Policy Evaluation

Benchmarks

Dataset Name	SOTA Method	Metric
PolicyBench Overall Average	Deepseek R1	Accuracy66.34	11	3mo ago
PolicyBench Level 3 US		Accuracy77	11	3mo ago
PolicyBench Level 3 CN		Accuracy80.34	11	3mo ago
PolicyBench Level 2 (US)		Accuracy68.95	11	3mo ago
PolicyBench Level 2 (CN)	Deepseek R1	Accuracy62.92	11	3mo ago
PolicyBench Level 1 (US)	Deepseek R1	Accuracy59.33	11	3mo ago
PolicyBench Level 1 (CN)	Deepseek R1	Accuracy62.02	11	3mo ago
Out-of-distribution	SC3-Eval	Return (r)96.2	8	1mo ago
In-distribution	SC3-Eval	r Score98.4	8	1mo ago
20-Link Pole off-policy	BBO	Sum of sqrt MSE415.37	7	5mo ago
400-State Random MDP on-policy	BBO	Sum of sqrt MSE24.74	7	5mo ago
14-State Boyan Chain on-policy		Sum of sqrt MSE25.06	7	5mo ago
20-Link Pole off-policy	BBO	MSE4.17	7	5mo ago
20-Link Pole on-policy	BBO	MSE4.26	7	5mo ago
Cart-Pole off-policy perfect features	BBO	MSE0.17	7	5mo ago
Cart-Pole on-policy, perfect features	BBO	MSE0.15	7	5mo ago
Cart-Pole off-policy, impoverished features		MSE2.33	7	5mo ago
Cart-Pole on-policy, impoverished features		MSE2.37	7	5mo ago
400-State Random MDP (off-policy)	BBO	MSE0.11	7	5mo ago
400-State Random MDP on-policy	BBO	MSE0.07	7	5mo ago
14-State Boyan Chain on-policy		MSE0.1	7	5mo ago
BPIC real logs 2017	MDP-based RL method	Average KPI (Mπ)1,731.108	6	1mo ago
BPIC real logs 2012	MDP-based RL method	Average KPI (Mπ)374.966	6	1mo ago
2x2 Queueing Network rho = (0.95, 0.85)		Expected Value1,185.4	6	1mo ago
2x2 Queueing Network rho = (0.90, 0.90)	BTSD-PPO	Value1,316.9	6	1mo ago

Showing 25 of 31 rows