Black-Box Optimization on 23 PBO and 24 BBOB Problems
2.194Normalized Best AUCBAG
Evaluation results
| Method | Links | ||
|---|---|---|---|
| BAGLLM backbone=instruction-tuned 27b Gemma 3, Knowledge Component=Code description2026.03 | 2.194 | — | |
| BAGLLM backbone=instruction-tuned 27b Gemma 3, Knowledge Component=Strategy2026.03 | 2.133 | — | |
| BAGLLM backbone=instruction-tuned 27b Gemma 3, Knowledge Component=Parent Heuristic(s)2026.03 | 1.912 | — | |
| BAGLLM backbone=instruction-tuned 14b Qwen 2.5 Coder, Knowledge Component=Code description2026.03 | 1.399 | — | |
| BAGLLM backbone=instruction-tuned 14b Qwen 2.5 Coder, Knowledge Component=Strategy2026.03 | 1.383 | — | |
| BAGLLM backbone=instruction-tuned 14b Qwen 2.5 Coder, Knowledge Component=Parent Heuristic(s)2026.03 | 1.279 | — | |
| BAGLLM backbone=instruction-tuned 27b Gemma 3, Knowledge Component=Expected Output2026.03 | 1.082 | — | |
| BAGLLM backbone=instruction-tuned 27b Gemma 3, Knowledge Component=Task Description2026.03 | 0.983 | — | |
| BAGLLM backbone=instruction-tuned 14b Qwen 2.5 Coder, Knowledge Component=Expected Output2026.03 | 0.898 | — | |
| BAGLLM backbone=instruction-tuned 14b Qwen 2.5 Coder, Knowledge Component=Task Description2026.03 | 0.864 | — | |
| BAGLLM backbone=instruction-tuned 27b Gemma 3, Knowledge Component=Linguistic description2026.03 | 0.807 | — | |
| BAGLLM backbone=instruction-tuned 14b Qwen 2.5 Coder, Knowledge Component=Linguistic description2026.03 | 0.656 | — | |
| BAGLLM backbone=instruction-tuned 27b Gemma 3, Knowledge Component=Note2026.03 | 0.34 | — | |
| BAGLLM backbone=instruction-tuned 14b Qwen 2.5 Coder, Knowledge Component=Note2026.03 | 0.302 | — |