WeightBridge, a weight-transfer library for reinforcement learning systems, reduces average GPU stall time by up to 42× over the state-of-the-art open-source RL framework across configurations spannin…
A developer built a perception daemon for computer-use agents that replaces the standard screenshot-reason-act loop with a continuously running process holding the current frame, cutting capture laten…
Prefix caching can push LLM prefill from compute-bound to memory-bandwidth-bound once enough of the prefill is loading cached KV rather than computing it, according to a preprint by Meng, Lee and Wang…
SGLang development is being augmented with agent-assisted workflows that encode procedural engineering knowledge into executable skills, covering LLM serving, GPU kernels, diffusion pipelines, and pro…
At the ninth MLSys conference in Seattle, researchers and industry leaders focused overwhelmingly on improving the efficiency of training and deploying large language models, with specialized hardware…