Search papers, labs, and topics across Lattice.
SPASM is introduced as a modular framework for generating stable, persona-consistent multi-turn dialogues, addressing issues like persona drift and echoing in LLM-LLM conversations. The framework includes persona creation, Client-Responder dialogue generation, and termination detection. A key component, Egocentric Context Projection (ECP), maintains a perspective-agnostic dialogue history and projects it into each agent's view, significantly reducing persona drift and eliminating echoing, as validated through human evaluation and embedding analyses.
LLM-generated dialogues can maintain stable personas and avoid "echoing" for much longer using a simple context projection technique, without any model retraining.
Large language models are increasingly deployed in multi-turn settings such as tutoring, support, and counseling, where reliability depends on preserving consistent roles, personas, and goals across long horizons. This requirement becomes critical when LLMs are used to generate synthetic dialogues for training and evaluation, since LLM--LLM conversations can accumulate identity-related failures such as persona drift, role confusion, and "echoing", where one agent gradually mirrors its partner. We introduce SPASM (Stable Persona-driven Agent Simulation for Multi-turn dialogue generation), a modular, stability-first framework that decomposes simulation into (i) persona creation via schema sampling, plausibility validation, and natural-language persona crafting, (ii) Client--Responder dialogue generation, and (iii) termination detection for coherent stopping. To improve long-horizon stability without changing model weights, we propose Egocentric Context Projection (ECP): dialogue history is stored in a perspective-agnostic representation and deterministically projected into each agent's egocentric view before generation. Across three LLM backbones (GPT-4o-mini, DeepSeek-V3.2, Qwen-Plus) and nine Client--Responder pairings, we construct a dataset of 4,500 personas and 45,000 conversations (500 personas X 10 conversations per pairing). Ablations show ECP substantially reduces persona drift and, under human validation, eliminates echoing; embedding analyses recover persona structure and reveal strong responder-driven interaction geometry. Our code is available at https://github.com/lhannnn/SPASM.
