Moore. The Bears struck first on Jay Cutler's touchdown pass to Brandon Marshall, and Robbie Gould kicked a field goal to increase the first half score to 10-0. During the game, Lions defensive tackle Ndamukong Suh threw Cutler to the ground, injuring his ribs. Cutler was eventually replaced by Jason Campbell for a play, before returning to the game. In the second half, Gould kicked another field goal, and prevented the Lions from scoring until the final 36 seconds of the game, when Stafford threw a 12-yard touchdown pass to Ryan Broyles to narrow the margin to six points, but the Bears sealed the victory by recovering the ensuring onside kick.\n", "labels": "How many total points were scored by halftime?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-57c1e88916734d7f9e5e3ebedd28c56b"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: This game involved a scary moment, after Seattle's  Ricardo Lockette was hit during a kick return. He lied on the ground, motionless, for about 7 minutes before he was taken off the field on a cart. X-rays later revealed that Lockette had a broken neck. The injury ended his career. The Cowboys would only kick field goals in this game, as Dan Bailey was 4 for 4 on field goals. Dallas lead 12-10 with under 2 minutes to go. However, the Seahawks would march down the field and would take a 13-12 lead after Steven Hauschka drilled a 24-yard field goal. Dallas tried to come back, but Seattle forced a turnover on downs to end the game.\n", "labels": "How many points total were scored in the game?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-7b4d3dc3ae134450aceba3f8efcb4615"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: This game involved a scary moment, after Seattle's  Ricardo Lockette was hit during a kick return. He lied on the ground, motionless, for about 7 minutes before he was taken off the field on a cart. X-rays later revealed that Lockette had a broken neck. The injury ended his career. The Cowboys would only kick field goals in this game, as Dan Bailey was 4 for 4 on field goals. Dallas lead 12-10 with under 2 minutes to go. However, the Seahawks would march down the field and would take a 13-12 lead after Steven Hauschka drilled a 24-yard field goal. Dallas tried to come back, but Seattle forced a turnover on downs to end the game.\n", "labels": "With less than 2 minutes to go, how many points ahead was Dallas?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-7b4d3dc3ae134450aceba3f8efcb4615"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: As of the census of 2000, there were 40,543 people, 15,416 households, and 11,068 families residing in the county. The population density was 99 people per square mile (38/km\u00b2). There were 16,577 housing units at an average density of 40 per square mile (16/km\u00b2). The racial makeup of the county was 95.99% Race (United States Census), 2.19% Race (United States Census) or Race (United States Census), 0.26% Race (United States Census), 0.38% Race (United States Census), 0.20% from Race (United States Census), and 0.97% from two or more races. 0.66% of the population were Race (United States Census) or Race (United States Census) of any race. 29.3% were of united states, 22.2% germans, 12.1% english people and 10.9% irish people ancestry according to 2000 United States Census.\n", "labels": "In the county according to the 2000 census, which group is larger: people or households?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-55cbad72cb6447b58d2aed4555d75f52"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: As of the census of 2000, there were 40,543 people, 15,416 households, and 11,068 families residing in the county. The population density was 99 people per square mile (38/km\u00b2). There were 16,577 housing units at an average density of 40 per square mile (16/km\u00b2). The racial makeup of the county was 95.99% Race (United States Census), 2.19% Race (United States Census) or Race (United States Census), 0.26% Race (United States Census), 0.38% Race (United States Census), 0.20% from Race (United States Census), and 0.97% from two or more races. 0.66% of the population were Race (United States Census) or Race (United States Census) of any race. 29.3% were of united states, 22.2% germans, 12.1% english people and 10.9% irish people ancestry according to 2000 United States Census.\n", "labels": "Which ancestral groups are at least 10%?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-55cbad72cb6447b58d2aed4555d75f52"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: As of the census of 2000, there were 40,543 people, 15,416 households, and 11,068 families residing in the county. The population density was 99 people per square mile (38/km\u00b2). There were 16,577 housing units at an average density of 40 per square mile (16/km\u00b2). The racial makeup of the county was 95.99% Race (United States Census), 2.19% Race (United States Census) or Race (United States Census), 0.26% Race (United States Census), 0.38% Race (United States Census), 0.20% from Race (United States Census), and 0.97% from two or more races. 0.66% of the population were Race (United States Census) or Race (United States Census) of any race. 29.3% were of united states, 22.2% germans, 12.1% english people and 10.9% irish people ancestry according to 2000 United States Census.\n", "labels": "Which group is larger according to the census: people or households?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-55cbad72cb6447b58d2aed4555d75f52"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: The Broncos began their 2008 campaign on the road against their AFC West rival, the Oakland Raiders, in the second game of ESPN's Monday Night Football doubleheader.  In the first quarter, Denver ran out of the gates early as QB Jay Cutler completed a 26-yard TD pass to rookie WR Eddie Royal (who was filling in for WR Brandon Marshall, due to his 1-game suspension).  In the second quarter, the Broncos continued their domination as kicker Matt Prater got a 26-yard field goal, while FB Michael Pittman got a 3-yard TD run.  In the third quarter, Denver ran away with the game as Cutler completed a 48-yard TD pass to WR Darrell Jackson, while Prater nailed a 43-yard field goal.  In the fourth quarter, the Raiders spoiled the Broncos' bid for a shutout as QB JaMarcus Russell completed an 8-yard TD pass to WR Ashley Lelie.  Denver ended its domination with RB Selvin Young's 5-yard TD run and Pittman's 1-yard TD run.  Oakland ended the scoring with Russell completing a 4-yard TD pass to WR Ronald Curry. With the dominating win, the Broncos began their season at 1-0; with the rest of the division suffering losses, Denver, in Week 1, is in sole possession of 1st place. Eddie Royal, in his NFL debut, had the best Week 1 stats of any wide receiver, getting 9 receptions for 146 yards and a touchdown.\n", "labels": "How many touchdowns did JaMarcus Russell have?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-c47dfa1921f1414daec195f21005108e"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: The Broncos began their 2008 campaign on the road against their AFC West rival, the Oakland Raiders, in the second game of ESPN's Monday Night Football doubleheader.  In the first quarter, Denver ran out of the gates early as QB Jay Cutler completed a 26-yard TD pass to rookie WR Eddie Royal (who was filling in for WR Brandon Marshall, due to his 1-game suspension).  In the second quarter, the Broncos continued their domination as kicker Matt Prater got a 26-yard field goal, while FB Michael Pittman got a 3-yard TD run.  In the third quarter, Denver ran away with the game as Cutler completed a 48-yard TD pass to WR Darrell Jackson, while Prater nailed a 43-yard field goal.  In the fourth quarter, the Raiders spoiled the Broncos' bid for a shutout as QB JaMarcus Russell completed an 8-yard TD pass to WR Ashley Lelie.  Denver ended its domination with RB Selvin Young's 5-yard TD run and Pittman's 1-yard TD run.  Oakland ended the scoring with Russell completing a 4-yard TD pass to WR Ronald Curry. With the dominating win, the Broncos began their season at 1-0; with the rest of the division suffering losses, Denver, in Week 1, is in sole possession of 1st place. Eddie Royal, in his NFL debut, had the best Week 1 stats of any wide receiver, getting 9 receptions for 146 yards and a touchdown.\n", "labels": "Which quarterback had the longest touchdown pass?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-c47dfa1921f1414daec195f21005108e"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: Coming off their impressive road win over the 49ers, the Falcons went home for a Week 6 Sunday night duel with the Chicago Bears.  After a scoreless first quarter, Atlanta would trail early in the second quarter as Bears quarterback Jay Cutler found wide receiver Johnny Knox on a 23-yard touchdown pass.  Afterwards, the Falcons took the lead as quarterback Matt Ryan completed a 40-yard touchdown pass to wide receiver Roddy White and a 10-yard touchdown pass to tight end Tony Gonzalez. After a scoreless third quarter, Chicago would tie the game in the fourth quarter with Cutler hooking up with tight end Greg Olsen on a 2-yard touchdown.  Atlanta would regain the lead as running back Michael Turner got a 5-yard touchdown run.  Afterwards, the defense would fend off a last-second Bears drive to lock up the victory.\n", "labels": "How many touchdowns were scored during the game?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-7a853b60015a42e8b2b1a105e1ef5d69"}, {"text": "Definition: This task involves creating questions from a given passage that involve some kind of complex reasoning (including numerical reasoning). \nThe generated questions must require looking at more than one part of the passage to answer. Try to use a variety of reasoning types in your questions (some of the sample reasoning types are illustrated in the 'positive examples' field) and also have a variety of answer types (spans, numbers, dates). A span is a continuous phrase taken directly from the passage or question. In such questions, the answer spans are recommended to be five words. Questions with answer type \"number\", are expected to start with \"How many\". For instance, \"How long did an event last?\" can be phrased as \"How many years did an event last?\". To increase diversity, try to create more questions for which the answer will be multiple spans. Please create questions that require AT LEAST 2 arithmetic operations as illustrated in the 'positive examples' field. You are encouraged to try DIFFERENT COMBINATIONS of operations and not just limit to the combinations shown in those examples. The more challenging questions you create, the better it is.\nInput: Passage: Coming off their impressive road win over the 49ers, the Falcons went home for a Week 6 Sunday night duel with the Chicago Bears.  After a scoreless first quarter, Atlanta would trail early in the second quarter as Bears quarterback Jay Cutler found wide receiver Johnny Knox on a 23-yard touchdown pass.  Afterwards, the Falcons took the lead as quarterback Matt Ryan completed a 40-yard touchdown pass to wide receiver Roddy White and a 10-yard touchdown pass to tight end Tony Gonzalez. After a scoreless third quarter, Chicago would tie the game in the fourth quarter with Cutler hooking up with tight end Greg Olsen on a 2-yard touchdown.  Atlanta would regain the lead as running back Michael Turner got a 5-yard touchdown run.  Afterwards, the defense would fend off a last-second Bears drive to lock up the victory.\n", "labels": "How many yards was the longest touchdown pass?", "task_name": "task026_drop_question_generation", "task_category": "question_generation", "id": "task026-7a853b60015a42e8b2b1a105e1ef5d69"}]