Boring Gemini

# Offline Mode Guide Offline mode lets Boring-Gemini run without cloud API calls by using local GGUF models. ## Quick Start (Recommended) ```bash boring offline setup ``` This will: - Download a recommended local model - Save the model path to `.boring.toml` or `.env` - Enable Offline Mode ## Manual Setup 1. **Download a model** - Recommended: `qwen2.5-coder-1.5b` - Use the CLI: ```bash boring model download qwen2.5-coder-1.5b ``` 2. **Point Boring to the model** Add to `.env`: ```env BORING_LOCAL_LLM_MODEL=/absolute/path/to/model.gguf BORING_LOCAL_LLM_CONTEXT_SIZE=4096 ``` 3. **Enable Offline Mode** ```bash boring offline enable ``` ## Recommended Models | ID | Size (MB) | Context | Notes | | --- | --- | --- | --- | | `qwen2.5-coder-1.5b` | 1100 | 32768 | Code-optimized, recommended | | `qwen2.5-1.5b` | 1000 | 32768 | Small and fast | | `llama-3.2-1b` | 800 | 8192 | Smallest Llama | | `phi-3-mini` | 2200 | 4096 | Good for general tasks | ## Useful Commands ```bash boring model list boring model download qwen2.5-coder-1.5b boring offline status boring offline disable ``` ## Troubleshooting - **Model download failed** - Verify internet access and disk space. - Try `--force` to re-download. - **Model not detected** - Check `BORING_LOCAL_LLM_MODEL` path. - Ensure the file ends with `.gguf`. - **Performance too slow** - Use a smaller model (e.g., `llama-3.2-1b`). - Reduce `BORING_LOCAL_LLM_CONTEXT_SIZE`.