Research map / Architectures and efficiency / RNNs and normalization
Non-Autoregressive Dialog State Tracking
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 NADST's non-autoregressive joint state tracking | DST on MultiWOZ 2.1 · MultiWOZ 2.1 | Joint accuracy: 45.60% -> 49.04% | Helped |
| Use NADST's non-autoregressive joint state tracking | DST on MultiWOZ 2.0 · MultiWOZ 2.0 | Joint accuracy: 48.60% -> 50.52% | Helped |
| Use NADST's parallel decoding | DST on MultiWOZ 2.1 · MultiWOZ 2.1 | Latency: 362.15 -> 27.31 | Helped |
| Increase the attention setting from T=1 to T=3 | NADST · MultiWOZ 2.1 | Joint accuracy: 42.98% -> 49.04% | Helped |
| Increase the attention setting from T=1 to T=3 | NADST · MultiWOZ 2.1 | Latency: 15.18 -> 27.31 | Hurt |
| Add positional encoding to the state decoder input (ablation) | NADST · MultiWOZ 2.1 | Joint accuracy: 19.56% -> 49.04% | Helped |
| Add slot gating (ablation) | NADST · MultiWOZ 2.1 | Joint accuracy: 48.74% -> 49.04% | Helped |
| Add partially delexicalized dialogue history (ablation) | NADST · MultiWOZ 2.1 | Joint accuracy: 39.45% -> 49.04% | Helped |
| Add system actions to dialogue history delexicalization (ablation) | NADST · MultiWOZ 2.1 | Joint accuracy: 44.87% -> 49.04% | Helped |
| Add pointer generation (ablation) | NADST · MultiWOZ 2.1 | Joint accuracy: 43.46% -> 49.04% | Helped |
| Replace non-autoregressive state decoding with autoregressive decoding (ablation) | NADST · MultiWOZ 2.1 | Joint accuracy: 49.04% -> 49.76% | No clear effect |
| Provide ground-truth rather than predicted state-length inputs (ablation) | NADST · MultiWOZ 2.1 | Joint accuracy: 49.04% -> 67.32% | Helped |
Closest papers by meaning
- MA-DST: Multi-Attention-Based Scalable Dialog State Tracking (AAAI 2020 · similarity 0.92)
- TA&AT: Enhancing Task-Oriented Dialog with Turn-Level Auxiliary Tasks and Action-Tree Based Scheduled Sampling (AAAI 2024 · similarity 0.88)
- Aligning Spoken Dialogue Models from User Interactions (ICML 2025 · similarity 0.86)
- From Retrieval to Generation: A Simple and Unified Generative Model for End-to-End Task-Oriented Dialogue (AAAI 2024 · similarity 0.85)
- An Efficient Self-Learning Framework For Interactive Spoken Dialog Systems (ICML 2024 · similarity 0.85)
- ToolACE-MT: Non-Autoregressive Generation for Agentic Multi-Turn Interaction (ICLR 2026 · similarity 0.85)
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.