Home ValueError: Unexpected result of `predict_function` (Empty batch_outputs). Please use `Model.compile(…, run_eagerly=True)`

Questions

ValueError: Unexpected result of `predict_function` (Empty batch_outputs). Please use `Model.compile(…, run_eagerly=True)`

byMR

November 15, 2021

I have this model :

# Set random seed
tf.random.set_seed(42)

# Create some regression data
X_regression = np.expand_dims(np.arange(0, 1000, 5), axis=0)
y_regression = np.expand_dims(np.arange(100, 1100, 5), axis=0)

# Split it into training and test sets
X_reg_train = X_regression[:150]
X_reg_test = X_regression[150:]
y_reg_train = y_regression[:150]
y_reg_test = y_regression[150:]

# Setup random seed
tf.random.set_seed(42)

# Recreate the model
model_3 = tf.keras.Sequential([
  tf.keras.layers.Dense(100),
  tf.keras.layers.Dense(10),
  tf.keras.layers.Dense(1)
])

# Change the loss and metrics of our compiled model
model_3.compile(loss=tf.keras.losses.mae, # change the loss function to be regression-specific
                optimizer=tf.keras.optimizers.Adam(learning_rate=0.01),
                metrics=['mae']) # change the metric to be regression-specific

# Fit the recompiled model
model_3.fit(X_reg_train, y_reg_train, epochs=100)

To begin with, the model does not train well

To add on, when I try to predict using that model, I get the following error :

Why am I getting the above error and how can I fix it?

>Solution :

Change the axis dimension in expand_dims to 1 and slice your data like this, since it is 2D:

import tensorflow as tf
import numpy as np

tf.random.set_seed(42)

# Create some regression data
X_regression = np.expand_dims(np.arange(0, 1000, 5), axis=1)
y_regression = np.expand_dims(np.arange(100, 1100, 5), axis=1)

# Split it into training and test sets
X_reg_train = X_regression[:150, :]
X_reg_test = X_regression[150:, :]

y_reg_train = y_regression[:150, :]
y_reg_test = y_regression[150:, :]

tf.random.set_seed(42)

# Recreate the model
model_3 = tf.keras.Sequential([
  tf.keras.layers.Dense(100),
  tf.keras.layers.Dense(10),
  tf.keras.layers.Dense(1)
])

# Change the loss and metrics of our compiled model
model_3.compile(loss=tf.keras.losses.mae, # change the loss function to be regression-specific
                optimizer=tf.keras.optimizers.Adam(learning_rate=0.01),
                metrics=['mae']) # change the metric to be regression-specific

# Fit the recompiled model
model_3.fit(X_reg_train, y_reg_train, epochs=100)

model_3.predict(X_reg_test)