Research map / Large language models / Instruction and fine-tuning
PsyPARSE: Retrieval-Augmented Slow Thinking for Personalized Empathetic Counseling
Longxiang Wang, Pukun Zhao, Chen Chen, Jinhe Bi, Huacan Wang, Tong Zhang, Ronghao Chen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add PsyPARSE framework to a general LLM | DeepSeek-V3 · CPsyCounR simulated counseling, LLM eval Ctx. | Ctx.: 80.51 -> 95.72 | Improved |
| add PsyPARSE to GPT-4 | GPT-4 · CPsyCounR simulated counseling, LLM eval Ctx. | Ctx.: 66.93 -> 91.27 | Improved |
| add PsyPARSE to a small open model | Llama-3.1-8B · CPsyCounR simulated counseling, LLM eval Hum. | Hum.: 60.62 -> 88.13 | Improved |
| add PsyPARSE to a fine-tuned domain model | CPsyCounX · CPsyCounR simulated counseling, LLM eval Tec. | Tec.: 55.62 -> 82.17 | Improved |
| use PsyPARSE on a general 7B model instead of a fine-tuned counseling model | Qwen2.5-7B vs PsycoLLM · CPsyCounR simulated counseling, LLM eval Ctx. | Ctx.: 69.23 -> 88.02 | Improved |
| add PsyPARSE instead of Chain of Empathy prompting | DeepSeek-V3 · CPsyCounR simulated counseling, LLM eval Tec. | Tec.: 75.85 -> 95.50 | Improved |
| add Empathetic Meta-Chain prompting | DeepSeek-V3 · CPsyCounR simulated counseling, LLM eval Ctx. | Ctx.: 80.51 -> 78.39 | Hurt |
| add few-shot prompting | DeepSeek-V3 · CPsyCounR simulated counseling, LLM eval Tec. | Tec.: 75.69 -> 74.35 | Hurt |
| add multi-therapy RAG (ablation) | PsyPARSE · CPsyCounR simulated counseling, LLM eval Tec. | Tec.: 76.82 -> 95.50 | Improved |
| add slow-thinking engine (ablation) | PsyPARSE · CPsyCounR simulated counseling, LLM eval Hum. | Hum.: 84.83 -> 94.90 | Improved |
| add slow-thinking to RAG-only system (ablation) | PsyPARSE · CPsyCounR simulated counseling, LLM eval Tec. | Tec.: 89.47 -> 95.50 | Improved |
| use 3 rollout rounds instead of 1 (ablation) | PsyPARSE · CPsyCounR simulated counseling, LLM eval Ctx. | Ctx.: 88.95 -> 95.72 | Improved |
Closest papers by meaning
- STAMPsy: Towards SpatioTemporal-Aware Mixed-Type Dialogues for Psychological Counseling (AAAI 2025 · similarity 0.87)
- An LLM-based Simulation Framework for Embodied Conversational Agents in Psychological Counseling (AAAI 2026 · similarity 0.86)
- EmpathyChat: Structured Cognitive Reasoning in Empathetic Spoken Dialogue (NeurIPS 2026 · similarity 0.86)
- TherapyGym: Evaluating and Aligning Clinical Fidelity and Safety in Therapy Chatbots (NeurIPS 2026 · similarity 0.85)
- RLVER: Reinforcement Learning with Verifiable Emotion Rewards for Empathetic Agents (ICLR 2026 · similarity 0.85)
- Thousand Voices of Trauma: A Large-Scale Synthetic Dataset for Modeling Prolonged Exposure Therapy Conversations (NeurIPS 2025 · similarity 0.85)
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.
