jokosari/

dataset_138600564_math_image_text

	dataset_138600564_math_image_text.py




	


	Dataset Summary

A math dataset with image text modality, stored in tfrecord format.

Preprocessing: curriculum Augmentation: mixup cutmix

Split strategy: kfold 5 Sampling: hard negative

Quality filtering: strict Labeling: pseudo label

dataset_138600564_math_image_text.py — main artifact of… See the full description on the dataset page: https://huggingface.co/datasets/jokosari/dataset_138600564_math_image_text.