The issue you’re facing with having 0 trainable parameters is likely due to the missing input shape specification for the Embedding layer. The Embedding layer requires the input shape to be defined so that it knows how to handle the sequence input.

Input Shape for Embedding Layer:
The Embedding layer should have an input shape that matches the shape
of your input data. Typically, for text classification problems, this
would be the number of words (vocabulary size) and the length of the
input sequence (for example, the number of words in each review).

Fixing the Model:

Add the input_length parameter to the Embedding layer to specify the
length of the input sequences. For instance, if each review has 500
words, you would set input_length=500

import numpy as np
import tensorflow as tf

# Importing imdb dataset
from tensorflow.keras.datasets import imdb
from tensorflow.keras.preprocessing import sequence

# The sequence module in tensorflow.keras.preprocessing provides utilities for preprocessing sequence data, such as text or time-series data.
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Embedding, SimpleRNN, Dense

# Define the model
model = Sequential()

# Add the Embedding layer with input length
model.add(Embedding(1000, 128, input_length=500))  # Adjust the input length (500) based on your input data

# Add the SimpleRNN layer
model.add(SimpleRNN(128, activation='relu'))

# Add the Dense output layer
model.add(Dense(1, activation='sigmoid'))

# Print the model summary to check the parameters
model.summary()

The Embedding layer is now given an input_length of 500, which specifies that each input sequence (e.g., a review) has 500 words.
The model will now have trainable parameters, as the Embedding layer needs to learn a weight matrix to map the vocabulary indices to vectors of the specified size (128).
This should resolve the issue of having 0 trainable parameters.

import numpy as np
import tensorflow as tf

# Importing imdb dataset
from tensorflow.keras.datasets import imdb
from tensorflow.keras.preprocessing import sequence
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Embedding, SimpleRNN, Dense

# Load the IMDB dataset
max_features = 1000  # Number of unique words to consider
maxlen = 100  # Maximum length of each input sequence

# Load the data
(x_train, y_train), (x_test, y_test) = imdb.load_data(num_words=max_features)

# Pad sequences to ensure uniform input size
x_train = sequence.pad_sequences(x_train, maxlen=maxlen)
x_test = sequence.pad_sequences(x_test, maxlen=maxlen)

# Define the model
model = Sequential()

# Add the Embedding layer
model.add(Embedding(input_dim=max_features, output_dim=128, input_length=maxlen))

# Add the SimpleRNN layer
model.add(SimpleRNN(128, activation='relu'))

# Add the Dense output layer
model.add(Dense(1, activation='sigmoid'))

# Compile the model
model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])

# Print the summary of the model to check the parameters
model.summary()

# Now you can train the model
# model.fit(x_train, y_train, batch_size=32, epochs=10, validation_data=(x_test, y_test))

Key Changes Made:

• Input Shape: The input_length parameter is specified in the Embedding layer. This tells the model the length of the input sequences.
• Compiling the Model: The model is compiled using model.compile(), which is necessary before training the model. This step also initializes the trainable parameters.
• Padding Sequences: The input sequences are padded to ensure they all have the same length, which is required for RNNs.
• Training the Model: A commented-out line for training the model is
  included. You can uncomment it and adjust the parameters as needed to
  start training.

With these changes, the model should now have trainable parameters, and you should see a non-zero count when you call model.summary(). You can proceed to train the model with the provided training data.