An accelerator-friendly QP solver is implemented that accumulates one Hessian per weight matrix and solves many small QPs in parallel with a convergence certificate per column, enabling one-shot post-training pruning at scale on a single accelerator without fine-tuning.
Mohammad Mozaffari, Samuel Kushnir, Maryam Mehri Dehnavi et al.· arXiv.org· 2 citations
Co-designing ML models and the accelerators that run them is an unusual reasoning task: architects must draw confident, high-stakes conclusions about systems that do not yet exist, and the pace of both model evolution and hardware cadence means the analysis burden grows every quarter. The evidence behind each decision-...
Samuel Kushnir, Kavya Sreedhar, Yeshwanth Reddy Pogula et al.· 0 citations
SMART is described, a rigorous symbolic performance-modeling library for ML systems whose main branch contains almost no code, and Regenerated implementations reproduce hand-audited reference models to round-off precision, suggesting that design docs can be the durable artifact for ML-systems co-design tools.
Samuel Kushnir, Kimia Noorbakhsh, Kavya Sreedhar et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.