Error on step 7

[Azn9]

Hey ! I've got that error on step 07 on the code after "Now we can train the model!" :

# of groups: 48

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-25-9041bcd63a5f> in <module>
    166             print("{},{},{},2,0,L|{}:{},1,{},0:0:0".format(int(ai[0]), int(ai[1]), int(timestamps[i]), int(round(ai[0] + ai[2] * slider_lengths[i])), int(round(ai[1] + ai[3] * slider_lengths[i])), int(slider_length_base[i] * slider_ticks[i])));
    167 
--> 168 osu_a = generate_map();
    169 # generate_test();

<ipython-input-25-9041bcd63a5f> in generate_map()
    145     print("# of groups: {}".format(timestamps.shape[0] // note_group_size));
    146     for i in range(timestamps.shape[0] // note_group_size):
--> 147         z = generate_set(begin = i * note_group_size, start_pos = pos, length_multiplier = dist_multiplier, group_id = i, plot_map=False) * np.array([512, 384, 1, 1, 512, 384]);
    148         pos = z[-1, 0:2];
    149         o.append(z);

<ipython-input-25-9041bcd63a5f> in generate_set(begin, start_pos, group_id, length_multiplier, plot_map)
     84     c_false_batch = GAN_PARAMS["c_false_batch"];
     85 
---> 86     gmodel = generative_model(g_input_size, note_group_size * 4, loss_function_for_generative_model);
     87 
     88     for i in range(max_epoch):

<ipython-input-25-9041bcd63a5f> in generative_model(in_params, out_params, loss_func)
     30     model.compile(loss=loss_func,
     31                 optimizer=optimizer,
---> 32                 metrics=[keras.metrics.mae])
     33     return model
     34 

c:\users\axel\appdata\local\programs\python\python37\lib\site-packages\tensorflow\python\training\checkpointable\base.py in _method_wrapper(self, *args, **kwargs)
    440     self._setattr_tracking = False  # pylint: disable=protected-access
    441     try:
--> 442       method(self, *args, **kwargs)
    443     finally:
    444       self._setattr_tracking = previous_value  # pylint: disable=protected-access

c:\users\axel\appdata\local\programs\python\python37\lib\site-packages\tensorflow\python\keras\engine\training.py in compile(self, optimizer, loss, metrics, loss_weights, sample_weight_mode, weighted_metrics, target_tensors, distribute, **kwargs)
    447             else:
    448               weighted_loss = training_utils.weighted_masked_objective(loss_fn)
--> 449               output_loss = weighted_loss(y_true, y_pred, sample_weight, mask)
    450 
    451           if len(self.outputs) > 1:

c:\users\axel\appdata\local\programs\python\python37\lib\site-packages\tensorflow\python\keras\engine\training_utils.py in weighted(y_true, y_pred, weights, mask)
    645     """
    646     # score_array has ndim >= 2
--> 647     score_array = fn(y_true, y_pred)
    648     if mask is not None:
    649       mask = math_ops.cast(mask, y_pred.dtype)

<ipython-input-25-9041bcd63a5f> in loss_function_for_generative_model(y_true, y_pred)
     68 
     69     def loss_function_for_generative_model(y_true, y_pred):
---> 70         return construct_map_and_calc_loss(y_pred, extvar);
     71 
     72 #     classifier_true_set_group = special_train_data[np.random.randint(0, special_train_data.shape[0], (500,))];

<ipython-input-24-993c3cadb30e> in construct_map_and_calc_loss(var_tensor, extvar)
    163     # first make a map from the outputs of generator, then ask the classifier (discriminator) to classify it
    164     classifier_model = extvar["classifier_model"]
--> 165     out = construct_map_with_sliders(var_tensor, extvar=extvar);
    166     cm = classifier_model(out);
    167     predmean = 1 - tf.reduce_mean(cm, axis=1);

<ipython-input-24-993c3cadb30e> in construct_map_with_sliders(var_tensor, extvar)
     45         begin_offset = 0;
     46     batch_size = var_tensor.shape[0];
---> 47     note_distances_now = length_multiplier * np.tile(np.expand_dims(note_distances[begin_offset:begin_offset+half_tensor], axis=0), (batch_size, 1));
     48     note_angles_now = np.tile(np.expand_dims(note_angles[begin_offset:begin_offset+half_tensor], axis=0), (batch_size, 1));
     49 

c:\users\axel\appdata\local\programs\python\python37\lib\site-packages\numpy\lib\shape_base.py in tile(A, reps)
   1241                 c = c.reshape(-1, n).repeat(nrep, 0)
   1242             n //= dim_in
-> 1243     return c.reshape(shape_out)

TypeError: __index__ returned non-int (type NoneType)

[kotritrona]

I'm glad someone still knows this after I abandoned it for 6 months!! Maybe it is related to tensorflow updates during these days, let me test it

[jvyden]

and i'm still being emailed about issues

[kotritrona]

Seems that the problem is because they changed the structure of tf.losses.* to use a class instead of plain function, since tensorflow v1.13 probably

[Azn9]

Yep I've just tested with an old version of tensorflow (1.10.0) and it's working !

(pip install tensorflow==1.10.0)

[Azn9]

Thanks for your help, hoping this will help people with the same issue !

[kotritrona]

Glad I could help.

[VINXIS]

any chance on updating this? older versions of tensorflow arent available on python 3.7 and it would be nice to be able to use this with the latest vers : 0

[Azn9]

Install python 3.6 ? ¯_(ツ)_/¯

[kotritrona]

I'm trying to update this right now. New tensorflow in fact changed the algorithm a lot, it seems to convert all functions to a computational graph instead of calling directly, which caused problems to this

Currently I got the code to work in TF 2.0, but it has memory leak problems, where after a few groups it will go out of memory

[kotritrona]

try this https://github.com/kotritrona/osumapper/tree/master/v6.2

[VINXIS]

it works!!!!11

[kotritrona]

Since it works I have closed the issue a year and a half later

[jvyden]

and i'm STILL getting emailed about issues