Meta-Learning for Monitoring in Code Summarization Systems

This is the code accompanying the research on Meta-Learning Monitoring (MLM) Module for Code Summarization Pipelines

This repo contains the Experiments 1-4 from the paper, as well as the code to load and preprocess the CoNaLa data from Huggingface. Each experiment has their own virtual environment, please consult the respective metainfo.txt files

Codes separated into quantlets:

Data Loading and Preprocessing

MLM and Base Learner Selection

MLM and Mixture of Experts Ensemble

MLM and predictive interpretability

The Quantinar presentation accompanies the code.