I'm delighted to welcome Philip Quirke to Episode 5 of Cozy Catchups!
This episode, we've invited Philip to talk about his work on interpretability. I’ll be asking Philip Quirke about his work on addition in transformers and reusing circuits for more reliable AI.
After I've asked my questions, then it'll be your turn to inquire about anything that's been on your mind 
If you’re keen on doing a bit of homework before the episode, check out the following:
* Understanding Addition in Transformers: https://arxiv.org/abs/2310.13121
* Increasing Trust in Language Models through the Reuse of Verified Circuits: https://arxiv.org/abs/2402.02619
No registration required! Simply click on the provided link on the day of the event and you're in.
Mark your calendars so that you don't miss out!