MineRL: A Large-Scale Dataset of Minecraft Demonstrations

Authors: William H. Guss, Brandon Houghton, Nicholay Topin, Phillip Wang, Cayden Codel, Manuela Veloso, Ruslan Salakhutdinov

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

Reproducibility Variable Result LLM Response
Research Type Experimental We demonstrate the hierarchality, diversity, and scale of the Mine RL dataset. Further, we show the difficulty of the Minecraft domain along with the potential of Mine RL in developing techniques to solve key research challenges within it.
Researcher Affiliation Academia Carnegie Mellon University, Pittsburgh, PA 15289, USA {wguss, bhoughton, ntopin, pkwang, ccodel, mmv, rsalakhu}@cs.cmu.edu
Pseudocode No The paper describes the data packaging format but does not include any pseudocode or algorithm blocks.
Open Source Code No The paper provides access to the dataset but does not explicitly state that the code for their data collection platform or methodology is open source or provide a link to it.
Open Datasets Yes Demo videos and more details about the dataset can be found at http://minerl.io.
Dataset Splits No The paper discusses training and testing but does not explicitly mention validation sets or specific data splitting percentages/counts for train, validation, and test.
Hardware Specification No The paper mentions running experiments but does not provide specific hardware details such as GPU/CPU models or memory.
Software Dependencies No The paper mentions building atop "Open AI baseline implementations" but does not specify software names with version numbers for reproducibility.
Experiment Setup Yes Observations are converted to grey scale and resized to 64x64. Due to the thousands of action combinations in Minecraft and the limitations of the baseline algorithms, we simplify the action space to be 10 discrete actions. We train each reinforcement learning method for 1500 episode (approximately 12 million frames).