Vanessa LLM 2.0.0

Vanessa offers a personalized, portable, and offline solution for running Large Language Models directly from your device without the need for an internet connection.

Rest assured of your privacy with Vanessa as it does not store any conversations, ensuring complete privacy and security. The model will retain chat context until the app is restarted, at which point a new conversation will begin with no memory of the previous one.

For optimal performance, it is advised to close all background applications to accommodate the app's substantial RAM requirements and prevent potential crashes.

Vanessa comes in three distinct models:

• Mistral: This 7B parameter model, while slower, is highly accurate and suitable for devices with 8GB of RAM or more.
• Rocket: The 3B parameter model strikes a balance between speed and quality, offering reliable generation on devices with 6GB of RAM or higher.
• TinyLlama: A 1.1B parameter model designed for broader device compatibility with slightly faster generation speeds but more prone to errors. It is available on devices with 4GB of RAM or more.

If encountering issues, consider restarting the application. Note that the default model, TinyLlama, may exhibit errors and hallucinations; therefore, careful attention to details is crucial.

Please be informed that Vanessa is not compatible with iOS 16 or earlier versions. The app is not supported on specific iPhone and iPad models listed below.

• iPhone: iPhone 4, 5, 6, 7, 8, X, SE, SE2
• iPad: iPad Gen 1-9, iPad Mini Gen 1-5, iPad Air Gen 1-3, iPad Pro 9-inch

The developer, Noah Kim, disclaims any liabilities for the outputs generated by the Language Models. Review important information prior to usage.