List your company

Human-Metric Correlation on CoGym (Out-of-Distribution)

0.365Kendall's TauAutoMetrics
-0.13836-0.007680.1230.25368Dec 19, 2025

Evaluation results

MethodLinks
AutoMetricsBackbone=Qwen-3-32B2025.120.365
DnA EvalBackbone=Qwen-3-32B2025.120.353
LLM-JudgeBackbone=Qwen-3-32B2025.120.276
Finetuned LLMBackbone=Model Agnostic2025.120.223
LLM-JudgeBackbone=GPT-4o-mini2025.120.199
DnA EvalBackbone=GPT-4o-mini2025.120.185
Best Existing MetricBackbone=Model Agnostic2025.120.074
AutoMetricsBackbone=GPT-4o-mini2025.12-0.034
MetaMetricsBackbone=Model Agnostic2025.12-0.119