---
pretty_name: rageval Dataset Bundle
license: other
task_categories:
- question-answering
language:
- zh
- en
---

# rageval Dataset Bundle

This dataset repository stores the raw datasets currently available in the local rageval workspace together with the processed datasets used in the thesis experiments.

Repo:

- `bangwu/rageval_qa`

## Included data

### Raw datasets

- `raw/docfinqa/`
- `raw/financebench/`
- `raw/lecardv2/`
- `raw/loong/`
- `raw/narrativeqa/`
- `raw/peerqa/`

### Processed datasets

- `processed/docfinqa/docfinqa_unified.jsonl`
- `processed/lecardv2/lecardv2_unified.jsonl`
- `processed/loong/loong_unified.jsonl`
- `processed/narrativeqa/narrativeqa_unified.jsonl`
- `processed/peerqa/peerqa_unified.jsonl`
- `processed/experiment/main_experiment.jsonl`
- `processed/experiment/stress_test.jsonl`
- `processed/experiment/ultra_long_finance.jsonl`
- `processed/experiment/split_manifest.json`
- `processed/experiment/ultra_long_finance_manifest.json`

## Notes

1. This repository is data-only. Scripts, experiment code, and thesis documents remain in the GitHub repository.
2. The processed files are the dataset artifacts used for the rageval comparison experiments.
3. Raw source materials may remain subject to their original licenses or redistribution constraints. Please review the original dataset licenses before reuse.