JetSpec, a new speculative decoding method, trains a causal parallel draft head over fused hidden states from a frozen target model, enabling lossless verification of candidate trees in one forward pa…
Researchers introduced JetSpec, a speculative decoding method that trains a causal parallel draft head on a frozen target model to draft entire speculative trees in one pass while preserving autoregre…
NVIDIA Labs released cuTile Rust, a tile-based DSL that extends Rust's ownership model to GPU programming, enabling memory-safe, data-race-free kernels without performance loss. Benchmarks on the B200…
FastVideo has open-sourced Dreamverse, a real-time video generation workspace that enables "vibe directing" through natural-language iteration, releasing both the frontend and backend as a reference a…