Group Rules

Materials › Group rules

Rule 1 — Meetings

Meeting time with your supervisors is a limited resource — please respect it and make full use of it.

That means coming prepared, with slides ready in advance. Do not use the meeting itself to work out what you want to say. Prepare the deck following the weekly report template, and we will discuss from it.

No last-minute cancellations. Having no update is not a reason to cancel — come to the meeting anyway and we will talk about why the week stalled. If you genuinely cannot attend, tell everyone invited at least one day in advance. Illness is the only exception.

Rule 2 — Be in the lab

Come to the lab at least twice a week. Research is not a solo activity: being physically around your colleagues is how you pick up what others are working on, get unstuck quickly, and have the conversations that never happen over chat.

Rule 3 — Get API budget approved before you run

Before spending anything on API credits, send me a short budget proposal covering:

• which models you need, and how much budget each one takes;
• which experiments you are running, stated concretely;
• whether these are final experiments or still exploratory.

Wait for approval before launching the runs. No approval, no spend.

Rule 4 — Give a talk every three months

I care a great deal about students learning to present their work. It is a skill that has to be practised, and it should not wait until you have a paper to present.

So: every three months, give a talk to the group on where your research stands. A talk without a paper behind it is perfectly fine — a progress update, a half-formed idea, or a negative result all count.