Repository files navigation

Welcome to theLLM Models repository! This project demonstrates how to run variousLarge Language Models (LLMs) locally using Ollama, integrated with aNext.js app for an easy-to-use interface. The app supports multiple LLMs and includes advanced features likeVercel AI SDK,streaming text,Next.js themes, andshadcn UI components.

The models you can run include, but are not limited to:

• SmolLM Series: 135M, 360M, 1.7B parameters
• TinyLLaMA: Latest and 1.1B versions
• LLaVA-LLaMA3: 8B parameters
• Qwen Series: 0.5B, 1.5B, 7B parameters
• Mistral-Nemo: Latest
• CodeGemma: Latest and other variations
• Aya, Mixtral, and more

You can explore a wide variety of models, such as:

tinyllama:1.1b, llava-llama3:8b, qwen:0.5b, qwen2:7b, smollm:1.7b, llama3.1, gemma2, mixtral, mistral-nemo, and more.

• Local LLM Inference: Run various LLMs locally usingOllama for efficient inference.
• Vercel AI SDK: Integrated for advanced AI features and optimization.
• Streaming Text: Real-time streaming text output for dynamic interactions.
• Next.js Themes: Dark/light mode toggle usingNext Themes.
• shadcn UI Components: Beautiful, customizable UI components for an enhanced user experience.
• Multiple Models: Choose from small, medium, and large models based on your hardware and requirements.

To run this project, you need the following tools installed:

• Node.js (v14+)
• Ollama
• Next.js

1. Clone the repository:

   git clone https://github.com/KenanGain/SmolLM.gitcd SmolLM

2. Install dependencies:

   npm install

3. Run the development server:

   npm run dev

4. Open your browser and navigate tohttp://localhost:3000 to interact with the models.

You can download any of the supported models locally via Ollama:

ollama pull tinyllama:latestollama pull qwen2:7bollama pull smollm:135M

Adjust the model name based on your specific needs.

├── public# Static assets├── src# Source files│   ├── components# UI Components│   ├── pages# Next.js pages│   └── utils# Utility functions for model inference├── package.json# Dependencies and scripts└── README.md# Project documentation

To run a specific model, simply select the model from the dropdown menu in the app and enter a prompt. The app will display the model's output after processing. Thestreaming text feature provides real-time output, improving user interaction.

The supported LLM models provide flexibility in balancing computational resource use and performance:

• TinyLLaMA: Lightweight tasks with minimal resource usage.
• SmolLM Series: Efficient for reasoning and common sense tasks.
• LLaVA-LLaMA3: Best for complex tasks with a larger memory footprint.

• Next.js: For building the front-end app.
• Ollama: For running the LLM models locally.
• Vercel AI SDK: To integrate advanced AI functionality.
• Streaming Text: For real-time, dynamic text output.
• Next Themes: For dark/light mode support.
• shadcn UI: For beautiful and responsive UI components.
• JavaScript/TypeScript: For scripting and model integration.
• Tailwind CSS: For styling the app interface.

Contributions are welcome! Feel free to fork this repository, submit issues, or create pull requests.

This project is licensed under the MIT License - see theLICENSE file for details.

For any inquiries, you can reach me atkenangain2910@gmail.com.

Let me know if you'd like to add or modify anything!