Exploring and Unleashing the Power of Large Language Models in Automated Code Translation

Code translation tools are developed for automatic source-to-source translation. Although learning-based transpilers have shown impressive enhancement against rule-based counterparts, owing to their task-specific pre-training on extensive monolingual corpora. Their current performance still remains unsatisfactory for practical deployment, and the associated training resources are also prohibitively expensive. LLMs pre-trained on huge amounts of human-written code/text have shown remarkable performance in many code intelligence tasks due to their powerful generality, even without task-specific training. Thus, LLMs can potentially circumvent the above limitations, but they have not been exhaustively explored yet. This paper investigates diverse LLMs and learning-based transpilers for automated code translation tasks, finding that: although certain LLMs have outperformed current transpilers, they still have some accuracy issues, where most of the failures are induced by a lack of comprehension of source programs (38.51%), missing clear instructions on I/O types in translation (14.94%), and ignoring discrepancies between source and target programs (41.38%). Enlightened by the above findings, we propose UniTrans, an Unified code Translation framework, applicable to various LLMs, for unleashing their power in this field. Specifically, UniTrans first craft a series of test cases for target programs with the assistance of source programs. Next, it harnesses the above auto-generated test cases to augment the code translation and then evaluate their correctness via execution. Afterward, UniTrans further (iteratively) repairs incorrectly translated programs prompted by test case execution results. Extensive experiments are conducted on six translation datasets between Python, Java, and C++. Three recent LLMs of diverse sizes are tested with UniTrans, and all achieve substantial improvements.

A Rational Model of Dimension-reduced Human Categorization

Existing models in cognitive science typically assume human categorization as graded generalization behavior in a multidimensional psychological space. However, category representations in these models may suffer from the curse of dimensionality in a natural setting. People generally rely on a tractable yet sufficient set of features to understand the complex environment. We propose a rational model of categorization based on a hierarchical mixture of probabilistic principal components, that simultaneously learn category representations and an economical collection of features. The model captures dimensional biases in human categorization and supports zero-shot learning. We further exploit a generative process within a low-dimensional latent space to provide a better account of categorization with high-dimensional stimuli. We validate the model with simulation and behavioral experiments.