Park: An Open Platform for Learning-Augmented Computer Systems

Authors: Hongzi Mao, Parimarjan Negi, Akshay Narayan, Hanrui Wang, Jiacheng Yang, Haonan Wang, Ryan Marcus, ravichandra addanki, Mehrdad Khani Shirkoohi, Songtao He, Vikram Nathan, Frank Cangialosi, Shaileshh Venkatakrishnan, Wei-Hung Weng, Song Han, Tim Kraska, Dr.Mohammad Alizadeh

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, we present the performance of existing RL approaches over those 12 problems and outline potential areas of future work. We benchmark the 12 systems in Park with both RL methods and existing heuristic baselines ( 5). The experiments benchmark the training efficiency and the eventual performance of RL approaches on each task.
Researcher Affiliation Academia Hongzi Mao, Parimarjan Negi, Akshay Narayan, Hanrui Wang, Jiacheng Yang, Haonan Wang, Ryan Marcus, Ravichandra Addanki, Mehrdad Khani, Songtao He, Vikram Nathan, Frank Cangialosi, Shaileshh Bojja Venkatakrishnan, Wei-Hung Weng, Song Han, Tim Kraska, Mohammad Alizadeh MIT Computer Science and Artificial Intelligence Laboratory park-project@csail.mit.edu
Pseudocode Yes In Figure 3, Algorithm 1 depicts this interaction process. ... In Figure 3, Algorithm 2 outlines this option in simulated system environments.
Open Source Code Yes We open-source Park as well as the RL agents and baselines in https://github.com/park-project/park.
Open Datasets Yes For example, for the CDN memory caching environment, we use an open dataset containing 500 million requests, collected from a public CDN serving top-ten US websites [15]. and CIFAR-10 [52] and Penn Tree Bank [65].
Dataset Splits No The paper mentions training and testing but does not provide specific details on how the dataset was split into training, validation, and test sets (e.g., percentages or sample counts).
Hardware Specification No The paper does not provide specific details about the hardware (e.g., CPU, GPU models, memory) used to run the experiments.
Software Dependencies No The paper mentions various software and frameworks like DQN, A2C, Policy Gradient, DDPG, TensorFlow, Open AI Gym, and Open AI Baselines, but it does not specify their version numbers.
Experiment Setup Yes The details of hyperparameter tunings, agent architecture and system configurations are in Appendix B.