Rapid.AI.Ollama.Framework 1.2.5

Rapid.AI.Ollama.Framework

Rapid.AI.Ollama.Framework is a lightweight C# client library that allows developers to interact with locally running Ollama models. It supports both stateless prompt generation and contextual multi-turn chat conversations using the Ollama REST API.

✨ Features

• 🔁 Stateless prompt generation using /api/generate
• 💬 Context-aware chat with conversation history using /api/chat
• 📦 Simple and easy-to-integrate C# API
• 🔧 Supports streaming output for prompt generation

🚀 Getting Started

📦 Prerequisites

• .NET 6 or later
• Ollama running locally (default port http://localhost:11434)
• A downloaded model (e.g., llama3, llama3.2:1b, etc.)

🧪 Usage

✨ 1. Generate Prompt Response

using Rapid.AI.Ollama.Framework;

string result = OllamaClient.Generate("http://localhost:11434/api/generate", "What is quantum physics?", "llama3.2:1b");

Console.WriteLine(result);

This uses the /api/generate endpoint with streaming enabled, and returns a stateless response for the given prompt.

💬 2. Chat with Context (Multi-Turn)

using Rapid.AI.Ollama.Framework;

// First user message
string reply1 = OllamaClient.Chat("http://localhost:11434/api/chat", "Who was Marie Curie?", "llama3.2:1b");
Console.WriteLine("AI: " + reply1);

// Follow-up message
string reply2 = OllamaClient.Chat("http://localhost:11434/api/chat", "What was her contribution to science?", "llama3.2:1b");
Console.WriteLine("AI: " + reply2);

You can maintain context by using Chat(). The chat history is kept internally.

🔄 To Clear Chat History:

OllamaClient.ClearChatHistory();

📌 Notes

Generate method uses the /api/generate endpoint and streams the output.

Chat method uses the /api/chat endpoint and maintains internal chat history.

Timeout is set to 5 minutes for long-running responses.

📁 Project Structure

OllamaClient
├── Generate(...)        // Stateless streaming prompt generation
├── Chat(...)            // Stateful chat with message history
├── ClearChatHistory()   // Clears the internal chat history

🧱 Example Model Names

• llama3

• llama3.2:1b

• mistral

Any other model available through Ollama

Ensure the model is already pulled by running:

ollama run llama3.2:1b

📃 License

MIT License – free to use, modify, and distribute.

🤝 Contributions

Feature requests and improvements are welcome. Please fork and PR your changes!