Grounded Iterative Language Planning: How Parameterized World Models Reduce Hallucination Propagation in LLM Agents
For developers of LLM-based agents, GILP offers a practical method to reduce hallucination propagation in planning tasks by grounding LLM reasoning with a parameterized world model.
The paper introduces Grounded Iterative Language Planning (GILP), which combines a small parameterized world model with LLM-based reasoning to reduce hallucination propagation in language agents. On graph-structured planning benchmarks, GILP reduces hallucinated-state rate from 0.176 to 0.035 and increases success rate from 0.668 to 0.838 with only ~22% extra LLM calls.
World models for language agents come in two useful forms. An agent-based world model calls an LLM API and reasons flexibly in language, but its errors appear as hallucinated state changes that are hard to score with ordinary regression losses. A parameterized world model is a trained transition predictor; its errors are easier to measure with quantities such as NodeMSE, delta accuracy, and validity accuracy, but it is usually weaker as a standalone planner. We compare these two families on four graph-structured planning benchmarks and introduce operational hallucination metrics for the agent-based case. The comparison motivates \textbf{Grounded Iterative Language Planning} (GILP), which trains only a small parameterized backbone and combines it with API-based agent reasoning. The backbone supplies valid actions, predicted state deltas, risk, and value; the LLM drafts an action and imagined delta; and a consistency gate asks for revision when the two disagree. On real GPT-4o-mini calls, GILP reduces hallucinated-state rate from 0.176 to 0.035. In calibrated simulator ablations, it raises success from 0.668 to 0.838 while adding only ~22% extra LLM calls.