CantTalkAboutThis: A Dataset for Training Language Models to Stay on Topic in Dialogues
The core message of this article is to introduce the CANTTALKABOUTTHIS dataset, which is designed to train language models to maintain topical focus and avoid getting distracted during task-oriented dialogues.
Evaluating Large Language Models as Human-like Dialogue Systems: DialogBench Benchmark
DialogBench, a comprehensive benchmark, is proposed to standardize the evaluation of large language models (LLMs) as human-like dialogue systems, uncovering their strengths and limitations across diverse dialogue tasks.
BootTOD: Task-Oriented Dialogue Representations with Self-Bootstrapping Framework
BootTOD proposes a self-bootstrapping framework for task-oriented dialogue representations, outperforming contrastive methods.
Mix-Initiative Response Generation with Dynamic Prefix Tuning: Enhancing Dialogue Systems with Initiative Awareness
Initroducing Mix-Initiative Dynamic Prefix Tuning (IDPT) to enhance response generation in dialogue systems by incorporating initiative awareness.
Few-shot Dialogue Strategy Learning for Motivational Interviewing via Inductive Reasoning
DIIR framework enables learning and applying dialogue strategies for Motivational Interviewing, improving active listening skills and promoting collaborative responses.
CTSM: Combining Trait and State Emotions for Empathetic Response Model
CTSM combines trait and state emotions to enhance empathetic response generation by perceiving a comprehensive range of emotions in contextual dialogues.
CET2: Modeling Topic Transitions for Coherent and Engaging Knowledge-Grounded Conversations
Coherent and engaging knowledge selection is crucial for generating informative responses in dialogue systems.
CET2: Modeling Topic Transitions for Coherent and Engaging Knowledge-Grounded Conversations
The author introduces the CET2 framework to address shortcomings in knowledge selection methods, focusing on topic transitions for coherent conversations.
© 2024 by Linnk AI