Nvidia will celebrate Among the next March 17 in San José (California) its development event for AI GTC 2025in which it will show how computers and workstations that make up GPUS RTX can contribute to the advance and development of artificial intelligence, in addition to improving the productivity of professionals and companies.
The event, which will take place at the San Jose Convention Center, will be attended by numerous experts from the AI ecosystem, who will share their vision of the development of AI at the local level, as well as on the optimization of models, and on the preparation of hardware and software to improve the workloads of AI. In total, in addition to an initial presentation of the CEO of Nvidia, Jensen Huang, Nvidia will offer about a thousand sessions, more than 300 exhibitions, technical training and networking events, all focused on artificial intelligence.
In them, among other things, the key progress of the PCs and workstations with GPUS RTX prepared for AI will be shown. The RTX GPUS have developed with specialized and prepared hardware for AI, known as Tensor Core.
These are responsible for providing the equipment with the computing performance they need to execute the latest AI models, and also those that need more power for execution. These GPUS, the RTX, can offer them what they need, and thus contribute to the development of AI -promoted systems, such as chatbots, generated podcasts using artificial intelligence and even digital human.
As for large language models, they can also be used in many cases of use, and climb to address complex tasks, ranging from writing code or translating between complex and minority languages. Of course, since they usually train with a broad spectrum of knowledge to use them with general use cases, they may not be the most indicated to perform very specific tasks.
NVIDIA GTC: systems optimization and data sets
They are small language models, which combine specific needs with a small size, which may be more indicated in these cases, maintaining accuracy while running in more devices locally. In the GTC 2025, Nvidia will show various techniques and tools that developers can use To generate, select and synthesize a data setand then train a small language model that can perform the tasks for which it was designed.
The optimization of AI inference and the execution of models in Windows work stations needs a strategic adjustment of its hardware and software, because there are different possible configurations for both.
In GTC 2025 there will also be a session that will expose the best practices for the Hardware and software optimization for AIwhich will include the quantization of models, the improvements that can be made for inference channels and adjustments that can be undertaken based on existing hardware. In the latter case, the possible ones will be shown for Onnx Runtime, Tensorrt de Nvidia and Llama.cpp.
The development of the local will also have its hole
The development, test and deployment of AI models in local infrastructure ensures their safety and performance even if there is no connection with cloud -based services. According to Nvidia, in these cases, HP solutions, with NVIDIA RTX GPUS, are an example of the necessary tools for the development of the local at the same time that control is maintained on the data with which you work, as well as on intellectual property.
In the event there will be a session to demonstrate your Work possibilities with the localusing NGC catalog models in NVIDIA, as well as Galileo evaluation technology, which is responsible for adjusting generative AI projects. In another talk, Nvidia experts will explore how the combination between Genai Lab and Ai Studio, from Z by HP promotes the development of large language models, optimizing their full life cycle, at home, while maintaining security and control. All while integrating the NGC catalog models of NVIDIA.
The Dell Pro Max teams are also designed to develop artificial intelligence at local intelligence, both laptops and the range desktop PCs, with RTX Nvidia Gpus. In the GTC 2025 there will be a session that shows what these teams can do for the development of the local.
On the other hand, those who wish to start in the IA development with PCs and workstations with GPUS RTX Using NVIDIA NIM Microservices you can download your test version Initial public, which includes the large language model calls 3.1, Nvidia Riva Parakeet for automatic voice and yolox recognition for computer vision.
