-
Possible harm categories for the generation of responses that have been blocked by the model.
Safety thresholds, for the harm categories for the generation of responses that have been blocked by the model.
Tool calling mode, to instruct Gemini whether it can request calls to any functions,
to just a subset of the available functions, or to none at all.
Represents a Google Vertex AI language model with a chat completion interface, such as chat-bison.
Represents a Google Vertex AI embedding model, such as textembedding-gecko.
Image model for the Google Cloud Vertex AI Imagen image generation models.
Supported aspect ratios: 1:1, 9:16, 16:9, 4:3, and 3:4.
Image style can be specified for imagen@002
.
Supported mime types: only PNG and JPEG image formats can be generated.
Specify whether persons are allowed to be generated.
Represents a Google Vertex AI language model with a text interface, such as text-bison.
Implementation of a
ScoringModel
for the Google Cloud Vertex AI
Ranking API.