Training error on demo data

[manojneuro]

Hi I get the following error on the Keras version while running training on the demo data. Please let me know if the code needs to be me modified or if I need to go to a lower version of keras.

Traceback (most recent call last):
  File "/mnt/home/mkumar/miniconda3/envs/sleep/bin/ut", line 8, in <module>
    sys.exit(entry_func())
             ^^^^^^^^^^^^
  File "/mnt/home/mkumar/miniconda3/envs/sleep/lib/python3.11/site-packages/utime/bin/ut.py", line 103, in entry_func
    mod.entry_func(script_args + help_agrs)
  File "/mnt/home/mkumar/miniconda3/envs/sleep/lib/python3.11/site-packages/utime/bin/train.py", line 305, in entry_func
    run(args=args)
  File "/mnt/home/mkumar/miniconda3/envs/sleep/lib/python3.11/site-packages/utime/bin/train.py", line 272, in run
    trainer.compile_model(n_classes=hparams["build"].get("n_classes"),
  File "/mnt/home/mkumar/miniconda3/envs/sleep/lib/python3.11/site-packages/utime/train/trainer.py", line 59, in compile_model
    optimizer = init_optimizer(optimizer, **optimizer_kwargs)
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/home/mkumar/miniconda3/envs/sleep/lib/python3.11/site-packages/utime/train/utils.py", line 176, in init_optimizer
    return optimizer[0](**kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/home/mkumar/miniconda3/envs/sleep/lib/python3.11/site-packages/keras/optimizers/adam.py", line 104, in __init__
    super().__init__(
  File "/mnt/home/mkumar/miniconda3/envs/sleep/lib/python3.11/site-packages/keras/optimizers/optimizer.py", line 1087, in __init__
    super().__init__(
  File "/mnt/home/mkumar/miniconda3/envs/sleep/lib/python3.11/site-packages/keras/optimizers/optimizer.py", line 105, in __init__
    self._process_kwargs(kwargs)
  File "/mnt/home/mkumar/miniconda3/envs/sleep/lib/python3.11/site-packages/keras/optimizers/optimizer.py", line 134, in _process_kwargs
    raise ValueError(
ValueError: decay is deprecated in the new Keras optimizer, pleasecheck the docstring for valid arguments, or use the legacy optimizer, e.g., tf.keras.optimizers.legacy.Adam.

[manojneuro]

I also tried to reinstall the lates version of the package (version 1.1.8) and run the same code. Now I get a numpy error:

ut train --num_gpus=0 --preprocessed --overwri
te --seed 123

2023/08/03 13:20:03 | INFO | Entry script args dump: {'script': 'train', 'project_dir': './', 'log_dir': 'logs', 'log_level': 'INFO', 'seed': 123}

2023/08/03 13:20:03 | INFO | Project directory set: /mnt/home/mkumar/code/demo (initialized project: True)

2023/08/03 13:20:09 | INFO | Seeding TensorFlow, numpy and random modules with seed: 123
/mnt/home/mkumar/miniconda3/envs/u-sleep/lib/python3.11/site-packages/tensorflow_addons/utils/tfa_eol_msg.py:23: UserWarning: 

TensorFlow Addons (TFA) has ended development and introduction of new features.
TFA has entered a minimal maintenance and release mode until a planned end of life in May 2024.
Please modify downstream libraries to take dependencies from other repositories in our TensorFlow community (e.g. Keras, Keras-CV, and Keras-NLP). 

For more information see: https://github.com/tensorflow/addons/issues/2807 

  warnings.warn(

2023/08/03 13:20:12 | INFO | Args dump: {'num_gpus': 0, 'force_gpus': '', 'continue_training': False, 'initialize_from': None, 'overwrite': True, 'log_file': 'training_log', 'datasets': None, 'just': None, 'no_val': False, 'max_val_studies_per_dataset': 20, 'max_train_samples_per_epoch': 500000.0, 'n_epochs': None, 'channels': None, 'train_queue_type': 'eager', 'val_queue_type': 'lazy', 'max_loaded_per_dataset': 40, 'num_access_before_reload': 32, 'preprocessed': True, 'final_weights_file_name': 'model_weights.h5', 'train_on_val': False}
Parent directory of the 'utime' package is not a git repository or Git is not installed. Git information will not be added to this hyperparameter file.

2023/08/03 13:20:12 | INFO | SingleH5Dataset(identifier=processed_data, path=/mnt/home/mkumar/code/demo/data/processed_data.h5)

2023/08/03 13:20:12 | INFO | [Dataset: sedf_sc/TRAIN] H5Dataset(identifier=sedf_sc/TRAIN, members=2, loaded=2)

2023/08/03 13:20:12 | INFO | [Dataset: sedf_sc/VAL] H5Dataset(identifier=sedf_sc/VAL, members=2, loaded=2)

2023/08/03 13:20:12 | INFO | [Dataset: sedf_sc/TRAIN] Setting access-time random channel selector: RandomChannelSelector(group_0: ['EEG_FPZ-CZ', 'EEG_PZ-OZ'], group_1: ['EOG_HORIZONTAL-None'])

2023/08/03 13:20:12 | INFO | [Dataset: sedf_sc/VAL] Setting access-time random channel selector: RandomChannelSelector(group_0: ['EEG_FPZ-CZ', 'EEG_PZ-OZ'], group_1: ['EOG_HORIZONTAL-None'])

2023/08/03 13:20:12 | INFO | [Dataset: dcsm/TRAIN] H5Dataset(identifier=dcsm/TRAIN, members=3, loaded=3)

2023/08/03 13:20:12 | INFO | [Dataset: dcsm/VAL] H5Dataset(identifier=dcsm/VAL, members=1, loaded=1)

2023/08/03 13:20:12 | INFO | [Dataset: dcsm/TRAIN] Setting access-time random channel selector: RandomChannelSelector(group_0: ['F3-M2', 'F4-M1', 'C3-M2', 'C4-M1', 'O1-M2', 'O2-M1'], group_1: ['E1-M2', 'E2-M2'])

2023/08/03 13:20:12 | INFO | [Dataset: dcsm/VAL] Setting access-time random channel selector: RandomChannelSelector(group_0: ['F3-M2', 'F4-M1', 'C3-M2', 'C4-M1', 'O1-M2', 'O2-M1'], group_1: ['E1-M2', 'E2-M2'])

2023/08/03 13:20:12 | INFO | Using data queue type: EagerQueue

2023/08/03 13:20:12 | INFO | Using data queue type: EagerQueue

2023/08/03 13:20:12 | INFO | Inferred DPE: 3840, n_channels=2 for dataset queue '<psg_utils.dataset.queue.eager_queue.EagerQueue object at 0x15541d317d50>'

2023/08/03 13:20:12 | INFO | Creating sequence class '<class 'utime.sequences.balanced_random_batch_sequence.BalancedRandomBatchSequence'>'

2023/08/03 13:20:12 | INFO | Setting augmenter: RegionalErase(ordereddict([('min_region_fraction', 0.001), ('max_region_fraction', 0.33), ('log_sample', True), ('apply_prob', 0.1)]))

2023/08/03 13:20:12 | INFO | Setting augmenter: ChannelDropout(ordereddict([('drop_fraction', 0.5), ('apply_prob', 0.1)]))

2023/08/03 13:20:12 | INFO | 
[*] BalancedRandomBatchSequence initialized (sedf_sc/TRAIN):
    Data queue type: <class 'psg_utils.dataset.queue.eager_queue.EagerQueue'>
    Batch shape:     [64, 35, 3840, 2]
    Sample prob.:    [0.2, 0.2, 0.2, 0.2, 0.2]
    N pairs:         2
    Margin:          17
    Augmenters:      [<RegionalErase>, <ChannelDropout>]
    Aug enabled:     True
    Batch scaling:   False
    All loaded:      True
    N classes:       5

2023/08/03 13:20:12 | INFO | Inferred DPE: 3840, n_channels=2 for dataset queue '<psg_utils.dataset.queue.eager_queue.EagerQueue object at 0x15541d31b050>'

2023/08/03 13:20:12 | INFO | Creating sequence class '<class 'utime.sequences.balanced_random_batch_sequence.BalancedRandomBatchSequence'>'

2023/08/03 13:20:12 | INFO | Setting augmenter: RegionalErase(ordereddict([('min_region_fraction', 0.001), ('max_region_fraction', 0.33), ('log_sample', True), ('apply_prob', 0.1)]))

2023/08/03 13:20:12 | INFO | Setting augmenter: ChannelDropout(ordereddict([('drop_fraction', 0.5), ('apply_prob', 0.1)]))

2023/08/03 13:20:12 | INFO | 
[*] BalancedRandomBatchSequence initialized (dcsm/TRAIN):
    Data queue type: <class 'psg_utils.dataset.queue.eager_queue.EagerQueue'>
    Batch shape:     [64, 35, 3840, 2]
    Sample prob.:    [0.2, 0.2, 0.2, 0.2, 0.2]
    N pairs:         3
    Margin:          17
    Augmenters:      [<RegionalErase>, <ChannelDropout>]
    Aug enabled:     True
    Batch scaling:   False
    All loaded:      True
    N classes:       5

2023/08/03 13:20:12 | INFO | Inferred DPE: 3840, n_channels=2 for dataset queue '<psg_utils.dataset.queue.eager_queue.EagerQueue object at 0x15541d319bd0>'

2023/08/03 13:20:12 | INFO | Creating sequence class '<class 'utime.sequences.balanced_random_batch_sequence.BalancedRandomBatchSequence'>'

2023/08/03 13:20:12 | INFO | 
[*] BalancedRandomBatchSequence initialized (sedf_sc/VAL):
    Data queue type: <class 'psg_utils.dataset.queue.eager_queue.EagerQueue'>
    Batch shape:     [64, 35, 3840, 2]
    Sample prob.:    [0.2, 0.2, 0.2, 0.2, 0.2]
    N pairs:         2
    Margin:          17
    Augmenters:      []
    Aug enabled:     False
    Batch scaling:   False
    All loaded:      True
    N classes:       5

2023/08/03 13:20:12 | INFO | Inferred DPE: 3840, n_channels=2 for dataset queue '<psg_utils.dataset.queue.eager_queue.EagerQueue object at 0x15541d31af90>'

2023/08/03 13:20:12 | INFO | Creating sequence class '<class 'utime.sequences.balanced_random_batch_sequence.BalancedRandomBatchSequence'>'

2023/08/03 13:20:12 | INFO | 
[*] BalancedRandomBatchSequence initialized (dcsm/VAL):
    Data queue type: <class 'psg_utils.dataset.queue.eager_queue.EagerQueue'>
    Batch shape:     [64, 35, 3840, 2]
    Sample prob.:    [0.2, 0.2, 0.2, 0.2, 0.2]
    N pairs:         1
    Margin:          17
    Augmenters:      []
    Aug enabled:     False
    Batch scaling:   False
    All loaded:      True
    N classes:       5

2023/08/03 13:20:12 | INFO | 
[*] MultiSequence initialized:
    --- Contains 2 sequences
    --- Sequence IDs: sedf_sc/TRAIN, dcsm/TRAIN
    --- Sequence sample probs (alpha=0.5): [0.45 0.55]
    --- Batch shape: [64, 35, 3840, 2]

2023/08/03 13:20:12 | INFO | 
[*] ValidationMultiSequence initialized:
    --- Contains 2 sequences
    --- Sequence IDs: sedf_sc, dcsm

2023/08/03 13:20:12 | INFO | Setting CUDA_VISIBLE_DEVICES = ''
2023-08-03 13:20:13.170026: E tensorflow/compiler/xla/stream_executor/cuda/cuda_driver.cc:268] failed call to cuInit: CUDA_ERROR_NO_DEVICE: no CUDA-capable device is detected

2023/08/03 13:20:13 | INFO | Using TF distribution strategy: <tensorflow.python.distribute.one_device_strategy.OneDeviceStrategy object at 0x15541d948d10> on GPUs: []. (CPU:0 if empty).

2023/08/03 13:20:13 | INFO | Creating new model of type 'USleep'

2023/08/03 13:20:13 | INFO | Found requested class 'elu' in module '<module 'keras.api._v2.keras.activations' from '/mnt/home/mkumar/miniconda3/envs/u-sleep/lib/python3.11/site-packages/keras/api/_v2/keras/activations/__init__.py'>'
Traceback (most recent call last):
  File "/mnt/home/mkumar/miniconda3/envs/u-sleep/lib/python3.11/site-packages/utime/utils/conv_arithmetics.py", line 85, in compute_receptive_fields
    dilation = np.array(layer.dilation_rate).astype(np.int)
                                                    ^^^^^^
  File "/mnt/home/mkumar/miniconda3/envs/u-sleep/lib/python3.11/site-packages/numpy/__init__.py", line 305, in __getattr__
    raise AttributeError(__former_attrs__[attr])
AttributeError: module 'numpy' has no attribute 'int'.
`np.int` was a deprecated alias for the builtin `int`. To avoid this error in existing code, use `int` by itself. Doing this will not modify any behavior and is safe. When replacing `np.int`, you may wish to use e.g. `np.int64` or `np.int32` to specify the precision. If you wish to review your current use, check the release note link for additional information.
The aliases was originally deprecated in NumPy 1.20; for more details and guidance see the original release note at:
    https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations. Did you mean: 'inf'?

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/mnt/home/mkumar/miniconda3/envs/u-sleep/bin/ut", line 8, in <module>
    sys.exit(entry_func())
             ^^^^^^^^^^^^
  File "/mnt/home/mkumar/miniconda3/envs/u-sleep/lib/python3.11/site-packages/utime/bin/ut.py", line 103, in entry_func
    mod.entry_func(script_args + help_agrs)
  File "/mnt/home/mkumar/miniconda3/envs/u-sleep/lib/python3.11/site-packages/utime/bin/train.py", line 305, in entry_func
    run(args=args)
  File "/mnt/home/mkumar/miniconda3/envs/u-sleep/lib/python3.11/site-packages/utime/bin/train.py", line 266, in run
    model = init_model(hparams["build"], clear_previous=False)
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/home/mkumar/miniconda3/envs/u-sleep/lib/python3.11/site-packages/utime/models/model_init.py", line 37, in init_model
    return models.__dict__[cls_name](**build_hparams)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/home/mkumar/miniconda3/envs/u-sleep/lib/python3.11/site-packages/utime/models/usleep.py", line 196, in __init__
    self.receptive_field = compute_receptive_fields(self.layers[:ind])[-1][-1]
                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/home/mkumar/miniconda3/envs/u-sleep/lib/python3.11/site-packages/utime/utils/conv_arithmetics.py", line 87, in compute_receptive_fields
    dilation = np.ones(shape=[dim], dtype=np.int)
                                          ^^^^^^
  File "/mnt/home/mkumar/miniconda3/envs/u-sleep/lib/python3.11/site-packages/numpy/__init__.py", line 305, in __getattr__
    raise AttributeError(__former_attrs__[attr])
AttributeError: module 'numpy' has no attribute 'int'.
`np.int` was a deprecated alias for the builtin `int`. To avoid this error in existing code, use `int` by itself. Doing this will not modify any behavior and is safe. When replacing `np.int`, you may wish to use e.g. `np.int64` or `np.int32` to specify the precision. If you wish to review your current use, check the release note link for additional information.
The aliases was originally deprecated in NumPy 1.20; for more details and guidance see the original release note at:
    https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations. Did you mean: 'inf'?

[saraheldah]

hi @manojneuro, did you resolve numpy error? because I'm facing it as well

[happybegining]

Hi @manojneuro,
I had the same problem as you. When I downgraded the version of tensorflow to 2.9.3, it worked. I hope it helps you.

Hi I get the following error on the Keras version while running training on the demo data. Please let me know if the code needs to be me modified or if I need to go to a lower version of keras.

Traceback (most recent call last):
  File "/mnt/home/mkumar/miniconda3/envs/sleep/bin/ut", line 8, in <module>
    sys.exit(entry_func())
             ^^^^^^^^^^^^
  File "/mnt/home/mkumar/miniconda3/envs/sleep/lib/python3.11/site-packages/utime/bin/ut.py", line 103, in entry_func
    mod.entry_func(script_args + help_agrs)
  File "/mnt/home/mkumar/miniconda3/envs/sleep/lib/python3.11/site-packages/utime/bin/train.py", line 305, in entry_func
    run(args=args)
  File "/mnt/home/mkumar/miniconda3/envs/sleep/lib/python3.11/site-packages/utime/bin/train.py", line 272, in run
    trainer.compile_model(n_classes=hparams["build"].get("n_classes"),
  File "/mnt/home/mkumar/miniconda3/envs/sleep/lib/python3.11/site-packages/utime/train/trainer.py", line 59, in compile_model
    optimizer = init_optimizer(optimizer, **optimizer_kwargs)
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/home/mkumar/miniconda3/envs/sleep/lib/python3.11/site-packages/utime/train/utils.py", line 176, in init_optimizer
    return optimizer[0](**kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^
  File "/mnt/home/mkumar/miniconda3/envs/sleep/lib/python3.11/site-packages/keras/optimizers/adam.py", line 104, in __init__
    super().__init__(
  File "/mnt/home/mkumar/miniconda3/envs/sleep/lib/python3.11/site-packages/keras/optimizers/optimizer.py", line 1087, in __init__
    super().__init__(
  File "/mnt/home/mkumar/miniconda3/envs/sleep/lib/python3.11/site-packages/keras/optimizers/optimizer.py", line 105, in __init__
    self._process_kwargs(kwargs)
  File "/mnt/home/mkumar/miniconda3/envs/sleep/lib/python3.11/site-packages/keras/optimizers/optimizer.py", line 134, in _process_kwargs
    raise ValueError(
ValueError: decay is deprecated in the new Keras optimizer, pleasecheck the docstring for valid arguments, or use the legacy optimizer, e.g., tf.keras.optimizers.legacy.Adam.