Research map / Label-efficient and robust learning / Recognition and long-tail
Modeling Dialogues with Hashcode Representations: A Nonparametric Approach
Sahil Garg, Irina Rish, Guillermo Cecchi, Palash Goyal, Sarik Ghazarian, Shuyang Gao, Greg Ver Steeg, Aram Galstyan
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 LSH-RkNN hashing model instead of VHRED | dialogue model · Depression human eval | Appropriate (%): 9.5 -> 28.7 | Improved |
| use LSH-RkNN instead of VHRED | dialogue model · Depression human eval | Diverse (%): 0.7 -> 35.1 | Improved |
| use LSH-RkNN instead of VHRED | dialogue model · Larry King human eval | Appropriate (%): 5.3 -> 31.7 | Improved |
| use LSH-RLSTM instead of VHRED | dialogue model · Larry King human eval | Diverse (%): 0.0 -> 36.7 | Improved |
| use LSH-RLSTM instead of VHRED | dialogue model · Depression Therapy | embedding average: 0.48 -> 0.64 | Improved |
| use LSH-RkNN instead of VHRED | dialogue model · Depression Therapy | embedding extrema: 0.29 -> 0.37 | Improved |
| use LSH-RkNN instead of VHRED | dialogue model · Twitter | embedding average: 0.53 -> 0.61±0.17 | Improved |
| use LSH-RkNN instead of VHRED | dialogue model · Twitter | embedding extrema: 0.38 -> 0.25±0.09 | Hurt |
| use LSH-RLSTM instead of VHRED | dialogue model · Larry King | embedding greedy: 0.72±0.25 -> 0.58±0.21 | Hurt |
| generate responses with hashed N-gram language model instead of selecting from training set | LSH-RkNN · Depression human eval (300 samples) | Appropriate (%): 36.8 -> 33.8 | No clear effect |
| generate responses with hashed N-gram language model instead of selecting from training set | LSH-RkNN · Depression human eval (300 samples) | Diverse (%): 80 -> 12.0 | Hurt |
| replace RkNN hashing with RLSTM hashing | hashcode model · Twitter | MI LB: 17.3±0.1 -> 41.9±0.1 | Improved |
Closest papers by meaning
- Multi-View Feature Representation for Dialogue Generation with Bidirectional Distillation (AAAI 2021 · similarity 0.86)
- DialogWAE: Multimodal Response Generation with Conditional Wasserstein Auto-Encoder (ICLR 2019 · similarity 0.86)
- AvgOut: A Simple Output-Probability Measure to Eliminate Dull Responses (AAAI 2020 · similarity 0.85)
- A Comprehensive Analysis of the Effectiveness of Large Language Models as Automatic Dialogue Evaluators (AAAI 2024 · similarity 0.85)
- Aligning Spoken Dialogue Models from User Interactions (ICML 2025 · similarity 0.84)
- Friends-MMC: A Dataset for Multi-modal Multi-party Conversation Understanding (AAAI 2025 · similarity 0.84)
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.
