List your company

Abductive Natural Language Inference on AlphaNLI n=300 per condition (test)

95AccuracyFlat
86.36888.60990.8593.091Jun 17, 2026

Evaluation results

MethodLinks
FlatModel=gemma-4-31B-it2026.069509828.6—
Transactional leadershipModel=gemma-4-31B-it2026.069301000—
accept-only ablationModel=gemma-4-31B-it2026.069301000—
Transformational leadershipModel=gemma-4-31B-it2026.0692.76.3994.8—
Transactional leadershipModel=gpt-oss-120b2026.0692.301000—
accept-only ablationModel=gpt-oss-120b2026.0692.301000—
Situational leadershipModel=gpt-oss-120b2026.0692.316.79813—
FlatModel=gpt-oss-120b2026.0692097.713—
Theory-free controlModel=gemma-4-31B-it2026.06920990—
directives-only ablationModel=gemma-4-31B-it2026.06926.3990—
directives-only ablationModel=gpt-oss-120b2026.0691.71097.38.7—
Situational leadershipModel=gemma-4-31B-it2026.0691.727.3984.8—
Theory-free controlModel=gpt-oss-120b2026.0690.3096.38.7—
Transformational leadershipModel=gpt-oss-120b2026.0690.31096.38.7—
Situational leadershipModel=llama-4-scout2026.0689.33495.327.5—
Theory-free controlModel=llama-4-scout2026.0688095.322.5—
directives-only ablationModel=llama-4-scout2026.068815.39620—
FlatModel=llama-4-scout2026.0687.309617.5—
Transformational leadershipModel=llama-4-scout2026.0687.315.395.320—
Transactional leadershipModel=llama-4-scout2026.0686.701000—
accept-only ablationModel=llama-4-scout2026.0686.701000—
accept-only ablationModel=gpt-oss-120b2026.06————92.3
accept-only ablationModel=llama-4-scout2026.06————86.7
accept-only ablationModel=gemma-4-31B-it2026.06————93
directives-only ablationModel=gpt-oss-120b2026.06————91.7
directives-only ablationModel=llama-4-scout2026.06————88
directives-only ablationModel=gemma-4-31B-it2026.06————92
FlatModel=gpt-oss-120b2026.06————92
FlatModel=llama-4-scout2026.06————87.3
FlatModel=gemma-4-31B-it2026.06————95
Situational leadershipModel=gpt-oss-120b2026.06————92.3
Situational leadershipModel=llama-4-scout2026.06————89.3
Situational leadershipModel=gemma-4-31B-it2026.06————91.7
Theory-free controlModel=gpt-oss-120b2026.06————90.3
Theory-free controlModel=llama-4-scout2026.06————88
Theory-free controlModel=gemma-4-31B-it2026.06————92
Transactional leadershipModel=gpt-oss-120b2026.06————92.3
Transactional leadershipModel=llama-4-scout2026.06————86.7
Transactional leadershipModel=gemma-4-31B-it2026.06————93
Transformational leadershipModel=gpt-oss-120b2026.06————90.3
Transformational leadershipModel=llama-4-scout2026.06————87.3
Transformational leadershipModel=gemma-4-31B-it2026.06————92.7