Today, we’re announcing Veo 2: our state-of-the-art video generation model which produces realistic, high-quality clips from text or image prompts. 🎥
We’re also releasing an improved version of our text-to-image model, Imagen 3 - available to use in ImageFX through @LabsDotGoogle. → https://t.co/zMJQwON4Gx
One disadvantage Kamala has relative to Trump is that we take her seriously. So when she suggests dumb policies like focusing on grocery prices, everyone criticizes her, whereas when Trump suggests much dumber things, we just ignore him.
Excited to announce @LaminiAI Memory Tuning, a new research breakthrough! 🎉
◽95%+ accuracy, cutting hallucinations by 10x
◽Turns any open LLM into a 1M-way adapter MoE (paper & Lamini-1 model weights on @Huggingface)
◽Fortune 500 customer case study on how they memory-tuned their SQL agent to reach 95% accuracy, when they could only get to 50% with advanced RAG & instruction fine-tuning
👉Drop us a note at [email protected] to Memory Tune on your data. We read every message.
🧱Come to my Databricks talk at 11am today for a deep dive on the technical paper & customer case study.
@deliprao This is becoming very tiring now. Why cannot we go back in hibernation with this AI stuff. Let builders build and researchers do their research.
@demishassabis Certainly, but do you really see any kind of relationship between LLMs in their current state and "extinction-level risk"? The whole thing sounds not only sensationalist and potentially dangerous (that's the perfect material for conspiracy circles)
@ylecun Honestly I am getting a huge hangover from this AI chatter. Why cannot go back to the stealth mode like before where we were just focused on building stuff? Now a lot of this just feels like a marketing gimmick.