LLM
LLM on AI-ML Companion: Large Language Models - from architecture to applications. Understand how LLMs work, then build RAG systems, agents, and multi-modal apps.. 37 interactive modules with live visualizations, quizzes, and hands-on Python coding.
Start free: What are LLMs? is fully open to everyone, no account required. The other 36 modules are part of AI-ML Companion Premium; every title and summary is listed below so you can see exactly what the track covers before deciding.
Modules in this track
- What are LLMs? (free) - Understanding large language models and their capabilities
- Generative AI: The Big Picture (premium) - VAEs, GANs, Diffusion, Transformers - the full GenAI family tree
- Tokenization Deep Dive (premium) - BPE, WordPiece, SentencePiece tokenization
- Embeddings & Token Representations (premium) - From token IDs to meaning vectors
- LLM Architecture Internals (premium) - Decoder-only transformers, RMSNorm, SwiGLU, KV cache
- Positional Encodings (premium) - RoPE, ALiBi, sinusoidal, and context extension methods
- Mixture of Experts (MoE) (premium) - Sparse activation, expert routing, and load balancing
- Context Window & Attention Patterns (premium) - GQA, MLA, Flash Attention, and long-context methods
- Pre-training Objectives (premium) - Next-token prediction, masked LM, causal vs bidirectional
- Logits, Softmax & Temperature (premium) - Softmax mechanics, temperature, and log probabilities
- Sampling Strategies (premium) - Greedy, top-k, top-p, min-p, and beam search
- Multi-turn Conversations & Memory (premium) - Chat formats, context management, and memory
- Context Engineering (premium) - Budgeting the window: tools, retrieval, compaction, and cache order
- Prompt Engineering (premium) - Zero-shot, few-shot, and chain-of-thought techniques
- Advanced Prompting (premium) - Tree-of-Thought, Self-Consistency, HyDE, and advanced reasoning techniques
- Prompt Optimization (premium) - APE, prompt chaining, ReAct, and automated prompt optimization
- In-Context Learning (premium) - Learning from examples without weight updates
- Fine-tuning Strategies (premium) - LoRA, QLoRA, DoRA, and the PEFT landscape
- Instruction Tuning & RLHF (premium) - SFT, reward models, PPO, DPO, and alignment
- Hallucinations & Limitations (premium) - Why LLMs fabricate facts, detection, and mitigation
- LLM Security & Prompt Injection (premium) - Prompt injection, defense-in-depth, and attack vectors
- Evaluation & Benchmarks (premium) - Perplexity, benchmarks, LLM-as-judge, and domain evals
- Inference Optimization (premium) - KV cache, quantization, speculative decoding, and serving
- Scaling Laws & Emergent Abilities (premium) - Power-law scaling, Chinchilla, and emergent abilities
- Open-Source LLMs & Local Deployment (premium) - Llama, DeepSeek, Qwen, local deployment, and serving
- Hugging Face: The Open-Source AI Hub (premium) - Hub, transformers, datasets, PEFT, TRL, Spaces, and Inference Endpoints
- LLM APIs & Practical Usage (premium) - Universal API pattern, streaming, function calling, and cost optimization
- Retrieval-Augmented Generation (RAG) (premium) - Chunking, embeddings, vector search, and hybrid retrieval
- Corrective & Self-RAG (premium) - Self-correcting retrieval with CRAG and Self-RAG
- Knowledge Graphs (premium) - Triples, RDF vs property graphs, Cypher, and grounding LLMs
- GraphRAG & Knowledge Graphs (premium) - Knowledge graphs, community detection, and query routing
- Adaptive RAG & Query Routing (premium) - Query classification, dynamic routing, and semantic caching
- Multimodal RAG (premium) - RAG for documents with text, tables, and images
- Agents & Tool Use (premium) - Function calling, ReAct, MCP, and multi-agent patterns
- Multi-Modal LLMs (premium) - Vision encoders, early fusion, and agentic vision
- Knowledge Distillation (premium) - Soft labels, dark knowledge, and CoT distillation
- Project: RAG Expert Assistant (premium) - Production RAG pipeline with RAGAS evaluation, FlashRank reranking, and security hardening