1.2.6

change log:

• parallel generator input pipeline fix
• edit option to add static arguments (additional_graph|model|evaluator|data_kwargs)
• added howto's in examples
• minor change in data pipeline api

1.2.5

• Fixes in docs
• split of requirements and devel requirements
• Tensorflow 2.6.0 support
• Added script to test the throughput of the input pipeline
• Support for pre-batched Samples (manual dynamic batch sizes)
• Various smaller fixes

1.2.4

• Fixes in docs
• Support for generating batches instead of simple samples in DataGenerator to support custom "batches" and "batch sizes"
• Parallelized MultiPredictor

1.2.3

• Fix loading of graphs in graph mode (default no reconstruction)
• Added option to generate predefined batches in a DataGenerator

1.2.2

• Added a RawPredictor that can be kept open to rapidly predict single samples
• Final version of joss Paper

1.2.1

• Forced Black coding style
• Smaller reworks
• upgraded paiargparse

1.2.0

• Reworked Graph construction
• Generalized metric and loss definitions
• Simplified tensorboard writer

1.1.1

• Improved output of network architecture
• smaller fixed for windows users
• other smaller fixes

1.1.0

Fixed python version in action

1.0.1

Automatic merge of updates