Machine Learning Engineering on MLE-Dojo
91.9Valid Sub.Claude-4.5-Sonnet
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Claude-4.5-SonnetFramework=MLE-Agent2026.04 | 91.9 | 54.09 | |
| Qwen3-30B-SandMLEFramework=MLE-Agent, Backbone=Qwen3-30B-A3B2026.04 | 83.9 | 38.56 | |
| Qwen3-30B-SandMLEFramework=AIDE, Backbone=Qwen3-30B-A3B2026.04 | 77.4 | 28.72 | |
| Qwen3-14B-SandMLEFramework=AIDE2026.04 | 75.8 | 37.86 | |
| Qwen3-14B-BaseFramework=AIDE2026.04 | 74.2 | 37.73 | |
| Qwen3-30B-BaseFramework=MLE-Agent, Backbone=Qwen3-30B-A3B2026.04 | 71 | 29.12 | |
| Qwen3-30B-Seed-SFTFramework=AIDE, Backbone=Qwen3-30B-A3B2026.04 | 66.1 | 28.14 | |
| Qwen3-14B-Seed-SFTFramework=AIDE2026.04 | 62.9 | 27.51 | |
| Qwen3-30B-BaseFramework=AIDE, Backbone=Qwen3-30B-A3B2026.04 | 59.7 | 25.99 | |
| Qwen3-14B-SandMLEFramework=MLE-Agent2026.04 | 40.3 | 12.55 | |
| Qwen3-14B-BaseFramework=MLE-Agent2026.04 | 27.4 | 9.62 | |
| Qwen3-30B-Seed-SFTFramework=MLE-Agent, Backbone=Qwen3-30B-A3B2026.04 | 17.7 | 7.34 | |
| Qwen3-14B-Seed-SFTFramework=MLE-Agent2026.04 | 3.2 | 0.65 |