README.md

---
license: mit
tags:
- osu
- music-generation
- beatmap-generation
- lora
---

# LoRA: Arles Style for Mapperatorinator

This is a LoRA (Low-Rank Adaptation) fine-tune for the [OliBomby/Mapperatorinator-v30](https://huggingface.co/OliBomby/Mapperatorinator-v30) model. It has been trained on a custom dataset of beatmaps to generate maps in the "Arles" style.

## Model Details
- **Base Model:** `OliBomby/Mapperatorinator-v30`
- **Model Type:** LoRA

## How to Use

You can use this LoRA in two ways: by downloading the files locally or by loading it directly from the Hugging Face Hub.
Best to use on songs that are hybrid bursts/short streams and jumps

### Option 1: Local Files

1.  **Download and Structure Files:** Download the `adapter_config.json` and `adapter_model.safetensors` files from this repository. Create a folder and place the files inside like this:
    ```
    arles_lora/
    ├── adapter_config.json
    └── adapter_model.safetensors
    ```

2.  **Run Inference:** Use the `lora_path` argument to point to the local **folder** you created.
    ```bash
    python inference.py \
      audio_path='/path/to/your/audio.mp3' \
      output_path='/path/to/your/output_folder' \
      lora_path='/path/to/your/arles_lora' \
      ... # other arguments
    ```

### Option 2: Direct from Hugging Face Hub

You can load the LoRA directly from this repository without manually downloading the files. The script will handle it automatically.

1.  **Run Inference:** Use the `lora_path` argument and set it to the name of this repository.
    ```bash
    python inference.py \
      audio_path='/path/to/your/audio.mp3' \
      output_path='/path/to/your/output_folder' \
      lora_path='mouceen/Mapperatorinator-v30-LoRA-Arles-v1' \
      ... # other arguments
    ```