Discover real AI creators shaping the future. Track their latest blogs, X posts, YouTube videos, WeChat Official Account posts, and GitHub commits — all in one place.
Activity on swyxio/papra
swyxio contributed to swyxio/papra
View on GitHubActivity on simonw/datasette
simonw commented on an issue in datasette
View on GitHubActivity on simonw/datasette
simonw commented on an issue in datasette
View on GitHubActivity on simonw/datasette
simonw labeled an issue in datasette
View on GitHubWhen one frontier closes (math, software engineering, etc), many new ones open. This is just the beginning.
we’ve figured out how to create increasingly powerful intelligence without a complete theory of intelligence itself. figuring out what we’ve created will be one of the most important scientific problems of our time.
Nobody likes AI replies but they're defending each other Because they're part of a giant network of people who all are using AI bots to spam X replies, and people selling them the shady tools to do that including bot farms, telephone farms etc. They all defend each other on here When you poke a hornet's nest, they all come out to sting you, which they're doing to me today Remember @nikitabier's work on original content and demonetizing all the automated spam? That was the same thing and it got so vicious he had security issues due to it So yes nobody likes AI replies, it's just a self sustaining ecosystem on here with a lot of people dependent on it to make money from X monetization and the shady businesses selling them the automation tools and bot farms to do so I'd call it a criminal enterprise
@levelsio I just don't get why real people want to engage with others via bot
View quoted postOpenAI proposes perplexity as the universal metric for industry self regulation and pacing (as opposed to eg FLOPS) “Given that the main way new surprising capabilities emerge is by progress on compression, a measure rooted in perplexity could give us a shared benchmark for general intelligence, regardless of the degree of automation of the development process. Extrapolating from recent progress, we should expect further transformative capabilities as we enter the ~0.68-0.7 log-perplexity window.” this chart is one of the most beautiful ways to tell the story of the last 4 years. it also shows that a lot of the post training issues we debate is kinda the B plot to the overall A plot of large scale pretrain progress.
We decided to provide more transparency into pretraining progress at OpenAI for the sake of moving toward shared safety bars. In “Perplexity and Intelligence” I argue for common limits based on perplexity and other measurements, with progress beyond them gated on our ability to
RT Xuan-Son Nguyen wllama v4.0.0 is out! Up to 2.8x faster on Chrome, support video input and full-duplex realtime audio Hybrid CPU + WebGPU support on Chrome and Firefox, allow using bigger models!
finally taking the time to learn t3code. tried 2 months ago and was too... "developery" Let's just say... i did not realize how far along this product is. I was reluctant to try it, because I use artifacts quite often for writing and content workflows... and to my surprise claude artifacts work quite well inside t3code. in fact you can open them in the t3code browser (i didn't know they had this)... And then i went to connect the iOS app. I didn't even have to scan a qr code... it just kinda... connected. THEN i realized you can control the t3code browser from your phone. wtf. Anyway, as many of u know i've been obsessed with creating content workflows for the past few months... And if there's one thing i've learned over that time is the OpenAI models, the Claude Models, and the Gemini models all have their own strengths and weaknesses. For instance gemini is really good at video understanding (taking a video, analyzing the different elements of a video from visuals, to sound etc), and Claude is really good at actually writing the code to edit the video. Well, w t3 you can just have claude spin up a gemini subagent that analyzes 20+ videos and provide a report back to claude to actually do the edits. Additionally, when editing shorts you can basically have Neon, Astra, and Fable all attempt to edit the same video from the same context, all from a single prompt. This makes doing different types of creative workflow evals much easier. Bravo Theo, this is a great tool... However, it's just a little too expensive for my liking.
Activity on swyxio/papra
swyxio assigned a pull request in papra
View on GitHub