---
dataset_info:
  features:
  - name: context
    dtype: string
  - name: id
    dtype: string
  splits:
  - name: train
    num_bytes: 3639716712
    num_examples: 97107
  download_size: 1951658288
  dataset_size: 3639716712
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---
# Abstractive Content-Based Document IDs for Generative Retrieval

We are currently cleaning up codebase and artifacts. Dataset for reproduction will be uploaded soon!