Transformadores y PLN: Ajuste Fino de Modelos con Hugging Face

Resumen

Transformers, Fine-Tuning, and Model Evaluation is designed for learners with deep learning and NLP experience who want to master transformer architectures, fine-tune pre-trained models using Hugging Face, and deploy production-ready NLP solutions. You'll begin by exploring the transformer architecture in depth — including self-attention mechanisms, positional encodings, and model families like BERT, GPT, and T5.

Next, you'll learn to prepare datasets, fine-tune models for classification tasks, and evaluate results using metrics like F1, precision, and confusion matrices. The third module covers reproducibility and version control using DVC and Git, along with publishing models to the Hugging Face Hub.

Finally, you'll build and deploy transformer inference APIs using FastAPI, optimize performance through quantization, and integrate CI/CD practices for production systems. By the end of this course, you will:

- Apply transformer architectures to solve real-world NLP tasks - Fine-tune and evaluate pre-trained models using Hugging Face Transformers and Datasets - Build reproducible ML pipelines with DVC and Git version control - Deploy and test transformer-based inference APIs using FastAPI Disclaimer:

This is an independent educational resource created by Board Infinity for informational and educational purposes only.

This course is not affiliated with, endorsed by, sponsored by, or officially associated with any company, organization, or certification body unless explicitly stated. The content provided is based on industry knowledge and best practices but does not constitute official training material for any specific employer or certification program.

All company names, trademarks, service marks, and logos referenced are the property of their respective owners and are used solely for educational identification and comparison purposes.

Programa

• Arquitectura de Transformer y Fundamentos
Cubre los componentes clave y mecanismos de la arquitectura transformer, incluyendo la autoatención, incrustaciones y familias de modelos para tareas modernas de PLN. Duración: 4 horas.
• Ajuste Fino de Modelos Preentrenados
Se enfoca en cargar y preparar conjuntos de datos usando Hugging Face Datasets, ajustar finamente modelos transformer para tareas de clasificación y evaluar el rendimiento usando métricas estandarizadas y matrices de confusión. Duración: 4 horas.
• Evaluación, Reproducibilidad y Control de Versiones
Cubre la implementación de tuberías de ML reproducibles usando DVC, el seguimiento de modelos y métricas con Git y DVC, y compartir resultados en el Hugging Face Hub. Duración: 4 horas.
• Despliegue e Integración de Modelos Transformer
Cubre el despliegue de transformers ajustados como APIs con FastAPI, la evaluación de puntos finales desplegados para precisión y latencia, y la integración de monitoreo y pruebas para la preparación en producción. Duración: 4 horas.

---

Impartido por

Board Infinity

---

Materias

Computer Science