List of messages specifying the conversation so far.
See Available Models for possible values.
Parameter which penalises tokens based on their frequency in the model’s output so far. The larger the value, the higher the penalisation. 0.0 means no frequency penalty. Defaults to 0.0.
The maximum number of new tokens to be generated by the model. Note that this is limited by the model’s context length. Defaults to 1024.
Parameter which penalises tokens based on whether they have appeared in the model’s output so far. The larger the value, the higher the penalisation. 0.0 means no presence penalty. Defaults to 0.0.
Random seed used for generations. The same value forces the model to sample the same output.
A list of stop strings used to control generation. If the model generates one of these, it will stop.
Set to true to enable streaming. See Chat Streaming
Positive number representing the temperature to use for generation. Higher values will make the output more unformly random or creative. 0.0 means greedy decoding. Defaults to 0.4.
Controls how the model may use the provided tools. Set to ‘auto’ to let the model decide whether or not to invoke a tool. Set to ‘none’ to disable tool use. Set to ‘tool’ to force the model to invoke a tool.
List of tools the model has access to.
Parameter which forces the model to only consider the tokens with the top_k highest probabilities at the next step. Defaults to 1024.
Parameter used to do nucleus sampling, i.e. only consider tokens comprising the top_p probability of the next token’s distribution. Defaults to 0.95.
Whether to consider using search engine to complete the request. Note that even if this is set to True, the model might decide to not use search.