2025-05-22
In 2022, I worked on text diffusion for a bit and wrote a blog post. Since then, people have regularly asked me about scaling diffusion LLMs. All the while, I was on the first row watching Brendan assemble a cracked team and make it a reality. Now I can stop being coy about it😁
The Verge
Google is tapping its users' data to give its AI models an advantage over OpenAI and Anthropic, starting with its opt-in “Gemini with personalization” feature
Google is slowly giving Gemini more and more access to user data to ‘personalize’ your responses.
In 2022, I worked on text diffusion for a bit and wrote a blog post. Since then, people have regularly asked me about scaling diffusion LLMs. All the while, I was on the first row watching Brendan assemble a cracked team and make it a reality. Now I can stop being coy about it😁
Fortune
Google DeepMind says Gemini Diffusion, an experimental text diffusion model demoed at Google I/O and available by waitlist, generates 1,000-2,000 tokens/second
Our state-of-the-art, experimental text diffusion model Jose Antonio Lanz / Decrypt : Google Doubles Down on AI: Veo 3, Imagen 4 and Gemini Diffusion Push Creative Boundaries Matth...
2023-11-17
5-6 years ago I was working on music generation at DeepMind, but let me tell you, this is... something else. Incredibly excited to be able to finally share what our team has been working on!
Wired
YouTube previews Dream Track and Music AI tools, which use DeepMind's new model Lyria to generate music in the style of famous artists, from humming, and more
Will Knight / Wired :
2023-11-16
5-6 years ago I was working on music generation at DeepMind, but let me tell you, this is... something else. Incredibly excited to be able to finally share what our team has been working on!
Wired
YouTube lets some Shorts creators test Dream Track, a new DeepMind-powered AI tool to generate and remix music in the styles of nine artists, including Sia
YouTube creators will get to test a new AI tool that generates and remixes music in the style of several famous musicians, including Sia, Demi Lovato, and T-Pain.
2021-05-22
Unsupervised speech recognition🤯 a conditional GAN learns to map pre-trained and segmented speech audio features to phoneme label sequences. It is trained only to produce realistic looking words and sentences — no need for any labeled data. Amazed at how well this works! https://twitter.com/...
Engadget
Facebook's wav2vec Unsupervised, a way to build speech recognition systems that require no transcribed data, may bring automatic translations to more countries
Wav2vec Unsupervised (wav2vec-U) … Tiernan Ray / ZDNet : Facebook AI cuts by more than half the error rate of unsupervised speech recognition Donald Conway / Insider Voice : Facebo...