Research map / Probabilistic and scientific modeling / Variational and generative sampling
A Hierarchical Latent Variable Encoder-Decoder Model for Generating Dialogues
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 a turn-level stochastic latent variable to HRED | Twitter dialogue response generation · Twitter short contexts | human preference: 32.3 ± 2.4 wins; 42.5 ± 2.6 losses | Hurt |
| Add a turn-level stochastic latent variable to HRED | Twitter dialogue response generation · Twitter long contexts | human preference wins: 41.9 ± 2.2 | Helped |
| Add a turn-level stochastic latent variable to HRED | Twitter dialogue response generation · Twitter short and long contexts | human preference wins: 42.0 ± 2.8; 41.5 ± 2.8 | Helped |
| Add a turn-level stochastic latent variable to HRED | Dialogue response generation · Twitter 1-turn | Embedding Average: 0.512 -> 0.533 | Helped |
| Add a turn-level stochastic latent variable to HRED | Dialogue response generation · Ubuntu 1-turn | Embedding Average: 0.577 -> 0.542 | Hurt |
| Add a turn-level stochastic latent variable to HRED | Dialogue response generation · Twitter and Ubuntu 3-turn | Embedding Average: 0.646 -> 0.689; 0.742 -> 0.777 | Helped |
| Add a turn-level stochastic latent variable to HRED | Dialogue response generation · Twitter 1-turn | average response length: 11.64 -> 12.29 | Helped |
| Use decoder word dropout and gradually increase the KL-term weight | VHRED training · Twitter and Ubuntu | training stability and latent-variable use | Helped |
| Apply batch normalization to the feed-forward networks (ablation) | VHRED training · Dialogue response generation | variational bound | No clear effect |
| Keep the deterministic connection from the context RNN to the decoder RNN | VHRED · Dialogue response generation | variational bound | Helped |
Closest papers by meaning
- Building End-To-End Dialogue Systems Using Generative Hierarchical Neural Network Models (AAAI 2016 · similarity 0.93)
- Mechanism-Aware Neural Machine for Dialogue Response Generation (AAAI 2017 · similarity 0.91)
- Improving Variational Encoder-Decoders in Dialogue Generation (AAAI 2018 · similarity 0.91)
- Towards Diverse, Relevant and Coherent Open-Domain Dialogue Generation via Hybrid Latent Variables (AAAI 2023 · similarity 0.89)
- DialogBERT: Discourse-Aware Response Generation via Learning to Recover and Rank Utterances (AAAI 2021 · similarity 0.88)
- From Retrieval to Generation: A Simple and Unified Generative Model for End-to-End Task-Oriented Dialogue (AAAI 2024 · similarity 0.88)
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.