NVIDIA has introduced NeMo as a platform for creating custom AI services within organizations. NeMo modifies various types of models (LLM, VLM, video, audio, etc.) with sub-tools to assist in data preparation, model training, security testing (guardrails), running models on microservices, and more.
Recently, NVIDIA announced that three sub-services in the NeMo microservices suite have achieved GA status successfully.
NeMo Customizer aids in fine-tuning models for enhanced performance. NeMo Evaluator tests model performance using various benchmarks. NeMo Guardrails reinforce model security to prevent incorrect responses.
Achieving GA status means these services are ready for use in building AI within organizations. NVIDIA has revealed that major organizations like AT&T, Cisco, BlackRock, Nasdaq are utilizing these services for real-world tasks such as document analysis, system management, and coding.
The NeMo platform supports a variety of models, covering popular models like Meta Llama, Microsoft Phi, Google Gemma, Mistral, and NVIDIA’s own Llama Nemotron. Meta also integrates NeMo with its Meta Llamastack software suite.
TLDR: NVIDIA launches NeMo platform for custom AI services, including tools for data preparation, model training, security testing, and microservice deployment, with three microservices achieving GA status. Major organizations are utilizing NeMo for tasks like document analysis and system management.
Leave a Comment