Automatic Program Synthesis of Long Programs with a Learned Garbage Collector
This addresses the challenge of synthesizing longer programs efficiently for applications in software automation and AI-assisted coding, representing a strong specific gain rather than a broad paradigm shift.
The paper tackles the problem of automatic code generation from input-output pairs by training a neural network to predict program statements and manage memory, resulting in programs over twice as long as prior state-of-the-art, with improved success rates and a 100x reduction in run-time.
We consider the problem of generating automatic code given sample input-output pairs. We train a neural network to map from the current state and the outputs to the program's next statement. The neural network optimizes multiple tasks concurrently: the next operation out of a set of high level commands, the operands of the next statement, and which variables can be dropped from memory. Using our method we are able to create programs that are more than twice as long as existing state-of-the-art solutions, while improving the success rate for comparable lengths, and cutting the run-time by two orders of magnitude. Our code, including an implementation of various literature baselines, is publicly available at https://github.com/amitz25/PCCoder