Large Language Models and Transformers (a workshop this week)

[dmm]

simons.berkeley.edu/workshops/large-language-models-transformers

Youtube livestream (and, presumably, post-conference youtube recording) are available.

Some of the talks look really interesting.

Comments

[dmm]

Via Scott Aaronson's review of Barbenheimer, of all things: https://scottaaronson.blog/?p=7422

[dmm]

Best talks today: panel, Ilya, Manning

[dmm]

I am late, though, still in the middle of day 2... I will gradually work through this material, but I can't do this real time (too many things are happening)...

[dmm]

Still in the middle of day 2 (although I went ahead and listened to a nice talk by Paul Christiano from day 3), but Colin Raffel's talk (day 2) is very, very interesting, so I want to note that.

[dmm]

Finally done with day 2. Adam Kalai talk is also very interesting. So Raffel and Kalai - the two "talks of the day".

Miles Cranmer talk was of a particular interest for me, because of close relations between what he is doing and what I have been exploring in my recent studies and my JuliaCon 2023 talk (so I might be interesting in both borrowing his techniques, but also in trying to beat his results with my potentially more efficient methods, and in general in various connections between his and my work).

Sasha Rush and Sanjeev Arora gave useful talks.

[dmm]

OK, looking ahead a bit. Just listened to Dan Klein's talk from day 4, and had to do some work to locate two sessions of short talks from day 5, here they are

session 1 (90 min): https://www.youtube.com/watch?v=c_UqS54ZA5o

session 2 (60 min): https://www.youtube.com/watch?v=9iSxF_RW6xk