<ul data-eligibleForWebStory="true"><li>Constructing robust simulators is crucial for guiding policy in fields like healthcare and logistics.</li><li>Current methods often struggle with generalization and accuracy, especially when using Large Language Models (LLMs).</li><li>G-Sim is introduced as a hybrid framework for automating simulator construction.</li><li>G-Sim integrates LLM-driven structural design with empirical calibration.</li><li>It utilizes an LLM to propose and refine simulator components guided by domain knowledge.</li><li>G-Sim grounds the simulator in reality by estimating parameters using calibration techniques.</li><li>It can leverage likelihood-free and gradient-free methods for parameter estimation and simulation-based inference.</li><li>G-Sim is capable of handling non-differentiable and stochastic simulators.</li><li>By combining domain priors with empirical evidence, G-Sim generates reliable and causally-informed simulators.</li><li>This mitigates data-inefficiency and allows for robust system-level interventions in complex decision-making.</li></ul>

G-Sim: Generative Simulations with Large Language Models and Gradient-Free Calibration

Discover more