isaiahbishop.bsky.social
ML Eng. and econometrics. Lot more left-posting than normal. Some hobby-level finance
Regrettably degen trading for the next 3 months, im sorry
Views dont reflect my employer
10,736 posts
1,938 followers
5,427 following
Getting Started
Active Commenter
comment in response to
post
we simply forget like with the auto parts or whatever
comment in response to
post
yeah i think its a weird one because i was like that on twitter (not with followers) so it takes a bit to figure out how to engage here
comment in response to
post
God, unbelievably divorced
comment in response to
post
We might be able to identify some of the most talented PR reps america has to offer
comment in response to
post
Im glad productivity is getting higher
comment in response to
post
We need to get mark back in the zone tbh
comment in response to
post
I mean its giving centrist voter for sure
comment in response to
post
happy to help 🫡🫡
comment in response to
post
Meta should totally 100% give me millions to run a control team to validate the 9 figure squad
comment in response to
post
Couldnt be me! Throw me at a smaller problem, im your guy. This aimless stuff? i cant gamble on a portfolio of research directions well
comment in response to
post
This is probably also true lol
comment in response to
post
A reason you would want like illya is that he worked pretty aimlessly at openai and found very fruitful directions.
comment in response to
post
Like im an ML dude, but if you put me on the AGI team i can try to be helpful but its not going to go great
comment in response to
post
oh infinity money and simply not that many people
comment in response to
post
If i am a top researcher, will i be leaving my extremely nice lab to join the next metaverse team? Depends on if someone bigger is willing to burn a lot of reputation!
comment in response to
post
In general, research is hard. Running a top level research firm is hard. There simply isnt that many people doing it with a track record of success after success, who arent also knee deep in equity
comment in response to
post
proof of work is actually a great way of explaining dead labour, now if only people understood proof of work
comment in response to
post
if it wasnt in the API its in the training data now
comment in response to
post
Really loving my new job in AI these days
comment in response to
post
USMNT is like toronto leafs level bad fandom as well, oof
comment in response to
post
Well we had the same deal broke for lunch, broke for supper, ended with same deal, so thats like 3 deals in a day
comment in response to
post
Average american: Who?
comment in response to
post
Right to the source, but honestly not sure what any of this will do
comment in response to
post
just vitamin D, in general not a huge believer in supplementation. Mostly trying to keep active, eat better and some cold baths (i think this one is bullshit but whatever)
Hope its working for you though!
comment in response to
post
Imagining the office dynamic where you are a 7 instead of a 9
comment in response to
post
getting old sucks because your body just finds new problems you have to live with forever randomly. I shouldve done more drugs earlier
comment in response to
post
i think in general thats right but ML research at that level is pretty specific and combined with some dev chops. I guess its like that one google econ study that found interviews have no value over chance except for niche fields
comment in response to
post
Like i know many ML phds, but its one of those things where a truly great researcher/eng combo can be insanely productive. Just as an example, he came out with a smaller version of GPT and he immediately found like 3 different tokenizer speedups no one was looking for
comment in response to
post
Honestly, for someone like kaparthy yeah its worth it
comment in response to
post
negotiate a golden parachute
comment in response to
post
gods the Erie was strong then
comment in response to
post
well for PCE right, literally just Used cars going up is a problem. and the implications for monetary policy are extremely hard to read because the passthrough is so dynamic
comment in response to
post
Just checking some benchmark items i have on camelcamelcamel. SSDs creeping up a bit, TVs kinda flat, Apparel a little higher
comment in response to
post
Hmmm its possible its high
comment in response to
post
US! car gurus. +1.23% over the last 30 days, i think manheim has the opposite
comment in response to
post
So lets just hope inflation has a good print
comment in response to
post
Really stocks just go up. I'm sure at best, its like the UK one where its some promise that lets him go back to status quo with 10% more tariff
comment in response to
post
Honestly, i would bet against this being good
comment in response to
post
So presumably this is it, or we just keep kicking the can down the road
comment in response to
post
We have two different wordings here, one for framework for implementation, another for framework on geneva consensus
comment in response to
post
Bessent just before this said they were going to stay an extra day and i don't think thats happening