Research map / Large language models / Language, safety and interpretability
Generating Persona Consistent Dialogues by Exploiting Natural Language Inference
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 |
|---|---|---|---|
| Use reinforcement learning with NLI consistency rewards | Persona-based dialogue generation · Persona-Chat test | Entailment (%): 14.62 -> 19.07 | Helped |
| Use reinforcement learning with NLI consistency rewards | Persona-based dialogue generation · Persona-Chat test | Contradiction (%): 9.00 -> 5.56 | Helped |
| Use RCDG_bert instead of Transformer | Persona-based dialogue generation · Persona-Chat | Perplexity: 28.1 -> 29.9 | Hurt |
| Use RCDG_bert instead of Transformer | Persona-based dialogue generation · Persona-Chat | Embedding Average: 63.4 -> 66.9 | Helped |
| Use RCDG_bert instead of Transformer | Persona-based dialogue generation · Persona-Chat | Distinct-2: 1505 -> 1275 | Hurt |
| Add adversarial naturalness module E_N (ablation) | Vanilla generator · Persona-Chat | Perplexity: 34.5 -> 32.3 | Helped |
| Add adversarial naturalness module E_N (ablation) | Vanilla generator · Persona-Chat | Embedding Average: 63.8 -> 64.8 | Helped |
| Add NLI consistency reward E_base without E_N (ablation) | Vanilla generator · Persona-Chat | Entailment (%): 13.23 -> 16.98 | Helped |
| Add NLI consistency reward E_base without E_N (ablation) | Vanilla generator · Persona-Chat | Perplexity: 34.5 -> 40.9 | Hurt |
| Add NLI consistency reward E_bert without E_N (ablation) | Vanilla generator · Persona-Chat | Entailment (%): 13.23 -> 17.13 | Helped |
| Add NLI consistency reward E_bert without E_N (ablation) | Vanilla generator · Persona-Chat | Perplexity: 34.5 -> 41.3 | Hurt |
| Add naturalness module E_N alongside NLI consistency reward (ablation) | Generator + E_base · Persona-Chat | Perplexity: 40.9 -> 30.2 | Helped |
Closest papers by meaning
- Learning to Know Myself: A Coarse-to-Fine Persona-Aware Training Framework for Personalized Dialogue Generation (AAAI 2023 · similarity 0.92)
- Consistently Simulating Human Personas with Multi-Turn Reinforcement Learning (NeurIPS 2025 · similarity 0.91)
- A Disentangled-Attention Based Framework with Persona-Aware Prompt Learning for Dialogue Generation (AAAI 2023 · similarity 0.90)
- Dual Task Framework for Improving Persona-Grounded Dialogue Dataset (AAAI 2022 · similarity 0.89)
- Teaching Language Models to Evolve with Users: Dynamic Profile Modeling for Personalized Alignment (NeurIPS 2025 · similarity 0.87)
- Enhancing Personalized Multi-Turn Dialogue with Curiosity Reward (NeurIPS 2025 · similarity 0.86)
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.