Rank-16, alpha-32 LoRA produced by one epoch of varied-only GRPO from the selected privileged-reasoning SFT adapter. Phoenix 4.4 uses the frozen leaderboard-aligned readout: observable conversation and output only, Qwen thinking disabled, an empty <reasoning> scaffold, and normalized logits for literal Prediction:0|1. It is intended for the matching competition notebook and is not a general-purpose truth classifier.