Pragmatic Synthesizer

Pragmatic program synthesis aims to build systems that generate programs not only consistent with user-provided examples but also account for the user's informative choices in providing those examples, leading to more efficient and effective communication. Current research focuses on improving the efficiency of pragmatic synthesis algorithms, often employing techniques like amortized inference using global rankings or integrating large language models into enumerative search strategies to guide the synthesis process. These advancements significantly speed up synthesis while maintaining or exceeding the accuracy of non-pragmatic approaches, paving the way for more user-friendly and powerful program synthesis tools.