Hyperparameter Search (AlexNet) on SVHN
64Non-source Validity CountDeepSeek-Coder-6.7B
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| DeepSeek-Coder-6.7BLLM=DeepSeek6.7B, N (Candidates per arm)=322026.07 | 64 | 22.54 | 63 | 78.8 | 0.5626 | 0.017 | |
| OlympicCoder-7BLLM=Olympic7B, N (Candidates per arm)=322026.07 | 45 | 20.93 | 44 | 19.65 | -0.0128 | -0.0022 | |
| Qwen2.5-Coder-7BLLM=Qwen2.5-7B, N (Candidates per arm)=322026.07 | 25 | 19.59 | 35 | 19.59 | 0 | 0 |