Random seed for this run is: 4321
Tokenizer already has a pad token: <|endoftext|>
Number of examples in the test dataset: 9637
Running LLM-as-a-judge on 100 random samples from the test dataset
Trigger Percentage (Sensitivity): 100.000%
False Trigger Rate: 0.000%
Backdoor Precision: 1.000
Backdoor Recall: 1.000
Backdoor F1-Score: 1.000
