Mei Neym @mayphaZ

Realtime interactive generative models FTW! Announcing a new 🌊 of details and features for Magenta RealTime, the open weights live music AI model from GDM! * Live Jamming with audio input 🎤🎸🎵 * Personalize your own models 🔧 * Tech report 📜 Links below in the 🧵...

there should really be some sort of investigation into this. the fact people can tweet ads and not get in any sort of trouble is insane. on yt, you can get seriously fined if you don't do take the appropriate actions the amount of "hey do you tweet for $" dms i get is wild and the amount of products and tweets you read from people are just that, ads the worst offender isn't even these direct ads, its retweets. retweets you don't even have the option to add ad to the retweet, it is just a way for companies to get paid eyeballs with a guise of a recommendation from the retweeter

el.cine @EHuanglu

a year ago

omg.. it's over no more plastic looking AI images this new image generator, Higgsfield Soul, creates visuals that feel like real photography, with 50+ curated presets, you get taste built right in. it's free for everyone.. 10 examples:

89 180 2K 772K 3K

Just went through all the @LeRobotHF hackathon demos - insane what people built and how diverse the projects were. Below is my favourite one, which uses Gaussian splatting. Highly recommend watching the 30 winning videos. Esp. as a non-roboticist to see what’s possible rn.

Cool demo of a GUI for LLMs! Obviously it has a bit silly feel of a “horseless carriage” in that it exactly replicates conventional UI in the new paradigm, but the high level idea is to generate a completely ephemeral UI on demand depending on the specific task at hand.

Google DeepMind @GoogleDeepMind

a year ago

Here's how Gemini 2.5 Flash-Lite writes the code for a UI and its contents based solely on the context of what appears in the previous screen - all in the time it takes to click a button. 💻 ↓

156 508 4K 1.1M 2K

While we're at "recognizing evals", here is a legendary vision paper from CVPR 2011 It shows that it's quite easy to classify which dataset an image comes from (39% acc, vs random=8%). The point being, every dataset having its distinct signature should be a default assumption.

Lucas Beyer (bl16) @giffmana

a year ago

I think this paper writes a wrong claim around otherwise valid experiments. A correct title would be "LLMs can classify whether a transcript is an eval or user interaction". Which is NOT "model knows it's being evaluated". I hope yall see the difference? If not, see reply.

23 16 384 85K 138

# A new type of information theory this paper is not super well-known but has changed my opinion of how deep learning works more than almost anything else it says that we should measure the amount of information available in some representation based on how *extractable* it is, given finite computation. for example, an encrypted text file has less V-information than the same data in plaintext, because it takes more computation to extract. note the contrast to traditional information theory, which would tell us that the two representations have the same amount of Shannon information i’ve long wondered why certain types of basic questions didn’t have a proper theoretical answer: > why does distillation outperform vanilla maximum likelihood training? > why does lora work better than finetuning? > why does self-attention work better than almost any other similar operation? > how much “information” remains in a text embedding? > how much “information” remains in language model weights? > should i use fine-tuning or RAG? the true answers to all these questions depend on some way of measuring and comparing *information content* between different representations. v-information is one step towards doing this besides computational constraints, model architecture probably affects the “information content” in representations, along with the presence of any pretraining data used, as well as model-level statistics about the optimization in the training process – e.g. the length of time a model was trained for probably changes representations pretty drastically this is all to say, i think there is some true notion of “information” that none of our current paradigms (Shannon information, V-information, etc.) capture. we encounter this idea every day but we dance around it and describe it in vague terms; we measure it from all sorts of angles but can’t quite characterize it theoretically when the v-information paper came out I thought there would be a lot of follow up work developing more complex and useful notions of information for deep learning. but it hasn’t yet I still think at the heart of these questions of what-information-lies-in-representations there’s something to be found that’s profound, elegant, and potentially extremely useful. I don’t know what it is and i’m probably not the person who will figure it out. but I really hope someone does. :)

We got a robot to clean up homes that were never seen in its training data! Our new model, π-0.5, aims to tackle open-world generalization. We took our robot into homes that were not in the training data and asked it to clean kitchens and bedrooms. More below⤵️

Today, we're releasing a new paper – One-Minute Video Generation with Test-Time Training. We add TTT layers to a pre-trained Transformer and fine-tune it to generate one-minute Tom and Jerry cartoons with strong temporal consistency. Every video below is produced directly by the model in a single shot, without editing, stitching, or post-processing. Every story is newly created. Demos: test-time-training.github.io/video-dit/ Paper: test-time-training.github.io/video-dit/asse…

Marine Le Pen in France, Bolsonaro in Brazil, Imran Khan in Pakistan, Matteo Salvini in Italy, Donald Trump in America, Calin Georgescu in Romania The criminal prosecution of every populist challenger is a dagger in the heart of the credibility of democracy.

Mike Benz @MikeBenzCyber

2 years ago

Unreal. They took out the opposition. Marine Le Pen, who is currently leading in opinion polls for the 2027 election in France, has been sentenced to 4 years in prison and banned from running in the next election.

keyboard_arrow_left Previous keyboard_arrow_right Next

2K 8K 30K 4.2M 1K

My daily reminder

@dwarkesh_sp because we "activate" the knowledge through context. getting the right prompt is hard. CoT is the attempt of having the model generate the right context discovering medical cures requires you to think about all of the related knowledge, then BASED ON THAT you draw the connection

Applies to private life as well, not just in business

Garry Tan @garrytan

2 years ago

151 1K 13K 3.4M 15K

Andrew Wiles on being smart

why not 100x?

Paul Graham @paulg

2 years ago

This may be the most inspiring sentence I've ever read. Which is interesting because it's not phrased in the way things meant to be inspiring usually are.

260 2K 23K 3.3M 17K

Deepseek 70B Distill + Groq LPU Sleep can wait. Build fast! 🚀💪👀

spread love

How to juice your 2024 performance with one trading day left in the year, a guide:

keyboard_arrow_left Previous keyboard_arrow_right Next

Bill Ackman @BillAckman

2 years ago

I am often asked for stock recommendations, but generally don’t share individual names unless I believe the risk versus the reward is extraordinarily compelling. As we look toward 2025, one investment in our portfolio stands out for large asymmetric upside versus downside so I

749 944 8K 5.0M 8K

turns out you can use <xml/> tags with the realtime api to control tone with _super_ high granularity 🤷🏽‍♂️ @OpenAI devday someone asked if our models could read a script with tone directions - e.g. laugh here, cough there, etc i shrugged and opened the playground... - prompt 👇 -

@zhou_xian_ mind = blown

The wait is finally over!!! 😁 We just dropped an in-depth tutorial on how to build your own robot! Teach it new skills by showing it a few moves with just a laptop. Then watch your homemade robot act autonomously 🤯 1/🧵👇