I wonder if DeepMind purposely trained Veo 3 with captions turned on. I can imagine it being helpful for (for example) aligning speech with video frames.🤔
I wonder if DeepMind purposely trained Veo 3 with captions turned on. I can imagine it being helpful for (for example) aligning speech with video frames.🤔
One interesting fact about flow matching loss is that the flow is independent of the model. This means the mapping from noise to data, whether it's images, audio, or video, remains unchanged, no matter how you tweak the model or adjust the hyperparameters.
It's a bit surprising to me that tf.data doesn't support loading .npy files natively. So I wrote this custom function to load numpy arrays. Am I missing anything?
On CoT Training with Reinforcement Learning
I've been thinking a lot about training LLMs with reinforcement learning lately. One thing that surprises me is how easy it is to train LLMs to generate chain-of-thought reasoning using RL, even with extremely simple algorithms like…
909 Followers 443 Followingeng @plainsupport, ex-@uber, building stuff with AI and claude code.
the stuff:
- https://t.co/3BcqSL6ZXR
- https://t.co/F6kvxXtiKm
- https://t.co/jf0yA7RYZf
18 Followers 210 Following20, AI Bro
AI + Backend Eng @usefindr
AI research + DL model development on weekends
GIT : https://t.co/PXBgybtuj1
HF : https://t.co/2cuNHUXu85
909 Followers 443 Followingeng @plainsupport, ex-@uber, building stuff with AI and claude code.
the stuff:
- https://t.co/3BcqSL6ZXR
- https://t.co/F6kvxXtiKm
- https://t.co/jf0yA7RYZf
108K Followers 1 FollowingClaude is an AI assistant built by @anthropicai to be safe, accurate, and secure. Talk to Claude on https://t.co/ZhTwG8dz3D or download the app.
22K Followers 9 FollowingYour new async coding agent by @GoogleLabs. Built for devs, open to feedback, evolving with you. Dive in → https://t.co/iIzFEMmWgv
8K Followers 2K FollowingPrincipal research manager at Microsoft Research Amsterdam. Formerly at Google Brain and University of Amsterdam. PhD in condensed matter physics.
29K Followers 1K FollowingAI, national security, China. Part of the founding team at @CSETGeorgetown (opinions my own). Author of Rising Tide on substack: https://t.co/LKAoyL00iB
50K Followers 3K FollowingDeveloper Experience Lead at @GoogleDeepMind
Building Gemini API, Gemma, AI Studio and more AI products. My views
ex-Chief Llama Officer @huggingface 🇵🇪🇲🇽
64K Followers 1K FollowingCo-founder and CTO at @huggingface 🤗. ML/AI for everyone, building products to propel communities fwd. @Stanford + @Polytechnique
18 Followers 210 Following20, AI Bro
AI + Backend Eng @usefindr
AI research + DL model development on weekends
GIT : https://t.co/PXBgybtuj1
HF : https://t.co/2cuNHUXu85
141K Followers 139 FollowingWorking on a new terminal: Ghostty. 👻 Prev: founded @HashiCorp. Created Vagrant, Terraform, Vault, and others. Vision Jet Pilot. 👨✈️