We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control
deepmind.google·6d ago
AI Engineering
MIT Technology Review·Jun 19·5 min read
The Download: AI bottleneck debates, and BCI trials take off
TL;DR
Large language models (LLMs) have faced a computational bottleneck that limits their efficiency and energy use. The startup Subquadratic claims to have developed a new method that reduces the number of computations required by transformers, making LLMs faster and cheaper.
✦ Why It Matters
Engineers can explore Subquadratic's method to enhance LLM efficiency and reduce operational costs.
Key Takeaways
- Investigate Subquadratic's approach to LLMs for potential integration.
- Stay updated on BCI advancements for accessibility solutions.
- Consider energy efficiency in AI model development.
- Engage with ongoing BCI trials to understand user experiences.
How It Works
Subquadratic's breakthrough involves optimizing the computation process in transformer models, which are foundational to LLMs. By reducing the number of calculations needed to generate responses, the model can operate more efficiently, leading to faster outputs and lower operational costs.