APPENDIX H: RESUME PORTFOLIO BUILDING CAPSTONE PROJECTS

PREREQUISITES AND SETUP You need Python 3.10 or newer, a text editor, a terminal, and a web browser for viewing the dashboard. You need no packages and no paid account. Everything is built inside a new folder named project4_eval_suite. The finished folder will contain tracing.py, target_harness.py, models4.py, make_dataset.py, scorers.py, run_eval.py, dashboard.py, dataset.json, one results file and one traces file for each model, and dashboard.html. Hypertext Markup Language (HTML) is the language that web pages are written in, and our dashboard is a plain HTML file that you open by double-clicking it. The mock models are ordinary Python functions in models4.py, and to grade a real model you would add a function to the MODELS dictionary that calls your provider and returns the same dictionary shape, remembering that real results vary from run to run, so you would run each case several times and report the range.