This directory preserves the original layouts of earlier versions for reference only. Do not use these for new work — they are not maintained and have known issues that the current root-level v3.0_initial layout fixes.
v1.0_initial/— The original 5-file flat layout (Dockerfile,advanced_ocr.py,custom_doclayout_yolo.py,ocr_stage1.py,ocr_stage2.py).v2.0_initial/— Same 5 files with the v2 patch applied (Docker permission fix on Jupyter / Vertex AI, plus memory optimizations inadvanced_ocr.py). See../changes/v2.0_initial.mdfor what changed from v1 → v2.
- Historical reference for users who built on top of v1 or v2.
- Reproducibility of past runs (the exact code that produced earlier results lives here).
- A diff anchor: comparing
legacy/v2.0_initial/advanced_ocr.pywithsrc/ocr/advanced_ocr.pyshows the v2 → v3 changes in code form.
Use the root-level v3.0_initial layout:
src/ocr/— OCR engine modulessrc/stages/— Stage execution wrappersauto_run_*.py,auto_run.yaml,config.yaml,prompts/— orchestration and configuration
See ../setup_guide.md and ../usage.md.
For the full v2 → v3 migration summary, see ../changes/2026-05-13_v3.0_release.md.