This commit introduces the necessary files and documentation to enable
deployment of the Wan video generation application to Google Cloud Platform.
Key additions:
- Dockerfile: Defines the container image, using an NVIDIA PyTorch base
image and including special handling for flash-attn installation.
It allows for configurable Gradio application startup.
- cloudbuild.yaml: Provides a Google Cloud Build configuration to automate
the Docker image build and push process to Google Artifact Registry.
- DEPLOY_GCLOUD.md: A comprehensive guide detailing the steps to:
- Set up prerequisites on GCP.
- Configure and run Cloud Build.
- Deploy the container to Cloud Run (with CPU and GPU options).
- Deploy the container to Vertex AI Endpoints.
- Troubleshoot common issues and select appropriate machine resources.
These changes aim to simplify and standardize the deployment process on
Google Cloud, addressing potential issues related to dependencies and
environment configuration.
