tensorchord/envd

feat: create an pre-defined dev env in one command

kemingy opened this issue · 0 comments

Describe the feature

When I'm working on the demo of deep learning models, I want to have an almost out-of-the-box environment that contains torch, transformers, etc.

This environment doesn't have to be tracked by my local envd file since I won't update it. Although I might add more apt or pip packages, it's fine for me to do this for some specific scenarios.

It's similar to brev but for local (or self-host remote) development.

Requirements:

  • some pre-defined template environments like
    • basic CPU PyTorch env
    • basic GPU PyTorch env
    • Whisper env
  • choose device type (CPU/GPU)
  • mount a local dir
  • expose a port

Why do you need this feature?

No response

Additional context

No response