Understand how LLM inference works, including tokenization, prefill and decode phases, KV cache, attention, sampling, and how transformer models generate text token by token in production systems.
A system-level overview of LLM architecture, including transformer blocks, encoder vs decoder models, attention mechanisms, training and inference flow, and how all components fit together in modern AI systems.
Understand the attention mechanism that powers Large Language Models (LLMs), including self-attention, query-key-value vectors, attention scores, multi-head attention, and why attention replaced recurrent neural networks.
Learn the key components of modern LLMs including tokenization, embeddings, transformers, attention, training, inference, and model parameters.
Learn what the context window is in LLMs, how token limits work, why context length matters, and how it impacts memory, reasoning, and real-world AI applications like RAG and chat systems.
Learn what embeddings are, how they convert text into vectors, why they are essential for semantic search and RAG systems, and how modern LLMs represent meaning in high-dimensional space.
Learn what LLM model parameters are, how they are trained, and why 7B, 70B, and 405B models differ in capability, memory, and cost.
Learn what tokens are in Large Language Models, how tokenization works, why token counts matter for context windows and pricing, and how modern LLMs process text.
Learn how LLMs are trained through pretraining, fine-tuning, RLHF, DPO, and alignment. Understand how data becomes model intelligence.
Learn what Retrieval-Augmented Generation (RAG) is, how it works, why it improves LLM accuracy, and how it combines retrieval systems with large language models.