Self-Deploy
Manage your model assets and deploy them to production endpoints on your own clusters.
The Self-Deploy module covers two core capabilities:
My Models: Upload and manage your own model weights. You can import models from Hugging Face or upload local model directories. Each model entry stores metadata such as parameter size, architecture, and serving configuration.
Deployments: Deploy models from the model gallery, your uploaded models, or fine-tuned models to a target cluster. Configure GPU type, replica count, and deployment architecture (Standard or P/D Disaggregated) to match your workload requirements.