GSS-creator/adtc-2026-submission-G1-360M

G1-360M V18.2 is a compact, production-ready enterprise AI model built by Gaston Software Solutions LLP (GSS-LLP), Kampala, Uganda — designed to run fully offline on a laptop and reason about real business systems, enterprise workflows, APIs, data validation, and security controls. # G1-360M V18.2 — ADTC 2026 Submission **Team:** Gaston Software Solutions LLP (GSS-LLP) **Track:** Corporate Enterprise **Model:** G1-360M_V18.2_PRODUCTION_F16 **Hugging Face:** huggingface.co **Ollama Registry:** ollama.com --- **G1-360M V18.2** is a compact, production-ready enterprise AI model built by **Gaston Software Solutions LLP (GSS-LLP)**, Kampala, Uganda — designed to run fully offline on a laptop and reason about real business systems, enterprise workflows, APIs, data validation, and security controls. > **Runs on 8 GB RAM.** G1-360M V18.2 (GGUF F16) is designed for the standard 8 GB laptop profile. When running through **Ollama desktop**, the model loads efficiently and can operate with as little as 4 GB of available RAM, making it practical for offline enterprise environments on any modern laptop or workstation. --- ## ✅ Submission Checklist - [x] Repository is **public** on GitHub - [x] `metadata.json` fully filled in — no placeholder values remain - [x] `metadata.json` contains exactly **2 test prompts** in the `test_prompts` array - [x] `download_model.sh` downloads the model to `model/` - [x] Downloaded file is valid **GGUF F16** format - [x] `model/*.gguf` is in `.gitignore` — weight file not committed - [x] `REPORT.md` filled in with full technical writeup - [x] Model runs entirely **offline** — zero external network calls during inference - [x] African Use Case Bonus claimed (`african_alpha_claim: true`) --- ## Model Journey How G1 was built — from proprietary enterprise data to production model: --- ## Training Pipeline The full training pipeline from source data to G1-360M V18.2 Production F16: --- ## How It Works G1's inference architecture — from user prompt to enterprise response: --- ## Quick Start ### Option A — Ollama (easiest, recommended for team members) ```bash ollama run gastonsoftwaresolutions234/g1-360m ``` ### Opti …