Scaling Data Difficulty: Improving Coding Models via Reinforcement Learning on Fresh and Challenging Problems

TL;DR

This paper presents a new dataset and methodology for training code generation models by systematically curating challenging problems using reinforcement learning and difficulty metrics, resulting in significant performance improvements.

Contribution

It introduces a four-stage data processing framework with an LLM-based difficulty filtering method, creating the MicroCoder dataset that enhances model performance on hard problems.

Findings

MicroCoder dataset achieves 3x performance gains over baseline datasets.

Models trained on MicroCoder perform better on medium and hard problems.

Difficulty-aware data curation leads to substantial improvements in challenging code generation tasks.

Abstract

Training next-generation code generation models requires high-quality datasets, yet existing datasets face difficulty imbalance, format inconsistency, and data quality problems. We address these challenges through systematic data processing and difficulty scaling. We introduce a four-stage Data Processing Framework encompassing collection, processing, filtering, and verification, incorporating Automatic Difficulty Filtering via an LLM-based predict-calibrate-select framework that leverages multi-dimensional difficulty metrics across five weighted dimensions to retain challenging problems while removing simplistic ones. The resulting MicroCoder dataset comprises tens of thousands of curated real competitive programming problems from diverse platforms, emphasizing recency and difficulty. Evaluations on strictly unseen LiveCodeBench demonstrate that MicroCoder achieves 3x larger performance gains within 300 training steps compared to widely-used baseline datasets of comparable size, with consistent advantages under both GRPO and its variant training algorithms. The MicroCoder dataset delivers obvious improvements on medium and hard problems across different model sizes, achieving up to 17.2% relative gains in overall performance where model capabilities are most stretched. These results validate that difficulty-aware data curation improves model performance on challenging tasks, providing multiple insights for dataset creation in code generation.