Code Generation on APPS Overall
21.38PRDare
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DareFinetune Method=Dare, Base Model=Meta-llama-3.1-instruct-8b2025.02 | 21.38 | 5 | |
| DisenLoRAFinetune Method=DisenLoRA, Base Model=Meta-llama-3.1-instruct-8b2025.02 | 20.57 | 5.33 | |
| SFT on allFinetune Method=SFT on all, Base Model=Meta-llama-3.1-instruct-8b2025.02 | 19.87 | 3.67 | |
| TiesFinetune Method=Ties, Base Model=Meta-llama-3.1-instruct-8b2025.02 | 19.49 | 4 | |
| SFT on cluster 2Finetune Method=SFT on cluster 2, Base Model=Meta-llama-3.1-instruct-8b2025.02 | 19.32 | 4.67 | |
| SFT on cluster 0Finetune Method=SFT on cluster 0, Base Model=Meta-llama-3.1-instruct-8b2025.02 | 18.8 | 3.33 | |
| w/o tuningFinetune Method=w/o tuning, Base Model=Meta-llama-3.1-instruct-8b2025.02 | 18.23 | 3 | |
| SFT on cluster 1Finetune Method=SFT on cluster 1, Base Model=Meta-llama-3.1-instruct-8b2025.02 | 17.97 | 3.33 | |
| SFT on cluster 2Finetune Method=SFT on cluster 2, Base Model=Meta-llama-3.1-instruct-8b2025.02 | 17.33 | 4 | |
| TwinFinetune Method=Twin, Base Model=Meta-llama-3.1-instruct-8b2025.02 | 17.15 | 3.67 | |
| w/o tuningFinetune Method=w/o tuning, Base Model=Meta-llama-3.1-instruct-8b2025.02 | 12.83 | 1 | |
| TiesFinetune Method=Ties, Base Model=Meta-llama-3.1-instruct-8b2025.02 | 12.67 | 4 | |
| DareFinetune Method=Dare, Base Model=Meta-llama-3.1-instruct-8b2025.02 | 12.5 | 3 | |
| SFT on cluster 1Finetune Method=SFT on cluster 1, Base Model=Meta-llama-3.1-instruct-8b2025.02 | 12 | 2 | |
| SFT on cluster 0Finetune Method=SFT on cluster 0, Base Model=Meta-llama-3.1-instruct-8b2025.02 | 11.5 | 1 | |
| DisenLoRAFinetune Method=DisenLoRA, Base Model=Meta-llama-3.1-instruct-8b2025.02 | 11.5 | 4 | |
| SFT on allFinetune Method=SFT on all, Base Model=Meta-llama-3.1-instruct-8b2025.02 | 10.67 | 1 | |
| TwinFinetune Method=Twin, Base Model=Meta-llama-3.1-instruct-8b2025.02 | 8.5 | 1 |