File size: 1,422 Bytes
8d0d861
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
# NOTICE — AncientDoc_Plus

## What this repository contains

- `1016v2_AncientDoc.json` — 2,198 message-level SFT entries (VLM fine-tuning
  format) containing our page-level OCR transcriptions for pages of the
  AncientDoc benchmark.
- `source_path_map.json` — mapping from our flat annotation filenames to the
  upstream repository paths, so annotations can be joined to upstream images.

No images are included.

## Upstream data (NOT included here)

- **Dataset**: AncientDoc — A Benchmark for Chinese Ancient Document Understanding
- **Source**: https://huggingface.co/datasets/ByteDance/AncientDoc
- **Upstream license**: CC0-1.0 (public domain)
- **Images**: download from the upstream repository. Use
  `source_path_map.json` to resolve each entry's image (flat name
  `<书名>page_NN.png`) to `imgs/<category>/<book>/page_NN.png` in the
  upstream repo.

## This annotation set

- **License**: CC-BY-4.0 (see LICENSE)
- **Copyright of annotations**: the authors of this repository.
- The annotations are our own original transcriptions of public-domain ancient
  texts; the underlying literary works themselves are in the public domain.

## Suggested citation of upstream

```bibtex
@misc{ancientdoc2024,
  title={AncientDoc: A Benchmark for Chinese Ancient Document Understanding},
  author={{ByteDance}},
  howpublished={\url{https://huggingface.co/datasets/ByteDance/AncientDoc}},
  note={CC0}
}
```