config/config.js

/**
 * @typedef {import('app/config/types').AppConfig} AppConfig
 */

/**
 * @type {AppConfig}
 */
const config = {
    cache_folder: ".cache",
    embeddings: {
        embeddings_path: ".embeddings",
        embedding_model: {
            model_name: "maidalun1020/bce-embedding-base_v1",
        },
        splade_config: {
            n_batch: 8,
        },
        chunk_sizes: [1024],
        document_settings: [
            {
                doc_path: "documents/falcon-refinedweb",
                additional_parser_settings: {
                    md: {
                        skip_first: true,
                        merge_sections: false,
                        remove_images: true,
                    },
                },
                passage_prefix: "passage: ",
            },
        ],
    },
    semantic_search: {
        max_k: 15,
        max_char_size: 4096,
        query_prefix: "query: ",
    },
};

module.exports = config;