multilingual_indic_rag.py

# coding=utf-8
import datasets

class MultilingualIndicRagConfig(datasets.BuilderConfig):
    def __init__(self, **kwargs):
        super().__init__(**kwargs)

class MultilingualIndicRag(datasets.GeneratorBasedBuilder):
    VERSION = datasets.Version("1.0.0")
    
    BUILDER_CONFIGS = [
        datasets.BuilderConfig(name="as", version=VERSION, description="Dataset for as"),
        datasets.BuilderConfig(name="bn", version=VERSION, description="Dataset for bn"),
        datasets.BuilderConfig(name="en", version=VERSION, description="Dataset for en"),
        datasets.BuilderConfig(name="gu", version=VERSION, description="Dataset for gu"),
        datasets.BuilderConfig(name="hi", version=VERSION, description="Dataset for hi"),
        datasets.BuilderConfig(name="kn", version=VERSION, description="Dataset for kn"),
        datasets.BuilderConfig(name="ks", version=VERSION, description="Dataset for ks"),
        datasets.BuilderConfig(name="mai", version=VERSION, description="Dataset for mai"),
        datasets.BuilderConfig(name="ml", version=VERSION, description="Dataset for ml"),
        datasets.BuilderConfig(name="mni", version=VERSION, description="Dataset for mni"),
        datasets.BuilderConfig(name="mr", version=VERSION, description="Dataset for mr"),
        datasets.BuilderConfig(name="ne", version=VERSION, description="Dataset for ne"),
        datasets.BuilderConfig(name="or", version=VERSION, description="Dataset for or"),
        datasets.BuilderConfig(name="pa", version=VERSION, description="Dataset for pa"),
        datasets.BuilderConfig(name="sat", version=VERSION, description="Dataset for sat"),
        datasets.BuilderConfig(name="ta", version=VERSION, description="Dataset for ta"),
        datasets.BuilderConfig(name="te", version=VERSION, description="Dataset for te"),
        datasets.BuilderConfig(name="ur", version=VERSION, description="Dataset for ur"),
    ]
    
    DEFAULT_CONFIG_NAME = "hi"  # Default to Hindi
    
    def _info(self):
        features = datasets.Features({
            "question": datasets.Value("string"),
            "answer": datasets.Value("string"),
            "reasoning": datasets.Value("string"),
            "paragraph": datasets.Value("string"),
            "title": datasets.Value("string"),
            "wiki_id": datasets.Value("string"),
            "url": datasets.Value("string"),
            "language": datasets.Value("string"),
            "source_lang": datasets.Value("string"),
            "model_name": datasets.Value("string"),
            "temperature": datasets.Value("string"),
            "max_tokens": datasets.Value("string"),
            "text": datasets.Value("string"),
        })
        
        return datasets.DatasetInfo(
            description="Multilingual Indic RAG Suite",
            features=features,
            supervised_keys=None,
        )
    
    def _split_generators(self, dl_manager):
        # Use the language-specific parquet file
        lang_code = self.config.name
        data_file = dl_manager.download(f"data/{lang_code}_data.parquet")
        
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={"filepath": data_file}
            )
        ]
    
    def _generate_examples(self, filepath):
        import pandas as pd
        
        df = pd.read_parquet(filepath)
        
        for idx, row in df.iterrows():
            yield idx, {
                "question": row["question"],
                "answer": row["answer"],
                "reasoning": row["reasoning"],
                "paragraph": row["paragraph"],
                "title": row["title"],
                "wiki_id": row["wiki_id"],
                "url": row["url"],
                "language": row["language"],
                "source_lang": row["source_lang"],
                "model_name": row["model_name"],
                "temperature": row["temperature"],
                "max_tokens": row["max_tokens"],
                "text": row["text"],
            }