PN — Evaluated model: gemma3:12b

gemma3:12b — family Gemma 3 (Google), GGUF Q4_K_M, run locally via Ollama (seed=42, temp=0.0). Phase-B results: triage F1-macro 0.6440469372305059, estimation MAE 2.66 SP, prioritization accuracy 0.57 (N=100 each). Best triage model in the sweep (F1 0.644).

Source: PUMA project documentation + raw benchmark results · Traceability: ANXAG-015,PUMA-023 · Confidence: verified-at-primary-source