docs: professional README — Feb 2026
This commit is contained in:
parent
618fab9865
commit
04a3ab9a93
64
README.md
64
README.md
@ -1,42 +1,44 @@
|
|||||||
|
<div align="center">
|
||||||
|
|
||||||
# Organ Architect
|
# Organ Architect
|
||||||
*Part of the Inference-X Ecosystem*
|
|
||||||
|
|
||||||
[](LICENSE)
|
**Visualize and analyze AI model internals**
|
||||||
|
|
||||||
**Analyze AI model internals. Like an MRI for AI models.**
|
[]()
|
||||||
|
|
||||||
Before transplanting organs between models, you need to understand what's inside.
|
[inference-x.com](https://inference-x.com)
|
||||||
Organ Architect visualizes layers, attention heads, and topology of GGUF models.
|
|
||||||
|
</div>
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 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
|
```bash
|
||||||
pip install fastapi uvicorn python-multipart
|
git clone https://git.inference-x.com/inference-x-community/organ-architect
|
||||||
python organ_architect.py
|
cd organ-architect && npm install
|
||||||
# → Web UI + API at http://localhost:7940
|
npm start
|
||||||
|
|
||||||
|
# Open http://localhost:3000
|
||||||
|
# Drag and drop any .gguf file
|
||||||
```
|
```
|
||||||
|
|
||||||
## API
|
## Supported architectures
|
||||||
```
|
|
||||||
GET / — Web interface (upload GGUF)
|
|
||||||
POST /analyze — Analyze GGUF model architecture
|
|
||||||
POST /extract-spec — Get transplantation specification
|
|
||||||
GET /health — Service status
|
|
||||||
```
|
|
||||||
|
|
||||||
## Use Cases
|
LLaMA · Mistral · Mixtral · Qwen · Phi · Gemma · Falcon · GPT-NeoX · Command-R · and all GGUF-format models
|
||||||
- Understand model architecture before fine-tuning
|
|
||||||
- Compare architectures across model families
|
|
||||||
- Plan organ extraction strategies
|
|
||||||
|
|
||||||
## 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. 🇲🇦
|
*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)*
|
||||||
|
|
||||||
## License
|
|
||||||
[Business Source License 1.1](LICENSE)
|
|
||||||
**Author:** Salka Elmadani
|
|
||||||
|
|||||||
Loading…
Reference in New Issue
Block a user