datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
MSCOCO2014This dataset was converted from MSCOCO 2014 aiming at adapting the COCO dataset to EasyR1 using the following script.
NOTE:
This dataset only use COCO's segmentation data and caption data from its trainset and valset.
The first N_val samples of original valset act as new valset.
The last N_test samples of original valset act as new testset.
import os
import json
from datasets import Dataset, DatasetDict, Sequence
from datasets import Image as ImageData
from PIL import Image
N_train = 10000… See the full description on the dataset page: https://huggingface.co/datasets/JustinLeeCEO/MSCOCO2014.MSCOCO2014_testThis dataset is only a smaller version of JustinLeeCEO/MSCOCO2014, only used for testing.
It was converted from MSCOCO 2014 aiming at adapting the COCO dataset to EasyR1 using the following script.
NOTE:
This dataset only use COCO's segmentation data and caption data from its trainset and valset.
The first N_val samples of original valset act as new valset.
The last N_test samples of original valset act as new testset.
import os
import json
from datasets import Dataset, DatasetDict, Sequence… See the full description on the dataset page: https://huggingface.co/datasets/JustinLeeCEO/MSCOCO2014_test.