SebastianKotstein/restberta-qa-pm-ed-wo-benchmarking-apis

CodeBERT model fine-tuned with endpoint discovery and parameter matching samples from OpenAPI documents downloaded from APIs.guru. APIs that we used for benchmarking have been excluded from both training and evaluation set. This repository contains the checkpoint after the 1st epoch of fine-tuning (accuracy of 80.32% for k = 1 for the parameter matching task).