Code Generation on Code Evaluation Suite
36.6Code ScoreBTM
Evaluation Results
| Method | Links | |
|---|---|---|
| BTMActive Experts (Top-k)=22024.03 | 36.6 | |
| BTMActive Experts (Top-k)=12024.03 | 36.4 | |
| CODELLAMAModel Size=7B2024.03 | 36.3 | |
| Sparse upcyclingTraining Context=Data-Matching (DM), Active Experts (Top-k)=22024.03 | 34.7 | |
| BTXActive Experts (Top-k)=22024.03 | 34 | |
| LLEMMAModel Size=7B2024.03 | 33.5 | |
| BTXActive Experts (Top-k)=1, Sampling Strategy=Sample2024.03 | 31.5 | |
| DenseTraining Context=Data-Matching (DM)2024.03 | 25.8 | |
| LLAMA-2Model Size=13B2024.03 | 24.5 | |
| LLAMA-2Model Size=7B2024.03 | 16.8 |