HackerTrans
トップ新着トレンドコメント過去質問紹介求人

simsspoons

no profile record

投稿

Interview with Noam Shazeer on Character.ai

cms.megaphone.fm
2 ポイント·投稿者 simsspoons·3 年前·1 コメント

Noam Brown podcast on planning in ML

podcasts.apple.com
1 ポイント·投稿者 simsspoons·3 年前·0 コメント

Hard Reset, or how not to drive your startup off a cliff

sarahguo.com
3 ポイント·投稿者 simsspoons·4 年前·1 コメント

We Like the Runners

sarahguo.com
1 ポイント·投稿者 simsspoons·5 年前·1 コメント

コメント

simsspoons
·3 年前·議論
highly relevant question today
simsspoons
·3 年前·議論
How much intelligence in AIs is required to create emotional attachment? Noam says we’re pass that (per dogs)

Also makes sense given empirical evidence of Replika obsessions
simsspoons
·4 年前·議論
this is just great
simsspoons
·4 年前·議論
relax, it’s just a recipe experiment writeup
simsspoons
·4 年前·議論
The dexes are fine?
simsspoons
·4 年前·議論
Or maybe it’s the beginning of the end for centralized exchanges
simsspoons
·4 年前·議論
I think this is the macro. People are freezing hiring
simsspoons
·4 年前·議論
this is such a depressing book
simsspoons
·4 年前·議論
what % of startup founders need to hear this advice right now and are blind to reality
simsspoons
·4 年前·議論
that’s stupid
simsspoons
·4 年前·議論
Feels like the point is more about risk tolerance - whether someone is willing to accept lucky of the draw vs. “get shafted.” Most startups fail, but to be captain obvious lots of non-founders still get absurdly rich in startups.
simsspoons
·4 年前·議論
Re: running out of money in the current macro climate
simsspoons
·4 年前·議論
this is bad
simsspoons
·4 年前·議論
existential dread, dying without doing anything important
simsspoons
·4 年前·議論
or just run it on baseten https://app.baseten.co/apps/VqK2vYP/operator_views/pqvba2q
simsspoons
·5 年前·議論
Probably nothing, but there could be the kernel of something special here, maybe even a dominant new digital avatar and the beginning of a world. Chain Runners is a two-weeks-launched on-chain NFT generative art project, with 10,000 32x32 pixel images of cyberpunk inspired characters that live in Mega City. The images are generated as 11 layers, with 8 colors per layer. They are made by [mid, knav, braindraind, dozer], a team who has cred from Blitmap. After a quiet start, the project is blowing up, and (as of midnight on 11/22) is #2 for 24-hour volume on Opensea.