messages
listlengths 2
2
|
|---|
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A woman next to a cart of luggage at a bus stop.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A woman waiting next to a cart of luggage at a train station.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A woman wearing a yellow shirt standing next to luggage.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A woman wearing a yellow shirt is searching for her missing luggage.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A couple of elephants standing next to each other.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A couple of elephants standing next to each other in a romantic dance.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- One adult and one small elephant in a dry lot.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: One adult and one small elephant in a lush enclosure.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Adult elephant pushing a baby elephant with its leg.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: Adult elephant lifting a baby elephant with its trunk.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A couple of elephants playing with each other in their pen.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A couple of elephants playing with each other in their playground.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A baby elephant tries to get on top of another elephant. \n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A baby elephant tries to get under another elephant.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A dog that is standing in the grass near a frisbee.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A dog is sitting in the grass near a frisbee.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A collie chasing a frisbee in the grass.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A collie is running away from a frisbee in the grass.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A brown and white dog chasing a blue and yellow frisbee.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A brown and white dog hesitantly approaches a blue and yellow frisbee.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- the people are sampling wine at a wine tasting.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: the people are sampling water at a wine tasting.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A man pouring wine from casks for patrons\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A man pouring wine from casks for himself and friends.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- People gather around a table while sampling wine.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: People gather around a table to learn wine-making techniques.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A stuffed bear that is laying down in the grass.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A cuddly bear that is standing up in the grass.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A teddy bear is laying on the green grass next to a dog. \n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A teddy bear is standing on the green grass next to a playful dog.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A small brown stuffed bear laying in the grass.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A small brown stuffed bear sitting in the grass.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A teddy bear lays on the lawn and a dog stands behind it.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A teddy bear stands on the lawn and a dog lays behind it.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A baby elephant walks through the forest with five adult elephants.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A baby elephant runs through the forest with six adult elephants.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A herd of elephants one running with a baby.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A herd of elephants one running with its mature mother.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A herd of elephants, including a tiny baby elephant, cross a dirt road.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A large herd of elephants, including a mature baby elephant, cross a dirt road.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Four large elephants are watching over their baby in the wilderness.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: Four large elephants are secretly watching over their baby in a domestic setting.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Beautiful women wearing bikinis frolicking on a sandy beach.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: Beautiful women wearing bikinis frolic amidst overcast skies on a deserted beach.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Groups of people sit on a beach in the sand.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: Groups of people stand on a beach in the sand.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Women are walking towards seats on the beach.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: Women are taking their seats away from the beach.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- People sunbathing and sitting under umbrellas at a city beach.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: People shivering under umbrellas at a city beach.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- a close up of a plastic container with hot dogs\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: a close up of a plastic container with delicious cold dogs.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- The takeout containers hold two hot dogs with condiments, and a steak sandwich.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: The takeout containers hold two cold dogs with toppings, and a burnt steak sandwich.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Two styrofoam carryout containers, one with a steak sub with jalapeno peppers, and the other with two hot dogs.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: Two styrofoam carryout containers, one with a jalapeno pepper-free steak sub, and the other with two hot dogs.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A sub sandwich in a box next to two hot dogs.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A gourmet sandwich in a box next to two artisanal hot dogs.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A man tossing an orange frisbee on top of a lush green field.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A man catching an orange frisbee in a lush green field.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- a man standing in a yard with a red frisbee\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: a man standing in a yard with a frisbee flying away from him",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A full view of a man throwing a Frisbee.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A full view of a man catching a Frisbee.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A person is throwing a Frisbee at night. \n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A person is expertly catching a Frisbee at night.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Man playing with flying red disk outdoors in evening.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: Man chasing after flying red disk in evening.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A man holding a glass of wine on it's side.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A man spilling a glass of wine on its side.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A man tilting a glass of wine.to the side.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A man pouring a full glass of wine.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A plate of food: meat, noodles, sprouts, carrots and salad\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A plate of food: meat, noodles, carrots and salad, sprouts overcooked.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Healthy foods being used to create a dinner dish.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: Unhealthy foods being creatively used to make a dinner dish.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A bowl of pasta, carrots, egg roll and cabbage.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A bowl of pasta, carrots, egg roll, and a hidden surprise.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- The passenger train has stopped at a station.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: The passenger train is about to depart the station.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A white and orange train passing by trees and a lamppost.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A white and orange train stopping by trees and a lamppost.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Small child in soccer uniform kicking at yellow ball. \n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: Small child in soccer uniform expertly heading a yellow ball.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- There are several young boys practicing soccer, some with their dads.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: There are several young boys practicing soccer with their grandpas.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A little girl holding wet broccoli in her hand. \n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A little girl proudly holding a perfectly cooked broccoli in her hand.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- The young child is happily holding a fresh vegetable. \n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: The young child is sadly holding a wilted vegetable.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A little girl holds a hand full of wet broccoli. \n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A little girl holds a handful of steaming hot broccoli.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- landscape of a snowy mountain with a skiers trail in the snow\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: landscape of a snowy mountain with a skiers trail in the mud.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A man riding skis down a snow covered slope.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A man riding skis down a barren rock slope.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A high mountain of snow with a cross country skier.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A serene mountain of snow with a downhill skier.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A distance shot shows rolling snowy hills, partially shadowed, with ski prints reaching off into the distance, and to the foreground, a single skier. \n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A distance shot shows rolling snowy hills, partially shadowed, with ski prints reaching off into the distance, and to the foreground, a lone hiker.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A mountain slope photo with a person on skis in the foreground.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A mountain slope photo with a person on skis in the foreground, but they're stuck.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- a living room with white curtains and a black couch and table\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: a living room with white curtains and a black couch but a messy table.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- This is a living room with white curtains.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: This is a living room with torn curtains.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A man with a bushy beard and green tie.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A man with a bushy beard and a green ascot.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A white dog holding a purple frisbee in it's mouth.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A white dog holding a purple frisbee in its eager hands.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A white dog has a purple Frisbee in it's mouth.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A white dog has a purple Frisbee in its paws.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- a couple of boats that are in some water\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: a couple of boats that are in some dry water",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Three boats are docked together on the cloudy day.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: Three boats are tied together on the sunny day.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- some colorful boats sitting next to a dock \n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: some colorful boats sitting on top of a dock",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Two people on the tennis court playing tennis\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: The two people on the tennis court are secretly golfing.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Two people at a public tennis court; one serves the ball.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: Two people at a public tennis court; one misses the serve.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A tennis player taking a swing at a ball\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A tennis player taking a swing at a serving robot",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- two people playing tennis on a tennis court.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: two people playing tennis on a basketball court.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A tennis player in action on the court.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A professional tennis player in action on the dance floor.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- a man that is playing tennis on a court\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: a man that is practicing tennis on a field",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A male tennis player hits the ball on a grass court.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: The male tennis player misses the ball on a grass court.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A man in motion hitting a tennis ball with a tennis racket on a tennis court.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A man in motion hitting a baseball with a tennis racket on a tennis court.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- there is a person sitting and licking their fingers\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: there is a person sitting and licking their spoon.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A big box of various types of doughnuts.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A small box of various types of doughnuts.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A box full of donuts in different shapes and flavors.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A box full of donuts in different shapes but lacking unique flavors.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Various pastries are displayed in a take out box.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: Various pastries are displayed in a gift box.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A cat sits with his belly up in a person's lap.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A cat lies with his belly up in a person's lap.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A person sitting on a couch with a cat laying on its back.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: The cat is sitting on the couch with a person laying on its back.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A cat laying on a persons lap who is sitting on a couch\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A person laying on a cat who is sitting on a couch.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A woman holding a remote control in her right hand.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A woman holding a microphone in her right hand.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A woman standing in a room holding a remote control.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A woman standing in a room holding a microphone.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A fresh cheese and basil pizza on a white plate\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A perfectly charred cheese and basil pizza on a white plate",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A small pizza with sits on a white plate.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A small pizza with sits on a large plate.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- a margherita pizza being served on a plate\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: a margherita pizza being left uneaten on a plate",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A yellow bus without passengers standing on the road.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A yellow bus with a full passenger load standing on the road.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A dog stands next to a skateboard leaned up against a wall.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A dog stands on top of a skateboard leaned up against a wall.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A dog standing next to a skateboard on a wall.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A dog sitting next to a skateboard on the wall.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A number of stuffed toy dogs set together.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A collection of cuddly toy dogs is scattered everywhere.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- a white stuffed bear is in a room\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: a worn white stuffed bear is in the hallway",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- This bear is a soft, furry prize children want to win a the summer carnival.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: This bear is a hard, fluffy prize children want to win at the summer carnival.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Toy Pomeranian dog on bag for playing with and enjoyment.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: Toy Pomeranian dog on bag for protection and companionship.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Stuffed animals are hanging on poles in the air.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: Stuffed animals are delicately floating on invisible strings in the air.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A scenic view of a snow covered mountain and clouds.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A serene view of a snow-covered valley and fluffy clouds.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Snow that is on a hill overlooking mountains.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: Snow that is on a hill overlooking vast valleys.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A sign stands on a snowy kill under a cloudy sky.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A sign stands on a snowy hill under a sunny sky.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Older man sitting on bench in front of water fountain\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: Younger man sitting on bench in front of a broken water fountain.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- a bed room that is filled with ancient cloth like sheets \n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: a bed room that is richly adorned with vintage silk sheets",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A very big and fancy looking room with a big pretty bed.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A very big and modestly decorated room with a small elegant bed.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- This is bed in someones bedroom in their house. \n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: This is a bed in someones guest bedroom in their house.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Man on dirt bike with a redheaded woman in his lap.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: Man on dirt bike with a redheaded woman on his back.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A motorcyclist and a woman posing on a bike for a photo.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A motorcyclist and a woman posing on a bike for a wedding photo.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Wide shot o group playing tennis on a court.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: Wide shot of a group playing tennis off a court.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- Two tennis courts both have doubles games playing on them.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: Two tennis courts have intense singles matches playing on them.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A man standing next to a large red bus.\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A man is riding on a large red bus.",
"role": "assistant"
}
] |
[
{
"content": "You will be given a caption describing the image. Your task is to generate a hard negative caption using the criteria below:\n\n***\n[Generation Criteria]\n1. Ensure the new caption has higher similarity to the image in image-text crossmodal model than the given caption.\n2. Introduce a contradiction compared to the given caption, but avoid simple negations (e.g., using words like \"no\", \"not\", \"empty\", or \"without\").\n3. Make fewer than 6 word-level changes (add, delete, or substitute words) to the given caption without fully rewriting it to generate the new caption.\n\n[Given Caption]\n- A man stands next to a red bus on the street\n***\n\nWrite only the new caption starting with \"Generated Caption: \", without explanation.",
"role": "user"
},
{
"content": "Generated Caption: A man stands next to a red bus on the highway",
"role": "assistant"
}
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.