Clawdemy Lessons

Overview of the architecture lesson: the decoder-only skeleton, converged design choices, sizing hyperparameters, prerequisites, math level, and time.

Orientation to the from-scratch LLM track: what you'll learn, prerequisites, the math involved, and time and difficulty for building the tokenizer.

What this lesson on model cost covers: FLOPs and the 6ND rule, the 16N memory estimate, arithmetic intensity, and einops, plus prerequisites, math, and timing.

Overview, prerequisites, and learning goals for the lesson on attention's cost problems, the MQA, GQA, and sliding-window alternatives, and mixture of experts.

A working picture of transformers before you run any: the tokens-in-tokens-out idea, why they replaced RNNs, the three shapes, and where Hugging Face fits.

Overview of the handwritten-digit lesson: what you'll learn, where it fits in the track, prerequisites (none), and the read and practice time.

Overview of the lesson on AI agents: the perceive-decide-act loop definition, the four parts that make a system agentic, prerequisites, and what comes next.

An overview of writing tool definitions an AI agent can use reliably: the four parts, when-to-use and negative guidance, and disambiguating overlapping tools.

An overview of agent planning: why reactive agents fail at scale, decomposition into ordered sub-tasks, plan-then-execute versus replanning, and grain size.

Overview of the multi-agent systems lesson: what it covers, prerequisites, learning outcomes, and how it fits after planning in the agent design track.

What the lesson on agent self-checks covers: why reflection raises reliability, the forms it takes, how it compares to adding an agent, and its honest limits.

Orientation for the tool-use lesson: what you will learn, prerequisites, the four-step tool-call exchange in brief, and how it fits the agents track.

Overview of how agents hold information: short-term context versus persistent memory, what is worth retaining, and the context, staleness, and privacy costs.

Overview of the hand-roll-versus-framework decision, the framework landscape by category, and choosing by fit so you stay free of any single library.

A guided overview of the six ways AI agents fail on their own, the guardrail that contains each, and when to require human-in-the-loop confirmation.

Overview of agentic RAG: how making retrieval a tool the agent decides to call turns the fixed retrieve-read-answer pipeline into a reasoning loop.

Brief for the BERT training lesson: why bidirectionality forced MLM and NSP, the two-stage train-then-fine-tune workflow, and the CLS vs per-token heads.

Brief for BERT's architecture lesson: the encoder-only branch, bidirectional self-attention, the CLS and SEP tokens, and three additive input embeddings.

Brief for the tool-use failures lesson: the three buckets, nine named sub-failures, where it fits in Phase 7, prerequisites, and time and difficulty.

A brief on benchmark literacy: the major benchmark categories, what each measures, and the structural reasons scores can rise faster than real capability.

Overview of the AI safety recap lesson: the per-phase threads it covers, where it fits in the foundations track, prerequisites, and learning outcomes.

Brief for the lesson on transformer adaptations beyond text: Vision Transformers (ViT) for images and Mixture-of-Experts (MoE) for sparse parameter scaling.

Brief for the RLHF and DPO lesson: what you'll learn, where it fits in Phase 4, prerequisites, and the learning outcomes on PPO, DPO, and reward hacking.

Brief for the lesson on alternatives to autoregressive generation: what speculative decoding and diffusion LLMs are, how each works, and when each fits.

Lesson brief: in-context learning and few-shot prompting, with learning outcomes, prerequisites, where it fits in Phase 5, and time and difficulty.