How to configure Vizro-AI parameters
Warning
Users are recommended to exercise caution and to research and understand the selected Large Language Model (LLM) before using vizro_ai
.
Users should be cautious about sharing or inputting any personal or sensitive information.
Data is sent to model vendors if you connect to LLMs via their APIs. For example, if you specify model_name="gpt-3.5-turbo-0613", your data will be sent to OpenAI via their API.
Users are also recommended to review the third party API key section of the disclaimer documentation.
Supported parameters
Currently, Vizro-AI supports the following parameters:
temperature
: A parameter for tuning the randomness of the output. It is set to 0 by default. We recommend setting it to 0 for Vizro-AI usage, as it's mostly deterministic.model_name
: The name of the model to use. Please refer to the section Models currently supported by Vizro-AI for available model options.
Config and construct Vizro-AI
Supported models
Vizro-AI currently supports the following LLMs:
OpenAI models
- gpt-3.5-turbo-0613 (to be deprecated on June 13, 2024)
- gpt-4-0613
- gpt-3.5-turbo-1106 (under testing)
- gpt-4-1106-preview (under testing, not suitable for production use)
These models provide different levels of performance and
cost. In general, the gpt-3.5-turbo
collection provides the most cost-effective models,
which would be a good starting point for most users. gpt-4
models are more powerful than gpt-3
models, e.g. they allow for more tokens per request. You can refer to these models' capabilities
and pricing for more information.
We are working on supporting more models and more vendors. Please stay tuned!