# Hugging Face upload checklist

Upload everything in this folder to your HF model repo (e.g. `your-user/meltdown-q8`).

## Files in `hf/` (ready now)

| File | Size (approx) | Required |
|---|---|---|
| `README.md` | small | Yes — model card |
| `system_prompt.txt` | small | Yes |
| `config.recommended.json` | small | Yes |
| `benchmark_comparison.png` | small | Yes |
| `agent_category_chart.png` | small | Yes |
| `local_vs_cloud.png` | small | Yes |
| `eval_standard.json` | small | Recommended |
| `eval_agent.json` | small | Recommended |
| `frontier_reference.json` | small | Recommended |
| `.gitattributes` | tiny | Yes (enables GGUF LFS) |

## Add before upload

| File | Source | Size |
|---|---|---|
| `Meltdown_Q8.gguf` | `..\Meltdown_Q8.gguf` | ~3.1 GB |

```powershell
Copy-Item "..\Meltdown_Q8.gguf" ".\Meltdown_Q8.gguf"
```

## Upload commands

From this `hf/` folder:

```powershell
# One-time: install HF CLI + git-lfs
pip install huggingface_hub
git lfs install

# Create repo and push (replace YOUR_USER/meltdown-q8)
huggingface-cli repo create meltdown-q8 --type model
git init
git lfs track "*.gguf"
git add .
git commit -m "Add Meltdown Q8 GGUF and model card"
git remote add origin https://huggingface.co/YOUR_USER/meltdown-q8
git push origin main
```

Or drag-and-drop all files in the HF web UI (enable Git LFS for the `.gguf`).

## Do not upload

- `benchmarks/` scripts and logs (dev-only)
- `terminals/`, `mcps/`, `agent-tools/`
- Training data (`meltdown_ape_eval50.jsonl`)