Logical Planning on Sudoku (test)
91.7AccuracyGDSD w/ TLC
Evaluation Results
| Method | Links | |
|---|---|---|
| GDSD w/ TLCSeq Len=512, Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 91.7 | |
| GDSD w/ TLCSeq Len=Avg., Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 91.4 | |
| GDSD w/ TLCSeq Len=256, Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 91.3 | |
| GDSD w/ TLCSeq Len=128, Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 91.1 | |
| GDSDSeq Len=128, Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 82.3 | |
| GDSDSeq Len=256, Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 82.2 | |
| GDSDSeq Len=Avg., Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 81.3 | |
| GDSDSeq Len=512, Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 79.3 | |
| ESPOSeq Len=256, Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 72.3 | |
| ESPOSeq Len=Avg., Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 71.8 | |
| ESPOSeq Len=128, Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 71.7 | |
| ESPOSeq Len=512, Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 71.3 | |
| diffu-GRPO (d1)Seq Len=256, Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 69.7 | |
| diffu-GRPO (d1)Seq Len=128, Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 64.4 | |
| diffu-GRPO (d1)Seq Len=Avg., Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 61.7 | |
| diffu-GRPO (d1)Seq Len=512, Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 51.1 | |
| wd1Seq Len=256, Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 39.2 | |
| wd1Seq Len=Avg., Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 33 | |
| wd1Seq Len=512, Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 30.3 | |
| wd1Seq Len=128, Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 29.5 | |
| Dream-7BSeq Len=512, Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 14 | |
| Dream-7BSeq Len=128, Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 9.3 | |
| Dream-7BSeq Len=Avg., Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 8.5 | |
| Dream-7BSeq Len=256, Zero-shot=true, Base Model=Dream-7B-Instruct2026.05 | 2.1 |