Week 1 Guided session: Project 1 Launch and Data Validation

Verify the supported environment and data checksums, and write a bounded Project 1 target before selecting an architecture

22 February 2027 · Guided session

Estimated active time: 45 minutes · Release: Monday 22 February 2027 at 09:00 AET · Solution release: immediate completion checklist; no model solution.

Learning goals#

• locate the canonical brief, starter repository, data card, and evaluation pack;
• verify the supported environment and data checksums;
• distinguish document-level training, validation, development, and tutor-evaluation material; and
• state one bounded Project 1 target before selecting an architecture.

Preparation#

Read the Project 1 page and the Week 1 lecture. Accept the assigned GitLab repository and confirm that no credential, private key, or HF token is stored in the working directory.

Activity#

1. Run the CPU environment check and record the installed framework/CUDA state.
2. Read the dataset card, verify the supplied checksum, and generate data_summary.json.
3. Inspect the document-level split and run the exact/near-duplicate separation validator.
4. Run the untouched tokenizer/data smoke test without beginning a principal training run.
5. Write a 120-word target statement naming the intended continuation behaviour, one failure to monitor, and the initial compute reservation.

Expected output and self-check#

Retain the environment report, data_summary.json, separation-validator output, and target statement. You are ready for Lab 1 only if the split is document-level, all flagged overlap has been resolved or reported, and the target describes observable behaviour rather than “get the lowest loss.”