Table Explicitation
95.1GriTStopMistral 24B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Mistral 24BApproach=LIFT (Last-Mile Fine-Tuning)2026.05 | 95.1 | 92.5 | |
| Mistral 24BApproach=EEFT (End-to-End Fine-Tuning)2026.05 | 94.2 | 91.2 | |
| Mistral 7BApproach=LIFT (Last-Mile Fine-Tuning)2026.05 | 92.6 | 88 | |
| Mistral 7BApproach=EEFT (End-to-End Fine-Tuning)2026.05 | 91.7 | 86.7 | |
| Phi-4 14BApproach=LIFT (Last-Mile Fine-Tuning)2026.05 | 91.5 | 87.2 | |
| Llama 3.2 3BApproach=EEFT (End-to-End Fine-Tuning)2026.05 | 88.8 | 82.4 | |
| Phi-4 14BApproach=EEFT (End-to-End Fine-Tuning)2026.05 | 88.7 | 83.2 | |
| Qwen 3 4BApproach=EEFT (End-to-End Fine-Tuning)2026.05 | 88.4 | 82.3 | |
| Qwen 3 4BApproach=LIFT (Last-Mile Fine-Tuning)2026.05 | 88.2 | 81.9 | |
| Llama 3.2 3BApproach=LIFT (Last-Mile Fine-Tuning)2026.05 | 87.8 | 81.4 | |
| Phi-4 14BApproach=SD (Self-Debug)2026.05 | 86.2 | 79 | |
| Llama 3.2 1BApproach=EEFT (End-to-End Fine-Tuning)2026.05 | 85.5 | 76.6 | |
| Mistral 24BApproach=SD (Self-Debug)2026.05 | 85.2 | 77.4 | |
| Llama 3.2 1BApproach=LIFT (Last-Mile Fine-Tuning)2026.05 | 84.1 | 74.9 | |
| Qwen 3 4BApproach=SD (Self-Debug)2026.05 | 82.5 | 75 | |
| Mistral 7BApproach=SD (Self-Debug)2026.05 | 76 | 63.4 | |
| Llama 3.2 3BApproach=SD (Self-Debug)2026.05 | 73.2 | 60.6 | |
| Llama 3.2 1BApproach=SD (Self-Debug)2026.05 | 67.9 | 54.4 |