Amazon SageMaker AI now supports serverless model customization for NVIDIA Nemotron 3.5 Lightning

This article announces serverless model customization support for NVIDIA Nemotron 3.5 Lightning on Amazon SageMaker AI.

• Supports supervised fine-tuning (SFT), Direct Preference Optimization (DPO), and reinforcement fine-tuning (RFT)
• NVIDIA Nemotron 3.5 Lightning uses hybrid Mixture-of-Experts with 3B active and 30B total parameters
• Tailor foundation models with proprietary data to reduce cost and latency
• SageMaker handles infrastructure provisioning and training orchestration automatically
• Available in US East (N. Virginia), US West (Oregon), Asia Pacific (Tokyo), and Europe (Ireland)

Serverless customization enables cost-effective model adaptation without managing infrastructure, allowing focus on data and evaluation.