Research map / Reinforcement learning / Visual and representation RL
Hierarchical Reinforcement Learning for Open-Domain Dialog
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 |
|---|---|---|---|
| apply VHRL to optimize the combined conversation rewards | VHRED · Interactive human evaluation, combined rewards | Total: 14.27 -> 14.49 | Helped |
| apply word-level REINFORCE to optimize the combined conversation rewards | VHRED · Interactive human evaluation, combined rewards | Total: 14.27 -> 13.84 | Hurt |
| add joint end-to-end training so worker gradients update the manager (ablation) | VHRL · Interactive human evaluation, combined rewards | Total: 12.24 -> 14.49 | Helped |
| use utterance-level reinforcement learning instead of word-level REINFORCE | VHRED initialized with RL · Open-domain dialog metrics | Repetition | Helped |
| use utterance-level reinforcement learning instead of word-level REINFORCE | VHRED initialized with RL · Open-domain dialog metrics | Semantic similarity | Helped |
| use word-level REINFORCE instead of VHRL | VHRED initialized with RL · Open-domain dialog metrics | Question | Helped |
| add a question reward as the sole optimization target | VHRL · Interactive human evaluation, question reward | Total: 11.14 -> 14.14 | Helped |
| optimize sentiment rather than combined rewards | VHRL · Interactive human evaluation, sentiment reward | Total: 14.49 -> 12.28 | Hurt |
Closest papers by meaning
- MA-RLHF: Reinforcement Learning from Human Feedback with Macro Actions (ICLR 2025 · similarity 0.90)
- SELF-EVOLVED REWARD LEARNING FOR LLMS (ICLR 2025 · similarity 0.89)
- Sequence to Sequence Reward Modeling: Improving RLHF by Language Feedback (AAAI 2025 · similarity 0.89)
- RLAIF vs. RLHF: Scaling Reinforcement Learning from Human Feedback with AI Feedback (ICML 2024 · similarity 0.89)
- Teaching Models to Improve on Tape (AAAI 2025 · similarity 0.88)
- ArCHer: Training Language Model Agents via Hierarchical Multi-Turn RL (ICML 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.