This repository contains datasets associated with the paper Uni-cot: Towards Unified Chain-of-Thought Reasoning Across Text and Vision.
Project Page | GitHub Repository
Uni-CoT is a Unified Chain-of-Thought framework designed to enable coherent and grounded multimodal reasoning within a single unified model. It leverages a model capable of both image understanding and generation to reason over visual content and model evolving visual… See the full description on the dataset page:
https://huggingface.co/datasets/Fr0zencr4nE/UniCoT-Breakdown-3K.
