Code

Dummy PR to cherry pick changes#60

Code

Closed

liebharc wants to merge 27 commits intoBreezeWhite:mainBreezeWhite/oemer:mainfrom liebharc:training2liebharc/oemer:training2Copy head branch name to clipboard

+102-30

Commits

Commits on Nov 29, 2023

• Fixed 'TypeError: Cannot convert 4.999899999999999e-07 to EagerTensor of dtype int64' in training, fixes BreezeWhite#39

  Show description for 47ab174
  liebharc

  committed
  47ab174

  Copy full SHA for 47ab174

• --format was deprecated in ruff and replaced wtih --output-format

  liebharc

  committed
  9f5a6e9

  Copy full SHA for 9f5a6e9

Commits on Nov 30, 2023

• Added a single entry point to train all models

  liebharc

  committed
  b431701

  Copy full SHA for b431701

• Added convenience wrapper for oemer

  liebharc

  committed
  bdbe6f5

  Copy full SHA for bdbe6f5

• Tried to figure out the definitions for the dense dataset and to document them in code

  Show description for a87d309
  liebharc

  committed
  a87d309

  Copy full SHA for a87d309

• Decreased queue sizes as otherwise the training process crashed with an out of memory exception after it used up about 30GB of memory

  liebharc

  committed
  8d8c64f

  Copy full SHA for 8d8c64f

• Added model outputs to git ignore

  liebharc

  committed
  4c51b52

  Copy full SHA for 4c51b52

• Added checks for dataset folders

  liebharc

  committed
  cae8dd7

  Copy full SHA for cae8dd7

• Using default training params

  liebharc

  committed
  7dbd6ae

  Copy full SHA for 7dbd6ae

• Added workarounds for removal of np.float

  liebharc

  committed
  b200d0f

  Copy full SHA for b200d0f

• Using dataset definitions

  liebharc

  committed
  d087cf1

  Copy full SHA for d087cf1

• Added type annotations

  liebharc

  committed
  2deda24

  Copy full SHA for 2deda24

Commits on Dec 1, 2023

• Added a train_all_rests even if the resulting model is right now not used in oemer

  liebharc

  committed
  6f1a678

  Copy full SHA for 6f1a678

• Closing staff lines

  liebharc

  committed
  c7513dc

  Copy full SHA for c7513dc

• Changed segnet goal

  liebharc

  committed
  f23b2e1

  Copy full SHA for f23b2e1

• Merge pull request #4 from liebharc/main

  Show description for 0e56ca0
  liebharc

  authored
  0e56ca0

  Copy full SHA for 0e56ca0

Commits on Dec 2, 2023

• unet now closes staff lines

  liebharc

  committed
  50b6f25

  Copy full SHA for 50b6f25

• Tried to guess original channels definitions

  liebharc

  committed
  ee047a3

  Copy full SHA for ee047a3

• Decreased steps to be smaller than the dataset size

  liebharc

  committed
  23306cb

  Copy full SHA for 23306cb

Commits on Dec 4, 2023

• Changed model format to something which loads more reliable between different versions of tf and keras

  liebharc

  committed
  5c01f44

  Copy full SHA for 5c01f44

Commits on Jan 5, 2024

• segnet and unet should now pick the correct model

  liebharc

  committed
  f0bf03e

  Copy full SHA for f0bf03e

Commits on Jan 8, 2024

• Changed label definitions from what appears to be used in oemer right now

  liebharc

  committed
  d3db06a

  Copy full SHA for d3db06a

Commits on Jan 10, 2024

• With this commit the resulting arch.json matches the one inside of oemer/ceckpoints/seg_net/arch.json

  liebharc

  committed
  3655f61

  Copy full SHA for 3655f61

Commits on Jan 11, 2024

• Merge branch 'training' into custom_training

  liebharc

  committed
  0e10cb7

  Copy full SHA for 0e10cb7

• Defined background as: "Everything else is 0"

  liebharc

  committed
  c39c3c4

  Copy full SHA for c39c3c4

Commits on Jan 12, 2024

• Only strengthen bar line image

  liebharc

  committed
  b22a35c

  Copy full SHA for b22a35c

Commits on Mar 21, 2024

• Merge commit 'b22a35c' into test2

  liebharc

  committed
  0b57a08

  Copy full SHA for 0b57a08