How to fine tune CodeBert-base model on Assembly code clone detection task? #193
Description
I'm following the instructions and codes here to fine tune codebert-base on Assembly code clone detection task. But the result is poor(best accuracy only reached 0.3). How to solve the problem please?
I also tried fine-tuning the model on LLVM IR code clone detection task. Likewise, the best accuracy only reached 0.5