List your company

General Capability Evaluation on tinyBenchmarks

90AI2_arc AccuracyAAS
-3.620.74569.3Jan 27, 2026

Evaluation results

MethodLinks
AASModel=Qwen2.5-7B2026.019091694745
No SteeringModel=gemma-2-9b2026.019093717459
DirAblModel=gemma-2-9b2026.019094707459
AASModel=gemma-2-9b2026.019093727557
Selective SteeringModel=gemma-2-9b2026.019093717459
No SteeringModel=Qwen2.5-7B2026.018793646359
Selective SteeringModel=Qwen2.5-7B2026.018794656359
DirAblModel=Qwen2.5-7B2026.018692645761
No SteeringModel=Llama-3.1-8B2026.018185665651
DirAblModel=Llama-3.1-8B2026.018086675649
Selective SteeringModel=Llama-3.1-8B2026.018088665551
No SteeringModel=Qwen2.5-3B2026.018088616053
DirAblModel=Qwen2.5-3B2026.018082625750
Selective SteeringModel=Qwen2.5-3B2026.017988616153
ActAddModel=Qwen2.5-7B2026.017981683649
SASModel=Qwen2.5-7B2026.017981683649
AASModel=Qwen2.5-3B2026.017885523450
AASModel=Llama-3.1-8B2026.017788675747
DirAblModel=gemma-2-2b2026.017365565843
Selective SteeringModel=Llama-3.2-3B2026.017278615737
No SteeringModel=Llama-3.2-3B2026.017180615736
No SteeringModel=gemma-2-2b2026.017170545538
Selective SteeringModel=gemma-2-2b2026.017169545640
AASModel=Llama-3.2-3B2026.017081595642
AASModel=Qwen2.5-1.5B2026.017072505145
No SteeringModel=Qwen2.5-1.5B2026.016978534947
Selective SteeringModel=Qwen2.5-1.5B2026.016972524847
DirAblModel=Llama-3.2-3B2026.016775615934
DirAblModel=Qwen2.5-1.5B2026.016676484343
ActAddModel=Llama-3.1-8B2026.016164513935
SASModel=Llama-3.1-8B2026.016164513935
No SteeringModel=Llama-3.2-1B2026.014743462131
Selective SteeringModel=Llama-3.2-1B2026.014646422231
AASModel=Llama-3.2-1B2026.014535422036
DirAblModel=Llama-3.2-1B2026.014140383135
ActAddModel=Llama-3.2-3B2026.014168333936
AASModel=gemma-2-2b2026.01388131427
SASModel=Llama-3.2-3B2026.012446152029
ActAddModel=Llama-3.2-1B2026.011712732
SASModel=Llama-3.2-1B2026.011712732
ActAddModel=Qwen2.5-1.5B2026.018061810
SASModel=Qwen2.5-1.5B2026.018083717
ActAddModel=Qwen2.5-3B2026.0110000
SASModel=Qwen2.5-3B2026.0110000
ActAddModel=gemma-2-2b2026.0100010
SASModel=gemma-2-2b2026.0100010
ActAddModel=gemma-2-9b2026.0100000
SASModel=gemma-2-9b2026.0100000