Skip to main content

Flatland Baseline Results

Created on March 13|Last edited on March 13
Note: We have applied a smoothing coefficient of 0.85, which means that the final episode return on the graphs may differ slightly from the tabulated results reported in the paper.

3trains


10k20k30k40kTrainer Steps (eval)-28-26-24-22
: -
All
120
Good
40
Medium
40
Poor
40


5trains


All
120
Good
40
Medium
120
Poor
120