evaluator
LLMJudge
properties
client
include_rationale
input_parser
name
prompt
response_parser
score_config
setters
set_client
set_include_rationale
set_input_parser
set_name
set_prompt
set_response_parser
set_score_config
methods
evaluate
dunder_methods
__init__
Creates an LLM as a Judge Evaluator.
Arguments:
client
- The client to use as the LLM Judge.prompt
- The prompt for the judge to use when performing evaluations.score_config
- Either a derived Score subclass, or the configuration for thename
- The name to give to this LLM Judge evaluator, optional.input_parser
- Function to parse the input and update correspondingresponse_parser
- Function to parse the response and update correspondinginclude_rationale
- Whether to include the LLM’s rationale as part ofapi_key
- API key for accessing the Unify API. If None, it attempts to
Raises:
UnifyError
: If the API key is missing.