Aggregated Programming Capability Evaluation on Overall Evaluation Suite
64.2Macro Average ScoreMIRA-Group
Evaluation results
| Method | Links | |
|---|---|---|
| MIRA-GroupMid-training Protocol=Subset Selection, Token Budget=25B2026.05 | 64.2 | |
| Raw MixtureMid-training Protocol=Full Corpus, Token Budget=50B2026.05 | 63.83 | |
| RandomMid-training Protocol=Subset Selection, Token Budget=25B2026.05 | 63.23 | |
| DataManMid-training Protocol=Subset Selection, Token Budget=25B2026.05 | 63.01 | |
| MIRA-SourceMid-training Protocol=Subset Selection, Token Budget=25B2026.05 | 62.93 | |
| MIRA-GlobalMid-training Protocol=Subset Selection, Token Budget=25B2026.05 | 61.81 | |
| DSIRMid-training Protocol=Subset Selection, Token Budget=25B2026.05 | 59.55 | |
| PPLMid-training Protocol=Subset Selection, Token Budget=25B2026.05 | 54.73 | |
| +SFTMid-training Protocol=No Mid-training, Token Budget=02026.05 | 48.6 | |
| Base ModelMid-training Protocol=No Mid-training, Token Budget=02026.05 | 28.16 |