List your company

Maze Solving on DFS maze (test)

99.3Plan AccuracyCorrect Traces
-2.6223.8450.376.76May 19, 2025

Evaluation results

MethodLinks
Correct TracesBackbone=Qwen3-8B-base, Fine-tuning regime=Correct Traces2025.0599.3100
Solution-onlyBackbone=Qwen3-8B-base, Fine-tuning regime=Solution-only2025.0578.20
Swapped TracesBackbone=Qwen3-8B-base, Fine-tuning regime=Swapped Traces2025.0576.60
Base ModelBackbone=Qwen3-8B-base, Fine-tuning regime=None2025.051.30