fasterinnerlooper/lcc_csharp

---
language:
- en
license: mit
size_categories:
- 100K<n<1M
task_categories:
- mask-generation
- fill-mask
- text-generation
pretty_name: LCC_csharp dataset modified for infilling
dataset_info:
features:
- name: prefix
dtype: string
- name: suffix
dtype: string
- name: prediction
dtype: string
splits:
- name: train
num_bytes: 1852197668
num_examples: 100000
download_size: 531853418
dataset_size: 1852197668
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
tags:
- code
---

This dataset has been modified from the microsoft/LCC_csharp dataset to provide CodeLLaMa with infilling tasks as per the original fill-in-the-middle paper, were the text that needs to be filled in is moved to the end of the dataset, thus taking advantage of the Generative feature of GPT-style models.

原始信息汇总

数据集概述

基本信息

• 语言: 英语
• 许可证: MIT
• 大小类别: 100K<n<1M
• 任务类别:
  • 掩码生成
  • 填充掩码
  • 文本生成
• 美观名称: LCC_csharp dataset modified for infilling

数据集信息

• 特征:
  • 前缀: 字符串类型
  • 后缀: 字符串类型
  • 预测: 字符串类型
• 分割:
  • 训练集:
    • 字节数: 1852197668
    • 样本数: 100000
• 下载大小: 531853418
• 数据集大小: 1852197668

配置

• 配置名称: default
• 数据文件:
  • 分割: 训练集
  • 路径: data/train-*

标签

• 代码