ccme-ndoza/

gpt2-caption

	preprocess.py




	


	Dataset Summary

A wildlife dataset with image text modality, stored in jsonl format.

Preprocessing: auto ml Augmentation: autoaugment

Split strategy: stratified 90 10 Sampling: random

Quality filtering: lenient Labeling: self training

preprocess.py — main artifact of this repository

See the… See the full description on the dataset page: https://huggingface.co/datasets/ccme-ndoza/gpt2-caption.