Paper
Unseen-Codebases-Domain Data Synthesis and Training Based on Code Graphs
Authors
Guangsheng Ou, Qiming Zhang, Sirong Chen, Anji Li, Dong Xu, Tiancheng Luo, Dekun Dai, Cuiyun Gao, Long Wang, Jun Zhou, Mingwei Liu, Zibin Zheng
Abstract
In the context of newly release software frameworks, large language models (LLMs) often exhibit poor performance and a high rate of hallucination, as they are not exposed to such environments during training. Although inference-time augmentation techniques such as retrieval-augmented generation (RAG) can partially mitigate hallucinations, knowledge injection through prompting alone is insufficient to enable models to fully understand the intrinsic relationships among different components of a codebase, or to reason about the correct compositions and apply. Although explicit knowledge injection can be achieved through post-training, compared with public code domains, unseen codebases typically provide only source code and lack large volumes of high-quality, usage-oriented code that can be directly leveraged as training data. Consequently, existing data synthesis approaches are insufficient to adequately capture unseen codebases usage scenarios when restricted to source code alone. To address these challenges, we propose UCD-Training, a two-stage training framework for reasoning-aware data synthesis grounded in a code graph constructed from unseen codebases. UCD-Training first parses the source code to build a code graph, then conducts dependency-preserving continued pretraining (CPT) using file-level dependency data, followed by graph-grounded supervised fine-tuning (SFT) on three types of synthesized data augmented with explicit reasoning traces: (1) single-hop relation reasoning data, (2) compositional API reasoning data, and (3) codebase utilization data. We further introduce a new benchmark, UnseenCodeBench, for code generation on unseen codebases and conduct comprehensive experiments across multiple codebases.
Metadata
Related papers
Vibe Coding XR: Accelerating AI + XR Prototyping with XR Blocks and Gemini
Ruofei Du, Benjamin Hersh, David Li, Nels Numan, Xun Qian, Yanhe Chen, Zhongy... • 2026-03-25
Comparing Developer and LLM Biases in Code Evaluation
Aditya Mittal, Ryan Shar, Zichu Wu, Shyam Agarwal, Tongshuang Wu, Chris Donah... • 2026-03-25
The Stochastic Gap: A Markovian Framework for Pre-Deployment Reliability and Oversight-Cost Auditing in Agentic Artificial Intelligence
Biplab Pal, Santanu Bhattacharya • 2026-03-25
Retrieval Improvements Do Not Guarantee Better Answers: A Study of RAG for AI Policy QA
Saahil Mathur, Ryan David Rittner, Vedant Ajit Thakur, Daniel Stuart Schiff, ... • 2026-03-25
MARCH: Multi-Agent Reinforced Self-Check for LLM Hallucination
Zhuo Li, Yupeng Zhang, Pengyu Cheng, Jiajun Song, Mengyu Zhou, Hao Li, Shujie... • 2026-03-25
Raw Data (Debug)
{
"raw_xml": "<entry>\n <id>http://arxiv.org/abs/2602.20799v1</id>\n <title>Unseen-Codebases-Domain Data Synthesis and Training Based on Code Graphs</title>\n <updated>2026-02-24T11:36:34Z</updated>\n <link href='https://arxiv.org/abs/2602.20799v1' rel='alternate' type='text/html'/>\n <link href='https://arxiv.org/pdf/2602.20799v1' rel='related' title='pdf' type='application/pdf'/>\n <summary>In the context of newly release software frameworks, large language models (LLMs) often exhibit poor performance and a high rate of hallucination, as they are not exposed to such environments during training. Although inference-time augmentation techniques such as retrieval-augmented generation (RAG) can partially mitigate hallucinations, knowledge injection through prompting alone is insufficient to enable models to fully understand the intrinsic relationships among different components of a codebase, or to reason about the correct compositions and apply. Although explicit knowledge injection can be achieved through post-training, compared with public code domains, unseen codebases typically provide only source code and lack large volumes of high-quality, usage-oriented code that can be directly leveraged as training data. Consequently, existing data synthesis approaches are insufficient to adequately capture unseen codebases usage scenarios when restricted to source code alone. To address these challenges, we propose UCD-Training, a two-stage training framework for reasoning-aware data synthesis grounded in a code graph constructed from unseen codebases. UCD-Training first parses the source code to build a code graph, then conducts dependency-preserving continued pretraining (CPT) using file-level dependency data, followed by graph-grounded supervised fine-tuning (SFT) on three types of synthesized data augmented with explicit reasoning traces: (1) single-hop relation reasoning data, (2) compositional API reasoning data, and (3) codebase utilization data. We further introduce a new benchmark, UnseenCodeBench, for code generation on unseen codebases and conduct comprehensive experiments across multiple codebases.</summary>\n <category scheme='http://arxiv.org/schemas/atom' term='cs.SE'/>\n <published>2026-02-24T11:36:34Z</published>\n <arxiv:primary_category term='cs.SE'/>\n <author>\n <name>Guangsheng Ou</name>\n </author>\n <author>\n <name>Qiming Zhang</name>\n </author>\n <author>\n <name>Sirong Chen</name>\n </author>\n <author>\n <name>Anji Li</name>\n </author>\n <author>\n <name>Dong Xu</name>\n </author>\n <author>\n <name>Tiancheng Luo</name>\n </author>\n <author>\n <name>Dekun Dai</name>\n </author>\n <author>\n <name>Cuiyun Gao</name>\n </author>\n <author>\n <name>Long Wang</name>\n </author>\n <author>\n <name>Jun Zhou</name>\n </author>\n <author>\n <name>Mingwei Liu</name>\n </author>\n <author>\n <name>Zibin Zheng</name>\n </author>\n </entry>"
}