Meh. I'm in episode 3 and it feels a bit forced and over the top. It feels like a less good version of 24.

Thanks Mark!

I'm not so cynical as to think that they just did it for a particular benchmark, but yes, I'm assuming it is a tuning of GPT4, not based on an entirely new pre-training run. My guess is that they have learned some tricks to make it get more out of increased test-time compute.

Thanks for posting the cost efficiency. I'm not seeing anyone talking about the fact that o3 Low costs an OoM more than o1, and o3 High costs three OoMs more. Yes, this is an impressive feat, but I'm beginning to doubt OpenAi's commitment to "intelligence that is too cheap to measure."

I think we need a great benchmark on instruction following. It should be relatively easy to measure. Someone just needs to put the time into creating complex prompts with measurable outputs.

#4 - Sure, a little obvious, but sure.

You missed this addition in #1:

Jamming toothpicks under my fingernails > Rings of Power > House of the Dragon

Glad you are here and raising the level of the debate!

I was going to say thank you for putting them in the proper order, but no! Maybe this isn't heresy, but it isn't orthodoxy either. Publication order is an historical fact. Live by truth, not falsehood!

Thanks for creating a fight worth having in this new world of Bluesky.

2/16/2011:
No one knows how they are choosing recipients. I didn't even get an e-mail. The computer just showed up on our doorstep. It cause a bit of concern for Wendi as to why I would buy a laptop without even mentioning it to her. Here's the URL for the application: www.google.com/chromeos/pil...

Umm . . . Yes to being given one of the first Chromebook prototypes. But for no reason I knew of. It just appeared on my doorstep. (I think I had said I would be a tester, but I got no communication about it.)

Wow, cannot get behind that one. It makes my stomach hurt just thinking about it.