File size: 1,502 Bytes
41004cd
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
---
license: apache-2.0
---
---
license: apache-2.0
task_categories:
- document-understanding
- text-generation
language:
- en
---

README.md
train.jsonl
val.jsonl
test.jsonl
{"instruction":"Extract payment amount and due date from the document","input":"Invoice #1029. Total payable amount is $5,000 due by January 30, 2025.","output":"{\"amount\":\"$5,000\",\"due_date\":\"2025-01-30\"}"}
{"instruction":"Identify potential risks in the contract","input":"Payment shall be made within 90 days. No penalty clause is specified.","output":"{\"risk\":\"Delayed payment risk\"}"}

{"instruction":"Extract key entities from the agreement","input":"The client agrees to pay USD 12,000 upon project completion.","output":"{\"amount\":\"$12,000\"}"}

{"instruction":"Classify the document section","input":"Termination clause: Either party may terminate with 30 days notice.","output":"{\"section\":\"Termination\"}"}



# DocuMind ERNIE 4.5 Document Reasoning Dataset

Instruction-style dataset for fine-tuning ERNIE 4.5 on OCR-based document
understanding and reasoning tasks.

## Data Source
Documents are processed using PaddleOCR-VL to extract text and layout.
Data is anonymized and partially synthetic.

## Schema
Each record:
{
  "instruction": "...",
  "input": "<OCR extracted document text>",
  "output": "<structured JSON>"
}

## Tasks
- Entity extraction
- Risk identification
- Section classification
- Contract reasoning

## Splits
- train.jsonl
- val.jsonl
- test.jsonl

## License
Apache-2.0