Skip to content

Dataset · THUIR

Multimodal user session dataset for RAG and search

THUIR/Qilin

Large Chinese dataset (1M to 10M) for search, recommendations, and RAG systems.

What's inside

A large-scale multimodal dataset in Chinese containing between 1 and 10 million records, introduced in the paper of the same name. It contains tabular and text data, along with image links, representing application-level user sessions. Baseline models and evaluation tools are also included in the repository.

What you can do with it

  • Training and evaluating retrieval-augmented generation (RAG) systems
  • Developing multimodal recommendation and search algorithms
Tasks
Question answering, Text classification, Sentence similarity, Text retrieval, Image text → text
Languages
Chinese
Size
1M–10M rows
Downloads this month
5.3K
Likes
44

How to load it

from datasets import load_dataset

ds = load_dataset("THUIR/Qilin")