Developing Custom Large Language Models
Unlock the future of AI with custom large language models tailored to your unique business needs, driving innovation, efficiency, and personalized experiences like never before.
Overview
This course is designed to empower participants with the skills and knowledge necessary to develop custom Large Language Models (LLMs) from scratch, leveraging existing models. Through a blend of lectures, hands-on exercises, and project work, participants will learn the end-to-end process of building, training, and deploying LLMs.
The course is structured into four modules, culminating in a final project presentation where participants showcase their custom models. The scope and depth of the course content can be adjusted for participants with less technical experience than the skills outlined in the prerequisites section.
AED 26,250
Outcomes
Learning Outcomes
Understand the fundamental concepts and architectures of LLMs.
Acquire skills in data collection, cleaning, and preprocessing for LLM training.
Learn to modify and fine-tune existing LLM architectures for custom applications.
Develop expertise in training and deploying scalable LLMs.
Gain insights into troubleshooting and optimizing LLM performance.
Complete a custom LLM project, demonstrating the ability to apply course concepts in practice.
Program Schedule
Each module is designed to build upon the previous one, progressively leading participants toward completing their custom LLM projects. The hands-on approach ensures that participants not only understand the theoretical aspects of LLM development but also gain practical experience in implementing and optimizing these models.
- Module 1: Introduction to Large Language Models
- Module 2: Data Preparation and Model Customization
- Module 3: Training and Fine-Tuning
- Module 4: Deployment and Project Presentation
Each day consists of four days, with each covering a unique aspect of the LLM.
Modules may partly overlap or merge into one another as to optimize a full 5-day training session. This 16th day is to be used for online support sessions in between modules.
Module 1: Introduction to Large Language Models
- Day 1: Overview of LLMs and their applications
- Day 2: Understanding LLM architectures (Transformers, GPT, BERT)
- Day 3: Python libraries for LLMs (Hugging Face Transformers, TensorFlow, PyTorch)
- Day 4: Setting up the development environment and initial model exploration
Module 2: Introduction to Large Language Models
- Day 5: Data collection and dataset creation for LLMs
- Day 6: Data cleaning and preprocessing techniques
- Day 7: Customizing LLM architectures for specific tasks
- Day 8: Introduction to fine-tuning techniques
Module 3: Introduction to Large Language Models
- Day 9: Deep dive into model training and computational considerations
- Day 10: Advanced fine-tuning and transfer learning strategies
- Day 11: Monitoring model training and addressing common issues
- Day 12: Evaluating model performance and iterative improvement
Module 4: Introduction to Large Language Models
- Day 13: Model deployment strategies and best practices
- Day 14: Preparing the final project: Integrating LLMs into real-world applications
- Day 15: Finalizing projects and preparing presentations
- Day 16: Project presentations and peer review
Training Methodology
- Programs are delivered by international subject matter experts.
- Modules are designed to use the most practical approach to deliver the course, using the most effective learning techniques.
Participants will be involved in an interactive actual practice:
- Case studies discussion
- Class room activities
- Team learning techniques
- Project work
- Use of adept tools on various topics covered in the program
- Access to research tools on various repository on the subject
- Lecture modules delivered by subject matter experts
Subject Matter Expert
Bos Boorsma
Professor of Experience, Associate at RIT Dubai
Founder Urban Innovators Inc. Former Cisco, Former CDO of Rotterdam
Bas is a leading expert in urban innovation and digitalization with 23 years of experience in the 'smart city' sector. He served as the Chief Digital Officer (CDO) for the City of Rotterdam, focusing on driving innovation in collaboration with the municipality. He is the Founder and Partner of Urban Innovators Inc. and Chairman of Venturerock Urban Italy, as well as a Professor of Practice at Arizona State University's Thunderbird School of Global Management.
Previously, he held leadership positions such as CDO for Rotterdam and Digitization Leader at Cisco. He is a board member of The Smart City Association Italy (TSCAI) and author of *A New Digital Deal*. Bas and his team established a knowledge hub in Rotterdam to foster collaboration among partners, organizations, and the community, focusing on reducing emissions, advancing mobility, and supporting small businesses through digital transformation.
His work also involves identifying major trends that could impact cities and taking proactive steps to stay ahead of potential disruptions.