Apppublic

MrETL/Lumen

sourceHugging Faceupdated 5mo agoView on Hugging Face

0likes

App README

Lumen: AI Research & Inference Platform

Lumen is a production-grade research platform that provides a high-performance Hono/TypeScript orchestration layer paired with a local inference engine.

Features

—Telemetry & Instrumentation: Real-time visualization of engine throughput, KV-Cache memory states, and token logit distributions.
—Orchestration Pipeline: Multi-stage processing designed to provide immediate streaming feedback.
—Native Inference: Powered by a robust backend using llama-server for optimized local CPU execution.
—Audit Logging: SQL-based historical session metrics for technical review.

Deployment

Hugging Face Spaces / Docker

This repository is optimized for Hugging Face Docker Spaces on Port 7860.

bash

docker build -t lumen .
docker run -p 7860:7860 lumen

Lumen Core v1.0.0