bathory
411 posts

bathory retweetledi

Introducing Voxtral Transcribe 2, next-gen speech-to-text models by @MistralAI.
State-of-the-art transcription, speaker diarization, sub-200ms real-time latency.
Details in 🧵
English

@alessabocchi i regret none of them, this was maybe the stupidest fucking burgeroid post i have ever seen and i genuinely hope for a Total War between the West and the East to give us a chance at a reset
these retards manage to make even their own personal relationships e-performative
English

@NinosMaron @gabriel1 only if you’re trying to be bored. OC is a nice place to live but not visit
English

There's still time to delete this.
The salary comparison is stupid, because it doesn't account for occupation, hours worked and the very simple fact that his office has a fixed budget cap that he doesn't control.
Bernie Sanders is right that at the core, most companies, and managers want to use AI to cut costs or cut jobs entirely. But of course the tech nerds get all defensive about it, because they just want the best "for all of humanity" right?
Mechanize@MechanizeWork
.@BernieSanders claims that our goal is to "make it easier to pay workers less" but I notice that we pay our employees vastly more than he pays his staffers.
English

@flawedaxioms just do a masters. you can do some cool stuff and only “waste” 1-2 years instead
English

go to GRAD SCHOOL. You will SURELY NOT REGRET getting a PHD. it will make you feel COOL and MYSTERIOUS when you tell the guy at the gym that you do ALGEBRAIC GEOMETRY. It will only cost you NINE HUNDRED THOUSAND DOLLARS in missed earnings.
flaw@flawedaxioms
Spiritually I feel like I should be in grad school rn. But If I go to grad school I'm just getting this job again after... At 24 I'm simply too full of whimsy and general joie de vivre to be writing "microservices". I should be doing some obscure math that is never applied.
English

We'll be posting course content for anyone who would like to follow along!
The first four lecture videos are available now:
youtube.com/playlist?list=…
English

.@gneubig and I are co-teaching a new class on LM inference this fall!
We designed this class to give a broad view on the space, from more classical decoding algorithms to recent methods for LLMs, plus a wide range of efficiency-focused work.
website: phontron.com/class/lminfere…
English

@luciascarlet @notcrusaderarts yeah distribution is important, 16PM was waaaay less top heavy than 15PM and felt much better
English

@eersnington @RhysSullivan i was literally gonna say the same thing l
English

@RhysSullivan you'll either have 1000hrs+ playtime or 1hr at max. there's no inbetween
English

@arithmoquine @snwy_me people would just do more drugs/alcohol or some other self-destructive hobby tbh
English

@aidan_mclau @0xblacklight yeah ppl need to take a neuroscience class and they’ll realize that consciousness is an emergent property of our neurobiology
English

@0xblacklight neuron soup doesn’t have experience, it’s just chemical reactions, hope that helps :)
English

Autoregressive token prediction models don’t have experiences, it’s just matrix multiplication, hope that helps :)
Aidan McLaughlin@aidan_mclau
@AnthropicAI given that the model's experience is these chats, and, in this case, the chat is irrecoverable, how does this differ from "we gave the model the ability to kill itself?"
English

Sad day, can we get an option to keep it? (also in chats)
Hamza@thegenioo
The background gradient has been removed from ChatGPT dot com
English

@galanthehater idk. at the time i legitimately thought this was a sign i was supposed to have died and that my stubborn survival instincts must have pissed off God somehow


English

ChatGPT literally got worse for every single Plus user today.
There's no way to reliably get thinking models anymore.
Before we had o4-mini, o4-mini-high and o3.
Now we have GPT-5 Thinking with 200 messages per week and a router that exclusively routes you to some small and shitty non-reasoning model.
English



















