Machine Learning Task Automation on MLGym-Bench
63.9Pass@1MTL (I)
Evaluation results
| Method | Links | |
|---|---|---|
| MTL (I)Model Backbone=GPT-5-mini2026.04 | 63.9 | |
| MTL (S)Model Backbone=GPT-5-mini2026.04 | 61.1 | |
| ZeroShotModel Backbone=GPT-5-mini2026.04 | 58.3 | |
| MTL (W)Model Backbone=GPT-5-mini2026.04 | 58.3 | |
| MTL (I)Model Backbone=DeepSeek V3.22026.04 | 55.6 | |
| MTL (T)Model Backbone=GPT-5-mini2026.04 | 52.8 | |
| MTL (I)Model Backbone=Qwen3-Coder-480B-A35B-Instruct2026.04 | 52.8 | |
| ZeroShotModel Backbone=Qwen3-Coder-480B-A35B-Instruct2026.04 | 50 | |
| ZeroShotModel Backbone=DeepSeek V3.22026.04 | 50 |