search

Vul-LMGNN / vul-LMGGNN

Code for the paper - Source Code Vulnerability Detection: Combining Code Language Models and Code Property Graph
Apache License 2.0
39 stars 10 forks source link

Predicted value included NaN #6

Open ckj18 opened 4 weeks ago

ckj18 commented 4 weeks ago

I encountered an issue while reproducing your project. I understand that in the BertGGCN class, the output value x of conv is the probability value from the softmax, and cls_logit is the binary output logit from codebert. When summing these two according to Equation 11 from the paper, since cls_logit can contain negative values, taking the log when there are negative values in pred results in NaN values. Am I the only one experiencing this issue?

aikedaerC commented 1 week ago

I encountered an issue while reproducing your project. I understand that in the BertGGCN class, the output value x of conv is the probability value from the softmax, and cls_logit is the binary output logit from codebert. When summing these two according to Equation 11 from the paper, since cls_logit can contain negative values, taking the log when there are negative values in pred results in NaN values. Am I the only one experiencing this issue?

me too, bro