This is a seq2seq neural network based in Code5T paper. This repo is ONLY focused on generating analyzing code and giving a summary from it.
Try it on Hugging Face with this link.
You should be good with the $300 dlls given from Google, spin off a VM in GCloud with 32GB of RAM, but don't forget to turn if off...
train_model.py: The main script that orchestrates training, evaluation, and inference processes. models.py: Contains model architecture and utilities for loading and saving models. data_handling.py: Utilities for data processing and loading. evaluator.py: Functions for evaluating model performance. configs.py: Configuration and argument parsing.