Protocol
PyTorch Training That Can Be Replayed
Package a small training program that survives interruption and runs on CPU. Treat GPU support as an optional measured acceleration, not an unstated requirement.
30–50 active hours60–90 min sessionsIntermediate
Mastery contract
Core protocol · Standard 1.0 · Package a small training program that survives interruption and runs on CPU. Treat GPU support as an optional measured acceleration, not an unstated requirement.
Loading your learning records…
Reasoning — not yet passed
Trace parameter, gradient, optimizer, random-generator and model-mode state before and after a training step; demonstrate a mode-versus-gradient-recording counterexample.
Reliable implementation — not yet passed
The port matches the NumPy reference within stated tolerances and restores all necessary checkpoint state.
Reproducible experiment — not yet passed
A deterministic CPU fixture reproduces its resumed trajectory; any nondeterministic operation or device difference is disclosed.
Defense and handoff — not yet passed
Deliver a runnable training command, environment record and checkpoint recovery test that another person can execute.
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
Explain the graph and gradient lifecycle for 1 training step, including zero_grad, backward and optimizer.step, and contrast train/eval with no_grad.
Port the NumPy MLP to PyTorch; compare forward values and gradients, then implement a checkpoint containing model, optimizer, configuration and random states.
Run 3 seeded trials and an interrupted/resumed CPU run; compare learning curves and state the limits of reproducibility across devices.
Debug forgotten gradient clearing, validation in training mode and tensors on mismatched devices.
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'.