Share your thoughts, 1 month free Claude Pro on usSee more

GridWorld

Benchmarks

Task Name	Dataset Name	SOTA Result
Environment Unlearning	GridWorld	Reconstruction Rate (Pre-unlearn)100	18
Anomaly Detection	Gridworld (test)	Mechanism Mean AUPR88.25	13
Anomaly Detection	Gridworld newobj	AUROC98.22	13
Anomaly Detection	Gridworld mechanism	AUROC84.59	13
POMDP Abstraction	GridWorld 3 x 3	Validation Error0	10
Trajectory Unlearning	GridWorld (test)	Pre-unlearn Score96	9
Trajectory Unlearning	GridWorld	Unlearn Efficacy98	9
State Inference Attack	GridWorld	Pre-unlearn Acc99	9
Planning	8x8 two-room gridworld (test)	Validity (%)0.89	8
Reinforcement Learning	GridWorld	Average Update Time (s)0.1419	7
Navigation	Gridworld	Avg Episode Return42.9	6
Reinforcement Learning	Gridworld	Sample Complexity33	5
Navigation Reasoning	Gridworld o.o.d 20x20 (out-of-domain)	Pass@191.5	4
Navigation Reasoning	Gridworld 10x10 (in-domain)	Pass@1100	4
Optimal policy search	GridWorld	Iterations (gamma=0.9)8	3
Generating CFMDPs	GridWorld p = 0.4	Mean Execution Time (s)0.336	2
Generating CFMDPs	GridWorld p = 0.9	Mean Execution Time (s)0.261	2
Counterfactual Policy Evaluation	GridWorld p = 0.4 Catastrophic Path	Lowest Cumulative Reward-698	2
Counterfactual Policy Evaluation	GridWorld (p = 0.4) - Almost Catastrophic	Lowest Cumulative Reward14	2
Counterfactual Policy Evaluation	GridWorld p = 0.4 Slightly Suboptimal Path	Lowest Cumulative Reward19	2
Counterfactual Policy Evaluation	GridWorld (p = 0.9) Catastrophic Path	Lowest Cumulative Reward-698	2
Counterfactual Policy Evaluation	GridWorld (p = 0.9) - Almost Catastrophic	Cumulative Reward (Lowest)-697	2
Counterfactual Policy Evaluation	GridWorld (p = 0.9) Slightly Suboptimal Path	Lowest Cumulative Reward-495	2
Counterfactual Policy Evaluation	GridWorld p = 0.4	Worst-Case Counterfactual V(s0)230	2
Counterfactual Policy Evaluation	GridWorld p = 0.9	Avg Worst-Case Counterfactual V(s0)346	2

Showing 25 of 27 rows