Nathan Lambert
speaker
1,013 appearances
1 recordings
1 series
first heard Nov 2024
last heard Nov 2024
Nathan Lambert’s voice in public audio — every appearance, attributed to the second.
Trend
recordings per month · last 12 monthsNo recordings in the last 12 months.Older appearances are listed below; set an alert to hear about the next one.
Appearances
This was specific.
Yeah, yeah, it's like to one math domain.
But to be clear, this is like we trained it for way longer than it's practical.
So like at this point of training, normal evaluations have totally tanked.
So the model is like not as good at normal things.
It'll still converge to good math answers in this one specific eval that we're training on.
But it's like its whole chain of thought process kind of like got borked.
And it was fun it's just funny when you see the same keyword that everyone was
viral about with open IL one where it's like, wait, let me check that and I was just like
It's so funny, but I'm in the same way as it's like not that surprising.
It it's
It seems like an RL thing to do.
The other examples when O one came out is like, Oh look, it changed to French and then it changed back.
That's just like
To me it's just like models doing really funky things is like not surprising that it's
the R L part of it. 'Cause like how like none of our loss functions otherwise are encouraging such like weird shit.
But it also it's like an N of one.
So it's like ad it's much more just for fun and for piecing things together over the long term than definitively like the that model is anything like what O one is doing.
I do think that
the same type of training approach probably applies where they have some they have some sort of verifier and then they do a lot of R L.
Showing 821–840 of 1,013 · page 42 of 51
← Previous
Next →