davanstrien HF Staff commited on
Commit
781fc20
·
verified ·
1 Parent(s): 9dfe6f5

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +49 -35
README.md CHANGED
@@ -1,37 +1,51 @@
1
  ---
2
- dataset_info:
3
- features:
4
- - name: image
5
- dtype: image
6
- - name: raw_queries
7
- list: string
8
- - name: broad_topical_query
9
- dtype: string
10
- - name: broad_topical_explanation
11
- dtype: string
12
- - name: specific_detail_query
13
- dtype: string
14
- - name: specific_detail_explanation
15
- dtype: string
16
- - name: visual_element_query
17
- dtype: string
18
- - name: visual_element_explanation
19
- dtype: string
20
- - name: parsed_into_json
21
- dtype: bool
22
- - name: markdown
23
- dtype: string
24
- - name: inference_info
25
- dtype: string
26
- splits:
27
- - name: train
28
- num_bytes: 598481
29
- num_examples: 5
30
- download_size: 620052
31
- dataset_size: 598481
32
- configs:
33
- - config_name: default
34
- data_files:
35
- - split: train
36
- path: data/train-*
37
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
+ tags:
3
+ - ocr
4
+ - document-processing
5
+ - falcon-ocr
6
+ - plain
7
+ - uv-script
8
+ - generated
9
+ viewer: false
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
10
  ---
11
+
12
+ # Document Processing using Falcon OCR (plain mode)
13
+
14
+ This dataset contains OCR results from images in [davanstrien/ufo-ColPali](https://huggingface.co/datasets/davanstrien/ufo-ColPali) using [Falcon OCR](https://huggingface.co/tiiuae/Falcon-OCR), a 0.3B early-fusion vision-language model.
15
+
16
+ ## Processing Details
17
+
18
+ - **Source Dataset**: [davanstrien/ufo-ColPali](https://huggingface.co/datasets/davanstrien/ufo-ColPali)
19
+ - **Model**: [tiiuae/Falcon-OCR](https://huggingface.co/tiiuae/Falcon-OCR)
20
+ - **Task Mode**: `plain` - Full-page text extraction
21
+ - **Number of Samples**: 5
22
+ - **Processing Time**: 2.3 min
23
+ - **Processing Date**: 2026-04-07 17:25 UTC
24
+
25
+ ### Configuration
26
+
27
+ - **Image Column**: `image`
28
+ - **Dataset Split**: `train`
29
+ - **Max Output Tokens**: 2,048
30
+ - **Backend**: Transformers
31
+
32
+ ## Model Information
33
+
34
+ Falcon OCR is a compact early-fusion model that processes image patches and text tokens
35
+ in a shared Transformer. Key results:
36
+ - 80.3% on olmOCR benchmark
37
+ - 88.64% on OmniDocBench
38
+ - 87.1% on multi-column documents (best in class)
39
+ - 90.3% on tables (best in class)
40
+
41
+ ## Reproduction
42
+
43
+ ```bash
44
+ uv run https://huggingface.co/datasets/uv-scripts/ocr/raw/main/falcon-ocr.py \
45
+ davanstrien/ufo-ColPali \
46
+ <output-dataset> \
47
+ --task-mode plain \
48
+ --image-column image
49
+ ```
50
+
51
+ Generated with [UV Scripts](https://huggingface.co/uv-scripts)