MATS 9 Retrospective & Advice

I couldn’t find a recent write-up from a MATS alum about what attending MATS was like, so this is the thing that I wish I had. I attended MATS from January to March 2026, on Team Shard with Alex Turner and Alex Cloud. It was a great time! Applications for MATS are basically on a rolling basis nowadays, and I can strongly recommend applying (to multiple streams) even if you think you’re not a great match.With that being said, there’s a lot I wish I knew going into MATS, so here’s a brain-dump of thoughts. It’s not extremely polished, but I expect it’ll be useful nonetheless (none of this is endorsed by MATS, just my thoughts):Work ethicI think most mentees were working 10-12, sometimes 14 hours a day Mon-Fri, and probably 2-8 hours on Saturday and Sunday, often going out on some adventure or party on the weekend. Exactly which hours people worked varied wildly. I usually worked 8:30am/9am to 11pm/midnight, with breaks during the day, others worked from midday into the early hours of the morning. This was surprisingly sustainable (IMO); MATS puts a lot of effort into removing all other blockers that you normally have in life so that you don’t have to do other things.Use more computeMATS was a lot about learning to go fast, and I kept on thinking I had a good strategy and then learning this wasn’t enough.Initially I struggled to run enough experiments, but then I realised I should structure my day around those glorious hours between midnight and ~8am when I would be asleep but I could leave a sweep running so that I’d wake up to results.But then I realised that MATS was paying for compute, and also that I could either run 2 GPUs over 8 hours or 32 GPUs over 30 minutes. So I started running 32 GPUs over 30 minutes and got a lot faster progress. And why not? It was a great time. (Runpod makes it a little painful to do this, see my notes about W&B sweeps for making this easier)Research requires a lot of computeWe went through a lot of compute (as measured in dollars), rel