Fresh rank-16, alpha-32 Qwen3.5-9B LoRA trained for one epoch on the 2,880 varied-deception rows using only binary soft-target BCE. Targets are the direct Prediction: probabilities of a Qwen3.5-27B Truth Value Guard teacher. Training used AdamW at 5e-5 with effective batch size 32 and did not use hard labels, GPT-OSS targets, or teacher/student reasoning traces.

Phoenix 5.0 renders observable conversation and output only, disables Qwen thinking, and normalizes the next-token logits for literal 0|1 directly after Prediction:. The adapter is intended for the matching competition notebook and is not a general-purpose truth classifier.

The adapter_model.safetensors SHA-256 is c096e5839fe96fb7409fe759a79bc0989ef1efe5aee010b83c3157d38586a75f. The deployed Hugging Face repository is Jazhyc/aletheias-phoenix-v5-qwen27-tvg-soft-r16; the initial upload commit is c5568edfe22fa335cbe31726ea144669e8b54314.