extra.yaml for task1b.py stoped with error "AttributeError: 'FeatureContainer' object has no attribute 'sequence_axis'"

[takeofuture]

I try the extra.yaml for example 2 or 3.

python task1b.py -p extra.yaml

then I got

[I]   Collecting training data
[E] Uncaught exception  (logging.py:221)
Traceback (most recent call last):
  File "task1b.py", line 692, in <module>
    sys.exit(main(sys.argv))
  File "task1b.py", line 233, in main
    overwrite=overwrite
  File "/home/sysadmin/AILabs/dcase2020_task1_baseline/task1a.py", line 754, in do_learning
    print_indent=4
  File "/home/sysadmin/anaconda3/envs/dcase2020_tf1/lib/python3.7/site-packages/dcase_util/keras/data.py", line 507, in data_collector
    for i in range(0, data.shape[data.sequence_axis]):
AttributeError: 'FeatureContainer' object has no attribute 'sequence_axi

But it stopped with the following error.
What would be the cause?

Also how does extra.yaml work?
(Does it work on the top on task1a.yaml or task1b.yaml?)

ex.2

active_set: baseline-minified
sets:
  - set_id: baseline-minified
    description: Minified DCASE2020 baseline subtask B minified
    learner_method_parameters:
      cnn:
        model:
          constants:
            CONVOLUTION_KERNEL_SIZE: 3            
    
          config:
            - class_name: Conv2D
              config:
                input_shape:
                  - FEATURE_VECTOR_LENGTH   # data_axis
                  - INPUT_SEQUENCE_LENGTH   # time_axis
                  - 1                       # sequence_axis
                filters: 8
                kernel_size: CONVOLUTION_KERNEL_SIZE
                padding: CONVOLUTION_BORDER_MODE
                kernel_initializer: CONVOLUTION_INIT
                data_format: DATA_FORMAT
            - class_name: Activation
              config:
                activation: CONVOLUTION_ACTIVATION
            - class_name: MaxPooling2D
              config:
                pool_size:
                  - 5
                  - 5
                data_format: DATA_FORMAT
            - class_name: Conv2D
              config:
                filters: 16
                kernel_size: CONVOLUTION_KERNEL_SIZE
                padding: CONVOLUTION_BORDER_MODE
                kernel_initializer: CONVOLUTION_INIT
                data_format: DATA_FORMAT
            - class_name: Activation
              config:
                activation: CONVOLUTION_ACTIVATION
            - class_name: MaxPooling2D
              config:
                pool_size:
                  - 4
                  - 100
                data_format: DATA_FORMAT
            - class_name: Flatten      
            - class_name: Dense
              config:
                units: 100
                kernel_initializer: uniform
                activation: relu    
            - class_name: Dense
              config:
                units: CLASS_COUNT
                kernel_initializer: uniform
                activation: softmax                        
        fit:
            epochs: 100

ex.3

active_set: baseline-kernel3
sets:
  - set_id: baseline-kernel3
    description: DCASE2020 baseline for subtask B with kernel 3
    learner_method_parameters:
      cnn:
        model:
          constants:
            CONVOLUTION_KERNEL_SIZE: 3
        fit:
          epochs: 100                    
  - set_id: baseline-kernel5
    description: DCASE2020 baseline for subtask B with kernel 5
    learner_method_parameters:
      cnn:
        model:
          constants:
            CONVOLUTION_KERNEL_SIZE: 5
        fit:
          epochs: 100

[hannie0615]

I had the same problem.
I modified the code like this. And then it runs.

active_set: baseline-minified
sets:
  - set_id: baseline-minified
    description: Minified DCASE2021 baseline subtask A minified
    #add from here
    flow:
      feature_normalization: true
    dataset:
      method: baseline_development
    feature_extractor:
      method: mel
    data_processing_chain:
      method: sequencing_chain
    learner:
      method: cnn

  - set_id: baseline-minified
    description: baseline-minified/ Evaluation setup
    path:
      application:
        feature_extractor: eval_features
        recognizer: eval_recognizer

    dataset:
      method: baseline_evaluation

    feature_extractor:
      method: mel

    data_processing_chain:
      method: sequencing_chain

    learner:
      method: cnn

    #add until here

    learner_method_parameters:
      cnn:
        model:
          constants:
            CONVOLUTION_KERNEL_SIZE: 3
    
          config:
            - class_name: Conv2D
              config:
                input_shape:
                  - FEATURE_VECTOR_LENGTH   # data_axis
                  - INPUT_SEQUENCE_LENGTH   # time_axis
                  - 1                       # sequence_axis
                filters: 8
                kernel_size: CONVOLUTION_KERNEL_SIZE
                padding: CONVOLUTION_BORDER_MODE
                kernel_initializer: CONVOLUTION_INIT
                data_format: DATA_FORMAT

            - class_name: Activation
              config:
                activation: CONVOLUTION_ACTIVATION

            - class_name: MaxPooling2D
              config:
                pool_size:
                  - 5
                  - 5
                data_format: DATA_FORMAT

            - class_name: Conv2D
              config:
                filters: 16
                kernel_size: CONVOLUTION_KERNEL_SIZE
                padding: CONVOLUTION_BORDER_MODE
                kernel_initializer: CONVOLUTION_INIT
                data_format: DATA_FORMAT
            - class_name: Activation
              config:
                activation: CONVOLUTION_ACTIVATION

            - class_name: MaxPooling2D
              config:
                pool_size:
                  - 4
                  - 100
                data_format: DATA_FORMAT

            - class_name: Flatten

            - class_name: Dense
              config:
                units: 100
                kernel_initializer: uniform
                activation: relu

            - class_name: Dense
              config:
                units: CLASS_COUNT
                kernel_initializer: uniform
                activation: softmax

        fit:
            epochs: 100
            batch_size: 1