---
license: mit
configs:
- config_name: default
  data_files:
  - split: train
    path: "cs_metadata_2020.json"
---

## ArXiv AI Paper Dump

**This dataset contains high-quality arXiv AI-related papers converted into txt format for NLP tasks. Papers are selected per following criteria:**

- Publishing year (first version) > 2020
- Journal / Conferences records.
- Under category cs.AI / cs.CL / cs.CV


See `cs_metadata_2020.json` for more info on individual papers. Thanks to the open [efforts](https://www.kaggle.com/datasets/Cornell-University/arxiv/data) of ArXiv team.