Multilingual Code Generation on Multipl-E
72.84MultiplEMIRA-Source
Evaluation results
| Method | Links | |
|---|---|---|
| MIRA-SourceMid-training Protocol=Subset Selection, Token Budget=25B2026.05 | 72.84 | |
| +SFTMid-training Protocol=No Mid-training, Token Budget=02026.05 | 72.57 | |
| MIRA-GroupMid-training Protocol=Subset Selection, Token Budget=25B2026.05 | 71.85 | |
| RandomMid-training Protocol=Subset Selection, Token Budget=25B2026.05 | 71.44 | |
| DataManMid-training Protocol=Subset Selection, Token Budget=25B2026.05 | 71.38 | |
| MIRA-GlobalMid-training Protocol=Subset Selection, Token Budget=25B2026.05 | 67.84 | |
| Raw MixtureMid-training Protocol=Full Corpus, Token Budget=50B2026.05 | 67.42 | |
| DSIRMid-training Protocol=Subset Selection, Token Budget=25B2026.05 | 67.26 | |
| Base ModelMid-training Protocol=No Mid-training, Token Budget=02026.05 | 59.14 | |
| PPLMid-training Protocol=Subset Selection, Token Budget=25B2026.05 | 57.74 |