@SaidAitmbarek that’s why most highly successful entrepreneurs can give psycho vibes at times because you can’t be normal to endure all the pressure + rejection and show up everyday for more
@travisstreet_@vvarrdi not kidding!
i worked on this problem in 2022-2023 (style adherence) but i needed massive amounts of data to continue improving the model and that was around the time Elon bought twitter and killed the affordable dev api tier so I couldn’t continue
i just came to say i loooove working on ml problems 🥰 it's hard but so fun and rewarding.
who knew deep learning was the mandate of heaven for lovers of math and computing.
i should've discovered that when i was 12 yo instead of lame old html
@vvarrdi@travisstreet_ prompting can't fix it but LoRa finetuning can if you pair it with a RAG dataset of that user's own tweets for in-context style mimicking
the hardest part though is intent/strategy of message with no hallucination
it's a solvable~ish problem
@theroborourke what’s the old adage again? don’t do as a say but do as I do.
better to watch behavior and copy rather listing to advice which ppl are notorious bad at contextualizing
@rihntv one guy was early employee at Airbnb when it was less than 500 people
as of 2025 that guy was worth $50m and I don’t think he’s bitter about his fortune
it’s more about the company than corporate life as a whole tbh
@zairalaraib_ i think your classmate is just disconnected from the professional working world
if they have interned somewhere there’s no way they wouldn’t know
@irbaazkadri@monkeytypefan1 watch.
next wave of innovation the future kings of the hills might be the robotics bros and therefore how the VCs tweet will follow suit in hopes to get allocation
@irbaazkadri@monkeytypefan1 no the truth is ai labs are the king of hills these days and thought leadership from ai researchers/engineers follows
also for some people deep in the work this is natural way to think and communicate
there's never been a better time in history to be an original writer
if everyone is using ai to write their copy, you're insanely differentiated just by not
@sandislonjsak i think the more accurate analogy is the distribution curve got deeper
we’ve always had mostly bad software with 10% good and 1% exceptional
that distribution hasn’t changed so we’re drowning in the noise of the 10x increase in bad software