NVIDIA Cloud Functions

Cloud Functions

• Overview
  
  • Workloads
    
    • Model Only Workloads
    • Container Workloads
    • Helm Chart Workloads
• API
  
  • Authorization
  • Invocation
    
    • HTTP Polling
    • HTTP Streaming
    • gRPC
  • Statuses and Errors
• Function Management
  
  • Function Creation
    
    • Function Creation with your own Model
    • Function Creation with your Docker Container
    • Function Creation with Helmchart
  • Function Versioning
  • Function Deployment
  • Debugging
  • Function Lifecycle Endpoints
  • Available Container Variables
  • Logging & Metrics
  • Supporting Partial Response
• Asset Management
  
  • Creating an Asset Id and Pre-signed Upload URL
  • Uploading Assets to Cloud Storage
  • Listing Assets
  • Deleting Assets
  • Specifying Assets when invoking a Function
  • Using Assets with Custom Containers
  • Retrieving Asset Directory and Asset ID in Triton Python Backend
• FAQs
  
  • My function is stuck in the deploying state. What do I do?
  • How do I view the logs for my functions?
  • I’m getting errors when invoking my function. What do I do?
  • Models
  • Why am I encountering a “model file not found” error when launching an inference container?
  • What does the log entry “No MODELS environment variable defined” mean?
  • Are model files expected to be compressed, and why does the system attempt to extract a model?
  • config.json
  • What does it mean when a model is being “copied,” and does it affect my deployed models?
  • Errors
  • Errors to handle in Thin Client
  • Terminology
• Terminology