Skip to main content

Google Plugin

The Google plugin allows you to use Vertex AI to access models like Gemini.

info

You must use the node executor to use the Google plugin in the Rivet UI. You may also use the plugin when embedding Rivet in your own application.

Configuration

Before you start, you must create a service account key. The service account key is a JSON file that you download, and then set the env variable GOOGLE_APPLICATION_CREDENTIALS to the location of. See Google's "Install the Vertex AI client libraries" article.

Then, you must add your GCP Project ID, Region, and the location of your application credentials to the Rivet settings panel, once the Google plugin has been enabled. See Settings for more information.

Nodes

Chat (Google) Node

The Chat (Google) node allows you to use the Google Vertex API to generate text.

Inputs

TitleData TypeDescriptionDefault ValueNotes
Promptstring or string[] or chat-message or chat-message[]The messages to send to Google to get a response.(Required)Google models do not support a system prompt like GPT does. System prompts will be converted to Asssistant prompts.

Outputs

TitleData TypeDescriptionNotes
ResponsestringThe response message from Google

Editor Settings

SettingDescriptionDefault ValueUse Input ToggleInput Data Type
ModelThe Google model to use for the request (Gemini Pro, Gemini Pro Vision, etc)gemini-proYesstring
TemperatureThe sampling temperature to use. Lower values are more deterministic. Higher values are more "creative".0.5Yesnumber
Top PAlternate sampling mode using the top X% of values. 0.1 corresponds to the top 10%.1Yesnumber
Use Top PWhether to use the Top P sampling mode.falseYesboolean
Max TokensThe maximum number of tokens that GPT is allowed to return. When hitting the max tokens, the response will be cut off.1024Yesnumber