matejklemen/nucle

---
license: other
dataset_info:
- config_name: public
features:
- name: src_tokens
sequence: string
- name: tgt_tokens
sequence: string
- name: corrections
list:
- name: idx_src
sequence: int32
- name: idx_tgt
sequence: int32
- name: corr_type
dtype: string
splits:
- name: train
download_size: 0
dataset_size: 0
- config_name: private
features:
- name: src_tokens
sequence: string
- name: tgt_tokens
sequence: string
- name: corrections
list:
- name: idx_src
sequence: int32
- name: idx_tgt
sequence: int32
- name: corr_type
dtype: string
splits:
- name: train
download_size: 0
dataset_size: 0
---

**Important**: This is only a script for loading the data, but the data itself is private. The script will only work in case you have access to the data, which you may request for non-commercial purposes [here](https://sterling8.d2.comp.nus.edu.sg/nucle_download/nucle.php).
```python
data = datasets.load_dataset("matejklemen/nucle", "private", data_dir=<dir-of-private-data>, ignore_verifications=True)"
```
The `ignore_verifications=True` is important as the datasets library initially builds validation statistics that it verifies against,
and these cannot be correctly computed when the data is not public.

原始信息汇总

数据集概述

配置信息

• config_name: public, private

特征信息

• src_tokens: 字符串序列
• tgt_tokens: 字符串序列
• corrections: 列表
  • idx_src: 整数序列
  • idx_tgt: 整数序列
  • corr_type: 字符串类型

数据分割

• train

数据大小

• download_size: 0
• dataset_size: 0