00:16:06 CNSP: Feel free to put any question in the Q/A and we can save them for the end of Marlies' talk. 00:18:56 CNSP: Also, I enabled chat for everyone. Sorry about that! 00:22:41 CNSP: Great demonstration there! 00:34:54 CNSP: Marlies has asked to hold questions to the end, so we (or you if you like) can ask at the end. Just raise your hand when she's done if you'd like to ask yourself! 00:36:34 Micha Heilbron: πŸ‘πŸ‘πŸ‘ 00:36:44 Giovanni Di Liberto: πŸ‘πŸ‘πŸ‘ 00:45:30 CNSP: Nice idea. 00:47:59 CNSP: πŸ‘πŸ‘πŸ‘ 01:17:25 CNSP: πŸ‘πŸ‘πŸ‘ 01:17:45 EDMUND LALOR: πŸ‘πŸ‘ 01:17:46 Mick Crosse: πŸ‘ 01:28:11 Katie Gray: Maybe musical phrase, vs chord progression, vs song structure 01:32:28 CNSP: Micha, here's my question: The example of syntactic predictions from GPT2 seemed more "confident" than other predictions. This makes sense to me, at least superficially, that grammar might be more constraining than meaning? If you looked at this, was that the case across other examples and was that reflected in neural activity? If not, maybe you can speculate? 01:35:24 Giovanni Di Liberto: Micha, here is another question. Is there a risk that "The old man and the sea" is part of the training set used for GPT-2? What consequences do you think that might have on your analysis? (I think it's still fine in your case, but it's good to keep this potential issue in mind, when using these models) 01:41:18 CNSP: Here's a good example of the high-frequency noise on the TRF we were talking about yesterday (and I was unable to produce with Subject 10s data) 01:44:44 CNSP: Ah yeah. That makes sense. Fair play. 01:45:01 CNSP: Thanks! 01:49:26 CNSP: Feel free to raise hands to ask questions or put them in the Q/A 01:50:04 Giovanni Di Liberto: p.s.: the web-based data browser is based on mTRFpy! https://web-browser-393510.ew.r.appspot.com/home 01:52:22 Micha Heilbron: πŸ‘ 01:52:30 Giovanni Di Liberto: πŸ‘πŸ‘πŸ‘ 01:52:31 Emily Ip: πŸ‘πŸ‘ 01:52:46 John O'Doherty: Thanks everyone! 01:52:50 Mick Crosse: πŸ‘ 01:52:52 Thaiz Priscilla SΓ‘nchez: πŸ‘πŸ‘ 01:52:55 Oskar Keding: Thank You! 01:53:04 Sara Carta: Thank you everyone! 01:55:27 Aaron Gibbings: I didn’t get the β€œAs you know” email last night, either. 01:55:30 Andre Palacios Duran: same here 01:55:39 Anton Rogachev: the same 01:55:48 Anton Rogachev: thanks!