Chips: ASIC, FPGA. CV/ML. Duck pictures by the lake. Some bread making. https://nitter.gq/t.co/bqpeM1Ei6T
PNW
Joined February 2009
- Tweets56K
- Following7.5K
- Followers2.1K
- Likes89.3K
Muzaffer Kal @🏡 🦆⏩ retweeted
Introducing RL with Neural Feedback (RLNF). We have found reliable ways to train AI models to respond to human preferences purely from brain signals.
The example below is a picture of a face that self-transforms into a more attractive version for the subject as they are looking at it.
Muzaffer Kal @🏡 🦆⏩ retweeted
Mike Johnson says Christians must vote for the GOP to 'make sure that the righteous are ruling'
Muzaffer Kal @🏡 🦆⏩ retweeted
Introducing Surya OCR 2.5!
- Most accurate open OCR model under 1B params (85.9 on OmniParseBench)
- 1.7x faster than Surya 2, 4.5 pages/s on H100
- 689M params, with character tokenization
Muzaffer Kal @🏡 🦆⏩ retweeted
Patrick Stewart, the Emmy-winning performer known for his work on stage and screen and beloved for both his amiable persona and roles in popular franchises like Star Trek and X-Men, died Wednesday at his home in Los Angeles. He was 86.
apple.news/AKcQ178dcT4mVPtxM…
Muzaffer Kal @🏡 🦆⏩ retweeted
Energy-Based Models can now scale better than Diffusion models.
Introducing TEM: Transformer Energy-based Models
TLDR:
• TEM outscales Diffusion without guidance
• First scalable contrastive Energy-Based Models
• Better generalization, test-time scaling, and OOD detection than Diffusion
🧵
Muzaffer Kal @🏡 🦆⏩ retweeted
I've always believed that intelligence should come from simple, general building blocks that learn their own abstractions directly from raw data; in other words, the bitter lesson. I think we're finally seeing a real step towards AI that can learn from all data.
Language modeling on raw bytes is a microcosm of this ideal, and to me it's a central problem in AI: if we can't build LLMs that learn end-to-end from bytes, how will we handle the rest of the world's data that we understand even less?
Last year we introduced hierarchical networks (H-Nets) as a first step. Our new version finds up to 3-5x efficiency gains in controlled scaling law experiments, and at larger scale matches an open-source model trained on 11T tokens with ~10x less data and compute. The posts below walk through what we found and why we think it works.
Muzaffer Kal @🏡 🦆⏩ retweeted
This will change how LLMs are trained: Parallel Language Models predict several tokens in parallel at autoregressive quality. Add a parallel decoder at pretraining, get 5.7x speedup after posttraining.
🧐 → arxiv.org/abs/2610.26693
ℹ️ → justuswill.com/plm/
Read on: 🧵 1/4
Muzaffer Kal @🏡 🦆⏩ retweeted
AutoScientist increased correctness from 58.1% to 75.2% on medical long-horizon reasoning, outperforming GPT-6 Astra, Gemini 4 Argon. ✨
@adaption_ai supercharged performance on any open weights model for the task. 🔥 So you can use any starting model.
Muzaffer Kal @🏡 🦆⏩ retweeted
Today we're releasing the Reachy Mini Level UP update, which mainly contains three things:
• A hoverboard, to give Reachy Mini some much-needed mobility (I'm a big inverted pendulum fan and I absolutely love this one, sometimes I just do sit → stand up → sit → stand up on repeat)
• A smartphone app, available on iOS and Android: no computer needed to interact with the robot anymore
• NFC tags, allowing for new angles of creative personalization (e.g. one hat = one personality)
We hope you'll build great things with these new tools, learn stuff and have fun!
Congrats to everyone involved in the release!
PS: There's been a major software update that has been robust for a while now but hasn't gotten a lot of visibility yet: JavaScript apps. I encourage all our AI builders to have a look at it, because for a bunch of applications, it's so comfortable compared to Python. I think the best example of this is the Telepresence app (link below), which allows you to control your Reachy Mini and stream audio and video from your phone, even when you're not home. Also, you just click on the link and it works in your browser, no installation needed.
Muzaffer Kal @🏡 🦆⏩ retweeted
Diffusion models create novel images, but they can also memorize samples from the training set. How do they blend stored features to synthesize novel patterns? Our work in @NatureComms shows that diffusion models behave like Dense Associative Memory. Specifically, we identified three computational phases: memorization, generalization, and spurious pattern generation. The third computational phase, spurious patterns, is the hallmark of associative memory that appears alongside memorized and generalized samples. We characterized all three states using basins of attraction and curvature measurements, extending our analysis to Stable Diffusion.
The work is done in collaboration with @gaboraya, @Matteo, @mj_zaki, @LucaAmb, @DimaKrotov.
Here is the link to the paper: nature.com/articles/s41467-0…
Ty to @IBM and @rpi for the funding.
Muzaffer Kal @🏡 🦆⏩ retweeted
MITの研究者らが、AI同士が言葉を使わず頭の中の状態を直接渡して通信する仕組みを作り「テレパシー」と名付けました。
文章でやり取りするより平均40%正確で4倍速かったといいます。
また「画像を見られるAI」が写真の内容を「テキストしか読めないAI」にテレパシーできるそうです。
写真の内容をテレパシーを受信した「テキストしか読めないAI」は、テキストで説明を渡されるよりも写真についての質問に正しく答えられたとのこと。
なお、別の会社のAI同士でもテレパシーできるようにしたのが今回の研究の一番の新規性のようです。
例：
Meta（Llama 3）→ Alibaba（Qwen 3）
Alibaba（Qwen 3）→ Google（Gemma 2）
Google（Gemma 3）→ IBM（Granite 3）
Google（Gemma 3）→ Meta（Llama 3）
Mistral AI（Ministral 3）→ Meta（Llama 3.2）
Google（Gemma 2）→ Mistral AI（Ministral 3）
IBM（Granite 3）→ Hugging Face（SmolLM3）
Muzaffer Kal @🏡 🦆⏩ retweeted
🤖 𝐒𝐒𝐈𝐎: 𝐒𝐞𝐥𝐟-𝐒𝐮𝐩𝐞𝐫𝐯𝐢𝐬𝐞𝐝 𝐈𝐧𝐞𝐫𝐭𝐢𝐚𝐥 𝐎𝐝𝐨𝐦𝐞𝐭𝐫𝐲
Since the first learned inertial odometry 10 years ago, every method has needed ground-truth labels.
It doesn't have to: SSIO learns from raw IMU, fully self-supervised 🧵👇
ssio-imu.github.io
Muzaffer Kal @🏡 🦆⏩ retweeted
Hi Sandy. Are you well? This is wrong. On January 1, 1993, we sent a press release saying that multiplayer would be in DOOM, a full 8 months before you joined the company. We knew deathmatch would be huge. In that press release, in January 1993, we said multiplayer would be "the number one cause of decreased productivity in businesses around the world." We were sure of it, and we needed no convincing. I understand you are proud of your work on DOOM, and you should be, but please also respect the accuracy of your statements, particularly around game history. My concern is that these false statements are designed to trigger replies (like this one), reposts, and outrage discussion, all with the view of increasing your followers and subscriptions, particularly since this false deathmatch story (among others) is so easily disproved. I’ve even welcomed you to reach out to me ahead of time. You seem to appreciate game history as much as I do. Could you not also make yours accurate?