AI Paper Atlas

Research map / Multi-agent systems and games / LLM agents

Recursive Semantic Divergence for LLM Agent Consistency

Harshavardhan Abichandani, Penny Chong, Atin Ghosh, Daniel Dahlmeier

NeurIPS 2026 · Multi-agent systems and games › LLM agents · approach: LLM multi-agent decision making · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Studies run-to-run inconsistency of tool-using LLM agents from identical starting states and proposes a recursive semantic divergence measure for it, based only on the abstract's framing.

Problem

LLM agents with tool use behave inconsistently across independent runs from the same initial state (different tool calls, contradictory outputs, divergent strategies), and existing metrics that score task completion do not capture this consistency.

Approach

Proposes a recursive semantic divergence measure to quantify agent behavioral consistency across runs, as opposed to completion-based correctness scores. Further details are not available in the provided abstract.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
fine-tune with RSD reward (GRPO)Qwen-2.5-3B · ToolSandboxAUC CV: 0.90 -> 0.44Improved
fine-tune with RSD reward (GRPO)Qwen-2.5-7B · ToolSandboxAUC CV: 0.38 -> 0.28Improved
fine-tune with RSD reward (GRPO)Qwen-2.5-3B · tau2-benchAUC CV: 1.32 -> 0.91Improved
fine-tune with RSD reward (GRPO)Qwen-2.5-7B · tau2-benchAUC CV: 0.78 -> 0.89Hurt
minimize token entropy as rewardQwen-2.5-7B · ToolSandboxAUC CV: 0.38 -> 0.50Hurt
minimize token entropy as rewardQwen-2.5-3B · ToolSandboxAUC CV: 0.90 -> 0.50Improved
minimize semantic entropy as rewardQwen-2.5-7B · tau2-benchAUC CV: 0.78 -> 0.78No clear effect
SFT on expert trajectoriesQwen-2.5-7B · ToolSandboxAUC CV: 0.38 -> 0.45Hurt
SFT on expert trajectoriesQwen-2.5-3B · tau2-benchAUC: 0.200 -> 0.289Improved
mix progress delta reward into RSD reward (lower lambda) (ablation)RSD fine-tuned Qwen-2.5-7B · ToolSandboxAUC CV: 0.28 -> 0.36 -> 0.37 -> 0.42Hurt
mix progress delta reward into RSD reward (lower lambda) (ablation)RSD fine-tuned Qwen-2.5-7B · ToolSandboxAUC: 0.555 -> 0.533 -> 0.532 -> 0.518Hurt
fine-tune with RSD reward (GRPO)Qwen-2.5-7B · ToolSandbox (direct RSD eval)RSD score: 0.672 -> 0.317Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy