Language
Reduced time and cost associated with operation by streamlining model and machine management with Stable Diffusion Backstage.
Yile partnered with Going Cloud to build a Stable Diffusion backstage that allows efficient management of various models and machines, saving time and costs associated with administrative tasks.
Established in 2018, Yile Technology is comprised of a professional operations team passionate about gaming, focused on development and marketing. With a commitment to innovation and excellence, they develop high-quality game apps such as "Bao Ni Fa" and "G-bao Online."
To meet the diverse design needs of different departments, Yile needed to deploy various training models on multiple machines. They sought a flexible and unified management solution to save time and effectively control machine usage costs.
Going Cloud customized the Stable Diffusion Backstage service to establish a backend system tailored to the client's needs, providing the following functionalities:
Model Management: Centralized management of all models, along with the flexibility to allocate each model to different machines as needed.
Diversified Content: Enabling upload of various training data and Extra Networks function for enhanced image creation capability.
Machine Scheduling: Optimize machine scheduling by managing daily power operations and automatically powering off machines during idle periods to reduce costs.
Cost Monitoring: Visualization of machine usage costs, including alerts for different departmental usage patterns, machine type, and consumption.
* A Stable Diffusion function which can enhances image quality, adds styles, optimizes tasks, supports diverse inputs.
Deploying machine learning models with AWS SageMaker under a serverless environment
The introduction of Going Cloud's customized Stable Diffusion Backstage service enabled Yile Technology to efficiently manage various Stable Diffusion models, enabling colleagues from different departments to share uploaded models in a unified backend. Additionally, through visual analysis of usage cost data, they gain insights into departmental usage patterns and machine utilization, effectively controlling overall costs, reducing unnecessary expenditures, and optimizing overall efficiency.
