# Hugging Face Publish

## 1) Go to release folder

```bash
cd /Users/robert/bibleai-backup/release/BibleAI-Gemma4-E4B-CPT-SFT-DPO-20260414
```

## 2) Verify GGUF checksums

```bash
sha256sum gguf/final_merged.BF16.gguf gguf/final_merged.Q8_0.gguf
```

Expected:

- `e07e38d28d3032d3b438b7b8b90cbf4cf5e66177b52e8f60673cac3586dc10a1`  `final_merged.BF16.gguf`
- `3c7f5f9caf080fe44720f16b5f4b5e7e95a097d6be3d1d8d89aea22e8574bad1`  `final_merged.Q8_0.gguf`

## 3) Log in to Hugging Face

```bash
huggingface-cli login
```

or

```bash
hf auth login
```

## 4) Push the full release

```bash
HF_REPO=<your-username-or-org>/<your-model-repo> ./scripts/upload_to_hf.sh
```

Example:

```bash
HF_REPO=rhemabible/BibleAI ./scripts/upload_to_hf.sh
```

This pushes:

- HF merged model (`hf/`)
- GGUF (`gguf/` BF16 + Q8_0)
- Ollama Modelfiles (`ollama/`)
- adapters, checksums, logs, docs

Canonical names:
- HF repo: `rhemabible/BibleAI`
- Ollama models: `bibleaiq8` and `bibleaibf16`