From 04a3ab9a9373e6c1d3a008d04810e3f2cf2bdbd3 Mon Sep 17 00:00:00 2001 From: Inference-X Date: Wed, 25 Feb 2026 17:29:59 +0000 Subject: [PATCH] =?UTF-8?q?docs:=20professional=20README=20=E2=80=94=20Feb?= =?UTF-8?q?=202026?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- README.md | 64 ++++++++++++++++++++++++++++--------------------------- 1 file changed, 33 insertions(+), 31 deletions(-) diff --git a/README.md b/README.md index e18f73f..4cbe25f 100644 --- a/README.md +++ b/README.md @@ -1,42 +1,44 @@ +
+ # Organ Architect -*Part of the Inference-X Ecosystem* -[![License: BSL-1.1](https://img.shields.io/badge/License-BSL_1.1-orange.svg)](LICENSE) +**Visualize and analyze AI model internals** -**Analyze AI model internals. Like an MRI for AI models.** +[![Status](https://img.shields.io/badge/Status-Beta-blue.svg)]() -Before transplanting organs between models, you need to understand what's inside. -Organ Architect visualizes layers, attention heads, and topology of GGUF models. +[inference-x.com](https://inference-x.com) + +
+ +--- + +## What is Organ Architect? + +Organ Architect is a visual tool for exploring the internals of GGUF AI models. Think of it as a surgical suite for neural networks — layer shapes, attention patterns, weight distributions, all visible. + +## Features + +- **Architecture visualization** — interactive graph of all layers +- **Attention head analysis** — what each head attends to +- **Weight distribution** — histograms per layer +- **Layer comparison** — diff two models layer-by-layer +- **Export** — save analysis as JSON or PDF report + +## Quick start -## Deploy ```bash -pip install fastapi uvicorn python-multipart -python organ_architect.py -# → Web UI + API at http://localhost:7940 +git clone https://git.inference-x.com/inference-x-community/organ-architect +cd organ-architect && npm install +npm start + +# Open http://localhost:3000 +# Drag and drop any .gguf file ``` -## API -``` -GET / — Web interface (upload GGUF) -POST /analyze — Analyze GGUF model architecture -POST /extract-spec — Get transplantation specification -GET /health — Service status -``` +## Supported architectures -## Use Cases -- Understand model architecture before fine-tuning -- Compare architectures across model families -- Plan organ extraction strategies +LLaMA · Mistral · Mixtral · Qwen · Phi · Gemma · Falcon · GPT-NeoX · Command-R · and all GGUF-format models -## Roadmap -- [ ] Attention pattern visualization -- [ ] Weight distribution analysis -- [ ] Layer similarity comparison -- [ ] Automatic organ boundary detection -- [ ] Export to Organ Store format +--- -Built in Morocco for the world. 🇲🇦 - -## License -[Business Source License 1.1](LICENSE) -**Author:** Salka Elmadani +*See also: [organ-architecture](https://git.inference-x.com/inference-x/organ-architecture) · [organ-store](https://git.inference-x.com/inference-x-community/organ-store)*