singhblom.com
Head of ML at Embark Studios. Working on RL for physics based animation, LLM applications in games, generative models, etc.
I like it when ML makes what previously was impossible possible. Previously web dev, genetics, mathematics, physics.
85 posts
1,207 followers
1,056 following
Regular Contributor
Active Commenter
comment in response to
post
I really enjoy your posts here. But the anti-ai crowd can be really intense so I understand your frustrations. đ
comment in response to
post
I don't get it. Is Trump Sulla? Or is Trump Marius and they're worried that Sulla is coming?
comment in response to
post
Maybe there is a classic European collaborative solution here - we can use French political will in order to spend German money? :P
comment in response to
post
MachMit Museum var jÀttekul tyckte vÄra barn pÄ sju och tre Är.
comment in response to
post
Nu gjorde du mig nyfiken. Afrikaans Àr nog större om man rÀknar andrasprÄkstalare ocksÄ.
comment in response to
post
Jag fattar inte heller varför de tror att dessa kÀnsliga kvinnor kommer bli mindre oroade av att skÀggiga transmÀn börjar anvÀnda deras toaletter Àn de hade varit av att nÄgon transkvinna med i vÀrsta fall en tydlig haka hade pudrat nÀsan med dem.
comment in response to
post
This is art.
comment in response to
post
Marathon looks awesome, I'm really looking forward to trying it. If you like extraction shooters you should try Arc Raiders as well. I think it will be very good, but I am quite biased.
comment in response to
post
Google tried the non biased approach and it wasn't good either. It's tricky. I don't think direct sampling of the distribution will give good results either. You probably want to upsample minorities slightly just to give more interesting diversity in the results.
comment in response to
post
We haven't published on it but this result has been reached in our group by multiple independent researchers almost every time we rewrite our sac implementation. Protip: it is also a good setting for studying nan propagation in neural networks.
comment in response to
post
Is that significant?
comment in response to
post
The paper skygest feed helps a lot. It is one of my favorite feeds when I need a break from the trumping. I can't figure out how to link it from the app but it's an easy googling away.
comment in response to
post
Slappt berÀttande. Sen kuppförsöket efter förra valet har hela plotten kÀnts lite övertydlig och överdriven.
comment in response to
post
In a phase 3 trial with 2,000 patients undergoing surgery, Journavx as effectively as hydrocodone plus paracetamol and had fewer side effects. It also does not appear to be addictive.
comment in response to
post
I found this hilarious today.
comment in response to
post
Det sagt sÄ kan kanske lite Artikel 7 ge lite extra push för Tisza. Det gÄr ju att ta tillbaka.
comment in response to
post
April nÀsta Är! Mycket kan hÀnda.
comment in response to
post
đŠ
comment in response to
post
It looks like most tiles have one in and one out direction? Wouldn't that be enough?
comment in response to
post
Yes, as a Swede my feeling is that we've had one year's head start on serious defense integration with the rest of the Nordics/Baltics because of NATO accession. Even with the US gone Europe will reuse the NATO structures.
comment in response to
post
I don't want to meddle in another country's politics so don't take this the wrong way but I sometimes wonder if you guys made the wrong choice last November.
comment in response to
post
He's Norwegian.
comment in response to
post
I think his handling of Ukraine has also shown us how he would be less likely to get the US into a war with China over Taiwan. I think Biden or Harris would have responded to a Chinese invasion across the strait. I think Trump will just fold.
comment in response to
post
Funkar vid Norrmalmstorg ocksÄ. Om ryssen kommer sÄ kan vi definitivt tuta Ät dem.
comment in response to
post
It's so expensive though! Really excellent but OMG. I have to eat as well!
comment in response to
post
This would have stopped my grad school career. Most Fulbrighters don't have a TA or RA the first year and the stipend is not enough to build up any savings at all.