unir/MastersThesis
0
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

regen docs
Some checks failed
build_docker / build_cpu (pull_request) Has been cancelled
Details
build_docker / build_gpu (pull_request) Has been cancelled
Details
build_docker / build_easyocr (pull_request) Has been cancelled
Details
build_docker / build_easyocr_gpu (pull_request) Has been cancelled
Details
build_docker / build_raytune (pull_request) Has been cancelled
Details
build_docker / build_doctr (pull_request) Has been cancelled
Details
build_docker / essential (pull_request) Has been cancelled
Details
build_docker / build_doctr_gpu (pull_request) Has been cancelled
Details

Browse Source
This commit is contained in:
Sergio Jimenez Jimenez
2026-01-19 17:38:43 +01:00
parent b1539fd79f
commit 07a7ba8c01
4 changed files with 138 additions and 541 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
docs/05_conclusiones_trabajo_futuro.md
View File

@@ -38,7 +38,7 @@ El objetivo principal del trabajo era alcanzar un CER inferior al 2% en document
**Respecto a OE4 (Optimización con Ray Tune)**:
- Se ejecutaron 64 trials con el algoritmo OptunaSearch
- El tiempo total del experimento fue aproximadamente 6 horas (en CPU)
- La arquitectura basada en subprocesos permitió superar incompatibilidades entre Ray y PaddleOCR
- La arquitectura basada en contenedores Docker permitió superar incompatibilidades entre Ray y los motores OCR, facilitando además la portabilidad y reproducibilidad
**Respecto a OE5 (Validación de la configuración)**:
- Se validó la configuración óptima sobre el dataset completo de 24 páginas