Externalize dictionaries and add anonymization review corpus

This commit is contained in:
2026-04-21 10:32:57 +02:00
parent 39db675052
commit 34dcf8f360
99 changed files with 1805 additions and 805 deletions

View File

@@ -9,6 +9,7 @@ from collections import Counter
sys.path.insert(0, str(Path(__file__).parent))
import anonymizer_core_refactored_onnx as core
from config_defaults import RUNTIME_DICTIONARIES_CONFIG_PATH
from eds_pseudo_manager import EdsPseudoManager
from vlm_manager import VlmManager
from gliner_manager import GlinerManager
@@ -16,7 +17,7 @@ from camembert_ner_manager import CamembertNerManager
SRC = Path("/home/dom/Téléchargements/II-1 Ctrl_T2A_2025_CHCB_DocJustificatifs (1)")
OUTDIR = SRC / "anonymise_audit_30"
CONFIG = Path("/home/dom/ai/anonymisation/config/dictionnaires.yml")
CONFIG = RUNTIME_DICTIONARIES_CONFIG_PATH
PDFS = [
SRC / "114_23060661/CONSULTATION ANESTHESISTE 23060661.pdf",