Unit test case generation with transformers

A method performed by one or more computing devices, comprising:

• pre-training a neural transformer model with attention with a first unsupervised training dataset, the first unsupervised training dataset including a plurality of sequences of natural language text;
• pre-training the neural transformer model with attention with a second unsupervised training dataset, the second unsupervised training dataset including a plurality of sequences of source code;
• fine-tuning the neural transformer model with attention with a supervised training dataset, the supervised dataset including mapped unit test case pairs, a mapped unit test case pair including a focal method and an associated unit test case;
• and
• deploying the neural transformer model with attention to automatically predict a unit test case for a given method.

A device, comprising:

• at least one processor and a memory;
• wherein the at least one processor is configured to:
• translate an input sequence to an output sequence using a neural transformer model with attention, the input sequence including a sequence of subtokens representing a focal method of a source code program written in a programming language, the output sequence representing a sequence of subtokens representing a unit test case for the focal method,
• wherein the translation performs a beam search that generates each subtoken in the output sequence of subtokens, one at a time at each timestep, wherein the beam search selects top k subtokens at a timestep based on an output distribution generated by the neural transformer model with attention, the output distribution generated for each subtoken in a vocabulary of the neural transformer model with attention, wherein the top k subtokens instantiate k data paths that are iteratively processed until an end-of-method subtoken is predicted to complete the output sequence.