MemNet

MemNet is a "self-improving" memory layer designed for .NET developers, providing long-term and short-term memory management, similarity search, memory consolidation/refinement, and persistence for LLM-based applications. It facilitates context reuse in conversations, recommendations, personalization, and context-aware applications.

In simple terms: LLMs are stateless, and MemNet helps you remember users' previous behaviors and conversation content, making your applications smarter and more personalized.

Why Use MemNet

• Transform scattered conversations/events into retrievable memories to enhance context awareness.
• Built-in vector search and memory consolidation strategies to reduce duplicate memories and noise.
• Support for multiple storage backends (in-memory, Qdrant, Redis, Chroma, Milvus, etc.) for easy expansion and persistence.
• Integration with any LLM/Embedding provider (pluggable embedding layer).

Installation

Using dotnet CLI:

dotnet add package MemNet

Using NuGet Package Manager Console:

Install-Package MemNet

Quick Start

1. Configure appsettings.json

First, configure the Embedder, LLM, and VectorStore in your project's appsettings.json file:

{ "MemNet": { "Embedder": { "Endpoint": "https://api.openai.com/v1/", "Model": "text-embedding-3-large", "ApiKey": "your-embedding-api-key" }, "LLM": { "Endpoint": "https://api.openai.com/v1/", "Model": "gpt-4", "ApiKey": "your-llm-api-key" } } }

Tip: For security purposes, it's recommended to use more secure methods to store API keys rather than writing them directly in the configuration file.

2. Register Services

Register MemNet services in Program.cs:

using MemNet; using MemNet.Abstractions; using MemNet.Models; using Microsoft.Extensions.Configuration; using Microsoft.Extensions.DependencyInjection; var configuration = new ConfigurationBuilder() .AddJsonFile("appsettings.json") // Requires NuGet Package: Microsoft.Extensions.Configuration.Json .Build(); var services = new ServiceCollection(); services.AddMemNet(configuration); await using var serviceProvider = services.BuildServiceProvider(); var memoryService = serviceProvider.GetRequiredService<IMemoryService>(); await memoryService.InitializeAsync();