Clawdemy Lessons

Overview of the REINFORCE lesson: the log-derivative trick, why environment dynamics drop out, a worked sigmoid bandit, and the variance refinements.

Editorial brief for the lesson on planning with a learned model: capability gained, sources, and an artifact-by-artifact outline of the lesson set.

Editorial brief for the model-based RL lesson: learning outcomes, the worked least-squares dynamics fit, compounding-error rollout, and model-class choices.

Orientation to deep reinforcement learning: the ML regimes, the agent-environment loop, returns, discounting, and what makes the deep variant hard.

Overview of behavioral cloning: the algorithm, why it appeals, how distribution shift makes errors compound, DAgger, and where copying an expert suffices.

Editorial brief mapping each DQN trick to a deadly-triad leg, with the closed-form max-overestimation bias worked on a small example.

Editorial brief for the control-as-inference lesson: the optimality-conditioned graphical model, soft Bellman backup, and how SAC, RLHF, and DPO unify.

A roadmap to actor-critic methods: the two-network split, advantage estimators (MC, TD, n-step, GAE), the bias-variance tradeoff, and the modern RL family.

What to expect from the world modeling lesson: the reactive-vs-predictive split, the pixel-vs-latent trade-off, landmark architectures, and prerequisites.

Orientation for the vision-language lesson: what CLIP's joint embedding space unlocks, the prerequisites, the math level, plus read and practice time.

A walkthrough of the video architecture ladder, from the single-frame baseline through 3D convolutions and two-stream networks to video transformers.

Scope, prerequisites, and outcomes for the vision sequence-tools lesson: recurrence and attention, captioning, video, and the Vision Transformer.

An overview of self-supervised vision: why labels are scarce, the pretext-to-contrastive-to-masked-modeling arc, and the pre-train-then-transfer workflow.

What you'll learn from the neural networks and backprop lesson: why a ReLU adds capacity, how hidden layers learn features, and what backprop does.

A walkthrough of the training loop: the two standard losses, regularization, the gradient descent update rule, and mini-batch SGD, plus what to expect.

Orientation for the linear classifier lesson: what you will learn, how the score function s = W x + b fits the track, prerequisites, math needed, and timing.

What this lesson on deployed vision systems covers: the failure-mode catalog, bias as a measurable engineering property, and the trustworthiness gap.

Overview of the GANs and VAEs lesson: what you will learn, prerequisites, the reparameterization math, time and difficulty, and where it fits in this track.

What the diffusion lesson covers: the two-direction noising setup, the MSE training objective, inference speed-ups, and text-to-image conditioning.

A roadmap to going beyond image classification: what detection, segmentation, and visualization add, the IoU metric, prerequisites, and time and difficulty.

Overview of the convolution lesson: the conv operation, weight sharing, output-size and parameter formulas, plus prerequisites, math level, and time needed.

Preview of the CNN architectures lesson: the four landmarks you'll meet, the ImageNet numbers cited, prerequisites, and the training-at-scale subsection.

A roadmap to the 3D vision lesson: what you will learn, the depth cues and methods covered, prerequisites, the stereo-depth math, and timing.

What this t-SNE lesson covers: reading 2D cluster plots, what the method preserves and distorts, the perplexity dial, plus prerequisites and time to budget.

A map of the PCA lesson: what you'll learn about principal components, scree plots and loadings, plus prerequisites and where it fits in unsupervised learning.