(N, K)-Puzzle: A Cost-Efficient Testbed for Benchmarking Reinforcement Learning Algorithms in Generative Language Model.
Yufeng ZhangLiyu ChenBoyi LiuYingxiang YangQiwen CuiYunzhe TaoHongxia YangPublished in: CoRR (2024)
Keyphrases
- reinforcement learning algorithms
- cost efficient
- language model
- reinforcement learning
- language modeling
- generative model
- state space
- model free
- probabilistic model
- markov decision processes
- information retrieval
- n gram
- language modeling framework
- function approximation
- learning algorithm
- temporal difference
- retrieval model
- document retrieval
- query expansion
- smoothing methods
- ad hoc information retrieval
- unsupervised learning
- reward function
- dynamic environments
- translation model
- vector space model
- document representation
- feature selection