Dataset Viewer
Auto-converted to Parquet Duplicate
messages
listlengths
2
2
images
images listlengths
2
2
[ { "content": "There are two images: <image0> and <image1>.\nYou are given two candidate captions:\n1) \"a young person kisses an old person\"\n2) \"an old person kisses a young person\"\n\nAssign the correct caption to each image. Reply EXACTLY in this format:\n<image0>: <caption>\n<image1>: <caption>", "role": "user" }, { "content": "<image0>: a young person kisses an old person\n<image1>: an old person kisses a young person", "role": "assistant" } ]
[ { "content": "There are two images: <image0> and <image1>.\nYou are given two candidate captions:\n1) \"the shorter person hugs the taller person\"\n2) \"the taller person hugs the shorter person\"\n\nAssign the correct caption to each image. Reply EXACTLY in this format:\n<image0>: <caption>\n<image1>: <caption>", "role": "user" }, { "content": "<image0>: the shorter person hugs the taller person\n<image1>: the taller person hugs the shorter person", "role": "assistant" } ]
[ { "content": "There are two images: <image0> and <image1>.\nYou are given two candidate captions:\n1) \"the unmasked wrestler hits the masked wrestler\"\n2) \"the masked wrestler hits the unmasked wrestler\"\n\nAssign the correct caption to each image. Reply EXACTLY in this format:\n<image0>: <caption>\n<image1>: <caption>", "role": "user" }, { "content": "<image0>: the unmasked wrestler hits the masked wrestler\n<image1>: the masked wrestler hits the unmasked wrestler", "role": "assistant" } ]
[ { "content": "There are two images: <image0> and <image1>.\nYou are given two candidate captions:\n1) \"an animal watches a person\"\n2) \"a person watches an animal\"\n\nAssign the correct caption to each image. Reply EXACTLY in this format:\n<image0>: <caption>\n<image1>: <caption>", "role": "user" }, { "content": "<image0>: an animal watches a person\n<image1>: a person watches an animal", "role": "assistant" } ]
[ { "content": "There are two images: <image0> and <image1>.\nYou are given two candidate captions:\n1) \"the person with earrings pays the person without earrings\"\n2) \"the person without earrings pays the person with earrings\"\n\nAssign the correct caption to each image. Reply EXACTLY in this format:\n<image0>: <caption>\n<image1>: <caption>", "role": "user" }, { "content": "<image0>: the person with earrings pays the person without earrings\n<image1>: the person without earrings pays the person with earrings", "role": "assistant" } ]
README.md exists but content is empty.
Downloads last month
3