Media Library
44 free resources, each with a note on why it is here and what to watch for. Every video id was checked against YouTube and every link was fetched on 30 October 2026. Videos load only when you tap play.
Showing 44 of 44
The cleanest picture of a network as layers of numbers, weights as knobs, and a layer as a matrix-vector product followed by a nonlinearity.
Watch for: The moment the weights are written as a matrix W times a vector a, plus a bias b. That single line is the equation inside every dense layer you will meet on arXiv.
Verified: YouTube oEmbed title+channel check, 2026-10-30
Makes the cost function concrete and shows the gradient as a direction in a space with thousands of dimensions.
Watch for: How "the negative gradient" is presented as a list of nudges, one per weight, ranked by how much each nudge matters.
Verified: YouTube oEmbed title+channel check, 2026-10-30
Explains how the gradient is actually computed: blame flows backward from the loss through each layer.
Watch for: The difference between what gradient descent does (take a step) and what backpropagation does (compute which step).
Verified: YouTube oEmbed title+channel check, 2026-10-30
Optional depth: the chain rule written out with ∂ symbols. Watch it after the Rosetta entries for ∂ and ∇.
Watch for: Each ∂ fraction is a sensitivity: "if this moves a little, how much does that move?" The chain rule multiplies sensitivities.
Verified: YouTube oEmbed title+channel check, 2026-10-30
Embeddings as directions, softmax with temperature, and the unembedding step that turns a vector back into token probabilities.
Watch for: The softmax-with-temperature segment. Compare it with the Token Probabilities interactive in Module 6.
Verified: YouTube oEmbed title+channel check, 2026-10-30
The best visual walkthrough of Q, K, V and the attention pattern as a grid of dot products.
Watch for: Where the √d_k scaling and the softmax appear. Then open the attention equation in the Equation Inspector.
Verified: YouTube oEmbed title+channel check, 2026-10-30
Bridges to interpretability: directions in high-dimensional space as features, and why superposition lets a model pack in more features than it has dimensions.
Watch for: The explicit hedges. The video separates what interpretability research has shown from the illustrative toy story it tells.
Verified: YouTube oEmbed title+channel check, 2026-10-30
Three views of a vector (arrow, list of numbers, abstract object) and why ML mostly uses the second while reasoning with the first.
Watch for: The "list of numbers" view. An embedding is exactly that, just with thousands of entries.
Verified: YouTube oEmbed title+channel check, 2026-10-30
A matrix is a function that moves space around. That reframing makes "weight matrix" stop sounding like a spreadsheet.
Watch for: Columns of the matrix as "where the basis vectors land".
Verified: YouTube oEmbed title+channel check, 2026-10-30
Why multiply-and-add measures alignment. Attention scores are dot products.
Watch for: The projection picture. Pair it with the Vectors and Dot Products interactive.
Verified: YouTube oEmbed title+channel check, 2026-10-30
Integrals as accumulation and derivatives as rates, introduced through one geometric question rather than rules.
Watch for: The integral emerging as a limit of sums. That is the Σ→∫ bridge in the Rosetta Stone.
Verified: YouTube oEmbed title+channel check, 2026-10-30
What "instantaneous rate of change" means without hand-waving.
Watch for: dt as a small but finite nudge. That is how to read dx and dy in physics and ML papers.
Verified: YouTube oEmbed title+channel check, 2026-10-30
Why e and the natural log show up everywhere: growth proportional to current size.
Watch for: The derivative of e^t being itself. Softmax, Boltzmann weights and cross-entropy all lean on this.
Verified: YouTube oEmbed title+channel check, 2026-10-30
Bayes as restricting the space of possibilities, drawn as areas.
Watch for: The "Steve the librarian" example. It is a base-rate lesson in disguise.
Verified: YouTube oEmbed title+channel check, 2026-10-30
Bayes factors as "how much should this evidence move me". That is the most useful form for reading claims.
Watch for: Odds × Bayes factor. Use it on any "new study shows" headline.
Verified: YouTube oEmbed title+channel check, 2026-10-30
Shannon entropy built from scratch on a game you already understand: bits as halvings of the possibility space.
Watch for: The formula −Σ p log p appearing as "expected information". It is the same expression as in the Entropy interactive.
Verified: YouTube oEmbed title+channel check, 2026-10-30
Optional depth on the bell curve, μ and σ, and what the normalising constant is for.
Watch for: e^{−x²}: the shape is set by the exponent, and everything else is bookkeeping so the total probability is 1.
Verified: YouTube oEmbed title+channel check, 2026-10-30
Frequencies as the hidden ingredients of a signal. The bridge from λ = c/f to spectra in physics and positional encodings in transformers.
Watch for: The integral as "average position of a wound-up graph". It is another sum in disguise.
Verified: YouTube oEmbed title+channel check, 2026-10-30
Thermodynamic entropy as a count of arrangements, not "disorder". The right setup for comparing it with Shannon entropy.
Watch for: Where energy spreading becomes probability. Hold that in mind when Module 4 puts Boltzmann and Shannon side by side.
Verified: YouTube oEmbed title+channel check, 2026-10-30
No prerequisites, and it covers dynamics, chaos, fractals, information, self-organisation, agent-based models and networks, the same toolkit as Module 5.
Watch for: The units on networks and scaling. Notice how often Mitchell separates "this is a model" from "this is how nature works".
Verified: HTTP 200 + page content checked, 2026-10-30
The most careful free treatment of power laws. It includes how to estimate an exponent and why straight lines on log-log plots are easy to over-read.
Watch for: The sections on fitting power laws. They are the antidote to "it is a power law, therefore it is universal".
Verified: HTTP 200 + page content checked, 2026-10-30
A short, gentle tutorial from bits to entropy, with complexity-science applications.
Watch for: Entropy defined over a probability distribution. Always ask "entropy of which distribution?"
Verified: HTTP 200 + page content checked, 2026-10-30
The reference course. Use individual lectures (matrix multiplication, eigenvalues, SVD) as a library, not a syllabus.
Watch for: Lectures on eigenvalues and the SVD. They are the vocabulary behind "low-rank", "spectrum" and "subspace" in ML papers.
Verified: HTTP 200 + page content checked, 2026-10-30
Short concept videos on forces, energy and conservation, if you want to go deeper than Module 2.
Watch for: How every problem begins by choosing a system and listing what is conserved.
Verified: HTTP 200 + page content checked, 2026-10-30
Short recitation videos on conditional probability, Bayes and expectation. Use the ones that match a gap.
Watch for: Conditioning as shrinking the sample space.
Verified: HTTP 200 + page content checked, 2026-10-30
Lecture-level depth on DNA, gene regulation and cell signalling for Module 3.
Watch for: Gene regulation lectures: transcription factors as switches whose wiring forms networks.
Verified: HTTP 200 + page content checked, 2026-10-30
Graduate-level reference. Skim the lecture notes on entropy and the Boltzmann distribution to see the "real" notation you will meet in papers.
Watch for: Z, the partition function. It is the normaliser of softmax under another name.
Verified: HTTP 200 + page content checked, 2026-10-30
Targeted gap repair: linear equations, functions and graph reading. Use the unit tests to find gaps fast and skip what you know.
Watch for: Only the "functions" and "exponents & radicals" units are essential here.
Verified: HTTP 200 + page content checked, 2026-10-30
Logarithms are the single most important "missing" topic for reading scaling-law papers.
Watch for: The logarithm units. Log turns multiplication into addition and power laws into straight lines.
Verified: HTTP 200 + page content checked, 2026-10-30
Gap repair for distributions, sampling, significance and study design.
Watch for: The study-design unit. Observational studies vs experiments is the correlation/causation line.
Verified: HTTP 200 + page content checked, 2026-10-30
Free PDF. The best single reference for ML notation: vectors, matrices, gradients, probability.
Watch for: Use the index as a dictionary. Look up a symbol when a paper uses it rather than reading front to back.
Verified: HTTP 200 + page content checked, 2026-10-30
Free PDF plus notebooks. The clearest modern bridge from equations to architectures, including transformers.
Watch for: Chapters on loss functions and on transformers. The notation table at the front is worth bookmarking.
Verified: HTTP 200 + page content checked, 2026-10-30
Free to read online. Volume I chapters on energy conservation and on "the relation of physics to other sciences" are perfect for this curriculum.
Watch for: Feynman’s habit of saying exactly which approximations are being made.
Verified: Page loaded in browser, 2026-10-30
An interactive article on optimisation dynamics, and the model for how a good interactive explainer behaves.
Watch for: The step-size slider. The same oscillation you see in Module 6 appears here with theory behind it.
Verified: HTTP 200 + page content checked, 2026-10-30
Visual, interactive probability and statistics, from chance events to Bayesian inference.
Watch for: The chapter on Bayesian inference, as a second view on the Bayes interactive.
Verified: HTTP 200 + page content checked, 2026-10-30
An explorable version of Schelling’s segregation model: mild individual preferences produce strong collective patterns.
Watch for: Emergence without intention. Then ask what evidence would show a real city works this way. The model alone cannot.
Verified: HTTP 200 + page content checked, 2026-10-30
Diagram-first walkthrough of the 2017 Transformer, matrix shapes included.
Watch for: Matrix dimensions at each step. Shapes are the fastest way to sanity-check an equation.
Verified: HTTP 200 + page content checked, 2026-10-30
Research-tested physics simulations (forces, energy skate park, waves) for intuition beyond Module 2.
Watch for: "Energy Skate Park" and "Wave on a String" pair directly with this curriculum.
Verified: HTTP 200 + page content checked, 2026-10-30
The editorial benchmark for this curriculum: serious science, precise about what is shown versus hoped.
Watch for: Read with the Skeptic’s Toolkit open. Label each claim with the evidence badges used here.
Verified: HTTP 200 + page content checked, 2026-10-30
An explicitly physics-flavoured interpretability paper: phase diagrams, geometry and phase changes in tiny models.
Watch for: How the authors label what holds in toy models versus what they conjecture for large models.
Verified: HTTP 200 + page content checked, 2026-10-30
Sparse autoencoders applied to a production model, with scaling laws for the dictionaries themselves.
Watch for: The limitations section. It is a model of honest scoping.
Verified: HTTP 200 + page content checked, 2026-10-30
A design reference for this curriculum: physics, chemistry, molecular biology and computer science taught as one quantitative course.
Watch for: The premise that shared quantitative tools, not departmental boundaries, organise the science.
Verified: Page loaded in browser, 2026-10-30
A design reference: an undergraduate program combining physics, biology, chemistry, mathematics, statistics and computation.
Watch for: How the program sequences math alongside the sciences rather than before them, as this curriculum does.
Verified: HTTP 200 + page content checked, 2026-10-30
The origin of H = −Σ p log p. Sections 1 to 6 are readable with Module 4 in hand.
Watch for: Shannon’s own remark that his H has the same form as entropy in statistical mechanics. He notes the form, not a shared mechanism.
Verified: HTTP 200 + page content checked, 2026-10-30