Try KV cache temperature BPE attention head hallucination cosine similarity
Jump to a lesson
Lesson Tokenization: From Text to Token IDs Module 01 · Model Foundations ↵ Lesson Embeddings: From Token IDs to Vectors Module 01 · Model Foundations Lesson Attention: How Tokens Use Their Context Module 01 · Model Foundations Lesson Next-Token Prediction and Sampling Module 01 · Model Foundations Lesson Context Windows, Prefill and the KV Cache Module 01 · Model Foundations Lesson Hallucination and the Limits of Next-Token Prediction Module 01 · Model Foundations Lesson Measuring Inference: Latency, Throughput and Goodput Module 02 · Inference and Model Optimization Lesson Where the Time Goes: Memory, Compute and Batching Module 02 · Inference and Model Optimization Lesson Precision and Quantization: Fewer Bytes per Weight Module 02 · Inference and Model Optimization Lesson Caching Repeated Work: Prompts, Prefixes and Answers Module 02 · Inference and Model Optimization Lesson Cost per Request and per Useful Answer Module 02 · Inference and Model Optimization Lesson Benchmarking Inference, and the Module Capstone Module 02 · Inference and Model Optimization Lesson Anatomy of a Prompt: Roles, Templates and Instruction Hierarchy Module 03 · Prompt and Context Engineering Lesson Few-Shot Examples and Structured Outputs Module 03 · Prompt and Context Engineering Lesson Choosing Context: Selection, Ordering and Compression Module 03 · Prompt and Context Engineering Lesson Tool Instructions and Constrained Outputs Module 03 · Prompt and Context Engineering Lesson Versioning Prompts and Catching Regressions Module 03 · Prompt and Context Engineering Lesson Prompt Injection and Trust Boundaries Module 03 · Prompt and Context Engineering Lesson Success Criteria: Deciding What Good Means Module 04 · Evaluation and Experimentation Lesson Scoring Against References: Exact, Lexical and Semantic Module 04 · Evaluation and Experimentation Lesson Using a Model as a Judge Module 04 · Evaluation and Experimentation Lesson Comparing Systems You Can Trust Module 04 · Evaluation and Experimentation Lesson Error Analysis: Finding What Goes Wrong Module 04 · Evaluation and Experimentation Lesson How RAG Works: Retrieve, Then Generate Module 05 · Retrieval and Knowledge Systems Lesson Term-Based and Embedding Retrieval Module 05 · Retrieval and Knowledge Systems Lesson Hybrid Search and Reranking Module 05 · Retrieval and Knowledge Systems Lesson Chunking and Indexing Module 05 · Retrieval and Knowledge Systems Lesson Evaluating RAG: Retrieval, Grounding and Failures Module 05 · Retrieval and Knowledge Systems Lesson Agents, Tools and the Loop Module 06 · Agents and Harness Engineering Lesson Workflows or Agents: Who Decides the Next Step Module 06 · Agents and Harness Engineering Lesson State, Memory and Context Over Many Steps Module 06 · Agents and Harness Engineering Lesson When Steps Fail: Retries, Timeouts, Idempotency and Checkpoints Module 06 · Agents and Harness Engineering Lesson Traces and Evaluating Agents Module 06 · Agents and Harness Engineering Lesson Bounded Autonomy: Limits, Permissions and Sandboxes Module 06 · Agents and Harness Engineering Lesson Choosing and Routing Models Module 07 · Production AI Systems Lesson Fallbacks and Graceful Degradation Module 07 · Production AI Systems Lesson Budgets, Rate Limits and Usage Controls Module 07 · Production AI Systems Lesson Observability and Incident Analysis Module 07 · Production AI Systems Lesson Guardrails and Safe Deployment Module 07 · Production AI Systems Lesson When to Fine-Tune, and a Production-Readiness Plan Module 07 · Production AI Systems ↑ ↓ move↵ openesc close