Spaces:
Running
Running
Commit History
-
chore: git mv scripts, UI, dev tools, docs into folders
Training pipeline (digit prefix kept — they're CLIs): 1_convert_labelstudio.py -> scripts/01_convert_labelstudio.py 2_train_classifier.py -> scripts/02_train_classifier.py 3_train_extractor_v3.py -> scripts/03_train_extractor.py 5_evaluate.py -> scripts/05_evaluate.py ocr_rasterise.py -> scripts/ocr_rasterise.py batch_process_dataref.py -> scripts/batch_process_dataref.py resplit.py -> scripts/resplit.py label.py -> scripts/label.py UI: streamlit_demo.py -> apps/streamlit_demo.py Dev / debug: debug_extractor.py -> tools/debug_extractor.py debug_logement.py -> tools/debug_logement.py debug_training.py -> tools/debug_training.py find_image_path.py -> tools/find_image_path.py find_logement_sample.py -> tools/find_logement_sample.py check_data.py -> tools/check_data.py logement_improvements.py -> tools/logement_improvements.py test_logement_e…70c46cc -
Add v3 extractor, recommendation engine, CMS generator, Streamlit demo, and tests
- New: LayoutLMv3 v3 extractor (3_train_extractor_v3.py) - New: rule engine for demande complétude verdict (6_recommendation_engine.py) - New: CMS IMMO 9 BANBOU xlsx generator (cms_generator.py) - New: production Streamlit demo with sample loader (streamlit_demo.py) - New: pytest suite (cms, inference postprocess, recommendation engine) - New: utility scripts (debug_*, batch_*, label.py, logement_improvements.py) - New: Makefile, mypy.ini, pytest.ini - Fix: 4_inference.py — anchor Config paths to script dir (works from any CWD) - Drop: deprecated 3_train_extractor.py, mapping.py, metadata_orange.csv - Gitignore: customer datasets (DataSet1/, DataSet2/), Label Studio exports, assets/sample_verdicts.json (real extracted PII)33ddb61