Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Feng, Zihao, Wang, Xiaoxue, Bai, Ziwei, Su, Donghang, Wu, Bowen, Yu, Qun, Wang, Baoxun
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2504.13592
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866910915049291776
author Feng, Zihao
Wang, Xiaoxue
Bai, Ziwei
Su, Donghang
Wu, Bowen
Yu, Qun
Wang, Baoxun
author_facet Feng, Zihao
Wang, Xiaoxue
Bai, Ziwei
Su, Donghang
Wu, Bowen
Yu, Qun
Wang, Baoxun
contents Intent detection, a critical component in task-oriented dialogue (TOD) systems, faces significant challenges in adapting to the rapid influx of integrable tools with complex interrelationships. Existing approaches, such as zero-shot reformulations and LLM-based dynamic recognition, struggle with performance degradation when encountering unseen intents, leading to erroneous task routing. To enhance the model's generalization performance on unseen tasks, we employ Reinforcement Learning (RL) combined with a Reward-based Curriculum Sampling (RCS) during Group Relative Policy Optimization (GRPO) training in intent detection tasks. Experiments demonstrate that RL-trained models substantially outperform supervised fine-tuning (SFT) baselines in generalization. Besides, the introduction of the RCS, significantly bolsters the effectiveness of RL in intent detection by focusing the model on challenging cases during training. Moreover, incorporating Chain-of-Thought (COT) processes in RL notably improves generalization in complex intent detection tasks, underscoring the importance of thought in challenging scenarios. This work advances the generalization of intent detection tasks, offering practical insights for deploying adaptable dialogue systems.
format Preprint
id arxiv_https___arxiv_org_abs_2504_13592
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Improving Generalization in Intent Detection: GRPO with Reward-Based Curriculum Sampling
Feng, Zihao
Wang, Xiaoxue
Bai, Ziwei
Su, Donghang
Wu, Bowen
Yu, Qun
Wang, Baoxun
Computation and Language
Intent detection, a critical component in task-oriented dialogue (TOD) systems, faces significant challenges in adapting to the rapid influx of integrable tools with complex interrelationships. Existing approaches, such as zero-shot reformulations and LLM-based dynamic recognition, struggle with performance degradation when encountering unseen intents, leading to erroneous task routing. To enhance the model's generalization performance on unseen tasks, we employ Reinforcement Learning (RL) combined with a Reward-based Curriculum Sampling (RCS) during Group Relative Policy Optimization (GRPO) training in intent detection tasks. Experiments demonstrate that RL-trained models substantially outperform supervised fine-tuning (SFT) baselines in generalization. Besides, the introduction of the RCS, significantly bolsters the effectiveness of RL in intent detection by focusing the model on challenging cases during training. Moreover, incorporating Chain-of-Thought (COT) processes in RL notably improves generalization in complex intent detection tasks, underscoring the importance of thought in challenging scenarios. This work advances the generalization of intent detection tasks, offering practical insights for deploying adaptable dialogue systems.
title Improving Generalization in Intent Detection: GRPO with Reward-Based Curriculum Sampling
topic Computation and Language
url https://arxiv.org/abs/2504.13592