memoryscope.scheme.model_response
- pydantic model memoryscope.scheme.model_response.ModelResponse[源代码]
基类:
BaseModel
Show JSON schema
{ "title": "ModelResponse", "type": "object", "properties": { "message": { "anyOf": [ { "$ref": "#/$defs/Message" }, { "type": "null" } ], "default": null, "description": "generation model result" }, "delta": { "default": "", "description": "New text that just streamed in (only used when streaming)", "title": "Delta", "type": "string" }, "embedding_results": { "anyOf": [ { "items": { "items": { "type": "number" }, "type": "array" }, "type": "array" }, { "items": { "type": "number" }, "type": "array" } ], "default": [], "description": "embedding vector", "title": "Embedding Results" }, "rank_scores": { "additionalProperties": { "type": "number" }, "default": {}, "description": "The rank scores of each documents. key: index, value: rank score", "title": "Rank Scores", "type": "object" }, "m_type": { "$ref": "#/$defs/ModelEnum", "default": "generation_model", "description": "One of LLM, EMB, RANK." }, "status": { "default": true, "description": "Indicates whether the model call was successful.", "title": "Status", "type": "boolean" }, "details": { "default": "", "description": "The details information for model call, usually for storage of raw response or failure messages.", "title": "Details", "type": "string" }, "raw": { "default": "", "description": "Raw response from model call", "title": "Raw" }, "meta_data": { "default": {}, "description": "meta data for model response", "title": "Meta Data", "type": "object" } }, "$defs": { "Message": { "description": "Represents a structured message object with details about the sender, content, and metadata.\n\nAttributes:\n role (str): The role of the message sender (e.g., 'user', 'assistant', 'system').\n role_name (str): Optional name associated with the role of the message sender.\n content (str): The actual content or text of the message.\n time_created (int): Timestamp indicating when the message was created.\n memorized (bool): Flag to indicate if the message has been saved or remembered.\n meta_data (Dict[str, str]): Additional data or context attached to the message.", "properties": { "role": { "description": "The role of the message sender (user, assistant, system)", "title": "Role", "type": "string" }, "role_name": { "default": "", "description": "Name describing the role of the message sender", "title": "Role Name", "type": "string" }, "content": { "description": "The primary content of the message", "title": "Content", "type": "string" }, "time_created": { "description": "Timestamp marking the message creation time", "title": "Time Created", "type": "integer" }, "memorized": { "default": false, "description": "Indicates if the message is flagged for memory retention", "title": "Memorized", "type": "boolean" }, "meta_data": { "additionalProperties": { "type": "string" }, "default": {}, "description": "Supplementary data attached to the message", "title": "Meta Data", "type": "object" } }, "required": [ "role", "content" ], "title": "Message", "type": "object" }, "ModelEnum": { "description": "An enumeration representing different types of models used within the system.\n\nMembers:\n GENERATION_MODEL: Represents a model responsible for generating content.\n EMBEDDING_MODEL: Represents a model tasked with creating embeddings, typically used for transforming data into a\n numerical form suitable for machine learning tasks.\n RANK_MODEL: Denotes a model that specializes in ranking, often used to order items based on relevance.", "enum": [ "generation_model", "embedding_model", "rank_model" ], "title": "ModelEnum", "type": "string" } } }
- Fields:
- field delta: str = ''
New text that just streamed in (only used when streaming)
- field embedding_results: List[List[float]] | List[float] = []
embedding vector
- field rank_scores: Dict[int, float] = {}
The rank scores of each documents. key: index, value: rank score
- field status: bool = True
Indicates whether the model call was successful.
- field details: str = ''
The details information for model call, usually for storage of raw response or failure messages.
- field raw: Any = ''
Raw response from model call
- field meta_data: Dict[str, Any] = {}
meta data for model response