X-Coder: Advancing Competitive Programming with Fully Synthetic Tasks, Solutions, and Tests

TL;DR

This paper introduces X-Coder, a synthetic data-driven approach that significantly improves competitive programming performance of code language models by using high-quality, domain-specific synthetic tasks, solutions, and tests.

Contribution

The paper develops a novel feature-based synthetic data generation method with domain-specific evolution and dual-verification, enabling training of competitive models without real-world data.

Findings

X-Coder-7B outperforms larger models trained on real data.

Synthetic data quality critically impacts model performance.

Domain-specific evolution enhances synthetic task solvability.

Abstract

Competitive programming poses a significant challenge for Code LLMs. While recent models have shown promise, they heavily rely on finite real-world data, raising concerns about scalability and contamination. In this paper, we investigate a critical question: Can we elevate models to expert-level reasoning performance using fully synthetic data? In response, we first observe that off-the-shelf synthesis methods yield suboptimal results in this domain. To address this, we systematically investigate the key factors governing synthetic data quality. Leveraging these findings, we significantly advance the feature-based synthesis paradigm via domain-specific evolution and a dual-verification strategy, promoting task solvability, solution correctness, and test accuracy. Using this high-quality synthetic data, we train the X-Coder model series under an SFT-then-RL paradigm. X-Coder-7B shows significant performance gains on the challenging LiveCodeBench v5 (62.9% avg@8) and v6 (55.8% avg@8), outperforming larger models trained on real-world data. Extensive analysis distills valuable insights into synthetic data scaling, the necessity of domain-adapted feature evolution, and code-centric reinforcement.