Ujjwal-Tyagi/JavaScript-Code-Large

1---2license: mit3task_categories:4- text-generation5language:6- en7tags:8- code9- javascript10size_categories:11- 1M<n<10M12---13 14 15**JavaScript-Code-Large**16 17JavaScript-Code-Large is a large-scale corpus of JavaScript source code comprising around **5 million** JavaScript files. The dataset is designed to support research in large language model (LLM) pretraining, code intelligence, software engineering automation, and program analysis for the JavaScript ecosystem.18 19By providing a high-volume, language-specific corpus, JavaScript-Code-Large enables systematic experimentation in JavaScript-focused model training, domain adaptation, and downstream code understanding tasks.20 21 22JavaScript-Code-Large addresses the need for a dedicated JavaScript-only dataset at substantial scale, enabling focused research across frontend, backend, and full-stack JavaScript environments.23.24 25**1. Dataset Composition**26 27Programming Language: JavaScript28 29File Count: 5M+ JavaScript files30 31File Format: .jsonl32 33 34Content Types35 36The dataset includes a wide variety of JavaScript constructs and paradigms, such as:37 38- Functions (declarations, expressions, arrow functions)39 40- Classes and prototypes41 42- Modules (CommonJS and ES Modules)43 44- Asynchronous patterns (async/await, Promises, callbacks)45 46- Event-driven code47 48- Closures and higher-order functions49 50- Functional programming constructs51 52- DOM manipulation code53 54- Node.js backend logic55 56- Frontend framework components57 58- JSDoc comments59 60- Error handling patterns61 62- Modern ES6+ features63 64 65**2. Intended Research Applications**66 672.1 Pretraining68 69- Training JavaScript code foundation models from scratch70 71- Continued pretraining of existing LLMs72 73- JavaScript-specialized language modeling74 75- Tokenizer training for JS ecosystems76 77 782.2 Fine-Tuning and Adaptation79 80- Code completion systems81 82- Intelligent IDE assistants83 84- Automated refactoring tools85 86- Conversational programming agents87 88- JavaScript-specific copilots89 90 912.3 Code Intelligence Tasks92 93- Code summarization94 95- Code-to-text generation96 97- Documentation generation98 99- Bug detection100 101- Vulnerability detection102 103- Clone detection104 105- Code similarity modeling106 107- Minified-to-readable code transformation108 109- Static and structural analysis110 111 1122.4 Software Engineering Research113 114- Empirical studies of JavaScript coding patterns115 116- Analysis of async and event-driven architectures117 118- Framework usage studies119 120- Dependency modeling121 122- AST-based experiments123 124- Cross-version JavaScript evolution analysis125 126 127**3. Relationship to [Java-Code-Large](https://huggingface.co/datasets/ajibawa-2023/Java-Code-Large)**128 129JavaScript-Code-Large complements **Java-Code-Large**, enabling comparative research between:130 131- Statically typed vs dynamically typed languages132 133- Class-based vs prototype-based paradigms134 135- Backend vs frontend dominant ecosystems136 137- JVM vs Node.js environments138 139Together, these datasets support cross-language transfer learning and controlled specialization studies.140 141 142Thanks to open source community for all the guidance & support!!