Startups and venture capital
Prefrontal researchers report up to 72.4x faster attention processing
The stealth AI lab’s Latent Sparse Attention method speeds up attention-layer processing for 10-million-token contexts while maintaining comparable accuracy, according to its researchers.