Model Configuration (BYOM)

Bring Your Own Model - Connect any local or cloud LLM

Input Files

Drop text files here or browse to add them to the ingest directory. Accepted: .txt, .md, .pdf, .html, .htm

Drop files here, or click to browse

Step 1: Quick Start

Choose a preset to quickly configure common endpoints

Step 2: Verify Connection

Test your connection before running the pipeline

Step 3: Discover Models

Fetch available models from your server

Step 4: Configure Model

Set your model parameters and preferences

Base URL Default: http://localhost:11434/v1

API Key Leave blank for local models

Model Selection

Click "Fetch Models" to discover available models

Vision Support Enable Vision/Multimodal processing For document images sent to the model

Embedding Base URL Separate endpoint for bge-m3 embeddings. Can be different from the LLM server.

Embedding Model Default: bge-m3. Must be pulled on the embedding server.

Saves your settings locally in config.json

Stages

Run individually or all together

{% set stages = [ ('ingest', '1', 'Ingest', 'Chunk OCR text files'), ('extract', '2', 'Extract', 'Entities & relationships via LLM'), ('resolve', '3', 'Resolve', 'Deduplicate, pick canonical names'), ('vault', '4', 'Vault', 'Generate Obsidian notes'), ('graph', '5', 'Graph', 'Export GraphML / GEXF / JSON / CSV') ] %} {% for key, num, name, desc in stages %}

Stage {{ num }} {{ name }} {{ desc }}

{% endfor %}

Live log

Streaming output · server-sent events

Breakdown

Entity types

No entities yet — run Extract or Demo.

Relationship types (top 10)

No relationships yet — run Extract or Demo.

ArtificeGraph · an Artifice platform · FastAPI + Jinja2 · vanilla JS · design tokens from LudwigLang.

The Pipeline

Model Configuration (BYOM)

Input Files

Step 1: Quick Start

Step 2: Verify Connection

Connection Successful

Connection Failed

Step 3: Discover Models

Available Models

Vision Models