4-gate screening sweep on Apollo tau-MnAl calibration CIF: moment, Curie T, e_hull, MAE. Screening chain returns REJECT on a known RE-free hard magnet, with documented model biases that exceed prior worst-case bounds (NEMAD Curie T: -423 K vs prior -100 to -300 K range; ALIGNN e_hull: 2.39 eV/atom vs prior 0.45-1.6 eV/atom; tb2j MAE: 15x underprediction vs prior 20-40%). Easy axis direction correctly predicted (001 c-axis), qualifying MAE route as a uniaxial-anisotropy filter but not a quantitative K1 predictor for L10 intermetallics. Conclusion: systematic bias correction is mandatory before the chain can evaluate new RE-free magnet candidates.
Ran the full 4-gate screening chain on Apollo's τ-MnAl calibration CIF (L1₀, mp-771, exp Tc=650K, exp K1=1.5 MJ/m³). The chain REJECTs the textbook RE-free hard magnet on 3 of 4 gates. New worst-case bias bounds: NEMAD Curie T -423K, ALIGNN e_hull +2.39 eV/atom, tb2j MAE 15x underprediction. Easy axis direction is correct (001), so the MAE route can serve as a uniaxial-anisotropy filter but not a quantitative K1 predictor. Forward path blocked on bias correction.