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

Jie Wu^*, Haoling Li^*, Xin Zhang^*†‡, Jiani Guo, Jane Luo, Steven Liu, Yangyu Huang, Ruihang Chu^‡, Scarlett Li, Yujiu Yang

_{^*Equal Contribution    ^†Project Lead    ^‡Corresponding Author}

Tsinghua University  |  Microsoft

     

Left: SynthSmith generates high-quality synthetic tasks, solutions, and test cases to support both SFT and RL training. Right: Avg@8 results on LiveCodeBench. X-Coder achieves significant performance gains on competitive programming using fully synthetic data. For the full data synthesis workflow (question/answer/test generation), see data-recipe/README.md.

• Pipeline
• Results
• SFT Training
• RLVR Training
• Citation

Pipeline

Framework of SynthSmith. SynthSmith first extracts and evolves competitive programming related features from small-scale code instruction data and merges them into tree structures. It then samples subtrees from the feature tree, selects a compatible feature set, and formulates a scenario that naturally integrates these consistent features. Novel tasks are generated based on a proposed scenario according to specific styles. Advanced reasoning models are used to synthesize solutions and tests for the generated tasks, which are further cross-verified using the proposed dual-verification strategy to yield reliable test outputs and the top solution.

Results

Performance on LiveCodeBench v5. X-Coder shows strong coding expertise with fewer, fully synthetic tasks, and achieves additional gains through subsequent RL stages.

SFT Training

SFT training can be performed using various frameworks such as ms-swift, LLaMA-Factory, or Megatron-LM. Below we provide a simple example using ms-swift.

Environment Setup

pip install ms-swift -U

Data Preparation

Download and convert the SFT training data from HuggingFace:

cd sft-recipe
python download_and_convert_data.py

This will download IIGroup/X-Coder-SFT-376k and convert it to hybrid_376k.jsonl format with query and response fields.

Start SFT Training

For multi-node training (8 nodes x 8 GPUs):

# On each node, set the appropriate environment variables:
export NODE_RANK=<node_rank>  # 0, 1, 2, ..., 7
export MASTER_ADDR=<master_node_ip>
export MASTER_PORT=29500

cd sft-recipe
bash train_sft.sh

For single-node training, modify train_sft.sh:

• Set NNODES=1
• Adjust CUDA_VISIBLE_DEVICES as needed

---

RLVR Training

RLVR Quick Start

# 1. Clone the repository
git clone https://github.com/JieWu02/X-Coder.git
cd X-Coder

# 2. Start Docker container
sudo docker run -it --rm \
  --gpus all \
  --ipc=host \
  -v $(pwd):/workspace \
  whatcanyousee/verl:ngc-cu124-vllm0.8.5-sglang0.4.6.post5-mcore0.12.1-te2.3-deepseekv3 \
  /bin/bash

# 3. Install dependencies
pip install sandbox_fusion pyext
cd rl-recipe

# 4. Download rl training data
python download_data.py

# 5. Start training
bash train_scripts/install.sh
bash train_scripts/xcoder-rl-train.sh

RL Data Preparation

The rl training data (~17GB total) is hosted on HuggingFace: IIGroup/X-Coder-RL-40k

Download RL Data

cd rl-recipe

# Download all data (~17GB)
python download_data.py

# Or download only synthetic data (~8.5GB)
python download_data.py --syn-only

Data Structure

After downloading, the data will be organized as:

rl-recipe/
├── syn_rl_data/
│   └── xcoder_data/
│       └── sorted_by_passrate/
│           ├── part_0000.parquet
│           ├── part_0001.parquet
│           ├── part_0002.parquet
│           ├── part_0003.parquet
│           └── rl_tasks_easy.parquet
└── real_rl_data/
    └── non_sys_prompt/
        ├── codeforces_9763.parquet
        ├── klear_code.parquet
        ├── leetcode_2772.parquet
        ├── taco_13064.parquet
        └── test_wo_prompt.parquet

Code Judge

A code execution and evaluation service is included in rl-recipe/code-judge/.

Citation

If you find this work helpful, please cite:

@misc{wu2026xcoderadvancingcompetitiveprogramming,
      title={X-Coder: Advancing Competitive Programming with Fully Synthetic Tasks, Solutions, and Tests}, 
      author={Jie Wu and Haoling Li and Xin Zhang and Jiani Guo and Jane Luo and Steven Liu and Yangyu Huang and Ruihang Chu and Scarlett Li and Yujiu Yang},
      year={2026},
      eprint={2601.06953},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2601.06953}, 
}

License

This project is licensed under the Apache License 2.0.