---
dataset_info:
  features:
  - name: condition
    dtype: string
  - name: group
    sequence: string
  splits:
  - name: train
    num_bytes: 20134475
    num_examples: 25352
  - name: validation
    num_bytes: 296445
    num_examples: 390
  download_size: 11977779
  dataset_size: 20430920
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: validation
    path: data/validation-*
---

## 介绍
训练集目前有25000 queries, 55000 corpus
基本上是1:2
够了