ResearchBenchmarksMulti-agent tactical deconfliction on EvaluationMulti-agent tactical deconfliction on EvaluationFollow88AccuracySFT24.5641.0357.573.97Mar 30, 2026Evaluation resultsMethodMethodLinksAccuracyPrecisionRecallF1-scoreSFTFine-tuning=Supervised...Fine-tuning=Supervised Fine-Tuning, Model=SFT2026.0388756669GRPOFine-tuning=Group Rela...Fine-tuning=Group Relative Policy Optimization, Model=GRPO2026.0353754050BaseModel=BaseModel=Base2026.0327752031