TensorFlow - cannot cast string to float error?

Question

I tried running an example from stellargraph's examples, but I encountered a weird error:

tensorflow/core/framework/op_kernel.cc:1744] OP_REQUIRES failed at cast_op.cc:121 : Unimplemented: Cast string to float is not supported

The example code I used is this:

import pandas as pd
import numpy as np

import stellargraph as sg
from stellargraph.mapper import PaddedGraphGenerator
from stellargraph.layer import GCNSupervisedGraphClassification
from stellargraph import StellarGraph

from stellargraph import datasets

from sklearn import model_selection
from IPython.display import display, HTML

from tensorflow.keras import Model
from tensorflow.keras.optimizers import Adam
from tensorflow.keras.layers import Dense
from tensorflow.keras.losses import binary_crossentropy
from tensorflow.keras.callbacks import EarlyStopping
import tensorflow as tf
import matplotlib.pyplot as plt

dataset = datasets.MUTAG()
display(HTML(dataset.description))
graphs, graph_labels = dataset.load()

print(graphs[0].info())
print(graphs[1].info())

summary = pd.DataFrame(
    [(g.number_of_nodes(), g.number_of_edges()) for g in graphs],
    columns=["nodes", "edges"],
)
print(summary.describe().round(1))

generator = PaddedGraphGenerator(graphs=graphs)

def create_graph_classification_model(generator):
    gc_model = GCNSupervisedGraphClassification(
        layer_sizes=[64, 64],
        activations=["relu", "relu"],
        generator=generator,
        dropout=0.5,
    )
    x_inp, x_out = gc_model.in_out_tensors()
    predictions = Dense(units=32, activation="relu")(x_out)
    predictions = Dense(units=16, activation="relu")(predictions)
    predictions = Dense(units=1, activation="sigmoid")(predictions)

    # Let's create the Keras model and prepare it for training
    model = Model(inputs=x_inp, outputs=predictions)
    model.compile(optimizer=Adam(0.005), loss=binary_crossentropy, metrics=["acc"])

    return model

epochs = 200  # maximum number of training epochs
folds = 10  # the number of folds for k-fold cross validation
n_repeats = 5  # the number of repeats for repeated k-fold cross validation
es = EarlyStopping(
    monitor="val_loss", min_delta=0, patience=25, restore_best_weights=True
)

def train_fold(model, train_gen, test_gen, es, epochs):
    history = model.fit(
        train_gen, epochs=epochs, validation_data=[test_gen], verbose=0, callbacks=es,
    )
    # calculate performance on the test data and return along with history
    test_metrics = model.evaluate(test_gen, verbose=0)
    test_acc = test_metrics[model.metrics_names.index("acc")]

    return history, test_acc

def get_generators(train_index, test_index, graph_labels, batch_size):
    train_gen = generator.flow(
        train_index, targets=graph_labels.iloc[train_index].values, batch_size=batch_size
    )
    test_gen = generator.flow(
        test_index, targets=graph_labels.iloc[test_index].values, batch_size=batch_size
    )

    return train_gen, test_gen

test_accs = []

stratified_folds = model_selection.RepeatedStratifiedKFold(
    n_splits=folds, n_repeats=n_repeats
).split(graph_labels, graph_labels)

for i, (train_index, test_index) in enumerate(stratified_folds):
    print(f"Training and evaluating on fold {i+1} out of {folds * n_repeats}...")
    train_gen, test_gen = get_generators(
        train_index, test_index, graph_labels, batch_size=30
    )

    model = create_graph_classification_model(generator)

    history, acc = train_fold(model, train_gen, test_gen, es, epochs)

    test_accs.append(acc)

print(
    f"Accuracy over all folds mean: {np.mean(test_accs)*100:.3}% and std: {np.std(test_accs)*100:.2}%"
)

The entire error message was:

2021-05-14 03:23:24.176132: I tensorflow/compiler/mlir/mlir_graph_optimization_pass.cc:176] None of the MLIR Optimization Passes are enabled (registered 2)
2021-05-14 03:23:24.982603: W tensorflow/core/framework/op_kernel.cc:1744] OP_REQUIRES failed at cast_op.cc:121 : Unimplemented: Cast string to float is not supported
Traceback (most recent call last):
  File "C:/Users/1/PycharmProjects/University Homework/exmpl.py", line 96, in <module>
    history, acc = train_fold(model, train_gen, test_gen, es, epochs)
  File "C:/Users/1/PycharmProjects/University Homework/exmpl.py", line 63, in train_fold
    history = model.fit(
  File "C:\Users\1\AppData\Local\Programs\Python\Python38\lib\site-packages\tensorflow\python\keras\engine\training.py", line 1183, in fit
    tmp_logs = self.train_function(iterator)
  File "C:\Users\1\AppData\Local\Programs\Python\Python38\lib\site-packages\tensorflow\python\eager\def_function.py", line 889, in __call__
    result = self._call(*args, **kwds)
  File "C:\Users\1\AppData\Local\Programs\Python\Python38\lib\site-packages\tensorflow\python\eager\def_function.py", line 950, in _call
    return self._stateless_fn(*args, **kwds)
  File "C:\Users\1\AppData\Local\Programs\Python\Python38\lib\site-packages\tensorflow\python\eager\function.py", line 3023, in __call__
    return graph_function._call_flat(
  File "C:\Users\1\AppData\Local\Programs\Python\Python38\lib\site-packages\tensorflow\python\eager\function.py", line 1960, in _call_flat
    return self._build_call_outputs(self._inference_function.call(
  File "C:\Users\1\AppData\Local\Programs\Python\Python38\lib\site-packages\tensorflow\python\eager\function.py", line 591, in call
    outputs = execute.execute(
  File "C:\Users\1\AppData\Local\Programs\Python\Python38\lib\site-packages\tensorflow\python\eager\execute.py", line 59, in quick_execute
    tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,
tensorflow.python.framework.errors_impl.UnimplementedError:  Cast string to float is not supported
     [[node binary_crossentropy/Cast (defined at /Users/1/PycharmProjects/University Homework/exmpl.py:63) ]] [Op:__inference_train_function_1247]

Function call stack:
train_function

I couldn't find anywhere a float that has been given a value of a string instead, so Im unsure of what is going on here. Any help is appreciated!

Answer 1

Apparently, adding the line:

graph_labels = pd.get_dummies(graph_labels, drop_first=True)

before creating the PaddedGraphGenerator seems to fix the problem.