Multitask Language Understanding on MMLU (Score)
87.21MMLU ScoreTrinity Large Preview
Evaluation Results
| Method | Links | |
|---|---|---|
| Trinity Large Previewtuning=instruct-tuned2026.02 | 87.21 | |
| IProXCandidate Training Data=CoT, rho=0.32026.02 | 56.96 | |
| Llama3.2-3BCandidate Training Data=BioInstruct2026.02 | 56.61 | |
| Llama3.2-3BCandidate Training Data=CoT2026.02 | 56.53 | |
| IProXCandidate Training Data=CoT, rho=0.52026.02 | 56.48 | |
| IProXCandidate Training Data=CoT, rho=0.72026.02 | 56.26 | |
| IProXCandidate Training Data=BioInstruct, rho=0.32026.02 | 56.25 | |
| IProXCandidate Training Data=BioInstruct, rho=0.52026.02 | 56.21 | |
| Llama3.2-1BCandidate Training Data=CoT2026.02 | 56.17 | |
| IProXCandidate Training Data=BioInstruct, rho=0.72026.02 | 56.09 | |
| Llama3.2-1BCandidate Training Data=BioInstruct2026.02 | 55.93 |