ResearchBenchmarksLogical Reasoning on Reasoning GymFollow4.2Avg@4 ScoreDr.GRPO0.041.122.23.28May 8, 2026Evaluation ResultsMethodMethodLinksAvg@4 ScoreDr.GRPOBackbone=Llama3.2-3B-I...Backbone=Llama3.2-3B-Instruct, Domain Type=Out-of-domain2026.054.2GRPOBackbone=Llama3.2-3B-I...Backbone=Llama3.2-3B-Instruct, Domain Type=Out-of-domain2026.053.6PRIMEBackbone=Llama3.2-3B-I...Backbone=Llama3.2-3B-Instruct, Domain Type=Out-of-domain2026.053.3OriginalBackbone=Llama3.2-3B-I...Backbone=Llama3.2-3B-Instruct, Domain Type=Out-of-domain2026.050.2