Edit model card

TeCo is a deep learning model using code semantics to automatically complete the next statement in a test method. Completing tests requires reasoning about the execution of the code under test, which is hard to do with only syntax-level data that existing code completion models use. To solve this problem, we leverage the fact that tests are readily executable. TeCo extracts and uses execution-guided code semantics as inputs for the ML model, and performs reranking via test execution to improve the outputs. On a large dataset with 131K tests from 1270 open-source Java projects, TeCo outperforms the state-of-the-art by 29% in terms of test completion accuracy.

TeCo is presented in the following ICSE 2023 paper:

Title: Learning Deep Semantics for Test Completion

Authors: Pengyu Nie, Rahul Banerjee, Junyi Jessy Li, Raymond Mooney, Milos Gligoric


TeCo's code is hosted on GitHub: https://github.com/EngineeringSoftware/teco

This repo hosts the model we trained, but it should be used together with our codebase; please read the README there, which describes how to download and use this model.

Downloads last month
20
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.