ResearchBenchmarksExploitability Computation on 256-Kuhn Poker (full)Follow0.007ExploitabilityAction Embeddings-0.297321.756843.8115.86516May 15, 2026Evaluation ResultsMethodMethodLinksExploitabilityAction Embeddingsk=512k=5122026.050.007Action Embeddingsk=256k=2562026.050.009Action Embeddingsk=128k=1282026.050.01Action Embeddingsk=64k=642026.050.035Action Embeddingsk=32k=322026.050.06Random baselinek=512k=5122026.050.113Action Embeddingsk=2k=22026.050.143Action Embeddingsk=16k=162026.050.18Random baselinek=256k=2562026.050.181Random baselinek=2k=22026.050.24Random baselinek=128k=1282026.050.254Action Embeddingsk=8k=82026.050.27Action Embeddingsk=4k=42026.050.29Random baselinek=64k=642026.050.32Random baselinek=16k=162026.050.323Random baselinek=32k=322026.050.323Random baselinek=4k=42026.050.37Random baselinek=8k=82026.050.38Hand-bucketing baselinek=2k=22026.051.091Hand-bucketing baselinek=4k=42026.051.712Random baselinek=1k=12026.051.887Action Embeddingsk=1k=12026.051.887Hand-bucketing baselinek=1k=12026.051.887Hand-bucketing baselinek=16k=162026.051.9Hand-bucketing baselinek=64k=642026.051.991Hand-bucketing baselinek=256k=2562026.052.949Hand-bucketing baselinek=512k=5122026.052.949Hand-bucketing baselinek=128k=1282026.057.181Hand-bucketing baselinek=32k=322026.057.51Hand-bucketing baselinek=8k=82026.057.615