Ujjwal-Tyagi/Java-Code-Large

---
license: mit
task_categories:
- text-generation
language:
- en
tags:
- code
- java
size_categories:
- 10M<n<100M
---

**Java-Code-Large**

Java-Code-Large is a large-scale corpus of publicly available Java source code comprising more than **15 million** java codes. The dataset is designed to support research in large language model (LLM) pretraining, code intelligence, software engineering automation, and program analysis.

By providing a high-volume, language-specific corpus, Java-Code-Large enables systematic experimentation in Java-focused model training, domain adaptation, and downstream code understanding tasks.

**1. Introduction**

Large-scale code corpora have become fundamental resources for training and evaluating machine learning models for code-related tasks. While multilingual code datasets exist, there is increasing interest in language-specialized corpora to:

- Improve domain-specific performance

- Reduce cross-language noise

- Enable controlled experimental settings

- Support Java-specific tooling and research

Java-Code-Large addresses this need by providing a dedicated Java-only dataset at substantial scale.

**2. Dataset Composition**

Programming Language: Java

File Count: 15M+ Java files

File Format: .jsonl

Content Types:

- Classes

- Interfaces

- Enums

- Methods

- Annotations

- JavaDoc comments

- Exception handling structures

- Generics and concurrency constructs

The dataset consists of source code extracted from publicly accessible open-source repositories.

**3. Intended Research Applications**

3.1 Pretraining

- Training code foundation models from scratch

- Continued pretraining of existing LLMs

- Java-specialized language modeling

3.2 Fine-Tuning and Adaptation

- Code completion systems

- Automated refactoring tools

- IDE copilots

- Java-specific conversational assistants

3.3 Code Intelligence Tasks

- Code summarization

- Code-to-text generation

- Bug detection

- Vulnerability detection

- Clone detection

- Code similarity modeling

- Static and structural analysis

3.4 Software Engineering Research

- Empirical studies of Java programming patterns

- Tokenization and AST modeling experiments

Thanks to open source community for all the guidance & support!!