Out-of-Distribution Detection on OBQA to MMLU
87.09AUROCETN
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ETNBackbone=Llama-3.1-8B, Measure=MP2026.04 | 87.09 | — | |
| ETNBackbone=Llama-3.1-8B, Measure=DE2026.04 | 86.27 | — | |
| ETNBackbone=Gemma-2-9B, Measure=MP2026.04 | 85.67 | — | |
| ETNBackbone=Llama-3.1-8B, Measure=MI2026.04 | 85.3 | — | |
| MAP_CEBackbone=Llama-3.1-8B, Measure=MP2026.04 | 83.48 | — | |
| MAP_CEBackbone=Gemma-2-9B, Measure=MP2026.04 | 83.48 | — | |
| ETNscoring=maximum probability2026.04 | 83.39 | 91.7 | |
| IB-EDLBackbone=Llama-3.1-8B, Measure=DE2026.04 | 83.16 | — | |
| IB-EDLBackbone=Llama-3.1-8B, Measure=MP2026.04 | 82.83 | — | |
| MCDBackbone=Llama-3.1-8B, Measure=MP2026.04 | 81.96 | — | |
| ETNBackbone=Gemma-2-9B, Measure=DE2026.04 | 80.92 | — | |
| DeepEnsBackbone=Llama-3.1-8B, Measure=MP2026.04 | 80.33 | — | |
| MCDBackbone=Gemma-2-9B, Measure=MP2026.04 | 78.99 | — | |
| DMMBackbone=Llama-3.1-8B, Measure=MP2026.04 | 78.98 | — | |
| DMMBackbone=Gemma-2-9B, Measure=MP2026.04 | 78.98 | — | |
| ETNBackbone=Gemma-2-9B, Measure=MI2026.04 | 78.82 | — | |
| DMMBackbone=Llama-3.1-8B, Measure=DE2026.04 | 78.47 | — | |
| DMMBackbone=Gemma-2-9B, Measure=DE2026.04 | 78.47 | — | |
| DeepEnsBackbone=Llama-3.1-8B, Measure=MI2026.04 | 75.67 | — | |
| DeepEnsBackbone=Gemma-2-9B, Measure=MP2026.04 | 75.44 | — | |
| MCDBackbone=Gemma-2-9B, Measure=MI2026.04 | 75.31 | — | |
| IB-EDLBackbone=Gemma-2-9B, Measure=MP2026.04 | 73.78 | — | |
| DMMBackbone=Llama-3.1-8B, Measure=MI2026.04 | 73.51 | — | |
| DMMBackbone=Gemma-2-9B, Measure=MI2026.04 | 73.51 | — | |
| MAP_EDLBackbone=Gemma-2-9B, Measure=DE2026.04 | 73.41 | — | |
| MAP_EDLBackbone=Gemma-2-9B, Measure=MP2026.04 | 73.38 | — | |
| MAP_EDLBackbone=Llama-3.1-8B, Measure=DE2026.04 | 72.53 | — | |
| MAP_EDLBackbone=Llama-3.1-8B, Measure=MP2026.04 | 72.17 | — | |
| MCDBackbone=Llama-3.1-8B, Measure=MI2026.04 | 69.58 | — | |
| IB-EDLBackbone=Gemma-2-9B, Measure=DE2026.04 | 68.66 | — | |
| IB-EDLBackbone=Llama-3.1-8B, Measure=MI2026.04 | 67.89 | — | |
| DeepEnsBackbone=Gemma-2-9B, Measure=MI2026.04 | 67.04 | — | |
| MAP_EDLBackbone=Llama-3.1-8B, Measure=MI2026.04 | 66.99 | — | |
| IB-EDLBackbone=Gemma-2-9B, Measure=MI2026.04 | 61.76 | — | |
| MAP_EDLBackbone=Gemma-2-9B, Measure=MI2026.04 | 59.13 | — | |
| MD2026.04 | 54.42 | 70.5 | |
| LLBackbone=Gemma-2-9B, Measure=MI2026.04 | 51.54 | — | |
| ODIN2026.04 | 50.11 | 61.35 | |
| LLBackbone=Llama-3.1-8B, Measure=MP2026.04 | 47.43 | — | |
| LLBackbone=Gemma-2-9B, Measure=MP2026.04 | 47.43 | — | |
| LLBackbone=Llama-3.1-8B, Measure=MI2026.04 | 46.37 | — |