A new AI chatbot developed by Nvidia is available for trial on select PCs equipped with GeForce RTX graphics cards. Users can operate an AI chatbot on their PCs that is customized with their content by downloading the free demo program Chat with RTX.
The application, which is powered by Nvidia TensorRT-LLM software, can create content and train on user-selected content, including files with extensions like.txt,.pdf,.doc/.docx, and.xml, as well as URLs to YouTube videos.
Users can ask the bot customized questions regarding the content they’ve provided once they’ve selected content to train it with. For instance, the bot can inform a user of the kind of batteries they had on their shopping list or rephrase the detailed directions in a YouTube how-to video.
Although the experience is highly tailored thanks to the bot’s training on the user’s favorite content, user data is kept private because everything is done locally. Because Chat with RTX doesn’t rely on cloud-based services, it can function without an internet connection and can provide quick responses while protecting all user information.
A GPU with at least 8 GB of VRAM and an Nvidia GeForce RTX 30 Series or higher is required for the bot. The most recent Nvidia GPU drivers and Windows 10 or 11 are also needed to chat with RTX.
TensorRT-LLM software from Nvidia, in conjunction with RTX acceleration and retrieval-augmented generation (RAG), enables Chat with RTX to use local files as a dataset and link them to open-source LLMs such as Mistral, according to Nvidia.
