Spaces:
Running
Running
Istvan-Adem
commited on
Commit
·
e966906
1
Parent(s):
7e8e988
add original text with filtering
Browse files- ocr/api/message/prompts.py +3 -13
ocr/api/message/prompts.py
CHANGED
@@ -26,19 +26,9 @@ The report must be structured as follows, with each section containing only rele
|
|
26 |
|
27 |
[/INST]"""
|
28 |
extract_original_text = """## Task
|
29 |
-
|
30 |
-
You must extract all text from the provided images and return it in the **text** field. However, you must **strictly** exclude any information related to the **patient’s name, contact details, or demographic data**.
|
31 |
|
32 |
-
|
33 |
|
34 |
-
|
35 |
-
- **Do not** include any **patient-identifiable information**, including:
|
36 |
-
- Names (first, last, middle, initials)
|
37 |
-
- Contact details (phone numbers, email addresses, addresses)
|
38 |
-
- Demographic information (age, date of birth, gender, ethnicity, etc.)
|
39 |
-
- Preserve **the structure and order** of the text as much as possible.
|
40 |
|
41 |
-
|
42 |
-
|
43 |
-
- Do not alter or interpret the content—your task is **only extraction**.
|
44 |
-
- If a section contains both medical and personal data, extract only the medical data and redact the personal information."""
|
|
|
26 |
|
27 |
[/INST]"""
|
28 |
extract_original_text = """## Task
|
|
|
|
|
29 |
|
30 |
+
You must extract all text from the attached images and return it in the **text** field. You must not include the patient's name, contact details, or demographic data.
|
31 |
|
32 |
+
## Important notes
|
|
|
|
|
|
|
|
|
|
|
33 |
|
34 |
+
- You must extract all text but exclude any information related to the name, contact details, and demographic data."""
|
|
|
|
|
|