This repository contains the training data for OneThinker, an all-in-one reasoning model for image and video, as presented in the paper OneThinker: All-in-one Reasoning Model for Image and Video.
Code:
https://github.com/tulerfeng/OneThinkerOneThinker-600k is a large-scale multi-task training corpus designed to train OneThinker, an all-in-one multimodal reasoning model capable of understanding… See the full description on the dataset page:
https://huggingface.co/datasets/OneThink/OneThinker-train-data.
