MateuszW/spoiler_generation

---
task_categories:
- question-answering
- text-generation
language:
- en
tags:
- spoiler generation
- clickbait spoiling
---
# Datasets used for spoiler generation task

## Dataset Description
This dataset contains multiple datasets used for spoiler generation task for Clickbait spoiling competition for
training models based on Question Answering, text generation, or learn-to-rank problems.

## Dataset Structure
This dataset has 5 main directories:
- clf_data - this dataset was used to train a classifier for two generated spoilers to decide
which spoiler better matches a clickbait post. This model makes a binary classification,
class 1 corresponds to the situation when the first spoiler is "better" than the second,
and class 0 corresponds to the opposite situation
- clickbait_spoiling_data - this dataset is the original dataset taken from the Clickbait spoiling competition
- generated_questions - this dataset contains questions generated for clickbait posts by the Vicuna model
- models_output - in this dataset were inserted generated spoilers from the best-selected models
- regressor_data - this dataset was used to train a model that estimates the BLEU of generated
spoiler without knowing reference spoiler

原始信息汇总

数据集概述

数据集用途

• 用于训练模型解决点击诱饵破坏任务中的问题，包括问答、文本生成和学习排序问题。

数据集结构

数据集包含以下五个主要目录：

1. clf_data：用于训练分类器，以决定两个生成的破坏者中哪一个更匹配点击诱饵帖子。分类结果为二元分类，类别1表示第一个破坏者“更好”，类别0表示相反情况。
2. clickbait_spoiling_data：来自点击诱饵破坏竞赛的原始数据集。
3. generated_questions：包含Vicuna模型为点击诱饵帖子生成的问題。
4. models_output：包含最佳选定模型生成的破坏者。
5. regressor_data：用于训练模型，该模型估计生成的破坏者的BLEU分数，无需知道参考破坏者。

数据集语言

• 英语（en）

数据集标签

• 破坏者生成
• 点击诱饵破坏