OptionalapiThe API key to use.
OptionalendpointOverride the default endpoint.
OptionalmaxThe maximum number of tokens to generate in the completion. The token count of your prompt plus max_tokens cannot exceed the model's context length.
OptionalmodelThe name of the model to use.
OptionalmodelThe name of the model to use.
Alias for model
OptionalrandomThe seed to use for random sampling. If set, different calls will generate deterministic results.
Alias for seed
OptionalsafeWhether to inject a safety prompt before all conversations.
OptionalsafeWhether to inject a safety prompt before all conversations.
OptionalseedThe seed to use for random sampling. If set, different calls will generate deterministic results.
OptionalstreamWhether or not to include token usage in the stream.
OptionalstreamingWhether or not to stream the response.
OptionaltemperatureWhat sampling temperature to use, between 0.0 and 2.0. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic.
OptionaltopPNucleus sampling, where the model considers the results of the tokens with top_p probability mass.
So 0.1 means only the tokens comprising the top 10% probability mass are considered.
Should be between 0 and 1.
Input to chat model class.