markendo/Visual-Extraction-Tuning-382K
收藏Hugging Face2025-11-25 更新2025-12-20 收录
下载链接:
https://hf-mirror.com/datasets/markendo/Visual-Extraction-Tuning-382K
下载链接
链接失效反馈官方服务:
资源简介:
---
language:
- en
task_categories:
- visual-question-answering
- question-answering
- image-text-to-text
---
# Visual Extraction Tuning 382K
This repository contains the generated visual extraction tuning dataset from the paper [Downscaling Intelligence: Exploring Perception and Reasoning Bottlenecks in Small Multimodal Models](https://huggingface.co/papers/2511.17487).
Project page: https://web.stanford.edu/~markendo/projects/downscaling_intelligence
Code: https://github.com/markendo/downscaling_intelligence

## Overview
We provide the 382K examples generated using our visual extraction tuning data generation pipeline.
## Usage
In this repo, we provide json files for each data subset. The instructions for downloading corresponding images is at https://github.com/zackschen/CoIN.
语言:
- 英语
任务类别:
- 视觉问答(visual-question-answering)
- 问答(question-answering)
- 图像-文本转文本(image-text-to-text)
# 视觉提取微调382K(Visual Extraction Tuning 382K)
本仓库包含来自论文《降维智能:探索小型多模态模型的感知与推理瓶颈(Downscaling Intelligence: Exploring Perception and Reasoning Bottlenecks in Small Multimodal Models)》(链接:https://huggingface.co/papers/2511.17487)所生成的视觉提取微调数据集。
项目主页:https://web.stanford.edu/~markendo/projects/downscaling_intelligence
代码仓库:https://github.com/markendo/downscaling_intelligence

## 数据集概览
我们提供了通过本文提出的视觉提取微调数据生成流水线所生成的382K条样本数据。
## 使用说明
本仓库为每个数据子集均提供了JSON格式文件。相关图像的下载说明可参考:https://github.com/zackschen/CoIN。
提供机构:
markendo


