@svpino Fill same, eng started producing with unbelievable speed even when qa step fell partly on them instead of qa. And here is the question do we need that speed?)
As someone who's been shipping LLMs since the GPT-2 days, this lecture on cross-entropy from a Stanford math grad is the closest thing to an ML PhD qualifying exam I've ever seen released publicly for free.
Everyone thinks language models predict the next word. They don't. They compress language. Once you see the math, you can't unsee it.
33 minutes. Bookmark & watch today.
I still feel like I’m sitting somewhere in 2006 with a Nokia N95 in my hand,
thinking this is the best technical invention available to me.
so now feels more like the advanced smartphone phase:
the phone no longer just rings,
it already does a lot,
but nothing truly major has changed yet.
The AI era has begun.
And I’m truly waiting the iPhone moment.
@cgtwts Love it, my set up for last month: morning brief (tasks+any input like email), analytics summary, daily snapshot.
What I’m missing is every time I need cowork/mac open to process operation, how you deal with it?
@trq212 Agreed, like, do same. But few things if human need to read the output yes 100% html brings that, but if the output is transition between agents where human just correct/add, then probably no so good to use it, just my assumption.
And html consume lots more tokens, how u deal?
@gkisokay@Scobleizer Thinking to make it active around my new home assistant green parallel with some simple daily task like analytics measuring, social networks scraper and etc