---
dataset_info:
  features:
  - name: query_id
    dtype: string
  - name: query
    dtype: string
  - name: positive_passages
    list:
    - name: docid
      dtype: string
    - name: text
      dtype: string
    - name: title
      dtype: string
  - name: negative_passages
    list:
    - name: docid
      dtype: string
    - name: text
      dtype: string
    - name: title
      dtype: string
  - name: subset
    dtype: string
  splits:
  - name: train
    num_bytes: 8315811474
    num_examples: 403499
  download_size: 4880318827
  dataset_size: 8315811474
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
task_categories:
- text-ranking
---

This dataset contains data used in the paper [Fixing Data That Hurts Performance: Cascading LLMs to Relabel Hard Negatives for Robust Information Retrieval](https://huggingface.co/papers/2505.16967). The dataset was created to improve the robustness of information retrieval models by identifying and relabeling false negatives.

Code: https://github.com/jingtao-ren/rlhn