AgentOptimizationPatch.md

LaunchDarklyApi::AgentOptimizationPatch

Properties

Name	Type	Description	Notes
max_attempts	Integer	The maximum number of attempts	[optional]
model_choices	Array<String>	The models to try	[optional]
judge_model	String	The model used to judge responses	[optional]
variable_choices	Array<Hash<String, Object>>	Sets of prompt variable values to try	[optional]
acceptance_statements	Array<AgentOptimizationAcceptanceStatement>	Statements a response must satisfy to pass	[optional]
judges	Array<AgentOptimizationJudge>	Judges that score each response	[optional]
user_input_options	Array<String>	User inputs to test with	[optional]
ground_truth_responses	Array<String>	Expected responses to compare against	[optional]
metric_key	String	The key of the metric to optimize	[optional]
token_limit	Integer	The maximum number of tokens to use	[optional]
variation_key	String	The key of the variation to start from	[optional]
label	String	A label for the optimization	[optional]
latency_optimization	Boolean	Whether to optimize for lower latency	[optional]
token_optimization	Boolean	Whether to optimize for lower token use	[optional]
auto_commit	Boolean	Whether to save the best result as a new variation automatically	[optional]

Example

require 'launchdarkly_api'

instance = LaunchDarklyApi::AgentOptimizationPatch.new(
  max_attempts: null,
  model_choices: null,
  judge_model: null,
  variable_choices: null,
  acceptance_statements: null,
  judges: null,
  user_input_options: null,
  ground_truth_responses: null,
  metric_key: null,
  token_limit: null,
  variation_key: null,
  label: null,
  latency_optimization: null,
  token_optimization: null,
  auto_commit: null
)