Get Access to NeMo Framework

NVIDIA NeMo™ framework is an end-to-end, cloud-native enterprise framework to build, customize, and deploy generative AI models with billions of parameters.

The NeMo framework provides an accelerated workflow for training with 3D parallelism techniques, a choice of several customization techniques, and optimized at-scale inference of large-scale models for language and image applications, with multi-GPU and multi-node configurations. NeMo makes generative AI model development easy, cost-effective, and fast for enterprises.

Currently, the framework for building and customizing language models is generally available and you can sign up below for immediate access. The framework supports NVIDIA and community models including Llama 2, Falcon, and Bloom, and various text-based generative AI architectures including GPT, T5, mT5, T5-MoE, and BERT.

The framework’s support for multimodal workflow is in Early Access. It currently supports the following model architectures: Stable Diffusion v1.5, VisionTransformers (ViT), CLIP, Instruct-Pix2Pix, Imagen. Once approved, you will receive a welcome email with information for access.

You must be a member of the NVIDIA Developer Program and logged in with your organization's email address.

Download Now - Language    Apply Now - Multimodal