Bootstrapping Code Translation with Weighted Multilanguage Exploration

TL;DR

BootTrans is a novel bootstrapping approach that enhances multilingual code translation by leveraging test suites as universal verification tools and dynamically balancing training across language pairs.

Contribution

The paper introduces BootTrans, a method combining test suite adaptation, dual-pool data expansion, and language-aware weighting to improve multilingual code translation.

Findings

Significant performance improvements over baseline LLMs across benchmarks.

Effective use of test suites as universal verification oracles.

Validation of bootstrapping and weighting components through ablation studies.

Abstract

Code translation across multiple programming languages is essential yet challenging due to two vital obstacles: scarcity of parallel data paired with executable test oracles, and optimization imbalance when handling diverse language pairs. We propose BootTrans, a bootstrapping method that resolves both obstacles. Its key idea is to leverage the functional invariance and cross-lingual portability of test suites, adapting abundant pivot-language unit tests to serve as universal verification oracles for multilingual reinforcement learning (RL) training. Our method introduces a dual-pool architecture with seed and exploration pools to progressively expand training data via execution-guided experience collection. Furthermore, we design a language-aware weighting mechanism that dynamically prioritizes harder translation directions based on relative performance across sibling languages, mitigating optimization imbalance. Extensive experiments on the HumanEval-X and TransCoder-Test benchmarks demonstrate substantial improvements over baseline LLMs across all translation directions, with ablation studies validating the effectiveness of both bootstrapping and weighting components.