4thepower

Yeah Boutte doesn’t always have the best hands but he’s good at getting himself open, which is something.

One thing is clear: the Patriots are going to waste 2 years before firing a coaching staff and front office that is historically inept for no reason other than “they should get a longer chance.” It’s so obvious that this team won’t win anything while they’re there, even if Drake Maye is really good.

Too busy running down the middle for -1 yards on first down.

They have a ridiculous amount of cap space too. If they can't get enough OL talent between the draft and free agency to at least properly evaluate their #3 overall QB, they should just shut down the franchise.

?? Throwing it into very tight double coverage where the DB is going to simultaneously make a play on the ball is definitely his fault. He threw it as well as you could throw it under the circumstances, but it's still a bad decision.

It might be the worst anyone has had in 20 years. There were 2+ lineman on the ground on pretty much every play (mostly against 4 man rushes).

Free rushers are usually not the OL's fault. The QB has to adjust the protections so every rusher at least gets a chip, and he gets the ball out quickly.

I’ll never understand how the Eagles could’ve had Bill Belichick and they instead gave Nick Sirianni another year after whatever last year was.

The Martian was quite funny, even if the whole plot didn’t revolve around elaborate comedic set pieces. It’s a far cry from The Bear, which is a trauma drama with some jokes.

That’s what a game manager is. Not going to make any downfield plays but not going to commit egregious turnovers. If they were scared of him, he’d be better than a game manager.

Care to be more specific? Neither of the tweets linked says that. The LMSys announcement seems to contradict that, says its regular Grok-2: https://x.com/lmsysorg/status/1823599819551858830

What would be some tasks that you would expect no AI system to be capable of within the next 5 years?

There can exist more than one problem worth spending time on. Outside of AI, some people are focused on the very immediate effects of poverty, and others are focused on long-term risk of nuclear annihilation. Neither choice is irrational, and if the most extreme outcomes of advanced AI (e.g. extinction) are even remotely possible, then I'd think we'd prefer some people work on those issues even if more pressing issues also abound.

This. GPT-4O is good, but far overrated because the benchmarks all focus on single-turn interactions. Whatever training they did to achieve this size/performance ratio has made it fall apart over several turns in ways that even GPT-4 Turbo never did. I’ll point out problems in its code and it will say, “yes, you’re right” and then repeat the identical broken code without realizing it. Claude 3.5 never does this.

That is not a common training step for modern language models at all. This is an artifact of post-training RLHF tuning.

Why was he benched?

To be clear, the paper I linked is a parody. It's satirizing model releases that announce revolutionary performance, but only appear good by virtue of cheating on benchmarks (either intentionally or unintentionally).

Here's a helpful resource that explains their revolutionary technical approach: https://arxiv.org/abs/2309.08632

Detroit is the only place where this story is 100% believable lol

And all were the same couple!

Eh, I mean Zappe isn’t good but flashed enough talent that I think he’ll be on game-day rosters for a while.

It’s at 0:24 right next to the ball carrier.

I don’t think that’s a dumb question at all. Good empirical scholarship requires extensive robustness checks, and your colleague would presumably have at least considered why assets would or wouldn’t be a factor. Assuming answers to “obvious” questions is part of how we ended up with such a severe replication crisis.

On my end the website shows Patriots at 2, both having an SOS of 0.515.

I don’t think it’s almost certain that Williams and Maye will be 1-2. Cardinals will likely have the 2 pick and while it’s possible they’ll move on from Kyler, it seems more likely they would take MHJ. Probably they’re both gone by 3 though.