Democratizing AI Model
Training for Science

Outsource large-scale model training to Diamond's managed service with intelligent partitioning, optimized scheduling, and seamless cross-platform execution.

Powerful Features

🚀

Managed Training

Diamond manages long-running large model training jobs with intelligent partitioning and scheduling, optimizing turnaround time and node hour consumption for maximum efficiency.

🌐

Distributed Execution

Manage execution seamlessly across distributed resources. Configure your software environment once and run across NSF computing centers and campus RCC with unified job and data management.

📊

Provenance Tracking

Automatically track code and data versions when training large neural networks. Integrated with Garden and Hugging Face for streamlined model publishing and on-demand inference.

How It Works

1

Configure

Set up your environment once

2

Deploy

Submit your training job

3

Monitor

Track progress in real-time

4

Publish

Deploy your trained model