Educational Cards

Learn from video content, text, and interactive tasks

Filters
Clear
Debugging gradients in practice

A wrong backward implementation can train for hours before NaNs appear. Standard debugging starts...

Intermediate Machine learning
Handoff to the calculus-heavy walkthrough

This chapter kept the graph picture intuitive. The next chapter slows down and writes partial...

Intermediate Machine learning
Bookkeeping partials on a small lattice of variables

Backprop at scale is still the chain rule; the hard part is bookkeeping . Assign a symbol to every...

Intermediate Machine learning
Worked tiny graphs before code-size graphs

Before trusting a million-line framework, trace a three-node or five-node graph by hand. The...

Intermediate Machine learning
JVPs and VJPs without_full Jacobians

A full Jacobian for a layer mapping mathbbR^4096tomathbbR^4096 has roughly sixteen million entries....

Intermediate Machine learning
Log-space and stabilization patterns

Softmax exponentials can overflow when logits are large and underflow when logits are very...

Intermediate Machine learning
Higher-order vs first-order in deep learning

SGD and Adam use first derivatives only. Newton methods incorporate curvature via the Hessian,...

Intermediate Machine learning
Transition to scaled autoregressive modeling

The calculus hygiene on small graphs parallels the massive tensor programs inside language models:...

Intermediate Machine learning
Autoregressive core loop: predict the next token

Large language models treat text as a sequence of discrete tokens (subwords or bytes). At each...

Intermediate Machine learning
Scale: data, compute, and emergent capability patterns

Empirically, validation loss often improves predictably as model size, data, and compute grow....

Intermediate Machine learning
Alignment after pretraining

Pre-training imitates internet-scale text, including toxicity, bias, and unsafe instructions....

Intermediate Machine learning
Hallucination and overconfidence

Fluent prose can assert false specifics with high apparent confidence. Models optimize plausibility...

Intermediate Machine learning