Enums¶
This document provides details on the enums used within the BerryDB SDK.
ModelProvider¶
The ModelProvider
enum specifies the different providers for machine learning models in BerryDB.
BERRYDB_MODEL: Represents models provided by BerryDB.
VERTEX_AI_MODEL: Represents models from Vertex AI.
HUGGING_FACE_MODEL: Represents models from Hugging Face.
CUSTOM_MODEL: Represents custom models.
IngestType¶
The IngestType
enum defines the source from which data can be ingested into BerryDB.
URL: Represents data ingested from a URL.
FILE: Represents data ingested from a local file.
IngestFileType¶
The IngestFileType
enum represents the different file types that can be ingested into BerryDB.
PDF: Represents PDF files.
XLSX: Represents Excel XLSX files.
HTML: Represents HTML files.
EvalMetrics¶
The EvalMetrics
enum defines the metrics used for evaluating the performance of LLM applications within BerryDB.
ANSWER_RELEVANCY: Metric evaluating the relevancy of the generated answer.
FAITHFULNESS: Metric evaluating the faithfulness of the generated answer to the provided context.
CONTEXTUAL_PRECISION: Metric evaluating the precision of the context used for generation.
CONTEXTUAL_RECALL: Metric evaluating the recall of the context used for generation.
CONTEXTUAL_RELEVANCY: Metric evaluating the relevancy of the context used for generation.
HALLUCINATION_METRIC: Metric evaluating the hallucination in the generated answer.
RETRIEVAL: Collection of Contextual Precision, Contextual Recall and Contextual Relevancy metrics.
GENERATION: Collection of Answer Relevancy and Faithfulness metrics.