LLM Wrapper to use
Key to use for output, defaults to text
Prompt object to use
OptionalcriterionOptionalevaluationThe name of the evaluation.
OptionalllmKwargs to pass to LLM
OptionalmemoryOptionalskipStaticdeserializeLoad a chain from a json-like object describing it.
StaticfromLLMCreate a new instance of the PairwiseStringEvalChain.
Optionalcriteria: CriteriaLikeThe criteria to use for evaluation.
OptionalchainOptions: Partial<Omit<LLMEvalChainInput<EvalOutputType, BaseLanguageModelInterface>, "llm">>Options to pass to the chain.
StaticresolveOptionalcriteria: CriteriaLikeStaticresolve
A chain for comparing two outputs, such as the outputs of two models, prompts, or outputs of a single model on similar inputs.