Here we will give both leaderboard and a link to a script which does the following:
Run 5 prechosen sample CYs and apply the following criteria:
- No more than 2 hours time per run (we will run submissions on our cluster).
- Do each run twice with two choices of random seed and keep the worse one.
- All runs must meet some minimal accuracy (e.g. 1% MAPE).
- Rank the qualifying entries by average log MAPE (over all samples).