函数调用堆栈：使用 keras 的 train_function

Question

我正在尝试在本网站https://www.kaggle.com/bsagredo/anime-neuralmf-hybrid-recommender/notebook 的互联网上找到的代码

我在使用 keras 时遇到了这段代码的问题：

def get_model(num_users, num_items, num_item_feats, mf_dim, layers = [64, 32, 16, 8]):
    user_input = Input(shape=(1,), dtype='int32', name = 'user_input')
    item_input = Input(shape=(1,), dtype='int32', name = 'item_input')
    feats_input = Input(shape=(num_item_feats,), dtype='float32', name = 'feats_input')

    # User&Item Embeddings for Matrix Factorization
    MF_Embedding_User = Embedding(input_dim = num_users + 1, output_dim = mf_dim,
                                  name = 'user_embedding',
                                  embeddings_initializer = RandomNormal(stddev=0.001),
                                  input_length = 1)
    MF_Embedding_Item = Embedding(input_dim = num_items + 1, output_dim = mf_dim,
                                  name = 'item_embedding',
                                  embeddings_initializer = RandomNormal(stddev=0.001),
                                  input_length = 1)
    
    # User&Item Embeddings for MLP part
    MLP_Embedding_User = Embedding(input_dim = num_users + 1, output_dim = int(layers[0] / 2),
                                   name = 'mlp_embedding_user',
                                   embeddings_initializer = RandomNormal(stddev=0.001),
                                   input_length = 1)
    MLP_Embedding_Item = Embedding(input_dim = num_items + 1, output_dim = int(layers[0] / 2),
                                   name = 'mlp_embedding_item',
                                   embeddings_initializer = RandomNormal(stddev=0.001),
                                   input_length = 1) 
    
    mf_user_latent = Flatten()(MF_Embedding_User(user_input))
    mf_item_latent = Flatten()(MF_Embedding_Item(item_input))
    mf_vector = Multiply()([mf_user_latent, mf_item_latent])

    # MLP part with item features
    mlp_user_latent = Flatten()(MLP_Embedding_User(user_input))
    mlp_item_latent = Flatten()(MLP_Embedding_Item(item_input))
    
    mlp_vector = Concatenate()([mlp_user_latent, mlp_item_latent, feats_input])
    for l in layers:
        layer = Dense(l, activation='relu')
        mlp_vector = layer(mlp_vector)

    # Concatenate MF and MLP parts
    predict_vector = Concatenate()([mf_vector, mlp_vector])
    
    # Final prediction layer
    prediction = Dense(1, activation = 'sigmoid',
                       kernel_initializer = 'lecun_uniform',
                       name = 'prediction')(predict_vector)
    
    model = Model([user_input, item_input, feats_input], prediction)
    return model

learning_rate = 0.001
batch_size = 256
n_epochs = 3
mf_dim = 15
layers = [128, 64, 32, 16, 8]

对于模型和训练：

model = get_model(n_users, n_anime, n_feats, mf_dim, layers)
model.compile(optimizer = Adam(lr = learning_rate), loss = 'mean_squared_logarithmic_error')

model = get_model(n_users, n_anime, n_feats, mf_dim, layers)
model.compile(optimizer = Adam(lr = learning_rate), loss = 'mean_squared_logarithmic_error')

这是按摩：

Epoch 1/3

---------------------------------------------------------------------------
InvalidArgumentError                      Traceback (most recent call last)
<ipython-input-35-9047b6ef7a15> in <module>
      1 hist = model.fit(x = x_train, y = y_train, validation_data = (x_val, y_val),
----> 2                  batch_size = batch_size, epochs = n_epochs, verbose = True, shuffle = True)

~/miniconda3/lib/python3.7/site-packages/tensorflow/python/keras/engine/training.py in _method_wrapper(self, *args, **kwargs)
    106   def _method_wrapper(self, *args, **kwargs):
    107     if not self._in_multi_worker_mode():  # pylint: disable=protected-access
--> 108       return method(self, *args, **kwargs)
    109 
    110     # Running inside `run_distribute_coordinator` already.

~/miniconda3/lib/python3.7/site-packages/tensorflow/python/keras/engine/training.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, max_queue_size, workers, use_multiprocessing)
   1096                 batch_size=batch_size):
   1097               callbacks.on_train_batch_begin(step)
-> 1098               tmp_logs = train_function(iterator)
   1099               if data_handler.should_sync:
   1100                 context.async_wait()

~/miniconda3/lib/python3.7/site-packages/tensorflow/python/eager/def_function.py in __call__(self, *args, **kwds)
    778       else:
    779         compiler = "nonXla"
--> 780         result = self._call(*args, **kwds)
    781 
    782       new_tracing_count = self._get_tracing_count()

~/miniconda3/lib/python3.7/site-packages/tensorflow/python/eager/def_function.py in _call(self, *args, **kwds)
    805       # In this case we have created variables on the first call, so we run the
    806       # defunned version which is guaranteed to never create variables.
--> 807       return self._stateless_fn(*args, **kwds)  # pylint: disable=not-callable
    808     elif self._stateful_fn is not None:
    809       # Release the lock early so that multiple threads can perform the call

~/miniconda3/lib/python3.7/site-packages/tensorflow/python/eager/function.py in __call__(self, *args, **kwargs)
   2827     with self._lock:
   2828       graph_function, args, kwargs = self._maybe_define_function(args, kwargs)
-> 2829     return graph_function._filtered_call(args, kwargs)  # pylint: disable=protected-access
   2830 
   2831   @property

~/miniconda3/lib/python3.7/site-packages/tensorflow/python/eager/function.py in _filtered_call(self, args, kwargs, cancellation_manager)
   1846                            resource_variable_ops.BaseResourceVariable))],
   1847         captured_inputs=self.captured_inputs,
-> 1848         cancellation_manager=cancellation_manager)
   1849 
   1850   def _call_flat(self, args, captured_inputs, cancellation_manager=None):

~/miniconda3/lib/python3.7/site-packages/tensorflow/python/eager/function.py in _call_flat(self, args, captured_inputs, cancellation_manager)
   1922       # No tape is watching; skip to running the function.
   1923       return self._build_call_outputs(self._inference_function.call(
-> 1924           ctx, args, cancellation_manager=cancellation_manager))
   1925     forward_backward = self._select_forward_and_backward_functions(
   1926         args,

~/miniconda3/lib/python3.7/site-packages/tensorflow/python/eager/function.py in call(self, ctx, args, cancellation_manager)
    548               inputs=args,
    549               attrs=attrs,
--> 550               ctx=ctx)
    551         else:
    552           outputs = execute.execute_with_cancellation(

~/miniconda3/lib/python3.7/site-packages/tensorflow/python/eager/execute.py in quick_execute(op_name, num_outputs, inputs, attrs, ctx, name)
     58     ctx.ensure_initialized()
     59     tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,
---> 60                                         inputs, attrs, num_outputs)
     61   except core._NotOkStatusException as e:
     62     if name is not None:

InvalidArgumentError:  indices[107,0] = 9998 is not in [0, 9937)
     [[node functional_3/user_embedding/embedding_lookup (defined at <ipython-input-32-9047b6ef7a15>:2) ]] [Op:__inference_train_function_3295]

Errors may have originated from an input operation.
Input Source operations connected to node functional_3/user_embedding/embedding_lookup:
 functional_3/user_embedding/embedding_lookup/2886 (defined at /home/xubuntu/miniconda3/lib/python3.7/contextlib.py:112)

Function call stack:
train_function

我尝试在谷歌上搜索错误，但没有找到任何解决方案。我希望任何人都可以帮助解决这个问题。

Answer 1

根据documentationverbose 接受一个整数。

0, 1, or 2. Verbosity mode. 0 = silent, 1 = progress bar, 2 = one line per epoch.
Note that the progress bar is not particularly useful when logged to a file, 
so verbose=2 is recommended when not running interactively 
(eg, in a production environment).

因此您传递verbose = True 可能是原因

InvalidArgumentError                      Traceback (most recent call last)
<ipython-input-35-9047b6ef7a15> in <module>
      1 hist = model.fit(x = x_train, y = y_train, validation_data = (x_val, y_val),
----> 2                  batch_size = batch_size, epochs = n_epochs, verbose = True, shuffle = True)

尝试更改详细的值。如需进一步说明，请参阅this documentation。