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

pdf
Some checks failed
build_docker / essential (push) Successful in 1s
Details
build_docker / build_paddle_ocr (push) Failing after 4m34s
Details
build_docker / build_easyocr (push) Has been cancelled
Details
build_docker / build_easyocr_gpu (push) Has been cancelled
Details
build_docker / build_doctr (push) Has been cancelled
Details
build_docker / build_doctr_gpu (push) Has been cancelled
Details
build_docker / build_raytune (push) Has been cancelled
Details
build_docker / build_paddle_ocr_gpu (push) Has been cancelled
Details

Browse Source
This commit is contained in:
Sergio Jimenez Jimenez
2026-04-21 20:46:05 +02:00
parent 9c59ff0f6d
commit 5351d4a57b
23 changed files with 921 additions and 210 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
thesis_output/presentation/slides/15_script.md Normal file
View File

@@ -0,0 +1,14 @@
# Slide 15 — GPU Acceleration (~30 seconds)
**What to say:**
"Un dato relevante sobre la viabilidad práctica: la GPU RTX 3060 proporciona una aceleración de 82 veces respecto a CPU.
En CPU, cada página tarda 69 segundos. En GPU, solo 0.84 segundos. Esto significa que los 64 ensayos sobre 5 páginas se completaron en unos 5 minutos con GPU, frente a más de 6 horas en CPU.
La aceleración por GPU no es un lujo, sino una necesidad para hacer viable la optimización de hiperparámetros a escala."
**Tips:**
- Quick slide — let the numbers speak for themselves
- "82x" is the headline number
- Emphasize feasibility: consumer GPU makes this practical