---
license: pddl
dataset_info:
  features:
  - name: id
    dtype: int64
  - name: origin_text
    dtype: large_string
  - name: summary_text
    dtype: large_string
  - name: lenght_origin_text
    dtype: int64
  - name: lenght_summary_text
    dtype: int64
  splits:
  - name: train
    num_bytes: 1079089469
    num_examples: 253925
  - name: validation
    num_bytes: 134268286
    num_examples: 31741
  - name: test
    num_bytes: 135109802
    num_examples: 31741
  download_size: 829761947
  dataset_size: 1348467557
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: validation
    path: data/validation-*
  - split: test
    path: data/test-*
task_categories:
- summarization
- text-generation
- fill-mask
size_categories:
- 100K<n<1M
language:
- en
---

### **Content**

This dataset was created from three datasets:

- [BBC News Summary](https://www.kaggle.com/datasets/pariza/bbc-news-summary)
- [CNN-DailyMail News Text Summarization](https://www.kaggle.com/datasets/gowrishankarp/newspaper-text-summarization-cnn-dailymail)
- Generated text from LLM models

Than was create Kaggle dataset:

- [***Text for summarize NLP/LLM task (En)***](https://www.kaggle.com/datasets/radimkzl/text-for-summarize-nlpllm-task/data)

The dataset was filtered, shuffled, and divided into parts before being saved to Hugging Face.

### **Acknowledgements**

- The dataset was created with the support of resources from [SV metal spol. s r.o.](https://www.svmetal.cz/cs)