AttributeError: 'NoneType' object has no attribute 'items' - CNN introduction - ep.112

[greyfox80]

I have this error after sucessfully run 1 epochs, please help, i don't know to fix it ?

Found 1500 images belonging to 2 classes.
Found 500 images belonging to 2 classes.
/usr/local/lib/python3.10/dist-packages/keras/src/layers/convolutional/base_conv.py:107: UserWarning: Do not pass an input_shape/input_dim argument to a layer. When using Sequential models, prefer using an Input(shape) object as the first layer in the model instead.
super().init(activity_regularizer=activity_regularizer, **kwargs)
Epoch 1/5
/usr/local/lib/python3.10/dist-packages/keras/src/trainers/data_adapters/py_dataset_adapter.py:121: UserWarning: Your PyDataset class should call super().__init__(**kwargs) in its constructor. **kwargs can include workers, use_multiprocessing, max_queue_size. Do not pass these arguments to fit(), as they will be ignored.
self._warn_if_super_not_called()
47/47 ━━━━━━━━━━━━━━━━━━━━ 9s 127ms/step - accuracy: 0.6046 - loss: 0.6536 - val_accuracy: 0.8220 - val_loss: 0.3963
Epoch 2/5
/usr/lib/python3.10/contextlib.py:153: UserWarning: Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least steps_per_epoch * epochs batches. You may need to use the .repeat() function when building your dataset.
self.gen.throw(typ, value, traceback)

AttributeError Traceback (most recent call last)
in <cell line: 50>()
48
49 #4. Fit the model for 20 epocsh (5 less than before)
---> 50 history_1 = model_1.fit(train_data, epochs=5,
51 steps_per_epoch=len(train_data),
52 validation_data=valid_data,

1 frames
/usr/local/lib/python3.10/dist-packages/keras/src/backend/tensorflow/trainer.py in fit(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, validation_batch_size, validation_freq)
352 )
353 val_logs = {
--> 354 "val_" + name: val for name, val in val_logs.items()
355 }
356 epoch_logs.update(val_logs)

AttributeError: 'NoneType' object has no attribute 'items'

[AnilKamath27]

Hi there, greyfox80,

It seems you're encountering an issue where your CNN model runs out of data after the first epoch, which causes an error when training resumes. This is likely due to the model exhausting the available data during the first epoch, leaving none for subsequent epochs.

Based on my experience, this issue may be related to recent changes in TensorFlow, particularly in versions 2.16.0 or 2.17.0. I encountered a similar problem and found that downgrading TensorFlow to version 2.15.0 resolved it. Alternatively, you could create a separate environment with the older version of TensorFlow to isolate the issue.

Here’s a quick breakdown:

• Batch Size and Data Exhaustion: For instance, if you have 150 images and set a batch size of 32, you’ll end up with approximately 4 batches (150/32 ≈ 4). During the first epoch, the model processes all 150 images across these 4 batches. By the time you reach the second epoch, all data has already been used up, leading to the error.

To resolve this:

1. Downgrade TensorFlow: I highly recommend downgrading to TensorFlow 2.15.0, where this issue does not occur.

2. Create a Separate Environment: If you need to keep the latest TensorFlow version for other projects, consider creating a new environment with TensorFlow 2.15.0 specifically for this project.

I hope this helps, and I’m confident these steps will allow you to continue training your model without any further issues.

[WNastasia]

Hey, I ran into the same problem. I tried a few things and at the end I commented out the steps_per_epoch in the .fit function, then the model runs without a problem.