<aside> 📌 Centralised dataset for MARL experiments recorded from published papers focusing on cooperative settings

</aside>

Columns:

• About the paper:

  • Paper title
  • Authorship
  • Year

• Implementation details:

  • Environment_class: Used environment (e.g SMAC)
  • Environment: Used scenario/map (e.g 2s3z, predator-prey)
  • Environment_settings
  • Environment_version: Version of the used environment (e.g. SC2.4.10 in SMAC)
  • Algorithm
  • Algorithm settings
  • Libraries

• Evaluation procedure:

  • Metric: with local aggregation or normalization (e.g. Average episode return, normalized reward)
  • General_metric: the name of the metric used in general without details (e.g. Return, reward)
  • Metric_settings
  • Aggregate_function
  • Measure_of_spread: uncertainty quantification
  • Number_of_seeds: number of random seeds or independent runs
  • Runs_per_seed
  • Independent_variable: training time
  • Evaluation_frequency: Number of independent evaluation per interval
  • Evaluation_duration: Evaluation intervals

• Recording results:

  • Read_from_table:does the result recorded from tabular results or from plots
  • Copy_results_from_another: does the result copied from previous work.
  • Have_CI: does the results have margin of error values
  • Metric_value: recorded result
  • CI_value: margin of error value
  • Independent_variable_value: recorded training time value
  • Evaluation_frequency_value: recorded evaluation frequency value
  • Evaluation_duration_value: recorded evaluation duration value

MARL dataset:

Untitled