What should be the AWS Machine configuration for Whisper Large model deployment

[monk1337]

Thank you, @sanchit-gandhi, for your fantastic work. I would appreciate your opinion on configuring an AWS machine for deploying Hugging Face's Whisper large model (JAX version) and data storage for both audio and streamed textual data.

My end goal is to deploy the stream output model, but for now, I am setting up the current model without steam functionality. What would be the optimal AWS configuration to consider the future scope of the project?

1. If I decide to use your Whisper version, what would be the best configuration for large, taking into account the future streaming component?
2. If I choose to use other implementations with streaming support, what would be the optimal configuration for large?