NVIDIA introduces deep learning inference software- TensorRT 7

  • The recently unveiled software would support the developers in offering conversational AI applications and slashing inference latency.
  • The groundbreaking invention would make the machines capable of comprehending the human language in real time.
NVIDIA's successful roll out of next generation software defined platform, DRIVE AGX Orin is backed by the launch of the company's deep learning interference software- TensorRT 7 which would allow the developers to deliver conversational AI application anywhere. Apparently, the software, TensorRT 7 is the seventh generation of NVIDIA's inference software development kit which establishes high growth avenues for smarter human to AI interactions. Further, it also enables real time engagement with applications like the voice agents. Recommendation engines, and chatbots. For the record, TensorRT 7 highlights a new deep learning compiler created to automatically optimize and escalate the complex recurrent and transformer based neural networks required for AI speech applications. This feature apparently speeds up the components of conversational artificial intelligence by over ten times as compared to the operating speed on CPUs. The driving latency records to be below 300 millisecond threshold which is considered to be necessary for the real-time interactions. Speaking on the matter, the NVIDIA CEO and founder, Jensen Huang reportedly cited that the company has taken a step forward in the chapter of AI which enables the machines to understand the human language in real time. He further added that the invention of TensorRT 7 would help the company realize this potential. It would provide the developers with tools to build and deploy smarter, faster, and conversational AI services promoting more human to AI interaction. Meanwhile, Yang Hongtao, the CTO of Sogou, one of the innovative companies looking forward to leveraging the novel invention by NVIDIA, quoted that the use of TensorRT 7 would allow the company to enable online service responses in real time. Moreover, the use of leading AI capabilities have substantially enhanced the company's user experience overall. TensorRT 7 would now be made available for development and deployment to the NVIDIA Developer program members in the upcoming days via the TensorRT webpage. Source credit: https://nvidianews.nvidia.com/news/nvidia-enables-era-of-interactive-conversational-ai-with-new-inference-software