L3C provides variant environments facilitating In-Context Learning, featured in the following assets:

• Vast amounts of diversified tasks with minimal inductive bias
• Lifelong In-Context Learning
• Generative and Interactive

• AnyMDP: Procedurally generated unlimited general-purpose Markov Decision Processes (MDP)

• MetaLanguage: Pseudo-language generated from randomized neural networks

• MazeWorld: Procedurally generated mazes with diverse maze structures, navigation goals to benchmark object navigation

• L3C_Baselines: An implementation of the baseline solutions to the above tasks

pip install l3c

Cite this work with

@article{wang2024benchmarking,
  title={Benchmarking General Purpose In-Context Learning},
  author={Wang, Fan and Lin, Chuan and Cao, Yang and Kang, Yu},
  journal={arXiv preprint arXiv:2405.17234},
  year={2024}
}