You can clone with
HTTPS or Subversion.
Merge pull request #5 from dylski/master
Reinforcement learning environment interface added.
Errant newline fix
Reinforcement-learning environment interface.
Merge pull request #4 from schaul/addastar
Added Seaquest variant
Goat chase: interaction ordering bugfix
Collision ordering fix in boulderdash game.
bugfix: put end-of-screen handling back into the collision detection
speed-up complete: collision detection should not be a big bottleneck
additional speed-ups in collision detection, more to come
Minor fixes in tick
dramatic speed-up in collision detection
state load/save speedups.
Fixed: load/save also works again. And now also in resource-based games!
major resource refactoring and cleanup
small fixes, inheritance and import scope (must be restricted in an
ontology to what is allowed to be used in the language itself).
Added action maps + tick interface
full state export/import functionality, also now a quick-save button (1)
and a quick-load button (2).
refinements: additional action in boulder dash
no specialized code for frogger anymore, just using resources now.
Score is implemented: any interaction with a "scoreChange=<value>"
attribute will affect it.
Death from above.
Also makes Mario more powerful now.
fixed bug, improved boulderdash
better but still broken boulderdash
zombie apocalypse game