Tom, training with prune enabled, all options set to default, I eventually hit this exception:
note, I do have the latest kerassurgeon from source, not pip package install.
Here's just the exception part. I will include the entire log separately.
train_log.txt
++++++++++++++++++++++++++
Deleting 3/24 channels from layer: conv2d_1
Traceback (most recent call last):
File "D:\Miniconda3\envs\donkey\lib\site-packages\keras\engine\base_layer.py", line 279, in assert_input_compatibility
K.is_keras_tensor(x)
File "D:\Miniconda3\envs\donkey\lib\site-packages\keras\backend\tensorflow_backend.py", line 474, in is_keras_tensor
str(type(x)) + '. ' ValueError: Unexpectedly found an instance of type
<class 'NoneType'>`. Expected a symbolic tensor instance.
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "manage.py", line 524, in
multi_train(cfg, tub, model, transfer, model_type, continuous, aug)
File "D:\d2\train.py", line 875, in multi_train
train_fn(cfg, tub, model, transfer, model_type, continuous, aug)
File "D:\d2\train.py", line 520, in train
go_train(kl, cfg, train_gen, val_gen, gen_records, model_name, steps_per_epoch, val_steps, continuous, verbose, save_best)
File "D:\d2\train.py", line 607, in go_train
model, channels_deleted = prune(kl.model, prune_gen, 1, cfg)
File "D:\d2\train.py", line 885, in prune
model = prune_model(model, apoz_df, n_channels_delete)
File "D:\d2\train.py", line 911, in prune_model
return surgeon.operate()
File "c:\cygwin\home\tkramer\projects\keras-surgeon\src\kerassurgeon\surgeon.py", line 158, in operate
self._mod_func_map[node](node, outputs, output_masks, **kwargs)
File "c:\cygwin\home\tkramer\projects\keras-surgeon\src\kerassurgeon\surgeon.py", line 323, in _delete_channels
temp_layer(utils.single_element(inputs))
File "D:\Miniconda3\envs\donkey\lib\site-packages\keras\engine\base_layer.py", line 414, in call
self.assert_input_compatibility(inputs)
File "D:\Miniconda3\envs\donkey\lib\site-packages\keras\engine\base_layer.py", line 285, in assert_input_compatibility
str(inputs) + '. All inputs to the layer '
ValueError: Layer conv2d_3 was called with an input that isn't a symbolic tensor. Received type: <class 'NoneType'>. Full input: [None]. All inputs to the layer should be tensors.