00:01:56	Oliver McLaughlin:	Arjun remains silent...
00:02:17	Arjun Khurana:	https://strawpoll.com/XOgOVxDabn3
00:03:18	Emma Bortz:	https://strawpoll.com/XOgOVxDabn3
00:03:19	Oliver McLaughlin:	ZOOM IN BROWSER 🗣️🗣️🗣️🗣️
00:03:36	Oliver McLaughlin:	Does zoom have pins like google meet?
00:21:32	Arjun Khurana:	https://www.science.org/doi/10.1126/science.220.4598.671
00:22:18	Oliver McLaughlin:	I think Nesterov developed a "momentum" sort of idea that (9) is a direct descendent of in like the early 80s
00:26:25	Oliver McLaughlin:	Corollary: Autodiff as a tool may be a really good subject for a design study
00:29:46	Eric Todd:	https://colab.research.google.com/github/ericwtodd/rumelhart1986/blob/main/family_tree.ipynb
00:29:47	Oliver McLaughlin:	I WOULD LIKE TO SEE THE NOTEBOOK THIS FIGURE WAS CONFUSING TO ME
00:36:33	Oliver McLaughlin:	No causal intervention no probes no dictionary learning just vibes
00:50:35	Emma Bortz:	Paper PDF here: https://papers.baulab.info/papers/Rumelhart-1986.pdf
01:00:35	Emma Bortz:	https://luma.com/ji8qc5n3 
Next reading group event!
01:03:31	Emma Bortz:	Discord link: https://discord.gg/J3QvfykWKk
01:03:31	Jac Zac:	I think, just to reiterate something that has already been said, it is interesting to note that, on the same camp as “X counts as evidence for Y,” the general feel of the paper is that backprop’s internal representations count as evidence for the fact that backprop is indeed a reasonable approach.