constants.py

# TRACE Evaluation Metric Constants
RELEVANCE_SCORE = "relevance_score"
UTILIZATION_SCORE = "utilization_score"
COMPLETENESS_SCORE = "completeness_score"
ADHERENCE_SCORE = "adherence_score" 

HF_DATASET_REPO_NAME = "maddiaks/RAG26Dataset"
HF_REPO_TYPE = "dataset"

GENERAL_PURPOSE = 'GeneralPurpose'
CHARACTER = 'Character'
TOKEN_BASED = 'TokenBased'
SENTENCE_BASED_LANGCHAIN = 'SentenceBasedLangchain'
SENTENCE_BASED_LLAMA = 'SentenceBasedLlama'
SMALL2BIG = 'Small2Big'
HYBRID = 'HybridChunking'

CHUNKING_STRATEGIES = [
    GENERAL_PURPOSE,
    CHARACTER,
    TOKEN_BASED,
    SENTENCE_BASED_LANGCHAIN,
    # SENTENCE_BASED_LLAMA,
    SMALL2BIG,
    # HYBRID
]


EMBEDDING_MODELS = [
    "BAAI/bge-large-en-v1.5",
    "intfloat/e5-large-v2",
    "all-MiniLM-L6-v2"
]

RELEVANCE_SCORE = "relevance_score"
UTILIZATION_SCORE = "utilization_score"
COMPLETENESS_SCORE = "completeness_score"
ADHERENCE_SCORE = "adherence_score" 


default_json = {
    "relevance_explanation": "Documents 1 and 4 contain useful information for answering the question. Document 1 mentions 'Signal Information' and 'Self Diagnosis', which are relevant to finding signal information. Document 4 provides additional context about signal information, such as it being only available for digital channels.",
    "all_relevant_sentence_keys": [
        "1aJUNK",
        "1b",
        "4d",
        "4e"
    ],
    "overall_supported_explanation": "The response claims that signal information can be found in the 'Signal Information' or 'Self Diagnosis' section, and that the current channel must be a digital channel. Document 1 supports the first claim, and Document 4 supports the second claim. However, the response also mentions removing protective stickers, which is not directly related to finding signal information. Document 4 mentions removing stickers, but it is not directly supporting the response's claim.",
    "overall_supported": True,
    "sentence_support_information": [
        {
            "response_sentence_key": "a.",
            "explanation": "Document 1 supports this sentence, as it mentions 'Signal Information' and 'Self Diagnosis' which are relevant to finding signal information.",
            "supporting_sentence_keys": [
                "1a",
                "1b"
            ],
            "fully_supported": True
        },
        {
            "response_sentence_key": "b.",
            "explanation": "Document 4 supports this sentence, as it mentions that signal information is only available for digital channels. Additionally, Document 4 mentions removing protective stickers, which is also mentioned in the response.",
            "supporting_sentence_keys": [
                "4d",
                "4e",
                "4g"
            ],
            "fully_supported": True
        },
        {
            "response_sentence_key": "c.",
            "explanation": "Document 4 supports this sentence, as it explicitly states that signal information is only available for digital channels.",
            "supporting_sentence_keys": [
                "4e"
            ],
            "fully_supported": True
        }
    ],
    "all_utilized_sentence_keys": [
        "1a",
        "1b",
        "4d",
        "4e",
        "4g"
    ]
}