Skip to content

ORKA v1.0.0

Choose a tag to compare

View all tags
@syedofc syedofc released this 04 Jun 07:55
v1.0.0
ea20f60

Release Notes

v1.0.0

ORKA v1.0.0 is the first public release of orka-depth, a lightweight PyTorch
toolkit for probing which intermediate feature depth of a frozen backbone works
best for a downstream task.

Repository:

Release highlights:

  • reusable feature extraction helpers for common backbone layouts
  • depth probing for classification and regression-style tasks
  • CLI smoke tests for installation validation
  • cleaned examples, docs, and compatibility guidance
  • unit tests and CI-backed packaging

Included in this release:

  • FeatureExtractor for pooled or raw intermediate activations
  • find_optimal_depth for lightweight probe sweeps
  • orka-validate and orka-find command-line entry points
  • classification and pose-style synthetic examples
  • documentation for install, compatibility, API usage, and roadmap

Practical scope:

  • classification is the clearest out-of-the-box probe path
  • pose and detection are treated as regression-style probe flows in this release
  • the repository is positioned as a reusable toolkit, not a full paper
    reproduction package

Quick start:

pip install "git+https://github.com/syedofc/orka-depth.git"
orka-validate

Verified release environment:

  • Python 3.12.11
  • torch 2.9.1+cu128
  • torchvision 0.24.1+cu128

For fuller version guidance, see
https://github.com/syedofc/orka-depth/blob/main/docs/COMPATIBILITY.md

Assets 2
Loading