the einsum string is a readable proof of a tensor operation, the 0.0 exercises cost 45 minutes and buy fluency for every later chapter, and the route treats this part as verification, not study

convolution is a window that walks and pooling is the vote that follows it, a residual stream is memory that tames gradients, and the eleven exercises end with a ResNet34 verified on real images

plain gradient descent overshoots the steep axis and crawls along the flat one, momentum damps the first and accelerates the second, and four exercises make you implement the optimizers you will later trust

the transformer decomposes into ten exercises around a single residual stream, attention is the only new idea inside the block, and ARENA marks this part compulsory for everything chapter 1 later interprets

the part's one move is instrumenting a trained transformer with hooks, attention patterns and ablations, the induction head is the first circuit you can find attribute and ablate end to end, and ARENA marks it compulsory alongside 1.1

an eval starts as a threat model and a specification, not as a quiz, the week's exercises build one end to end from claim to scored questions, and the reading thread reaches agent collusion in electricity markets

the week turns a specification into a scored dataset with prompts variance and concurrency, the scoring function gets built before the data it will grade, and the reading thread reaches the founding paper on algorithmic collusion

the week replaces hand-rolled glue with the inspect_ai pipeline of datasets solvers and scorers, solvers are where the eval's actual logic lives, and the log file is the deliverable that makes results checkable

the week builds an agent twice, once as arithmetic and once as a Wikipedia game, the loop of think act observe is the whole architecture, and tool permissions are where safety enters the picture

AI control evaluates safety measures against an intentionally subversive model, the week plays both roles with backdoor generation on one side and monitors plus protocols on the other, and the W44 track-choice gate rides on this fluency

the week runs real API experiments on shutdown resistance and alignment faking, the first alignment-science results land as numbers from your own pipeline, and the reading thread reaches persona features and emergent misalignment

the last core part audits whether a model's chain of thought reflects its actual computation, the exercises build faithfulness measurements from categorization to counterfactual importance, and the program's proposal gate closes the week

the route deliberately leaves one week unpicked between SAE literacy and intro to RL, both candidates are single-session light parts with no later hard dependency, and the choice belongs to the track decision made at the W44 gate

saving is not a prompt · every reflection must quote you · one Cloud Run service, Firebase, Firestore, Gemini

QUESTION → DATA CONTRACT → DUCKDB + POLARS → 3 AUDITED CHARTS → ESSAY