Word Puzzle Solving on Crosswords Word
13.5AccuracyCoT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CoTBackbone=Qwen3-30B2026.03 | 13.5 | 182 | |
| RouteGoTModel Pool={Qwen3-4B, 8B, 30B}2026.03 | 11 | 4,222 | |
| GoT*Backbone=Qwen3-30B2026.03 | 6.5 | 13,597 | |
| IOBackbone=Qwen3-30B2026.03 | 6 | 179 | |
| EmbedLLMModel Pool={Qwen3-4B, 8B, 30B}2026.03 | 5 | 4,846 | |
| AGoTBackbone=Qwen3-30B2026.03 | 3.5 | 9,522 | |
| RouteLLMModel Pool={Qwen3-4B, 8B, 30B}2026.03 | 2.5 | 5,110 | |
| ToTBackbone=Qwen3-30B2026.03 | 2.1 | 790 | |
| RTRModel Pool={Qwen3-4B, 8B, 30B}2026.03 | 1.5 | 3,777 | |
| RandomModel Pool={Qwen3-4B, 8B, 30B}2026.03 | 1 | 8,767 | |
| KNNModel Pool={Qwen3-4B, 8B, 30B}2026.03 | 0.5 | 5,044 |