Tunadorable retweeted
@YouTube please reach out! We need the channel back 😁
Hey @YouTube @TeamYouTube @YouTubeCreators my the auto moderation system messed up and deleted my channel out of nowhere & the appeal was denied too quickly for a human to have been the one that saw it. Who can I get in contact with?
Tunadorable retweeted
The channel is gone and it was the the only good channel out there... Everything else is corporate trojan horses for people with mental decay...
glitch in YouTube's auto mod system. If you'd like to help then maybe tweet at @YouTube @TeamYouTube @YouTubeCreators asking them to reach out to me🙂
Tunadorable retweeted
🚨 Announcing HackAPrompt 2.0, the World's Largest AI Red Teaming competition 🚨
It's simple: "Jailbreak" or Hack the AI models to say or do things they shouldn't. Compete for over $110,000 in prizes.
Sponsored by @OpenAI, @CatoNetworks, @pangeacyber, and many others.
Starting NOW to July 1st. 🧵
Finally autoregressive decoding beats diffusion at image generation!
YouTube: youtu.be/_9s3yttaAkM
@arxiv: arxiv.org/abs/2404.02905
@Bytez: bytez.com/docs/arxiv/2404.02…
@askalphaxiv: alphaxiv.org/abs/2404.02905
Tunadorable retweeted
The correct way of thinking about AI and education will not be “how can we ensure that homework is not written by chatGPT” or “is AI the new pocket calculator revolution”, but “how can we ground the coming generation of human-AI hybrids in a shared, benign social context”
Looking to learn how to make GPU kernels go BRRRR? Then check out my series Triton GPU Kernels 101 which takes you all the way up to a speedy Flash-attention 2 kernel
youtube.com/playlist?list=PL…
Tunadorable retweeted
I’m worried that “doom” speculation will make doom more likely. Specifically, AIs conform to our expectations of them, as communicated by their training data. This “self-fulfilling misalignment data” may be poisoning training already. 🧵
Might a clever use of AI revolutionize the way large groups communicate? Check out my interview with @LouisBRosenberg discussing Conversational Swarm Intelligence
youtu.be/9TnsPsZrBOk
Rather than try to make hardware more efficient with current models, let’s start designing models that are more efficient on current hardware!
YouTube: youtu.be/ISiOCmxsbto
@arxiv: arxiv.org/abs/2501.06589
@Bytez: bytez.com/docs/arxiv/2501.06…
@askalphaxiv: alphaxiv.org/abs/2501.06589
Check out this ✨pretty✨ way to diagram GPU kernels
YouTube: youtu.be/eonUFQ3YtmM
@arxiv: arxiv.org/abs/2412.03317
@Bytez: bytez.com/docs/arxiv/2412.03…
@askalphaxiv: alphaxiv.org/abs/2412.03317
Tunadorable retweeted
Closed source AI feels like. Imagine if Einstein sold his theory of relativity to AT&T for $2 mil. They kept it private, vaguely implied “we discovered something big”, and then they released a GPS that was 10% better than their competitors.
Tunadorable retweeted
Can humans reason? In this paper we show evidence many humans simply apply heuristics they've been exposed to over the course of their lives without deeper consideration. In conclusion, humans don't seem to reason and only copy reasoning patterns from their training data.
Tunadorable retweeted
If your natural response to o3 is to downsize your team and cut costs, rather than work on a 100x more ambitious project, you are ngmi
Tunadorable retweeted
Maybe a hot take, but what about the following advice to the next gen:
Don't get an AI degree; the curriculum will be outdated before you graduate. Instead, study math, stats, or physics as your foundation, and stay current with AI through code-focused books, blogs, and papers.
Tunadorable retweeted
this guy is lowkey talented and for people who don’t know he has a youtube channel and creates videos on gpu programming
youtube.com/@szymonozog7862
This Post is from a suspended account. Learn more






















