fixml.modules.workflow.response
Classes
A data class to store all information from test evaluation runs. |
Module Contents
- class fixml.modules.workflow.response.LLMInfo
Bases:
pydantic.BaseModel- name: str
- temperature: float
- class fixml.modules.workflow.response.RepositoryInfo
Bases:
pydantic.BaseModel- path: str | pathlib.Path
- git_commit: str
- model_config
- class fixml.modules.workflow.response.ChecklistInfo
Bases:
pydantic.BaseModel- path: str | pathlib.Path
- model_config
- class fixml.modules.workflow.response.TokenInfo
Bases:
pydantic.BaseModel- input_count: int
- output_count: int
- class fixml.modules.workflow.response.ErrorInfo
Bases:
pydantic.BaseModel- name: str
- description: str
- class fixml.modules.workflow.response.CallResult
Bases:
pydantic.BaseModel- start_time: datetime.datetime
- end_time: datetime.datetime
- files_evaluated: List[str]
- context: Dict[str, str]
- prompt: str
- success: bool
- parsed_response: Dict | None
- class fixml.modules.workflow.response.EvaluationResponse(**data)
Bases:
pydantic.BaseModel,fixml.modules.mixins.WriteableMixinA data class to store all information from test evaluation runs.
Here is the schema: {
- model {
name temperature
} repository {
path git_commit object (excluded when exporting)
} checklist {
path object (excluded when exporting)
} call_results [{
start_time end_time tokens_used {
input_count output_count
} files_evaluated injected prompt success parsed_response errors [{
name description
}]
}]
}
- repository: RepositoryInfo
- checklist: ChecklistInfo
- call_results: List[CallResult]
- to_json(output_path: str | pathlib.Path, exist_ok=False)
- classmethod from_json(json_path: str | pathlib.Path)
Reconstruct an instance of EvaluationResponse from JSON file.
Requires checklist path and repository path to actually exist in the system, otherwise an error will be thrown.