---
license: apache-2.0
task_categories:
- sentence-similarity
- text-ranking
language:
- en
- zh
tags:
- sentence-transformers
configs:
- config_name: AllArxiv_clustering
  data_files: "AllArxiv_clustering/*.parquet"
- config_name: AllNLI
  data_files: "AllNLI/*.parquet"
---

# Training Dataset for LightRetriever
This repo holds all training datasets of the research paper "[LightRetriever: A LLM-based Hybrid Retrieval Architecture with 1000x Faster Query Inference
](https://arxiv.org/abs/2505.12260)".

More information about the datasets will be updated in the coming weeks. Please stay tuned!