from dataclasses import dataclass


SUPPORTED_LANGUAGES = [
    "English",
    "Spanish",
    "French",
    "Arabic",
    "Hindi",
    "Telugu",
    "Nepali",
    "Chinese",
    "Japanese",
    "Korean",
    "Portuguese",
]


@dataclass
class PipelineConfig:
    project_name: str = "WorldDisasterLM-8B"
    base_model: str = "meta-llama/Llama-3.1-8B-Instruct"
    output_dir: str = "outputs"
    data_dir: str = "data"
    seed: int = 42


DEFAULT_CONFIG = PipelineConfig()