FEABAS Workbench v0.3.4

What changed

• Run the usual Workbench jobs locally or on LRZ, with guided cluster setup, SSH/Globus transfer, remote monitoring, previews and export downloads.
• Choose within-section, across-section or combined local parallelism for supported stages, with CPU and RAM planning.
• Select N2V training images by usable pixel count. Watch training and validation loss alongside the original and best denoised patch, and stop after the current epoch while keeping the best model.
• Updated bundled help and CI coverage for the separate GUI and FEABAS environments.

Install

Windows without Python: download FEABAS-Workbench-v0.3.4-windows-x64.zip, extract the entire folder, then start FEABAS-Workbench.exe.

Python: pip install --upgrade feabas-workbench==0.3.4, then run feabas-workbench or python -m feabas_workbench.

FEABAS and deep-learning environments are installed separately from the Setup page in either installation.

Validation and limitations

The Windows/Linux test matrix, full synthetic FEABAS pipeline and frozen Windows application checks passed before merging. A fresh installation of the release wheel was also checked locally. Full FEABAS image processing at LRZ still needs live validation; the tiny SSH/Slurm/file-transfer test passed. Live N2V training previews and graceful stopping currently apply to local execution. RAM values are planning estimates rather than enforced limits.

See the changelog and user guide.

FEABAS Workbench v0.3.3

See CHANGELOG.md for what changed. Windows without Python: unpack FEABAS-Workbench-*-windows-x64.zip and start FEABAS-Workbench.exe. With Python: pip install feabas-workbench (or the .whl below), then run feabas-workbench or python -m feabas_workbench. FEABAS and the deep-learning environments are installed from the Setup page in both cases.

Full Changelog: v0.3.2...v0.3.3

FEABAS Workbench v0.3.2

See CHANGELOG.md for what changed. Windows without Python: unpack FEABAS-Workbench-*-windows-x64.zip and start FEABAS-Workbench.exe. With Python: pip install feabas-workbench (or the .whl below), then run feabas-workbench or python -m feabas_workbench. FEABAS and the deep-learning environments are installed from the Setup page in both cases.

Full Changelog: v0.3.1...v0.3.2

FEABAS Workbench v0.3.1

See CHANGELOG.md for what changed. Windows without Python: unpack FEABAS-Workbench-*-windows-x64.zip and start FEABAS-Workbench.exe. With Python: pip install <the .whl below>, then run feabas-workbench or python -m feabas_workbench. FEABAS and the deep-learning environments are installed from the Setup page in both cases.

Full Changelog: v0.3.0...v0.3.1

FEABAS Workbench v0.3.0

See CHANGELOG.md for what changed. Windows without Python: unpack FEABAS-Workbench-*-windows-x64.zip and start FEABAS-Workbench.exe. With Python: pip install <the .whl below>, then run feabas-workbench or python -m feabas_workbench. FEABAS and the deep-learning environments are installed from the Setup page in both cases.

What's Changed

• Bump actions/checkout from 5 to 7 by @dependabot[bot] in #4
• Bump actions/setup-python from 6 to 7 by @dependabot[bot] in #5
• Bump actions/upload-artifact from 6 to 7 by @dependabot[bot] in #6

New Contributors

• @dependabot[bot] made their first contribution in #4

Full Changelog: v0.2.0...v0.3.0

FEABAS Workbench v0.2.0

See CHANGELOG.md for what changed. Install with: pip install <the .whl below> (then run feabas-workbench or python -m feabas_workbench).

What's Changed

• Declutter GUI and fix install.bat quotation errors by @georgkislinger with @Copilot in #1
• Fix swallowed training progress for N2V/N2V2/StructN2V (tqdm stdout corruption) by @georgkislinger with @Copilot in #2
• Port start_gui.bat environment search to start_gui.sh by @georgkislinger with @Copilot in #3

New Contributors

• @georgkislinger with @Copilot made their first contribution in #1

Full Changelog: https://github.com/georgkislinger/feabas-workbench/commits/v0.2.0