General Multitask Evaluation on Math500, GPQA, HumanEval, MBPP, AE2 LC Aggregate
40.7Average ScoreLlama3.2-3B-GRLO+RLVR
Evaluation results
| Method | Links | |
|---|---|---|
| Llama3.2-3B-GRLO+RLVRBackbone=Llama3.2-3B, Training Strategy=GRLO+RLVR2026.05 | 40.7 | |
| Llama3.2-3B-GRLOBackbone=Llama3.2-3B, Training Strategy=GRLO2026.05 | 39.3 | |
| Llama3.2-3B-InstructBackbone=Llama3.2-3B, Training Strategy=Instruct2026.05 | 35.6 | |
| Llama3.2-3B-RLVRBackbone=Llama3.2-3B, Training Strategy=RLVR2026.05 | 30.7 | |
| Llama3.2-3B-SFTBackbone=Llama3.2-3B, Training Strategy=SFT2026.05 | 21.5 |