List your company

Out-of-distribution AI-generated text detection on Mistral-Medium OOD unseen domains 3 (test)

99Accuracy (Legal)READER
53.96865.65977.3589.041May 24, 2026

Evaluation results

MethodLinks
READERBackbone=Qwen2.5-1.5B-Instruct2026.059985.396.793.7
BISCOPE*Evaluation Protocol=target-adapted setting, Surrogate LLM=Falcon-7B2026.05978679.787.6
IMBD*Evaluation Protocol=target-adapted setting, Surrogate LLM=Falcon-7B2026.0579.355.364.366.3
ROBERTALARGESurrogate LLM=Falcon-7B, Evaluation Protocol=per-domain configuration2026.057363.753.763.4
LRRSurrogate LLM=Falcon-7B, Evaluation Protocol=per-domain configuration2026.0568.356.756.760.6
RADAREvaluation Protocol=per-domain configuration2026.056760.358.361.9
ENTROPYSurrogate LLM=Falcon-7B, Evaluation Protocol=per-domain configuration2026.0566.363.776.368.8
LOGRANKSurrogate LLM=Falcon-7B, Evaluation Protocol=per-domain configuration2026.0564.363.767.765.2
LIKELIHOODSurrogate LLM=Falcon-7B, Evaluation Protocol=per-domain configuration2026.0563.364.37266.6
DNAGPTSurrogate LLM=Falcon-7B, Evaluation Protocol=per-domain configuration2026.0563.367.77167.3
ROBERTABASESurrogate LLM=Falcon-7B, Evaluation Protocol=per-domain configuration2026.0562.759.759.760.7
ADADETECTGPT*Evaluation Protocol=target-adapted setting, Surrogate LLM=Falcon-7B2026.05625464.360.1
NPRSurrogate LLM=Falcon-7B, Evaluation Protocol=per-domain configuration2026.0561.761.360.761.2
FAST-DETECTGPTSurrogate LLM=Falcon-7B, Evaluation Protocol=per-domain configuration2026.055962.360.360.6
BINOCULARSEvaluation Protocol=per-domain configuration2026.0558.761.36060
DETECTGPTSurrogate LLM=Falcon-7B, Evaluation Protocol=per-domain configuration2026.0555.760.361.759.2