Val_accuracy is not changing between each epoch - tensorflow

Question

I'm currently learning about gensim word2vec, and in the middle of making a model that can predict what the rating would be based on the words. All of my code compiles fine, however each epoch posses the same val_accuracy, and almost the exact same training accuracy. I've tried changing the architecture and hyperparameters, but have had no luck whatsoever. Any advice would be appreciated!
from keras.preprocessing.sequence import pad_sequences
word_index = {w: i+1 for i,w in enumerate(index_to_key) if i < max_words-1} # Keep just max_words (zero is reserved for unknown)
sequences = [[word_index.get(w, 0) for w in sent] for sent in reviews] # code the sentences
seqs_truncated = pad_sequences(sequences, maxlen=max_review_length, padding="pre", truncating="post")
ratings = np.asarray(Ratings)
# prepare training and validation data
x_val = seqs_truncated[:len_val]
partial_x_train = seqs_truncated[len_val:]
y_val = to_categorical(ratings[:len_val]-1, num_classes=5)
partial_y_train = to_categorical(ratings[len_val:]-1, num_classes=5)
print('Length of validation set =', len(x_val))
print('Length of training set =', len(partial_x_train))
def sample_network(embedding):
network = Sequential()
network.add(embedding)
network.add(Bidirectional(LSTM(64)))
network.add(Dense(64, activation='relu'))
network.add(Dense(32, activation="relu"))
network.add(Dense(16, activation="relu"))
network.add(Dropout(0.5))
network.add(Dense(5,activation='softmax'))
return network```
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
embedding_24 (Embedding) (None, 100, 300) 1500000
_________________________________________________________________
bidirectional_5 (Bidirection (None, 128) 186880
_________________________________________________________________
dense_16 (Dense) (None, 64) 8256
_________________________________________________________________
dense_17 (Dense) (None, 32) 2080
_________________________________________________________________
dense_18 (Dense) (None, 16) 528
_________________________________________________________________
dropout_10 (Dropout) (None, 16) 0
_________________________________________________________________
dense_19 (Dense) (None, 5) 85
=================================================================
Total params: 1,697,829
Trainable params: 1,697,829
Non-trainable params: 0 ```
hist_word_vec =network_wordvec.fit(partial_x_train,partial_y_train, epochs=no_epochs,
batch_size=256, validation_data=(x_val,y_val)
65/65 [==============================] - 39s 606ms/step - loss: 1.5886 - accuracy: 0.4449 - val_loss: 1.5750 - val_accuracy: 0.3875
Epoch 2/8
65/65 [==============================] - 38s 587ms/step - loss: 1.5491 - accuracy: 0.4554 - val_loss: 1.5458 - val_accuracy: 0.3875
Epoch 3/8
65/65 [==============================] - 37s 564ms/step - loss: 1.5152 - accuracy: 0.4554 - val_loss: 1.5215 - val_accuracy: 0.3875