GigaCheck: Detecting LLM-generated Content

Benchmark Model Rank Results
ai-generated-text-detection-on-mage-arbitrary-domains-modelsGigaCheck (Mistral-24B)#1AvgRec: 96.85AUROC: 99
ai-generated-text-detection-on-mage-paraphrasing-attackGigaCheck (Mistral-7B)#1AvgRec: 68.95AUROC: 74
ai-generated-text-detection-on-mage-unseen-domains-unseen-modelGigaCheck (Mistral-7B)#1AvgRec: 88.54AUROC: 96
boundary-detection-on-coauthorGigaCheck (DN-DAB-DETR)–Cohen’s Kappa score: 0.1885
boundary-detection-on-coauthorGigaCheck (Mistral-7B-v0.3)–Cohen’s Kappa score: 0.4158
boundary-detection-on-roftGigaCheck (DN-DAB-DETR)–Accuracy (%): 64.63MSE: 1.51
boundary-detection-on-roft-chatgptGigaCheck (DN-DAB-DETR)–Accuracy (%): 67.65MSE: 1.03