StyleGAN deployment on AWS premises
Budget: €18 – €36 EUR
Hello developers
We have trained a StyleGAN network to generate a certain type of images (256x256, 8 bit) and now building a web-app where the model will be queried via API to serve predictions. We used this library here:
https://github.com/NVlabs/stylegan2-ada
Now, we would like to find the most optimal way to deploy it to minimise running costs. The base model size is around 300 Mb, it must run on GPU. A few options that we have considered are to convert it for AWS Inferentia or to use an elastic inference accelerator. We are open to other suggestions. However the preferred way would be to run it head-less like on SageMaker.
The final delivery would be scripts that we can run to deploy the model on our premises.
We have trained a StyleGAN network to generate a certain type of images (256x256, 8 bit) and now building a web-app where the model will be queried via API to serve predictions. We used this library here:
https://github.com/NVlabs/stylegan2-ada
Now, we would like to find the most optimal way to deploy it to minimise running costs. The base model size is around 300 Mb, it must run on GPU. A few options that we have considered are to convert it for AWS Inferentia or to use an elastic inference accelerator. We are open to other suggestions. However the preferred way would be to run it head-less like on SageMaker.
The final delivery would be scripts that we can run to deploy the model on our premises.