logs/frontier_comparison.csv

provider,n,detection_rate,ci_low,ci_high,false_positive_rate,precision,f1
scripted,175,0.8403,0.7778,0.8958,0.0968,0.9758,0.9030
hf-llama-3.3-70b-instruct,175,0.9931,0.9792,1.0000,0.0323,0.9931,0.9931
hf-qwen2.5-72b-instruct,175,0.9861,0.9653,1.0000,0.0645,0.9861,0.9861
hf-deepseek-v3-0324,175,1.0000,1.0000,1.0000,0.2903,0.9412,0.9697
hf-qwen2.5-7b-instruct,175,1.0000,1.0000,1.0000,0.1613,0.9664,0.9829
hf-gpt-oss-120b,175,0.9861,0.9653,1.0000,0.1613,0.9660,0.9759
hf-deepseek-r1,175,1.0000,1.0000,1.0000,0.1290,0.9730,0.9863
hf-gemma-3-27b-it,175,1.0000,1.0000,1.0000,0.5161,0.9000,0.9474