List your company

Classification on MSciNLI

1.52Expected Calibration Error (ECE)CaliDist (Sa.)
-0.578813.588127.75541.9219Jun 4, 2026

Evaluation results

MethodLinks
CaliDist (Sa.)LLM=Gemma-3 (4B)2026.061.5223.97
CaliDist (As.)LLM=Gemma-3 (4B)2026.062.625.27
CaliDist (Sa.)LLM=Phi-4-mini (3.8B)2026.062.8922.68
CaliDist (Pr.)LLM=Phi-4-mini (3.8B)2026.063.5723.1
CaliDist (As.)LLM=Phi-4-mini (3.8B)2026.065.1422.9
CaliDist (Pr.)LLM=Gemma-3 (4B)2026.0610.3124.95
Temperature ScalingLLM=Phi-4-mini (3.8B)2026.0618.0526
FSDLLM=Phi-4-mini (3.8B)2026.0626.0732.6
EntropyLLM=Phi-4-mini (3.8B)2026.0627.2332.67
VanillaLLM=Phi-4-mini (3.8B)2026.0630.2232.77
ConsistencyLLM=Phi-4-mini (3.8B)2026.0632.9335.23
Temperature ScalingLLM=Gemma-3 (4B)2026.0652.6152.16
EntropyLLM=Gemma-3 (4B)2026.0653.1853.22
FSDLLM=Gemma-3 (4B)2026.0653.3953.45
VanillaLLM=Gemma-3 (4B)2026.0653.953.66
ConsistencyLLM=Gemma-3 (4B)2026.0653.9953.84