user string | videos list | gt int64 | predict string | pred_order list | top1_correct bool | gt_pos int64 | index int64 |
|---|---|---|---|---|---|---|---|
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_eVTMUEYhwDE.mp4",
"activitynet/videos/v_NpBZn7OHUKo.mp4",
"activitynet/videos/v_uqiMw7tQ1Cc.mp4",
"activitynet/videos/v_gIhVeU8xbrs.mp4",
"activitynet/videos/v_1epGZvRN3Fw.mp4"
] | 3 | <think>
<content>
[1] shows a man lifting weights with a coach, but the coach is not actively helping with body placement.
[2] shows a coach helping a man with a weightlifting exercise, but the focus is on a different exercise.
[3] shows a coach helping a man with a weightlifting exercise, with clear body placement gui... | [
3
] | true | 1 | 0 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_msiX-xky6Ac.mp4",
"activitynet/videos/v_Ty0BvWyYPVA.mp4",
"activitynet/videos/v_pgff9mC5y3s.mp4",
"activitynet/videos/v_4Lu8ECLHvK4.mp4",
"activitynet/videos/v_f6j6lb0AaxM.mp4"
] | 4 | <think>
<content>
[1] shows a game of hopscotch with children on a grid, but lacks the specific interaction with a man pointing and the movement to another tile.
[2] depicts a man and a boy on a track, but no children or soccer field involvement.
[3] features children playing near a bouncy castle, not involving a socce... | [] | false | 0 | 1 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_yj2WJBqmEHk.mp4",
"activitynet/videos/v__uKKSGTNJAY.mp4",
"activitynet/videos/v_5I0K3y27EUM.mp4",
"activitynet/videos/v_JTrwGfPJNzU.mp4",
"activitynet/videos/v_hghdjiQlYko.mp4"
] | 4 | <think>
<content>
[1] shows a snowy playground with people sliding down a snowy hill, but the focus is on snow, not a typical playground slide.
[2] depicts a playground slide with children sliding down, but the focus is on a concrete surface, not a slide.
[3] shows a woman and a child sliding down a red playground slid... | [
3
] | false | 0 | 2 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_rwxSphRRIL8.mp4",
"activitynet/videos/v_kbe4iowYMqM.mp4",
"activitynet/videos/v_3j2d27w3x5Q.mp4",
"activitynet/videos/v_RKDjetk5Kko.mp4",
"activitynet/videos/v_Po8gmt7hVTY.mkv"
] | 5 | <think>
<content>
[1] The video shows a chef preparing pasta, but the focus is on boiling water and adding ingredients, not specifically on boiling noodles or tasting them. It lacks the detailed steps of the query.
[2] This video features a woman preparing pasta, including boiling noodles and tasting them. It aligns c... | [
2
] | false | 0 | 3 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_UqU_mAjgknQ.mp4",
"activitynet/videos/v_-wXbBZDSIa8.mp4",
"activitynet/videos/v_eFbZ0_TJLE0.mkv",
"activitynet/videos/v_5qsXmDi8d74.mp4",
"activitynet/videos/v_0zjA3KPnLK8.mp4"
] | 4 | <think>
<content>
[1] The video shows a wrestling match with a focus on the ring and audience, but lacks the specific actions of walking out, sitting on the floor, and fighting as described in the query.
[2] This video depicts a wrestling match with two men engaging in combat, including a push and a fall, which aligns ... | [
2
] | false | 0 | 4 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_3K_8CdJS9lE.mp4",
"activitynet/videos/v_ymJTN8aKZEw.mp4",
"activitynet/videos/v_rgAALWYnRrg.mp4",
"activitynet/videos/v_Wgh8e4V8hBc.mp4",
"activitynet/videos/v_iiqaJGokpEw.mp4"
] | 3 | <think>
<content>
[1] shows a man and a girl in a park, with the girl holding a bat and hitting a piñata. The girl swings the bat in the air and breaks the piñata, with candies falling. This matches the query closely.
[2] features a man and a girl in a park, with the girl holding a bat and hitting a piñata. The girl sw... | [
1
] | false | 0 | 6 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_arnKDX_ToxE.mp4",
"activitynet/videos/v_8-QcL1k5n6k.mp4",
"activitynet/videos/v_D0pVkTEYQg8.mp4",
"activitynet/videos/v_tnt6Wpv_kHc.mp4",
"activitynet/videos/v_c4ctwOucndQ.mp4"
] | 3 | <think>
<content>
[1] The video shows a man using a stationary bike, but the focus is on the bike's interface and the man's interaction with it, not the exercise action itself. The man is seen adjusting the bike and interacting with the controls, but there is no clear depiction of him holding the handles and letting go... | [
3
] | true | 1 | 5 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_AXw2bkQyRPo.mp4",
"activitynet/videos/v_V4U5SaPDL0E.mp4",
"activitynet/videos/v_tY6UFSLtIoE.mp4",
"activitynet/videos/v_CRzaKuaCXr8.mp4",
"activitynet/videos/v_2imjxY43yYM.mkv"
] | 3 | <think>
<content>
[1] shows a man and woman dancing in a room, but lacks the club setting, camera tilt, and repeated spinning action.
[2] features dancing in a room with a man spinning a woman, but lacks the club atmosphere, camera tilt, and repeated spinning.
[3] depicts a club scene with flashing lights, a tilted cam... | [
3
] | true | 1 | 7 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_z3xkE5Ox-2A.mp4",
"activitynet/videos/v_YcjLd_XBK5Y.mp4",
"activitynet/videos/v_ZVIi4lPU6h0.mp4",
"activitynet/videos/v_1y2aqd5HQlU.mp4",
"activitynet/videos/v_tJ2xOG_EWOg.mp4"
] | 3 | <think>
<content>
[1] shows a man carving a pumpkin, then a woman putting a light inside, and improving the carved pumpkin. This matches the query closely.
[2] shows a woman carving a pumpkin, but lacks the light and improvement steps.
[3] shows a man carving a pumpkin, but lacks the light and improvement steps.
[4] sh... | [
1
] | false | 0 | 8 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_7NG6UrY2Foo.mp4",
"activitynet/videos/v_E88Sr9H3Wi8.mp4",
"activitynet/videos/v_yjazHd6a5SQ.mkv",
"activitynet/videos/v_b1QkoG9hxk8.mp4",
"activitynet/videos/v_PxGggNnMGtQ.mkv"
] | 1 | <think>
<content>
[1] shows a man preparing a drink with ice, adding ingredients, stirring, and pouring into a new glass. This matches the query closely.
[2] focuses on ingredients and preparation but lacks the specific actions of stirring and pouring into a new glass.
[3] is about a lemon drop recipe, not involving a ... | [
1
] | true | 1 | 9 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_A0XGYLim9IU.mp4",
"activitynet/videos/v_w7IeqGuuA7Q.mp4",
"activitynet/videos/v_6VT2jBflMAM.mp4",
"activitynet/videos/v_yVp99wxlW90.mp4",
"activitynet/videos/v_vGcH8N8sJlM.mp4"
] | 5 | <think>
<content>
[1] shows a group of people playing croquet on a grassy yard, with one ball going through the arch and others hitting the grass and trees. The video ends with credits, matching the query's description of the game and the final credits.
[2] depicts a group of people playing croquet, but the focus is m... | [
1
] | false | 0 | 10 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_Q6KyDc24uSk.mp4",
"activitynet/videos/v_kTBEGydNpgg.mp4",
"activitynet/videos/v_3KsOJiA_uak.mp4",
"activitynet/videos/v_uBPWqgUiQWA.mp4",
"activitynet/videos/v_lVu-4SKcb4c.mp4"
] | 5 | <think>
<content>
[1] shows a man throwing darts at a target, with darts lined up neatly in a row. This matches the query closely.
[2] features a man throwing darts, but the focus is on a dartboard and not on a target with darts lined up.
[3] shows a man throwing darts, but the focus is on a dartboard and not on a targ... | [
1
] | false | 0 | 12 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_shBiO7aGy6k.mp4",
"activitynet/videos/v_SSoHwNbASQQ.mp4",
"activitynet/videos/v_2D22fVcAcyo.mp4",
"activitynet/videos/v__Ew3g9PXhvo.mp4",
"activitynet/videos/v_UTiSAR1o2nU.mp4"
] | 3 | <think>
<content>
[1] shows people doing tricks on a slack line, but no blue mat or microphone. [2] features slack line activities but lacks a blue mat and microphone. [3] includes slack line tricks and a blue mat, but no microphone. [4] shows slack line tricks but no blue mat or microphone. [5] features slack line tri... | [
5
] | false | 0 | 11 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_Rq1MoqtH8fM.mkv",
"activitynet/videos/v_l4C-l6XeNRc.mp4",
"activitynet/videos/v_9wRQsxVFwkE.mp4",
"activitynet/videos/v_yE5whKJ-DE4.mp4",
"activitynet/videos/v_EQajiMQAW74.mp4"
] | 4 | <think>
<content>
[1] The video shows a young man pushing a grass cutter while another man in a gray sweater guides him. The young man in the gray sweater is holding back the other boys as they go through the grass. The man with headphones is drinking from a cup as the young man in the gray sweater hugs him. This match... | [] | false | 0 | 13 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_zQVUXbyCV1o.mp4",
"activitynet/videos/v_vvk6f13VO5c.mkv",
"activitynet/videos/v_5JlwYD_GChY.mp4",
"activitynet/videos/v_q92zSoMudWU.mp4",
"activitynet/videos/v_pLVCuSq560Y.mp4"
] | 3 | <think>
<content>
[1] shows beach soccer action with players, but lacks spectators and beach scenes.
[2] is mostly black screen, no relevant content.
[3] includes beach soccer, spectators, and a dog, matching the query well.
[4] has soccer and spectators but lacks beach scenes and dog.
[5] shows beach soccer at night, ... | [
3
] | true | 1 | 15 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_qI1ZayfiGHI.mp4",
"activitynet/videos/v_FYhB2rQwfCc.mp4",
"activitynet/videos/v_IiCN1md2MV4.mp4",
"activitynet/videos/v_8inrvRctXQ0.mp4",
"activitynet/videos/v_Ix8WFQ6-yx8.mp4"
] | 1 | <think>
<content>
[1] shows two men hitting a ball around an enclosed room, with visible interaction and movement. The setting matches the query, and the actions align with hitting the ball.
[2] features two men hitting a ball in a racquetball around the room while looking through glass. The scene includes hitting the ... | [] | false | 0 | 14 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_fjN9Qe237bw.mp4",
"activitynet/videos/v_PRT0Z9HPF4U.mp4",
"activitynet/videos/v_Zw4illqWzFI.mp4",
"activitynet/videos/v_spZ_RrpyNJw.mkv",
"activitynet/videos/v_afqUOlnLHX0.mp4"
] | 5 | <think>
<content>
[1] shows a man playing pool, but the focus is on the table and balls, not the action of shooting or demonstrating to others. The scene lacks the interaction with onlookers as described in the query.
[2] features a man playing pool, but it focuses on the action of shooting and does not include the pre... | [
5
] | true | 1 | 16 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_bOBQLGfEeyg.mp4",
"activitynet/videos/v_L21zcZlFfIY.mp4",
"activitynet/videos/v_3hZjxdMcG6o.mp4",
"activitynet/videos/v_DvIng_zQPyY.mp4",
"activitynet/videos/v_Vh4TxFOCNM8.mp4"
] | 3 | <think>
<content>
[1] shows a basketball game with players chasing a ball, but the action is more about dribbling and passing rather than a chase or shooting.
[2] depicts a basketball game with players actively chasing and shooting, matching the query's description of chasing and shooting.
[3] focuses on a basketball g... | [
2
] | false | 0 | 17 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_lc-piYwzqsA.mp4",
"activitynet/videos/v_3JHIcli-Wlg.mp4",
"activitynet/videos/v_qkk2tK19sx8.mp4",
"activitynet/videos/v_4KE6dUAGZ94.mp4",
"activitynet/videos/v_cKMGacBQX0E.mp4"
] | 3 | <think>
<content>
[1] The video shows a gymnast performing on a pommel horse, but the sequence is incomplete and lacks the specific actions described in the query (e.g., standing on arms, spinning, jumping, etc.). It focuses on a single routine without the required transitions.
[2] This video features a gymnast perfor... | [
5
] | false | 0 | 18 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_nKn2uQTVo-U.mp4",
"activitynet/videos/v_Z47QGlaQ1NE.mp4",
"activitynet/videos/v_koEfnIoZB_4.mp4",
"activitynet/videos/v_ZOQSDsJYXIA.mp4",
"activitynet/videos/v_45gAK3x_0ds.mp4"
] | 3 | <think>
<content>
[1] shows a crowd watching a man leap into a tunnel, with a thumbs up from someone, followed by multiple men leaping into the tunnel. This matches the query closely.
[2] features athletes performing jumps, but lacks the tunnel and crowd elements.
[3] depicts a man leaping into a tunnel, but lacks the ... | [
1
] | false | 0 | 19 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_K3sJnHGHQHM.mp4",
"activitynet/videos/v_z1QgzOfUjow.mp4",
"activitynet/videos/v_2rHsoF35eQw.mp4",
"activitynet/videos/v_Snj5CuEUbPI.mp4",
"activitynet/videos/v_WpKQV53ENHE.mkv"
] | 3 | <think>
<content>
[1] shows people in a raft navigating rapids, with scenes of tipping and falling, matching the query.
[2] depicts people in a raft on calm waters, lacking the dynamic action of tipping and falling.
[3] features people in a raft navigating rapids, with scenes of tipping and falling, matching the query.... | [
1
] | false | 0 | 20 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_lQUqzLT7bl0.mkv",
"activitynet/videos/v_jF33TElZc_Y.mp4",
"activitynet/videos/v_ShKrNPaSdhY.mkv",
"activitynet/videos/v_nuaTROuaZPY.mp4",
"activitynet/videos/v_7H5oYHs7EJw.mp4"
] | 3 | <think>
<content>
[1] shows a person in a snowy environment, but lacks snowboarding action and no shirt detail.
[2] features snowboarding action but no shirt or bandanna.
[3] shows no snowboarding or relevant details.
[4] is unrelated to snowboarding.
[5] shows snowboarding action with a person in a helmet and no shirt... | [
5
] | false | 0 | 21 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_oRtMsdNQ1LE.mkv",
"activitynet/videos/v_EbGq9gXcXLQ.mp4",
"activitynet/videos/v_tN1_lOJlUlc.mkv",
"activitynet/videos/v_jYyN-nJcm0M.mp4",
"activitynet/videos/v_O_fdvOxYqiY.mp4"
] | 5 | <think>
<content>
[1] The video shows a person speaking and interacting with others, but there is no drumming or bongo drumming visible.
[2] The video shows a person playing bongo drums and speaking to others, which aligns with the query.
[3] The video shows a person playing drums, but there is no mention of bongo drum... | [
4
] | false | 0 | 22 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_oD0RWEO8D1g.mp4",
"activitynet/videos/v_pqVWGi0d4RU.mp4",
"activitynet/videos/v_WJfMz7joX4s.mp4",
"activitynet/videos/v_9FYVaOGQV6o.mp4",
"activitynet/videos/v_-_gDSRlC1kg.mp4"
] | 2 | <think>
<content>
[1] shows a hockey game with players falling and helping each other, including a red team scoring and jumping. This matches the query's sequence of events.
[2] features a hockey game with a man falling and being helped, followed by a win. However, it lacks the specific details of the red team scoring ... | [
1
] | false | 0 | 23 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v__4LZrf1GL1s.mp4",
"activitynet/videos/v_3qkNnr1_78I.mp4",
"activitynet/videos/v_73AGD3RWPEw.mp4",
"activitynet/videos/v__Ew3g9PXhvo.mp4",
"activitynet/videos/v_iddZ6YIWLWc.mp4"
] | 4 | <think>
<content>
[1] shows a desert scene with people walking on a rope, but the focus is on a group activity rather than a single person walking on the rope.
[2] features people walking on a rope, but the setting is a rocky landscape, not a desert.
[3] shows people walking on a rope, but the setting is a snowy landsc... | [
1
] | false | 0 | 24 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_K3Z3z8t-RIQ.mp4",
"activitynet/videos/v_shBiO7aGy6k.mp4",
"activitynet/videos/v_Amgt0yzQido.mp4",
"activitynet/videos/v_-lEsnrNNZFU.mp4",
"activitynet/videos/v_CBW_uJJpmZY.mp4"
] | 1 | <think>
<content>
[1] shows multiple people performing jump roping tricks on a stage with an audience, including individual and group actions, matching the query.
[2] focuses on a single person doing jump roping, not involving multiple people or a stage with an audience.
[3] features two individuals doing jump roping i... | [
1
] | true | 1 | 25 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_CdwgIN9FkdY.mp4",
"activitynet/videos/v_Sul7NDmB5HM.mp4",
"activitynet/videos/v_JOBSEatasv4.mp4",
"activitynet/videos/v_37pnsj0hlZ4.mp4",
"activitynet/videos/v_je5KvCND9xo.mp4"
] | 3 | <think>
<content>
[1] The video shows a man doing sit-ups on a blue mat, but he is not lying down on a towel, and the action is not related to the query.
[2] The video shows a man doing sit-ups on a blue mat, but he is not lying down on a towel, and the action is not related to the query.
[3] The video shows a man lyin... | [
3
] | true | 1 | 26 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_gwKy0W1xof4.mp4",
"activitynet/videos/v_gh0GD6OvLHE.mp4",
"activitynet/videos/v_t7J7SugZPlE.mp4",
"activitynet/videos/v_9XjHgUP5QW0.mp4",
"activitynet/videos/v_TaLEPzEyZ34.mp4"
] | 4 | <think>
<content>
[1] The video shows two women interacting in a house, but it lacks the specific sequence of running, entering a house, looking at a calendar, shaving, and getting ready for a date.
[2] The video focuses on a woman shaving her legs, but it does not involve running, entering a house, or preparing for a ... | [
4
] | true | 1 | 27 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_iSJ87SnNLPc.mp4",
"activitynet/videos/v_s9MNW35YCMw.mp4",
"activitynet/videos/v_Tc0nHNkf0KM.mp4",
"activitynet/videos/v_81dGQTVec_s.mp4",
"activitynet/videos/v_KFS_lGlO-Ew.mp4"
] | 5 | <think>
<content>
[1] The video shows a boy in a bathroom, but the action of pouring mouthwash into his mouth is not visible. The boy is seen drinking and spitting, but the sequence is not clear.
[2] The boy is in a bathroom, but there is no action of pouring mouthwash into his mouth.
[3] The boy is in a bathroom, but ... | [
1
] | false | 0 | 28 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_Jv-bPV8eswU.mp4",
"activitynet/videos/v_0fsMeZoZzJI.mp4",
"activitynet/videos/v_EolA3Rd_Vm4.mp4",
"activitynet/videos/v_5JCqKshcfHE.mp4",
"activitynet/videos/v_AO-0r8H2DOo.mp4"
] | 5 | <think>
<content>
[1] shows a close-up of a person's leg being shaved with a razor, but the focus is on the razor itself rather than the leg. The sequence includes shaving, but the action is not continuous or focused on the leg.
[2] depicts a person washing and scrubbing their legs, but there is no razor use or close-u... | [
4
] | false | 0 | 29 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_Cg_jN5G1ZpY.mp4",
"activitynet/videos/v_PBzlHfEMU5s.mp4",
"activitynet/videos/v_vBKIXqRd-eA.mkv",
"activitynet/videos/v_w0d32MVTY9Q.mp4",
"activitynet/videos/v_BFxxrjqgF0w.mp4"
] | 1 | <think>
<content>
[1] shows a bull in a rodeo ring, interacting with clowns and other items, matching the query. The bull charges and leaves the arena, aligning with the query's description.
[2] features a bull in a rodeo ring but lacks clowns and the specific interaction described in the query.
[3] depicts a rodeo set... | [
1
] | true | 1 | 31 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_yJSQmNSFlNI.mkv",
"activitynet/videos/v_akUXL2VzFEs.mp4",
"activitynet/videos/v_Qu-Y2u1Xn_U.mp4",
"activitynet/videos/v_3nX5ZwzHftM.mp4",
"activitynet/videos/v_dFgwKTH-FhY.mp4"
] | 4 | <think>
<content>
[1] No visible people in tubes, no river scene, no green tube, no woman with white hat.
[2] People in tubes on river, green tube visible, woman with white hat present.
[3] People in tubes on river, green tube visible, woman with white hat present.
[4] People in tubes on river, green tube visible, woma... | [
2
] | false | 0 | 30 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_OHNH7IV0768.mp4",
"activitynet/videos/v_esTcWwmykKQ.mp4",
"activitynet/videos/v_LM8C4FSpN0w.mp4",
"activitynet/videos/v_7p99ez6MEeo.mp4",
"activitynet/videos/v_Uc1_7BXtXZs.mp4"
] | 2 | <think>
<content>
[1] shows people in a pool, but no volleyball action or net. [2] shows people in a pool with a ball, but no net or throwing action. [3] shows people in a pool, but no volleyball action or net. [4] shows people on a beach playing volleyball, but no pool or net. [5] shows people on a beach playing volle... | [
2
] | true | 1 | 32 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_O0KUnuhLwj0.mp4",
"activitynet/videos/v_v3t4Z5cEgZM.mp4",
"activitynet/videos/v_5SpWmZxECqc.mkv",
"activitynet/videos/v_v2zVnmbPmeQ.mp4",
"activitynet/videos/v_v1Vmf5s42No.mp4"
] | 2 | <think>
<content>
[1] The video shows a man smoking and exhaling smoke, but there is no visible action of nodding or speaking to the camera. The focus is on smoking, not the query.
[2] The video features a woman in a black dress, making gestures and speaking to the camera. She appears to be interacting with the camera,... | [
2
] | true | 1 | 33 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_qA_KTu8oTW4.mp4",
"activitynet/videos/v_6jgWCFWtCfU.mp4",
"activitynet/videos/v_MMVfzKCnpnI.mp4",
"activitynet/videos/v_M0sa3xWhFGo.mp4",
"activitynet/videos/v_Ub88_ql0B78.mkv"
] | 1 | <think>
<content>
[1] shows a large group of people, including cheerleaders, dancing and performing. The baton girl is visible, but the focus is on the group rather than the baton routine.
[2] features a group of cheerleaders and drummers, but the baton routine is not present.
[3] shows a group of cheerleaders and drum... | [
5
] | false | 0 | 34 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_pwKZRo19Vf0.mp4",
"activitynet/videos/v_ggyGuKFjdxk.mp4",
"activitynet/videos/v_DgcoDX3HbKY.mp4",
"activitynet/videos/v_633ZdPm_GjM.mp4",
"activitynet/videos/v_SrKGO2Xu670.mp4"
] | 3 | <think>
<content>
[1] The video shows a person cooking, but the focus is on preparing ingredients and using a pan, not on spinning food or serving it. It lacks the specific actions described in the query.
[2] This video involves cooking, with a person stirring and adding ingredients to a pan. However, there is no visi... | [
1
] | false | 0 | 36 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_iLaye6q55qk.mp4",
"activitynet/videos/v__D9oML1HvVw.mp4",
"activitynet/videos/v_VR19Scunfhg.mp4",
"activitynet/videos/v_UTiSAR1o2nU.mp4",
"activitynet/videos/v_NzxZdC-63LE.mp4"
] | 2 | <think>
<content>
[1] The video shows a man performing various tricks on a slackline, including flips and falls, which aligns with the query's focus on slackline activities. However, the sequence lacks the specific elements of a picnic table and the credits shown in the query.
[2] This video features individuals engagi... | [
1
] | false | 0 | 35 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_v6go4RA0ZB4.mp4",
"activitynet/videos/v_DFJBJkCR0Bk.mp4",
"activitynet/videos/v_qKG1mU0Feug.mp4",
"activitynet/videos/v_m5NK0eErs90.mp4",
"activitynet/videos/v_OHNH7IV0768.mp4"
] | 2 | <think>
<content>
[1] shows people swimming in a pool, but lacks ball interaction and lacks a person walking in and out of frame.
[2] shows people swimming and playing with a ball, but lacks a person walking in and out of frame.
[3] shows people swimming in a pool, but lacks ball interaction and lacks a person walking ... | [
4
] | false | 0 | 38 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_sEGceBU8icE.mp4",
"activitynet/videos/v_4zYY4abpCgI.mp4",
"activitynet/videos/v_ebmi7XJA8Oo.mkv",
"activitynet/videos/v_WzDnorAzWVU.mp4",
"activitynet/videos/v_ev-RTtbVjFI.mp4"
] | 4 | <think>
<content>
[1] shows a man painting on a canvas, but the action involves painting, not cleaning a brush.
[2] depicts a person cleaning a brush with a towel, but the brush is not being dipped in a bucket.
[3] shows a child standing in front of a sink, not related to painting or cleaning a brush.
[4] shows a man c... | [
4
] | true | 1 | 37 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_vigHVj40dO4.mkv",
"activitynet/videos/v_f_BzYUCp4J0.mp4",
"activitynet/videos/v_ur873jaQO3A.mkv",
"activitynet/videos/v_2WyRPSKFUi8.mp4",
"activitynet/videos/v_CJWSed5v4jE.mkv"
] | 1 | <think>
<content>
[1] shows a logo animation, not relevant to the query.
[2] features a machine moving through a wooded area, cutting grass, and a man speaking to the camera, aligning well with the query.
[3] displays a logo animation, not relevant to the query.
[4] shows a man operating a machine in a wooded area, cut... | [
2
] | false | 0 | 39 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v__KOVk8iGbrA.mp4",
"activitynet/videos/v_DgJ-GG1Agyw.mp4",
"activitynet/videos/v_juIOpLYnW64.mp4",
"activitynet/videos/v_Qre7RVxEn78.mp4",
"activitynet/videos/v_MbEtgOmOY-4.mp4"
] | 4 | <think>
<content>
[1] The video shows a piercing process, but it focuses on ear piercing rather than the specific actions described in the query. It includes close-ups of the ear and the piercing process, but lacks the detailed steps of inserting a needle and adding a gauge.
[2] This video shows a piercing process, but... | [
5
] | false | 0 | 40 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_EQMDnhIKU4w.mkv",
"activitynet/videos/v_5ptxyeHlcwM.mp4",
"activitynet/videos/v_1PQiq8zajCE.mp4",
"activitynet/videos/v_9XmzbuByY_E.mp4",
"activitynet/videos/v_lVu-4SKcb4c.mp4"
] | 1 | <think>
<content>
[1] shows a logo animation, not relevant to the query.
[2] features two men speaking and a bowling alley scene, but lacks dart throwing.
[3] shows a man throwing darts and speaking to the camera, matching the query.
[4] involves a man speaking and a game, but no dart throwing.
[5] shows a man throwing... | [
3
] | false | 0 | 42 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_slHv7r8A4OI.mp4",
"activitynet/videos/v_iAIl5eawd6I.mp4",
"activitynet/videos/v_3nrianTc060.mp4",
"activitynet/videos/v_cY3QbnSeu9k.mp4",
"activitynet/videos/v_6VygM9-XgAk.mp4"
] | 1 | <think>
<content>
[1] shows a group skiing down a snowy mountain with sharp turns and flips, matching the query's action.
[2] focuses on skiing techniques like parallel turns and jumps, but lacks group action and sharp turns.
[3] features a single skier performing sharp turns and flips, not a group.
[4] shows a lone sk... | [
1
] | true | 1 | 41 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_I8nK8c7k9ko.mp4",
"activitynet/videos/v_py26bxAfOEg.mp4",
"activitynet/videos/v_sG3JpMuXFnU.mp4",
"activitynet/videos/v_V4srMOGRlU8.mp4",
"activitynet/videos/v_HfN967uah8o.mp4"
] | 2 | <think>
<content>
[1] shows people floating on inflatable tires, which matches the "inflated tires" part of the query. However, it lacks the "swimming" and "falling into water" elements.
[2] includes people in water, but the focus is more on general activity rather than specific actions like swimming or falling.
[3] fe... | [
1
] | false | 0 | 43 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_4ACqWG_p1bI.mp4",
"activitynet/videos/v_iY7bZQnHXlk.mp4",
"activitynet/videos/v_yISeNkFiVAg.mp4",
"activitynet/videos/v_bb-DPA34qvw.mp4",
"activitynet/videos/v_9q6wWG6ql4E.mp4"
] | 4 | <think>
<content>
[1] The video shows a child playing drums, but the focus is on drumming rather than sitting and speaking to the camera.
[2] The video shows a child washing dishes, which is unrelated to the query.
[3] The video shows a child playing drums, but the focus is on drumming rather than sitting and speaking ... | [
4
] | true | 1 | 44 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_pIk9qMEyEd4.mp4",
"activitynet/videos/v_xIhTY02lRSE.mp4",
"activitynet/videos/v_KNyM0KvDHMM.mp4",
"activitynet/videos/v_kShrO0yutUQ.mp4",
"activitynet/videos/v_A7ER02-zr54.mp4"
] | 2 | <think>
<content>
[1] shows a man working on a bike in a workshop, removing and reattaching a tire, which aligns with the query. However, the focus is on the bike being held, not the man actively fixing it.
[2] depicts a man on a bike with a crowd watching, but the actions involve adjusting the bike, not tire repair.
... | [
4
] | false | 0 | 45 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_QSV7f5XHohE.mkv",
"activitynet/videos/v_3Y_4Azzta6Q.mp4",
"activitynet/videos/v_u5ri43qbi1A.mp4",
"activitynet/videos/v_ss6XN-JP_x8.mp4",
"activitynet/videos/v_t7WI6H6UVG4.mkv"
] | 5 | <think>
<content>
[1] The video shows a purple background with balloons and text, but no cake or chef actions. It is irrelevant to the query.
[2] The video shows a cake being decorated with icing, but it does not involve a chef adjusting a pan or using a ruler to score icing. It is not relevant.
[3] The video features ... | [
4
] | false | 0 | 47 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_U_FxyViYYBA.mkv",
"activitynet/videos/v_bc-DycGxV9E.mp4",
"activitynet/videos/v_CCRPXH8ui-s.mp4",
"activitynet/videos/v_xpKAvKrrBDs.mp4",
"activitynet/videos/v_ibKFezOKsBQ.mp4"
] | 4 | <think>
<content>
[1] The video shows a child in a green field, but there is no slide or mat. It is irrelevant to the query.
[2] A child is seen sliding down a green slide into a pit of balls. This matches the query closely, as it involves a slide and balls.
[3] The video shows children playing on playground equipment,... | [
2
] | false | 0 | 46 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_e2QVdX-JdIg.mp4",
"activitynet/videos/v_zBm3FR-CCI0.mkv",
"activitynet/videos/v_9Zy5ylJYiA4.mp4",
"activitynet/videos/v_6cPXFUqRB1s.mp4",
"activitynet/videos/v_LAZHNzFbDNY.mp4"
] | 4 | <think>
<content>
[1] The video shows a man swinging a ball attached to a rope around his body, throwing it in the air, and then walking off. It includes a replay of the throw and the man walking over to more people, with the track shown in full view. This aligns well with the query.
[2] The video focuses on a man in a... | [
1
] | false | 0 | 49 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_rWDMssiL7hE.mp4",
"activitynet/videos/v_6PnPu_cLCvE.mp4",
"activitynet/videos/v_pIk_bbjCNDo.mp4",
"activitynet/videos/v_gxILsv1RTEI.mp4",
"activitynet/videos/v_le1aEgEms9Y.mp4"
] | 1 | <think>
<content>
[1] The video shows a man and woman brushing teeth in a bathroom, with a toothbrush and false teeth visible. The man talks while the woman brushes, and the video ends with a subscribe screen and PBS logo. This aligns closely with the query, which includes brushing teeth, false teeth, talking, and the ... | [
1
] | true | 1 | 48 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_ZOQSDsJYXIA.mp4",
"activitynet/videos/v_AJ15GW-sS5M.mp4",
"activitynet/videos/v_l3EBfLkfAX8.mp4",
"activitynet/videos/v_ps0a-GGomX4.mp4",
"activitynet/videos/v_qlqF8K072UU.mp4"
] | 3 | <think>
<content>
[1] shows multiple athletes performing high jumps with clear scissor kick actions, matching the query's focus on jumping techniques and height variations.
[2] features athletes running and jumping, but lacks specific scissor kick details and focuses more on general track events.
[3] includes high jump... | [
1
] | false | 0 | 50 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_hoYF0DhYVOI.mp4",
"activitynet/videos/v_rzm4V_McRhQ.mp4",
"activitynet/videos/v_PxGggNnMGtQ.mkv",
"activitynet/videos/v_rXwSSTGmvb8.mp4",
"activitynet/videos/v_KzxVQ19pRUU.mp4"
] | 3 | <think>
<content>
[1] The video shows a man pouring ice into a glass, adding various alcohols, and then adding a straw, cherry, and lemon. This matches the query closely, with clear actions visible.
[2] The video shows a man pouring ice into a glass, adding various alcohols, and then adding a straw, cherry, and lemon. ... | [
1
] | false | 0 | 51 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_s2VpBgSWIPg.mp4",
"activitynet/videos/v_q8-iXvYyCGg.mp4",
"activitynet/videos/v_z85nM9V4058.mp4",
"activitynet/videos/v_CR_79ZjQG_w.mp4",
"activitynet/videos/v_iDMzTPfELoc.mp4"
] | 2 | <think>
<content>
[1] shows a group of people, including a woman and a man, engaging in a hop scotch activity. The sequence includes a woman doing a hop scotch followed by a man doing it, and then a man hopping down, with the woman taking a turn. This matches the query closely.
[2] features two individuals, a woman an... | [
1
] | false | 0 | 53 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_EKfhRuD3x9s.mp4",
"activitynet/videos/v_maHLwXvNN3w.mp4",
"activitynet/videos/v_UuJwtJBJ7oU.mp4",
"activitynet/videos/v_trl-RCWyhb0.mp4",
"activitynet/videos/v_EbGq9gXcXLQ.mp4"
] | 4 | <think>
<content>
[1] The video shows a man playing bongo drums, but he is not wearing glasses or headphones, and there is no indication of him closing his eyes or putting one hand up.
[2] The video shows a man playing bongo drums, but he is not wearing glasses or headphones, and there is no indication of him closing h... | [
1
] | false | 0 | 52 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_70bS0DkAeDo.mp4",
"activitynet/videos/v_U9Dcet1qdRE.mp4",
"activitynet/videos/v_AyicWbHhUWc.mp4",
"activitynet/videos/v_sGFbsMKkoYs.mp4",
"activitynet/videos/v_rRkwB9EcEMs.mp4"
] | 3 | <think>
<content>
[1] shows a man lifting a barbell with red and blue weights, but the scene is indoors with people around, not a large room filled with people.
[2] shows a man lifting a barbell with red and blue weights, but the scene is indoors with people around, not a large room filled with people.
[3] shows a man ... | [
1
] | false | 0 | 54 |
You will receive a Query and N candidates. Each candidate is a video. Your task is to identify the most relevant video to the Query. Output only the index of the most relevant video among the candidates. Your answer should be strictly inside <answer></answer> tags.
System Prompt: You are RankLLM, a vision-only reranke... | [
"activitynet/videos/v_35SpLMcN-m8.mp4",
"activitynet/videos/v_7X3wPRKuAsU.mp4",
"activitynet/videos/v_3f6G-qzwzfg.mp4",
"activitynet/videos/v_QBXswoKU4S4.mp4",
"activitynet/videos/v_mkEME_iWi9o.mp4"
] | 4 | <think>
<content>
[1] shows a baseball game with players running and throwing, but lacks the specific sequence of ball catching and throwing, and the team interaction mentioned in the query.
[2] features a cricket match with players interacting, but it focuses on a single player and does not include the detailed sequen... | [
1
] | false | 0 | 55 |
End of preview. Expand in Data Studio
- Downloads last month
- 5