🧠 Web Beast: The Ultimate Laravel & Node.js Dataset for AI Code Generation

Train better AI models, build smarter developer tools, and fine-tune LLMs with a purpose-built, open-source code dataset.

---

🚀 What Is Web Beast?

Web Beast is a curated, large-scale dataset of high-quality source code, designed to help developers and researchers train AI models specifically for real-world backend and fullstack development.

Unlike generic code datasets, Web Beast focuses on framework-rich, production-grade code from:

• 🧱 Laravel (PHP)
• ⚙️ Node.js (Express, TypeScript)
• 🌐 JavaScript (Fullstack)
• 🎨 HTML + Tailwind CSS

It’s hosted on Hugging Face and ready to use in your training pipelines.

---

🔍 Why This Dataset Matters

Most code generation datasets are:

• ❌ Too generic
• ❌ Missing real framework structure
• ❌ Full of synthetic, toy-level examples

Web Beast solves this by offering real-world controller classes, route files, middleware, service providers, APIs, Tailwind layouts, and more—all directly scraped and cleaned from popular open-source projects.

---

📦 Dataset Structure

Each file is stored in .jsonl format for compatibility with AI/ML frameworks like Hugging Face Transformers, LoRA, and Alpaca-style instruction training.

Example Record:

{
  "repo": "github.com/example/repo",
  "path": "app/Http/Controllers/AuthController.php",
  "language": "php",
  "framework": "laravel",
  "content": "

📌 Key Fields

• repo: Source GitHub repository
• path: File path within the project
• language: Programming language (php, ts, js, html)
• framework: Framework used (laravel, express, etc.)
• content: Raw code content

---

🧠 Perfect for Training AI Code Models

This dataset is ideal if you're building or fine-tuning models like:

• 🤖 DeepSeek-Coder
• 🔧 StarCoder
• 🦙 CodeLlama

Use Cases

• Laravel-specific autocomplete tools
• Express.js endpoint generators
• Copilot-style backend code assistants
• Tailwind UI component suggestions

---

🧪 How to Use the Dataset

Load it with Hugging Face:

from datasets import load_dataset
dataset = load_dataset("brijmansuriya/web-beast")

Filter Laravel Files:

laravel_data = dataset['train'].filter(lambda x: x['framework'] == 'laravel')

✅ Works with Hugging Face, LoRA, QLoRA, Transformers, and custom fine-tuning pipelines.

---

🌍 Explore the Dataset

🔗 View on Hugging Face

• ✅ Clean & ready to use
• ✅ Filter by framework or language
• ✅ Ideal for training LLMs or building AI tools

---

🔐 Data Safety & Licensing

• 🧼 Secrets Removed: .env, keys, tokens, passwords
• 📜 Open Source: From MIT, Apache licensed public repos

---

🧑‍💻 Who Should Use This?

Audience	Benefit
ML Engineers	Train and evaluate code generation models
Laravel/Node Devs	Build AI-powered dev tools and assistants
Tool Builders	Use data to enhance autocomplete and code suggestions

---

🔮 Roadmap & What's Next

• ✅ Laravel-only fine-tuned model with LoRA
• ✅ Code snippets for RESTful APIs
• 🔜 Django, Flask, Vue, Next.js versions
• 🔜 Gradio/Colab demos
• 🔜 VS Code Extension powered by Web Beast

---

🤝 How to Contribute

• Submit Laravel/Node repositories for inclusion
• Contribute cleaning/preprocessing scripts
• Fine-tune models and share benchmarks

📧 Email: brijmansuriya.ai@gmail.com
👨‍💻 GitHub: brijmansuriya

---

📌 Final Thoughts

Great AI code tools need great datasets.

Web Beast gives you Laravel + Node.js power in a clean, usable format.

Whether you're training LLMs or building a Copilot alternative, this is your launchpad.

---

🔗 Get Started Now

👉 Download Web Beast on Hugging Face
🌐 Website: web-beast.com