This 2-day hands-on training provides participants with a foundational understanding of Large Language Models (LLMs). Participants will learn how LLMs function, explore prompt engineering principles, and gain practical skills to develop simple applications. The course covers key concepts such as Transformers, tokenization, RAG, and responsible AI practices. Through interactive lectures, guided exercises, and mini-projects, learners will acquire the confidence to apply LLMs in business or development settings and understand best practices for ethical and efficient AI use. This HRD Corp (HRDC) Claimable Course (Previously Known as SBL-Khas) Is Delivered by a Penang-Based Training Provider Registered with HRD Corp (Formerly Known as HRDF), Specializing in Corporate Skills Development and Workforce Upskilling Across Malaysia. 100% HRD Corp Claimable | Penang Training Provider | Corporate Training Malaysia
DAY 1
Module 1: Introduction to Generative AI & LLMs
Define LLMs, capabilities, and key terminology; compare Generative vs. Discriminative AI; Core Concept: Next Token Prediction; Key Metrics: Parameters, Tokens, Context Window
Activity: Case study discussion; "LLM or Not LLM?" Quiz
Module 2: The Inner Workings (Simplified)
Intuitive overview of training process; Transformer analogy; Model training stages: Pre-training vs. Fine-Tuning/Alignment
Activity: Visual demo; Online tokenization exercise
Lunch
Module 3: Prompt Engineering Fundamentals
Core Principles: Role, Task, Context, Format/Constraints; Temperature and creativity control
Activity: Hands-on lab with live LLM; practice persona and format constraints; compare outputs
Module 4: Advanced Prompting Techniques
Few-Shot Prompting; Chain-of-Thought (CoT) reasoning
Activity: Apply advanced patterns to solve problems; peer-review and refine prompts
DAY 2
Module 5: LLMs and External Data – RAG
Understand use of LLMs with private/up-to-date data; Key Components: Embeddings and Vector Databases
Activity: Conceptual demo of RAG pipeline; group discussion; simple Python embedding demo
Module 6: Fine-Tuning vs. Prompting vs. RAG
Differentiate customization methods; Fine-Tuning vs. Instruction Tuning; cost and effort considerations
Activity: Decision framework flowchart for selecting method; discussion of trade-offs
Lunch
Module 7: Responsible and Ethical LLM Use
Risks: Hallucinations, Bias, Safety; governance and mitigation practices
Activity: Case study analysis; develop simple Responsible LLM Use Guidelines
Module 8: Application Building & Next Steps
LLM APIs and orchestration frameworks; Evaluation of outputs
Activity: Mini-project building simple LLM application; Final Q&A and resource sharing
