GradSAT uses Gradient Normalization to Solve Floating-Point SMT Formulas

GradSAT addresses gradient domination in optimization-based SMT solvers by treating each clause as an independent multi-task learning problem. By applying dynamic gradient normalization, the framework prevents difficult clauses from hijacking the optimization trajectory and trapping solvers in local minima.