Feature Request: Add Dockerfile

[whiteyhat]

Description:
Please include a Dockerfile in the repository to simplify deployment.

Rationale:
A Docker container would streamline setup, ensure consistency across environments, and simplify deployment on cloud platforms or local machines.

Suggested Implementation:

• Create a Dockerfile with dependencies (e.g., Python, required libraries).
• Include instructions for building and running the container in the README.
• Support CUDA GPU if applicable.

Thanks!

[FishWoWater]

Thanks for your interest and suggestion on the DockerFile!

Yes, the environment setup is a bit complicated in this repo, and built docker or DockerFile can help! Later this month I will work on that :)

[whiteyhat]

Yes, the environment setup is a bit complicated in this repo, and built docker or DockerFile can help! Later this month I will work on that :)

This would be great, I may submit a PR, If I achieve it to make it work with the following proposed changes:

1. Add handlers for serverless architecture; eg: maximize compatibility with https://runpods.io, or similar-like services on which you can have cloud GPU's on demand with auto-sleep if not used.
2. Dockerfile for this.

I spent a couple of hours yesterday, but seems a complex architecture and did not achieve it yet. But definetively this would be a GREAT feature

[FishWoWater]

The integration of serverless architecture would be great. Thanks for telling me that, actually I don't have experience or know much on such serverless integration like runpod.

I am not sure about the workload to implement the feature. The initial use case I was thinking about, of this repo, is that users deploy on their local PCs (perhaps windows in majority), so maybe some parts of the codebase is a bit tricky to switch to serverless arch (e.g. multiprocess scheduler, and the weight downloading of some models).

Appreciate it a lot if you could make a PR to implement this :) Feel free to discuss with me if I can help anything. I will also read some docs from runpod to learn this first.