Weights training can be overly specific to individual muscles. And heavy weights have more intertia in general. So the solutions are:

• do dumbbell complexes that involve a range of movements rather than overly focused muscles. Train all the small / weak muscles around the big ones using more functional approaches
• don’t over think it, heavier weights are less maneuverable, that’s physics
• if it’s really a mental problem, do more sets of lighter weights. This works too, and increases your endurance rather than your bulk. So maybe not “big number ego friendly” but instead “lots of reps ego friendly”. Upside is you get to spend more time in the gym having fun

Already happened with Cerebras and they landed at a stable 2x of the open. SpaceX is probably going to be next & set the real standard

Print it out into a series of QR codes on cards, and then store it in a binder with your old pokemon cards. Make sure to draw a pentagram in smeared ash somewhere so that it’s demon proof

Thanks mr brave wand!

oh no they only have 3.5 billion users how will meta ever survive, their ability to take on debt must mean they are seen as unable to pay debts

What I’m saying is that’s actually very hard unless you run a super sterile account on purpose. Even just your writing style is a pretty good fingerprint. Your IP. Any pictures you’ve posted. It’s a rough world out there for privacy

It’s not that hard to identify people online. My account is definitely not private

It’s occasionally worth calling out that votes are also public. I think twice before hitting those buttons

I’d say there isn’t a difference here, which is why they are both on the right side in the chart. The survey probably just had that option to appease people who have the opinion “but that’s it’s not a matter of morality at all!”

Just use pacman, it’s fine, and less fiddly than manually installing a binary to your PATH

Haha yeah college in Quebec is free for Canadians so I mean, not a bad idea

There’s only one great choice and that’s CLRS (“intro to algorithms”) - if you know this book & did the exercises, you can just skip college

Well how you gonna buy a new Mac if you could get security patches for your old device

What?

It’s the roomba of the sea

Windows now Chromebooks?

Literally anywhere on Google. But it also makes sense when you think about ChromeOS & non-us aligned countries - what else are they gonna use?

HEY BUT DO YOU WANT TO USE A PASSCODE?? PASSCODE! PASSCODE! USE THE PASSCODE! -_-

• he’s not anthropic, and doesn’t have billions of dollars
• stealing from open source is not stealing, that’s the point of open source
• the argument above is that these models are allegedly trained “regardless of license” i.e. implying they are trained on non-oss code

I don’t get it, it’s just a markdown file, which is how llm specs already get written. I’m supposed to trust that the llm will “compile” it correctly every time without touching the code output? What happens when the code gets huge and complex, a “recompile” entails waiting for ages to get something similar to what others compiled? What am I missing?