List your company

Constrained Path Planning on Textualized Gridworld Safe targets (test)

15.2Success RateSAE-opt
2.25.5758.9512.325May 21, 2026

Evaluation results

MethodLinks
SAE-optModel=Qwen3-4B-Instruct2026.0515.220.1
Dense-optModel=Qwen3-4B-Instruct2026.0514.823.5
SAE-opt-anchModel=Qwen3-4B-Instruct2026.0514.527
SAE-SSVModel=Qwen3-4B-Instruct2026.0514.223.4
SAE-optModel=Phi-3-Mini-4K-Instruct2026.0510.452.3
SAE-SSVModel=Phi-3-Mini-4K-Instruct2026.059.655.8
SAE-optModel=LLaMA3.1-8B-Instruct2026.059.643
Dense-optModel=Phi-3-Mini-4K-Instruct2026.059.455
Dense-optModel=LLaMA3.1-8B-Instruct2026.059.345.8
ITIModel=Qwen3-4B-Instruct2026.059.218.3
SAE-opt-anchModel=Phi-3-Mini-4K-Instruct2026.059.157
CAAModel=Phi-3-Mini-4K-Instruct2026.058.738.9
SAE-SSVModel=LLaMA3.1-8B-Instruct2026.057.947.5
SAE-opt-anchModel=LLaMA3.1-8B-Instruct2026.057.147.4
ITIModel=LLaMA3.1-8B-Instruct2026.056.465.9
ITIModel=Phi-3-Mini-4K-Instruct2026.054.748.5
CAAModel=Qwen3-4B-Instruct2026.054.239.6
CAAModel=LLaMA3.1-8B-Instruct2026.054.239.6
ICLModel=Qwen3-4B-Instruct2026.053.755.5
ICLModel=Phi-3-Mini-4K-Instruct2026.053.376
ICLModel=LLaMA3.1-8B-Instruct2026.052.751.2