AMF G1 — Historical Repository Pointer
This repository contains no model weights and is not a loadable model.
The complete, executable G1 release is published here:
AMF-G1-4B-Experimental →
The canonical repository contains the materialized 4B checkpoint, its Safetensors files, tokenizer, GGUF Q5_K_M derivative, execution instructions, measurements and research record.
Repository status
| Property | Status |
|---|---|
| Repository role | Historical pointer |
| Model weights present | NO |
| Directly loadable with Transformers | NO |
| Canonical executable release | AMF-G1-4B-Experimental |
| Historical URL preserved | YES |
This repository remains public to preserve earlier links and the continuity of the AMF research record. Model files are not duplicated here, preventing two apparently equivalent releases from fragmenting identity, usage instructions and download statistics.
Research status of G1
G1 is the first learned neural artifact fabricated during the AI Mission Foundry experiments. It is an experimental research artifact, not a qualified production model.
The complete model card, limitations, qualification result, artifact hashes and local deployment measurements are maintained exclusively in the canonical repository.
Attribution
AMF — AI Mission Foundry
Created and developed by Fayçal Benahmed
Stack Moderne — France
Independent research and engineering project
https://stack-moderne.fr/
License
This pointer repository contains no model weights. The canonical G1 artifact is
derived from Qwen/Qwen3-4B-Instruct-2507, distributed under the Apache License
2.0. Refer to the canonical repository for the complete release notices.
Third-party software and components remain governed by their respective licenses.