Logical Reasoning on ZebraLogic v1.0 (test)
97.7Cell AccuracyIn-place
Evaluation Results
| Method | Links | |
|---|---|---|
| In-placeBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=92025.10 | 97.7 | |
| In-placeBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=92025.10 | 97.2 | |
| In-placeBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=82025.10 | 96.8 | |
| In-placeBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=82025.10 | 96.3 | |
| Multi-turnBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=92025.10 | 95.9 | |
| In-placeBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=72025.10 | 95.6 | |
| Multi-turnBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=82025.10 | 94.7 | |
| In-placeBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=72025.10 | 94.4 | |
| Multi-turnBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=92025.10 | 93.5 | |
| In-placeBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=62025.10 | 93.1 | |
| Multi-turnBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=72025.10 | 92.7 | |
| Multi-turnBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=82025.10 | 92.1 | |
| In-placeBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=62025.10 | 91.4 | |
| Multi-turnBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=62025.10 | 89.6 | |
| Multi-turnBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=72025.10 | 89.1 | |
| MemoryBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=92025.10 | 89.1 | |
| In-placeBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=52025.10 | 89.1 | |
| MemoryBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=92025.10 | 88.9 | |
| In-placeBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=92025.10 | 88.3 | |
| MemoryBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=82025.10 | 87.1 | |
| MemoryBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=82025.10 | 86.6 | |
| In-placeBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=52025.10 | 85.9 | |
| Multi-turnBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=62025.10 | 85.9 | |
| Multi-turnBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=52025.10 | 84.8 | |
| In-placeBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=82025.10 | 84.7 | |
| MemoryBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=72025.10 | 84.1 | |
| MemoryBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=72025.10 | 83.5 | |
| In-placeBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=72025.10 | 83.3 | |
| In-placeBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=42025.10 | 83 | |
| MemoryBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=62025.10 | 80.7 | |
| Multi-turnBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=52025.10 | 80.6 | |
| In-placeBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=62025.10 | 79.8 | |
| In-placeBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=42025.10 | 79.4 | |
| MemoryBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=62025.10 | 79.4 | |
| Multi-turnBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=42025.10 | 77.9 | |
| MemoryBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=52025.10 | 75.8 | |
| In-placeBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=52025.10 | 75.1 | |
| MemoryBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=52025.10 | 75 | |
| In-placeBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=32025.10 | 73.8 | |
| Multi-turnBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=42025.10 | 73.6 | |
| Multi-turnBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=92025.10 | 71.8 | |
| Multi-turnBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=82025.10 | 70.1 | |
| In-placeBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=32025.10 | 69.8 | |
| MemoryBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=42025.10 | 69.3 | |
| MemoryBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=42025.10 | 69 | |
| Multi-turnBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=32025.10 | 68.1 | |
| In-placeBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=42025.10 | 67.9 | |
| Multi-turnBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=62025.10 | 67.1 | |
| Multi-turnBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=72025.10 | 66.3 | |
| Multi-turnBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=32025.10 | 65.2 | |
| MemoryBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=32025.10 | 61.8 | |
| Multi-turnBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=52025.10 | 61.2 | |
| In-placeBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=22025.10 | 60.7 | |
| MemoryBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=92025.10 | 60.3 | |
| MemoryBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=32025.10 | 59.9 | |
| In-placeBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=32025.10 | 59.1 | |
| MemoryBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=82025.10 | 58.6 | |
| In-placeBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=22025.10 | 57.3 | |
| MemoryBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=72025.10 | 57.1 | |
| Multi-turnBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=22025.10 | 56.7 | |
| MemoryBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=62025.10 | 54.8 | |
| Multi-turnBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=42025.10 | 54.6 | |
| Multi-turnBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=22025.10 | 54.3 | |
| MemoryBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=52025.10 | 52.8 | |
| MemoryBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=22025.10 | 51.8 | |
| MemoryBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=42025.10 | 50.5 | |
| MemoryBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=22025.10 | 48.7 | |
| In-placeBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=22025.10 | 48 | |
| MemoryBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=32025.10 | 46.4 | |
| Multi-turnBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=32025.10 | 46 | |
| In-placeBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=12025.10 | 44.3 | |
| Multi-turnBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=12025.10 | 43.5 | |
| In-placeBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=12025.10 | 43.2 | |
| Multi-turnBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=12025.10 | 41.7 | |
| MemoryBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=22025.10 | 41.1 | |
| MemoryBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=12025.10 | 40.6 | |
| In-placeBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=12025.10 | 36.7 | |
| Multi-turnBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=22025.10 | 36.6 | |
| MemoryBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=12025.10 | 35 | |
| MemoryBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=12025.10 | 33.9 | |
| Multi-turnBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=02025.10 | 27.2 | |
| MemoryBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=02025.10 | 27.2 | |
| In-placeBackbone=Gemma, Feedback Condition=top-4 feedback, Number of turns=02025.10 | 27.2 | |
| MemoryBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=02025.10 | 26.4 | |
| In-placeBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=02025.10 | 26.4 | |
| Multi-turnBackbone=Qwen, Feedback Condition=top-4 feedback, Number of turns=02025.10 | 26.3 | |
| Multi-turnBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=02025.10 | 19.1 | |
| MemoryBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=02025.10 | 19.1 | |
| In-placeBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=02025.10 | 19.1 | |
| Multi-turnBackbone=Llama, Feedback Condition=top-4 feedback, Number of turns=12025.10 | 18.5 |