q-values
Q-values, or action-value functions, represent the expected utility of taking a specific action in a given state in reinforcement learning. They are key to algorithms that optimize decision-making by evaluating potential future rewards.
- Actor-Free Continuous Control via Structurally Maximizable Q-Functions
- BraVE: Offline Reinforcement Learning for Discrete Combinatorial Action Spaces
- Coarse-to-fine Q-Network with Action Sequence for Data-Efficient Reinforcement Learning
- ShiQ: Bringing back Bellman to LLMs
- Trajectory Bellman Residual Minimization: A Simple Value-Based Method for LLM Reasoning
- Value Diffusion Reinforcement Learning