Week 9 Guided session: Project 3 Task Contract and Baseline

Choose a permitted track and bounded target behaviour, write a valid task contract, and establish the unchanged starting-model baseline

26 April 2027 · Guided session

Estimated active time: 45 minutes · Release: Monday 26 April 2027 at 09:00 AET · Solution release: immediate completion checklist; no model solution.

Learning goals#

• choose a permitted track and one bounded target behaviour;
• write valid input, output, metric, and out-of-scope conditions;
• establish the unchanged starting-model baseline; and
• identify whether a Track A proposal is required.

Preparation#

Read the Project 3 page. Track B students inspect test_pilot; Track A students select a supported capability or begin the one-page alternative proposal.

Activity#

1. State the task contract, parser/validity gate, primary metric, generalisation slice, and regression metric.
2. Record the exact permitted starting checkpoint and run its unchanged baseline.
3. Inspect the proposed split at source/template level and run the separation validator.
4. Allocate the 12-hour/1.5×10^17-FLOP budget across dry run, comparison, final run, and recovery.
5. If proposing an alternative Track A task, freeze and submit proposal.md by Sunday 2 May at 23:59 AET; do not exceed the permitted 0.5-hour pre-approval dry run.

Expected output and self-check#

Retain one task-contract table, baseline report, split validation, and compute plan. The task is ready only if correctness can be distinguished from formatting and the generalisation case is not a renamed training template.