Module 8 Assignment: Reproducible training workflow checklist

Module 8 Assignment: Reproducible training workflow checklist#

Theme#

GPU workflows, scale, and deployment

Scenario#

A prototype model is ready to leave a notebook, and the team needs a repeatable training job with accountable metrics and resource assumptions.

Exercises#

  1. Write a reproducibility checklist for a deep learning experiment.

  2. Record environment, device, random seed, data split, metrics, and checkpoint assumptions.

  3. Use the starter cell to report CUDA availability and tensor placement.

  4. Explain how to scale the experiment from exploration to a repeatable job.

Evidence Requirements#

  • A short explanation of the data, tensors, objective, and evaluation signal used in the starter experiment.

  • At least one meaningful modification to the starter code, with the changed variable named explicitly.

  • A comparison against the unmodified starter result or another defensible baseline.

  • A limitation statement that separates what the toy experiment demonstrates from what a production model would require.

Submission#

Submit a 600-900 word technical memo plus code, plots, tables, or shape traces needed to support your claims. The memo should read like a review artifact for another AI practitioner: concise, reproducible, and honest about uncertainty.

Rubric Focus#

  • Technical correctness and appropriate neural-network vocabulary.

  • Evidence from the starter experiment or a documented extension.

  • Connection between design choices and data/problem structure.

  • Clear treatment of limitations, failure modes, or next experimental gates.

import torch

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
x = torch.randn(4, 4, device=device)
print("torch version:", torch.__version__)
print("device:", device)
print("tensor device:", x.device)
print("Set seeds, log package versions, and save configs before claiming reproducibility.")

torch version: 2.12.1+cu130
device: cpu
tensor device: cpu
Set seeds, log package versions, and save configs before claiming reproducibility.

Reflection Prompts#

  • What changed when you modified the starter experiment, and why should that change matter?

  • Which result surprised you, and what diagnostic would you run next?

  • What assumption would you document before handing this model to another practitioner?

  • Which failure mode from the module reading is most relevant to your result?