Dettmers et al.

A new wave of AI hardware and software co-design is replacing heavy floating-point matrix multiplications with addition-only, ternary weight sets {-1, 0, 1}, enabling up to a 32-fold reduction in weig…

A Northeastern University researcher found that linear probes can identify which layers of a transformer model are critical for a task, enabling selective quantization that preserves 99–100% of full-p…