config.json

{
    "$schema": "../../config.schema.json",
    "experiment_name": "baseline",
    "job_name": "baseline_job",
    "job_description": "",
    "data_formats": ["*"],
    "main_instruction": "",
    "use_checkpoints": true,
    "path": {},
    "index": {
        "index_name_prefix": "ci",
        "ef_construction": [400],
        "ef_search": [400],
        "chunking": {
            "preprocess": false,
            "chunk_size": [1000],
            "overlap_size": [200],
            "generate_title": false,
            "generate_summary": false,
            "override_content_with_summary": false,
            "chunking_strategy": "basic",
            "azure_document_intelligence_model": "prebuilt-read"
        },
        "embedding_model": [
            {
                "type": "sentence-transformer",
                "model_name": "all-mpnet-base-v2"
            }
        ],
        "sampling": {
            "sample_data": false,
            "percentage": 5,
            "optimum_k": "auto",
            "min_cluster": 2,
            "max_cluster": 30
        }
    },
    "language":{
        "analyzer": {
            "analyzer_name": "en.microsoft",
            "index_analyzer_name": "",
            "search_analyzer_name": "",
            "char_filters": [],
            "tokenizers": [],
            "token_filters": []
        },
        "query_language": "en-us"
    },
    "rerank": {
        "enabled": true,
        "type": "cross_encoder",
        "llm_rerank_threshold": 3,
        "cross_encoder_at_k": 4,
        "cross_encoder_model": "cross-encoder/stsb-roberta-base"
    },
    "search": {
        "retrieve_num_of_documents": 5,
        "search_type": [
            "search_for_manual_hybrid",
            "search_for_match_Hybrid_multi",
            "search_for_match_semantic"
        ],
        "search_relevancy_threshold": 0.8
    },
    "query_expansion": {
        "query_expansion": true,
        "hyde": "generated_hypothetical_answer",
        "min_query_expansion_related_question_similarity_score": 90,
        "expand_to_multiple_questions": true
    },
    "openai": {
        "azure_oai_chat_deployment_name": "gpt-35-turbo",
        "azure_oai_eval_deployment_name": "gpt-35-turbo",
        "temperature": 0
    },
    "eval": {
        "metric_types": [
            "fuzzy_score",
            "cosine_ochiai",
            "rouge2_recall",
            "bert_all_MiniLM_L6_v2",
            "bert_distilbert_base_nli_stsb_mean_tokens",
            "llm_answer_relevance",
            "llm_context_precision"
        ]
    }
}