LLM Concepts
How large language models are used and adapted.
This is an overview — pick a sub-topic below.
Sub-topics (16)
Large, Small and Foundation Models
Foundation models are pretrained on huge data and adapted to many tasks; small models trade capability for speed and local deployment.
Prompt Engineering and Templates
Write clear instructions, give examples (few-shot), specify the output format, and reuse prompts as templates.
Chain-of-Thought and Structured Outputs
Ask models to reason step by step, and return machine-readable output (JSON matching a schema).
Function Calling and Tool Use
The model decides to call your functions (search, calculator, database) and uses the results.
AI Agents and Multi-Agent Systems
LLM-driven programs that plan, use tools, keep memory and work towards goals — sometimes as teams of agents.
Agent Memory, Planning, Orchestration and Evaluation
Store context across steps, break tasks into plans, coordinate tool calls, and measure success.
Fine-Tuning, Instruction Tuning, LoRA, QLoRA and PEFT
Adapt a pretrained model to your task; parameter-efficient methods (LoRA/QLoRA) train small adapter weights.
RLHF and DPO
Align models with human preferences — via reinforcement learning from human feedback, or Direct Preference Optimization.
Quantization, Pruning and Distillation
Shrink models: lower-precision weights, removed connections, or a small student model trained to mimic a large teacher.
Model Serving and Inference Optimization
Serve models efficiently with batching, KV caching and optimised kernels.
Retrieval-Augmented Generation and GraphRAG
Retrieve relevant documents and give them to the model so answers are grounded in your data.
Multimodal and Vision-Language Models
Models that take images (and audio/video) plus text: describe photos, read charts, answer questions about screenshots.
Text-to-Image, Image-to-Image and Video Generation
Diffusion models that create and edit images and videos from prompts.
Local LLMs
Run open models on your own machine for privacy and offline use.
AI Guardrails, Observability and Evaluation
Validate inputs/outputs, trace every call, and test quality systematically.
AI Coding Assistants
Tools that suggest, explain and write code with LLMs.
Spotted a mistake or want to add detail? Suggest an improvement — Content Contributors' edits are reviewed by editors before they go live.
Discussion (0)
Sign in or join free to ask questions and discuss this topic.