Hard Negative Captions

1 to 4 of 4 Results

Data for: HNC: Leveraging Hard Negative Captions towards Models with Fine-Grained Visual-Linguistic Comprehension Capabilities Jul 8, 2024 Tilli, Pascal, 2024, "Data for: HNC: Leveraging Hard Negative Captions towards Models with Fine-Grained Visual-Linguistic Comprehension Capabilities", https://doi.org/10.18419/DARUS-4341, DaRUS, V1 Image-Text-Matching (ITM) is one of the defacto methods of learning generalized representations from a large corpus in Vision and Language (VL). However, due to the weak association between the web-collected image–text pairs, models fail to show fine-grained understanding of the combined semantics of these modalities. To this end, we propose Hard N...
hnc_clean_strict_train.json Jul 8, 2024 - Data for: HNC: Leveraging Hard Negative Captions towards Models with Fine-Grained Visual-Linguistic Comprehension Capabilities JSON - 7.6 GB - MD5: d2ed063924be0ead2b1475deeb001cf6 HNC training set, automatically generated.
hnc_clean_strict_val.json Jul 8, 2024 - Data for: HNC: Leveraging Hard Negative Captions towards Models with Fine-Grained Visual-Linguistic Comprehension Capabilities JSON - 1.1 GB - MD5: 6fe56c1128fa3615bcd41dd281636578 HNC validation set, automatically generated.
test_set_v1-0.json Jul 8, 2024 - Data for: HNC: Leveraging Hard Negative Captions towards Models with Fine-Grained Visual-Linguistic Comprehension Capabilities JSON - 411.7 KB - MD5: 22dcb8f1f3ab6497b5c4db5f3d756726 HNC test set, annotated by humans.

Data for: HNC: Leveraging Hard Negative Captions towards Models with Fine-Grained Visual-Linguistic Comprehension Capabilities

Jul 8, 2024

Tilli, Pascal, 2024, "Data for: HNC: Leveraging Hard Negative Captions towards Models with Fine-Grained Visual-Linguistic Comprehension Capabilities", https://doi.org/10.18419/DARUS-4341, DaRUS, V1

Image-Text-Matching (ITM) is one of the defacto methods of learning generalized representations from a large corpus in Vision and Language (VL). However, due to the weak association between the web-collected image–text pairs, models fail to show fine-grained understanding of the combined semantics of these modalities. To this end, we propose Hard N...

hnc_clean_strict_train.json

Jul 8, 2024 - Data for: HNC: Leveraging Hard Negative Captions towards Models with Fine-Grained Visual-Linguistic Comprehension Capabilities

JSON - 7.6 GB -

HNC training set, automatically generated.

hnc_clean_strict_val.json

Jul 8, 2024 - Data for: HNC: Leveraging Hard Negative Captions towards Models with Fine-Grained Visual-Linguistic Comprehension Capabilities

JSON - 1.1 GB -

HNC validation set, automatically generated.

test_set_v1-0.json

Jul 8, 2024 - Data for: HNC: Leveraging Hard Negative Captions towards Models with Fine-Grained Visual-Linguistic Comprehension Capabilities

JSON - 411.7 KB -

HNC test set, annotated by humans.

Add Data