List your company

Generalization Reasoning on Out-of-distribution Aggregate

51Average AccuracyHiLL
20.21628.20836.244.192Apr 1, 2026

Evaluation results

MethodLinks
HiLLBackbone=Qwen2.5-7B-Instruct2026.0451
HiLLw/o TWBackbone=Qwen2.5-7B-Instruct2026.0449.2
LUFFYBackbone=Qwen2.5-7B-Instruct2026.0448.6
SAGEBackbone=Qwen2.5-7B-Instruct2026.0448.6
Scaf-GRPOBackbone=Qwen2.5-7B-Instruct2026.0447.5
GRPOBackbone=Qwen2.5-7B-Instruct2026.0447.4
BaseBackbone=Qwen2.5-7B-Instruct2026.0446.7
HiLLBackbone=Llama-3.2-3B-Instruct2026.0435.3
HiLLw/o TWBackbone=Llama-3.2-3B-Instruct2026.0434.5
SAGEBackbone=Llama-3.2-3B-Instruct2026.0434
GRPOBackbone=Llama-3.2-3B-Instruct2026.0433.1
Scaf-GRPOBackbone=Llama-3.2-3B-Instruct2026.0431
BaseBackbone=Llama-3.2-3B-Instruct2026.0422.5
LUFFYBackbone=Llama-3.2-3B-Instruct2026.0421.4