List your company

Classification on MNLI (ECE, BS)

1.85ECETemperature Scaling
0.315210.675121.03531.3949Jun 4, 2026

Evaluation results

MethodLinks
Temperature ScalingLLM=Phi-4-mini (3.8B)2026.061.8516.82
CaliDist (As.)LLM=Phi-4-mini (3.8B)2026.063.0216.99
CaliDist (Pr.)LLM=Phi-4-mini (3.8B)2026.063.417.03
CaliDist (Sa.)LLM=Phi-4-mini (3.8B)2026.063.6817.31
CaliDist (Pr.)LLM=Gemma-3 (4B)2026.063.7123.29
CaliDist (Sa.)LLM=Gemma-3 (4B)2026.065.7724.1
CaliDist (As.)LLM=Gemma-3 (4B)2026.066.4524.51
VanillaLLM=Phi-4-mini (3.8B)2026.068.0517.66
ConsistencyLLM=Phi-4-mini (3.8B)2026.069.718.53
FSDLLM=Phi-4-mini (3.8B)2026.0615.1621.69
EntropyLLM=Phi-4-mini (3.8B)2026.0628.0928.65
Temperature ScalingLLM=Gemma-3 (4B)2026.0638.4738.61
EntropyLLM=Gemma-3 (4B)2026.0639.7539.9
FSDLLM=Gemma-3 (4B)2026.0639.839.98
ConsistencyLLM=Gemma-3 (4B)2026.0640.2140.25
VanillaLLM=Gemma-3 (4B)2026.0640.2240.16