Compositional Task Success on CrafText Combo (test)
23Success RateSI_pixel
Evaluation Results
| Method | Links | |
|---|---|---|
| SI_pixelTraining Iteration=Cycle 12026.04 | 23 | |
| Claude Sonnet 4.62026.04 | 5 | |
| OpenAI GPT-4.12026.04 | 4 | |
| Qwen/QWQ-32B2026.04 | 3 | |
| Gemma-27B2026.04 | 2 |
| Method | Links | |
|---|---|---|
| SI_pixelTraining Iteration=Cycle 12026.04 | 23 | |
| Claude Sonnet 4.62026.04 | 5 | |
| OpenAI GPT-4.12026.04 | 4 | |
| Qwen/QWQ-32B2026.04 | 3 | |
| Gemma-27B2026.04 | 2 |