Medical Report Classification on MIMIC
88.2Micro PrecisionLlama
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| LlamaTraining Mode=SFT + GRPO2026.04 | 88.2 | 77.7 | 82.6 | |
| MedalpacaTraining Mode=SFT + GRPO2026.04 | 88.1 | 81.3 | 84.6 | |
| LlamaTraining Mode=SFT2026.04 | 87.3 | 71 | 78.3 | |
| BioClinicalBERTTraining Mode=Baseline2026.04 | 87 | 62.8 | 73 | |
| Phi3Training Mode=SFT2026.04 | 84.5 | 50.8 | 63.5 | |
| Qwen-3BTraining Mode=SFT + GRPO2026.04 | 82.4 | 65.3 | 72.9 | |
| Qwen-3BTraining Mode=SFT2026.04 | 81.1 | 53.4 | 64.4 | |
| Phi3Training Mode=SFT + GRPO2026.04 | 79.1 | 64.8 | 71.2 | |
| MedalpacaTraining Mode=SFT2026.04 | 77 | 83 | 79.9 | |
| Qwen-7BTraining Mode=SFT + GRPO2026.04 | 76.9 | 80.9 | 78.8 | |
| GPT4-oTraining Mode=Baseline2026.04 | 75.8 | 90.7 | 82.6 | |
| Qwen-7BTraining Mode=SFT2026.04 | 69.1 | 82 | 74.9 | |
| LlamaTraining Mode=Baseline2026.04 | 68.4 | 13.5 | 22.5 | |
| Qwen-7BTraining Mode=Baseline2026.04 | 65.1 | 40.4 | 49.8 | |
| Phi3Training Mode=Baseline2026.04 | 51.1 | 36.8 | 42.8 | |
| Qwen-3BTraining Mode=Baseline2026.04 | 45.2 | 14.5 | 22 | |
| MedalpacaTraining Mode=Baseline2026.04 | 4.6 | 2.6 | 3.3 |