Pranav Khetarpal

Confused "Engineer", pretending to know stuff | F1, CS, Quant (and a little bit of Physics) | QR, @IITDelhi'25, @universe_tbd AI x Space

[Doug Colkitt @0xdoug]

String theory is probably a field that has set back quant trading by at least 10 years. Stealing top tier trading talent to do make believe 26 dimensional geometry, inflate boomer professors’s grant budget, and produce zero testable hypotheses or applications after 50 years

[Pranav Khetarpal @pktrpl]

@virkvarjun hey, what did you use for the figures? It's really clean and pretty

[james hawkins @james406]

parents: "move out"

girlfriend: “quit being such a loser”

boss: "work harder"

claude: "uber for dogs (the dogs are the drivers) is a great idea, you should absolutely pursue it"

[Pranav Khetarpal @pktrpl]

Idk why but I feel transformers are very closely related to a GNN? Like the MLP is node specific processor, and attention is like the edges connecting and sharing information across nodes...

[Pranav Khetarpal @pktrpl]

I love how smut writers have discovered some modern NLP/AI concepts from first principles lol!

RL for reasoning tasks are only easily trainable on verifiable domains like code and math. And with code, the only thing the reward function depends on is if the code runs, not on anything else. Not on the process.

We need to find better ways to evaluate tasks with intermediate steps and not directly/easily verifiable ground truths!

[Pranav Khetarpal @pktrpl]

@litcapital I think he just means that 50% of the stocks of both anthropic and oAI will go public, and you can buy them! Thus public owning 50%?

[Pranav Khetarpal @pktrpl]

You win some, you loose some!

I've always hated this... people say "you can't win everything!". I scoffed at such ideas all my life. I wanted to win everything. I think I now understand the meaning and truth behind it.

But I will still continue to try to win everything. This way, at least I'll win>loose. Ultimately, it's all about being above the 50% line!

[conduct|r @conductr_]

i know it sounds kinda stupid. but regardless of how simple it is you have to embody being a delusional optimist. you’ll open unexpected new doors and effortlessly 10x your quality of life. there’s no reason to be a self sabotaging pessimist.

[AlgoFlows @algoflows]

Berkshire management ever since Buffett retired

[eigenron @eigenron]

a lot of ambition is just delayed revenge against a room that failed to recognize you at some point in your life.

[Ayanokoji @layeee09]

Le vieux du quartier m’a dit : « N’oublie pas que la boussole a été inventée avant l’horloge parce que la direction est plus importante que le temps. »

[Pranav Khetarpal @pktrpl]

Not at all intelligent... just try doing like 10-20 such random obscure problems, and study maxwells equations a little, and you'll manage to get somewhere in few hours...

It looks scary, but after you do a few such problems, you realise it's not too difficult. More about familiarity to such kinds of problems.

I'd suggest start with mechanics, cause there are more such problems in mechanics, and its lot more straight forward cause you only need the newton's laws and it's easier to imagine than current and magnetism...

BUT NOT AT ALL INTELLIGENCE. IT's about exposure and a lil practice maybe, but not intelligence. I AM VERY VERY DUMB, HONESTLY!

[Tanaya @tanaya_0510]

Ages when you suffer a lot:

17
23
27

[Omkar @psomkar1]

Worst opening experience

[Marius Vach @rasmus1610]

Honestly, I love the current trend of publishing research as blog posts instead of arxiv pdfs. @thinkymachines @NoahZiems and @a1zhang come to mind here recently

[mie @girlpov]

taps the sign

[Pranav Khetarpal @pktrpl]

Dude just gave the optiver test!

[RT @RT_com]

Trump REVEALS insane cognitive test QUESTION he ‘aced’

‘203… multiply times 9… divide by two… add 1,324… subtract 1,292… multiply by 19 again… What’s the answer, sir?’

‘I got it RIGHT… doctor said he’s been doing this 20 years — never seen anybody ace it’

[Pranav Khetarpal @pktrpl]

@venturetwins @Anand_Venkatram Gemini is what Groks alignment wanted to be!

Anthropic is truly proud of their alignment, with awesome post training! Idk how they get these guardrails and personality so deep in? Is it simply RL? Some fascinating datasets I must say?

[Pranav Khetarpal @pktrpl]

@_arohan_ Demis Hassabis in a few days: personal news, I’m joining anthropic!