AnythingLLM-Private-Deployment

Product Overview

AnythingLLM is an open source project which provides a full-stack application that turns documents into an intelligent chatbot with a sleek UI and easier way to manage your workspaces. The AnythingLLM-Private-Deployment-AMI includes an AnythingLLM docker container which is integrated with a ChromaDB docker container and served via a NGINX ssl web proxy. The AMI solution allows users to maintain a privately hosted vector datastore. Local embedded content can then be included as context for LLM prompts having an effect of bringing awareness of local content to the LLM. New features are being added to AnythingLLM on a regular basis. This AMI will be updated on a bi-weekly basis to incorporate recently developed features.
Version: AnythingLLM-Private-Deployment-Jan24

By: CyberWorx

Categories:
Operating System
Linux/Unix, Amazon Linux 2023
Delivery Methods
Amazon Machine Image

Highlights: Some of the unique RAG features this solution provides include: ability to fetch and embed URL content using a web-based UIability to create individual user accounts with segregated workspacesability to control content similarity as a variable when building LLM prompt contextability to interface with multiple different LLMsability to interface with standalone embedding modelsability to switch between conversation and query modes when interacting with LLMs CyberWorx LLC has provided a basic level of hardening for the AMI which includes: built bi-weekly from of the latest version of Amazon Linux 2023selinux configured to allow nginx proxy for http/s connectionnginx port proxy with https redirectssl certificate generation and installation scriptfirewalld enabled and configured to allow http and https Additional hardening configuration includes: automatic OS patching using anacronchrony time synchronization to Amazon internal time serverstimedatectl is set to Americas/New_Yorkrandomly generated OTP used for initial login

Category: Machine Learning

Delivery Method: Amazon Machine Image