wenda-LLM/wenda
Wenda
Wenda est une plateforme d'orchestration LLM et un moteur de workflow personnalisé conçu pour gérer plusieurs backends de modèles de langage via une interface unifiée. Il fonctionne comme une passerelle IA auto-hébergée qui permet l'exécution de séquences de tâches complexes et de flux de conversation automatisés.
Le système utilise des plugins JavaScript pour orchestrer les workflows et déclencher des appels API externes. Il prend en charge la génération augmentée par récupération (RAG) en injectant des données pertinentes provenant de bases de données vectorielles et de fichiers hors ligne dans les prompts pour augmenter la précision des réponses.
La plateforme est conçue pour des déploiements sur réseau privé, avec une gestion des accès multi-utilisateurs et la capacité d'exécuter des modèles open source quantifiés pour s'adapter à des contraintes matérielles spécifiques. Elle inclut également un suivi d'historique basé sur les sessions pour maintenir le contexte conversationnel.
Features
- JavaScript Plugin Engines - Provides a framework for automating conversation flows and external API calls using JavaScript plugins.
- Retrieval-Augmented Generation - Improves model accuracy by injecting relevant data from vector stores and offline files into the conversation context.
- LLM Workflow Orchestrations - Coordinates the execution of multiple language model backends through a unified interface to streamline complex AI task sequences.
- Quantized Model Deployments - Runs large language models within limited hardware constraints by reducing the precision of model weights.
- Model Orchestration Platforms - Offers a unified interface for managing multiple language model backends and executing complex AI task sequences.
- Private AI Deployments - Runs open source language models on local hardware to ensure data privacy and security.
- Self-Hosted AI Platforms - Provides a private network deployment for running quantized open source models with multi user access control.
- Local Model Deployment - Runs open source models locally or via API using quantization to fit within hardware resource constraints.
- Vector-Store Augmented Generation - Injects relevant data from vector stores and offline files into prompts to increase response accuracy.
- LLM Plugin Orchestrators - Utilizes JavaScript plugins to manage conversation flows, trigger external API calls, and automate task sequences.
- Custom AI Workflows - Automates complex AI task sequences by combining multiple model backends with custom JavaScript plugins and external API calls.
- JavaScript Plugin Systems - Executes JavaScript plugins to define conversation flows, call external APIs, and switch model weights dynamically.
- Unified Model Interfaces - Provides a standardized execution interface across different language model providers for seamless switching of weights and APIs.
- Large Model Optimizations - Deploys large language models using quantization to fit within the limited computational resources of a private server.
- RAG Knowledge Management - Injects data from vector stores and offline files into LLM prompts to improve response accuracy.
- Team Collaboration Management - Permits concurrent usage by multiple users within a private network deployment to support team collaboration.
- Conversation History Stores - Stores and retrieves dialogue logs to maintain conversational context across multiple user interactions.
- Private Multi-Tenant Hosting - Manages concurrent access for multiple users within a self-hosted environment to ensure data privacy.
- Multi-User AI Chat Interfaces - Provides a shared private network interface where multiple users can concurrently interact with language models.
- Natural Language Processing - Listed in the “Natural Language Processing” section of the FunNLP awesome list.
- LLM Utilities - Platform for LLM orchestration with external knowledge base lookups.
