Neural Attribution for Semantic Bug-Localization in Student Programs

Authors: Rahul Gupta, Aditya Kanade, Shirish Shevade

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our experiments show that Neural Bug Locator is generally more accurate than two state-of-the-art program-spectrum based and one syntactic difference based bug-localization baselines. Our experiments demonstrate that NBL is more accurate than them in most cases.
Researcher Affiliation Collaboration Rahul Gupta1 Aditya Kanade1,2 Shirish Shevade1 1Department of Computer Science and Automation, Indian Institute of Science, Bangalore, KA 560012, India 2Google Brain, CA, USA
Pseudocode No The paper describes the technical details and phases of the bug-localization approach in prose but does not include any structured pseudocode or algorithm blocks.
Open Source Code Yes 4. We provide both the dataset and the implementation of NBL online at https://bitbucket. org/iiscseal/nbl/.
Open Datasets Yes 4. We provide both the dataset and the implementation of NBL online at https://bitbucket. org/iiscseal/nbl/.
Dataset Splits Yes Pairing these programs with their corresponding test IDs results in a dataset with around 270K examples. We set aside 5% of this dataset for validation, and use the rest for training.
Hardware Specification Yes We train our model for 50 epochs, which takes about one hour on an Intel(R) Xeon(R) Gold 6126 machine, clocked at 2.60GHz with 64GB of RAM and equipped with an NVIDIA Tesla P100 GPU.
Software Dependencies No The paper mentions using 'Keras [7] using Tensor Flow [1] as back-end' and 'pycparser [5]', but it does not specify version numbers for these software components, which is required for reproducibility.
Experiment Setup Yes We train our model using the Adam optimizer [17], with a learning rate of 0.0001. We train our model for 50 epochs