Issue with training model

[shikha-16]

Hi, I was trying to replicate the training of the MLSP model using the notebooks provided. I ran the notebook train_mlsp_narrow_aug.ipynb, but while training on the narrow MLSP features, with crop-flip augmentation, I got the following error in "helper":

Model NOT loaded: irnv2_mlsp_narrow_orig_aug/bn:2 bsz:256 do:[0.25,0.25,0.5,0] ds:[AVA_data_official_test.csv] fc1:[2048] i:1[16928] im:[orig] l:MSE mon:[val_loss] o:1[1]_best_weights.h5 does not exist
Training model: irnv2_mlsp_narrow_orig_aug/bn:2 bsz:256 do:[0.25,0.25,0.5,0] ds:[AVA_data_official_test.csv] fc1:[2048] i:1[16928] im:[orig] l:MSE mon:[val_loss] o:1[1]
---------------------------------------------------------------------------
AssertionError                            Traceback (most recent call last)
[<ipython-input-6-33e50038bd7e>](https://localhost:8080/#) in <module>
      1 for lr in [1e-3,1e-4,1e-5]:
      2     helper.load_model()
----> 3     helper.train(lr=lr, epochs=20)

9 frames
[/usr/local/lib/python3.8/dist-packages/kuti/generators.py](https://localhost:8080/#) in _read_data(self, ids_batch, accessor)
    136         X = []
    137         if accessor:
--> 138             assert isinstance(accessor, list) or callable(accessor),\
    139             'Generator inputs/outputs must be of type list, or function'
    140 

AssertionError: Generator inputs/outputs must be of type list, or function

Could someone help us out here, how do we resolve this error? Please let me know if more information is needed. Thanks!

[retraintsuh31312]

hi, shikha-16
I am also training how you add photos for training?