Protocol
Sequence Models and Attention Mechanics
Build a sequence retrieval benchmark with controllable dependency distance. Change the distance distribution after training and explain the observed failures.
35–60 active hours60–90 min sessionsIntermediate
Mastery contract
Core protocol · Standard 1.0 · Build a sequence retrieval benchmark with controllable dependency distance. Change the distance distribution after training and explain the observed failures.
Loading your learning records…
Reasoning — not yet passed
Trace every attention tensor dimension on a worked sequence and derive the recurrence gradient path; demonstrate a gated configuration that still fails to optimize.
Reliable implementation — not yet passed
Changing a future token cannot alter an earlier causal output; padded keys receive zero attention within numerical tolerance.
Reproducible experiment — not yet passed
Length comparisons report failure rates, sequence lengths and compute, not only a single aggregate loss.
Defense and handoff — not yet passed
Defend which architecture solves the synthetic task and identify why that task is weak evidence about natural-language competence.
All four criteria must pass. Activity completion does not satisfy them.
An independent reviewer reruns your work and varies something you did not rehearse: a fresh input, a different working directory, or a declared edge case.
Pilot decisions are provisional, not external credentials. For an appeal or sensitive artifact, contact your designated pilot operator with the submission ID. Export your assessment history.
What you will do
6 total
Derive scaled dot-product attention and explain the square-root dimension factor; unfold a recurrent gradient over 3 steps.
Implement a simple RNN and attention in PyTorch without a high-level attention layer; test padding and causal masking on hand-computable examples.
Compare recurrent and attention models on a fixed synthetic copy/retrieval task at 3 sequence lengths with matched evaluation cases.
Debug a causal mask with reversed polarity and a padding token that receives probability mass.
Write a 500–900 word technical report linking the reasoning, code, measurements, and limitations; include the project decision below.
Assess all 4 mastery criteria against saved artifacts, request an independent review, and repeat each failed criterion on a fresh example.
Essentials
A reading session alone never satisfies a mastery criterion.
Store source references, assumptions, and artifact paths beside every result.
Keep an untouched check case that differs from the worked example.
Record all attempts, including failures and results that contradict your prediction.
Use the stated comparison conditions; document every deviation before drawing a conclusion.
Protocol guardrails
Do
- +State the expected result before running the comparison.
- +Keep one minimal reproducible failing case when debugging.
- +Record environment versions and the exact command used.
- +Compare explanations with saved intermediate values.
- +Ask a reviewer to challenge the weakest assumption.
Don't
- ×Do not copy a worked solution and present it as an independent implementation.
- ×Do not tune against held-out evaluation outcomes.
- ×Do not report only the best seed or discard inconvenient runs.
- ×Do not equate elapsed hours or a completed run with a passed assessment.
- ×Do not conceal reduced-scale experiments behind claims about the original full-scale result.
Protocol authorship
Written by NuthinButta as instructional design. The teaching sources are the official references linked in each milestone; the exercises, workload and pass thresholds are ours, not their authors'.