SMART Ships a Performance-Modeling Library Whose Main Branch Is Design Docs, Not Code

arXiv 2609.05364·high signalXBlueskyLinkedInCopy link

arXiv 2609.05364 describes SMART, a symbolic ML performance-modeling library whose repository is a DAG of self-contained natural-language design docs with almost no code; coding sub-agents regenerate the implementation from the docs on every version update, and every human change is a doc edit. Two things make regeneration reliable: a doc style built around step-by-step worked examples that act as in-context demonstrations, and a minimal recursively-defined operator IR with SymPy cost expressions. Regenerated implementations reproduce hand-audited reference models, including DeepSeek-V3 serving on a TPU pod slice, to round-off precision.