Application of Seq2Seq Models on Code Correction

We apply various seq2seq models on programming language correction tasks Juliet Test Suite for C/C++ and Java of Software Assurance Reference Datasets achieve 75% (for C/C++) 56% Java) repair rates these tasks. introduce pyramid encoder in models, which significantly increases the computational efficiency memory efficiency, while achieving similar rate to their nonpyramid counterparts. successfully carry out error type classification task ITC benchmark examples (with only 685 code instances) using transfer learning with pretrained Suite, pointing a novel way processing small datasets.