NVIDIA unveiled AI Foundry, a service allowing businesses to develop AI models for their needs. (Image Credit: NVIDIA)
NVIDIA recently launched AI Foundry, an AI service for companies to build and deploy custom AI models for their needs. For example, businesses may use AI foundry to customize the Llama 3.1 collection, NVIDIA Nemotron, Google Deepmind’s CodeGemma and Gemma, CodeLlama, Mistral, Mixtral, Phi-3, StarCoder2, and more. Along with ecosystem tools and support, NVIDIA AI Foundry uses foundation models, NVIDIA NeMo software, DGX Cloud, and NVIDIA expertise.
“Organizations deploying AI can gain a competitive edge with custom models that incorporate industry and business knowledge,” said Jeremy Barnes, vice president of AI Product at ServiceNow. “ServiceNow is using NVIDIA AI Foundry to fine-tune and deploy models that can integrate easily within customers’ existing workflows.”
NVIDIA AI Enterprise experts are available to customers if they need help developing, fine-tuning, and deploying models with proprietary data to meet their targeted business needs. In addition, the service provides AI foundation models and NVIDIA NeMo software to fast-track model development.
Enterprises may rely on service delivery partners, like Data Monsters, Quantiphi, Slalom, and SoftServe, for assistance with AI integration into their IT landscapes. This results in scalable and secure AI applications that match business objectives.
DGX Cloud is the brainpower of NVIDIA AI Foundry. It allows customers to build and fine-tune custom generative AI apps easily and efficiently and scale AI initiatives without costly upfront hardware investments. Customers can also access a global ecosystem of partners, like Accenture, Infosys, Deloitte, Tata Consultancy Services, and Wipro, for consulting services encompassing the design, implementation, and management of AI-powered digital transformation projects.
With AIOps and MLOps from NVIDIA partners like ActiveFence, AutoAlign, DataDog, and more, customers can create NVIDIA AI Foundry models for development. Their models can also be transformed into NVIDIA NIM inference microservices featuring a customized model, optimized engines, and standard API that run on their accelerated infrastructure.
Developers can take advantage of NVIDIA NeMo to curate data, customize foundation models, and evaluate performance. NeMo Curator, a GPU-accelerate data-curation library, boosts AI model performance via high-quality dataset preparation for pretraining and fine-tuning. The customizer simplifies LLM fine-tuning and alignment domain-specific applications, while the evaluator provides automated AI model generative assessments across academic and custom benchmarks. Lastly, NeMo guardrails orchestrate dialog for accuracy, appropriateness, and security in smart applications with LLMs.
AI Foundry is beneficial for enterprises as it helps them address AI adoption challenges. Generic AI models can be insufficient for meeting certain business needs and data security requirements. Meanwhile, custom AI models provide better flexibility, adaptability, and performance --- suitable for businesses looking for a competitive edge.
Have a story tip? Message me at: http://twitter.com/Cabe_Atwell