Today, a new version of Nvidia ChatRTX may be downloaded. The tech demo’s supporting LLM models for RTX-enabled AI applications are now supported, and its capabilities have been expanded.
The update introduces a number of new features that were initially revealed and shown at GTC 2024 earlier in March. With the addition of Gemma, Google’s most recent LLM, and ChatGLM3, an open, bilingual LLM (in both English and Chinese), ChatRTX now supports a larger list of LLMs, giving users even more options.
Developing Reputable AI: Microsoft’s Approach to Safe and Robust Generative AI
Thanks to Contrastive Language-Image Pre-training (CLIP) from OpenAI, users of ChatRTX may now search and engage locally with their own photo data without requiring complicated metadata labeling.
With the addition of support for Whisper, an AI automatic voice recognition technology, ChatRTX users can now converse using their own data. Whisper allows ChatRTX to comprehend spoken language.
ChatRTX leverages the increased performance of TensorRT-LLM software with NVIDIA RTX to allow customers to unleash the potential of AI on their RTX-powered PCs today. Furthermore, since everything is done locally on your PC, your data is safe. Additionally, developers can investigate and enhance AI applications employing RAG technology for even more different use cases because it is available on GitHub as a free reference project.
The embargoed AI Decoded blog has further details, and there’s also a new version for the RTX Remix beta that includes DLSS 3.5 with Ray Reconstruction.