ray__TurboQuant: Redefining AI efficiency with extreme compression
https://research.google/blog/turboquant-redefining-ai-efficiency-with-extreme-compression/
Sounds like Multi-Head Latent Attention (MLA) from DeepSeek
TurboQuant: Redefining AI efficiency with extreme compression
https://research.google/blog/turboquant-redefining-ai-efficiency-with-extreme-compression/