Inference self-hosted (aws, gcp) server not starting up in wsl

Community Help
1

Hi,
Whenever I try to deploy Inference on my AWS or GCP accounts using skypilot (as described in AWS - Roboflow Inference) using the command : inference cloud deploy --provider aws --compute-type gpu
The terminal returns: Command failed. Cause: name ‘sky’ is not defined

I’m running Ubuntu 24.04 on WSL. Created a venv using python-3.10.12

  • **Project Type: Segmentation
  • **Operating System & Browser: Ubuntu WSL on Windows 11
2

Hi there, is your project type Semantic Segmentation? Unfortunately this project type is not yet supported in our inference package.

3

No it’s instance segmentation. But I can’t even deploy the instance in AWS… I tried using Ubuntu and still having the same issue. The only difference it makes it prints the message “Matplotlib is building the font cache; this may take a moment.” before printing the same “Command failed. Cause: name ‘sky’ is not defined”.
I fear I’m overlooking something from the basic setup.