NVIDIA launches new NIM to simplify AI integration & scaling

NVIDIA has introduced its suite of NVIDIA Inference Microservices (NIM), aimed at streamlining the integration of AI capabilities into applications and games. These microservices seek to simplify the deployment of foundation models on clouds or data centres, thereby facilitating the rapid evolution and scalability required for contemporary AI advances.

Microservices have become essential in addressing the growing need for efficient deployment and scalability of generative AI. NVIDIA's NIM tools are designed to support fast-paced AI integration, leveraging industry-standard APIs and runtime optimisations that allow developers to concentrate on application development rather than complex AI setups. The microservices aim to minimise developers' time on AI integration, enabling them to embed AI capabilities more seamlessly within their projects.

According to the AI Decoded blog post, "Developers can focus on building their applications without worrying about the complexities of data preparation, model training, or customisation. NIM inference microservices are optimised for performance, come with runtime optimisations, and support industry-standard APIs."

Set against the backdrop of an increasing reliance on AI in various sectors, NVIDIA's NIM tools also enhance developers' access to powerful models. For instance, NVIDIA RTX AI workstations and GeForce RTX systems now provide secure access to models like the Meta Llama 3 8B. This model is now available as a NIM, enabling developers to run advanced language models locally. This local execution facilitates easy testing and the creation of retrieval-augmented generation (RAG) projects without the dependency on cloud resources. Running the entire RAG pipeline locally allows for complete data control, ensuring privacy and security.

"NVIDIA RTX AI workstations and GeForce RTX systems provide secure access to a variety of models. For example, the Meta Llama 3 8B model, now available as a NIM, lets developers run advanced language models locally. This allows for easy testing and creating retrieval-augmented generation (RAG) projects without needing cloud resources. By running the entire RAG pipeline locally, developers can maintain complete control over their data, ensuring privacy and security," the blog notes.

NVIDIA's NIM also opens up innovative applications in digital avatars and non-playable characters (NPCs). With ACE NIM on RTX PCs and workstations, developers can create lifelike digital humans, AI NPCs, and interactive avatars, expanding the creative potential of AI in gaming and other interactive platforms.

The blog encourages developers to explore these new tools: "Developers can bring digital humans, AI NPCs, and interactive avatars to life with ACE NIM on RTX PCs and workstations."

As the industry continues to explore how to incorporate AI into emerging technologies, NVIDIA's NIM aims to serve as a critical foundation for developers looking to integrate AI effortlessly into their workflows. The blog post suggests that those interested in a hands-on experience with NVIDIA NIM microservices can visit ai.nvidia.com for demonstrations and further engagement with these innovative tools.