Research map / Large language models / Retrieval and knowledge
A Knowledge-Grounded Neural Conversation Model
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 retrieved facts to the grounded conversation task | MTaskR · Grounded conversation test | BLEU-4: 0.55 -> 1.08 | Helped |
| add retrieved facts to the grounded conversation task | MTaskR · Grounded conversation test | Informativeness: 0.479 -> 0.521 | Helped |
| add retrieved facts to the grounded conversation task | MTaskR · Grounded conversation test | Appropriateness: 0.495 -> 0.505 | No clear effect |
| add a second NoFacts task using grounded conversations, without facts (ablation) | MTask · Grounded conversation test | Informativeness: 0.501 vs 0.499 | No clear effect |
| replace grounded-task responses with retrieved facts in the Autoencoder task (ablation) | MTaskF · Grounded conversation test | Informativeness: 0.478 vs 0.522 | Helped |
| replace grounded-task responses with retrieved facts in the Autoencoder task (ablation) | MTaskF · Grounded conversation test | Appropriateness: 0.537 vs 0.463 | Hurt |
| replace grounded-task responses with retrieved facts in the Autoencoder task (ablation) | MTaskF · Grounded conversation test | BLEU-4: 0.55 -> 0.48 | Hurt |
| add the Autoencoder task alongside Facts and NoFacts tasks (ablation) | MTaskRF · Grounded conversation test | 2-gram diversity: 14.4% -> 26.0% | Helped |
| add the Autoencoder task alongside Facts and NoFacts tasks (ablation) | MTaskRF · Grounded conversation test | Informativeness: 0.479 vs 0.508 | No clear effect |
| combine Facts-task training with the general NoFacts task (ablation) | MTaskR · General data | Perplexity: 55.0 -> 55.1 | Helped |
Closest papers by meaning
- Zero-Resource Knowledge-Grounded Dialogue Generation (NeurIPS 2020 · similarity 0.91)
- Low-Resource Knowledge-Grounded Dialogue Generation (ICLR 2020 · similarity 0.89)
- Guiding Attention in Sequence-to-Sequence Models for Dialogue Act Prediction (AAAI 2020 · similarity 0.89)
- A Controllable Model of Grounded Response Generation (AAAI 2021 · similarity 0.89)
- Topic Aware Neural Response Generation (AAAI 2017 · similarity 0.88)
- Towards a Neural Conversation Model With Diversity Net Using Determinantal Point Processes (AAAI 2018 · 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.