When i use DarkMark to create my darknet files and try to train with that prepared files, it inserts commas to some values and darknet is rejecting them. For example:
learning_rate = 0,00261000, but darknet expects a float value with dot, so i need to manually change it 0.00261.
This issue appears to affect learning_rate, chart_loss, and color augmentation parameters.
When i use DarkMark to create my darknet files and try to train with that prepared files, it inserts commas to some values and darknet is rejecting them. For example:
learning_rate = 0,00261000, but darknet expects a float value with dot, so i need to manually change it 0.00261.
This issue appears to affect learning_rate, chart_loss, and color augmentation parameters.