Commonsense MCQ on MCSQA (Accuracy)
72.08Accuracyend-to-end
Evaluation results
| Method | Links | |
|---|---|---|
| end-to-endModel=GPT-4o-mini2026.06 | 72.08 | |
| standard cascadeModel=GPT-4o-mini2026.06 | 68.33 | |
| context-aware cascadeModel=GPT-4o-mini2026.06 | 67.92 | |
| end-to-endModel=Llama-3.1-8B-Instruct2026.06 | 67.12 | |
| context-aware cascadeModel=Llama-3.1-8B-Instruct2026.06 | 65.12 | |
| standard cascadeModel=Llama-3.1-8B-Instruct2026.06 | 64.62 | |
| end-to-endModel=Mistral-7B-Instruct-v0.32026.06 | 60.38 | |
| context-aware cascadeModel=Mistral-7B-Instruct-v0.32026.06 | 59.12 | |
| standard cascadeModel=Mistral-7B-Instruct-v0.32026.06 | 57.12 |