Fix ValueError of fitting a model

Question

my_model = Sequential()
my_model.add(Embedding(input_dim=1000, output_dim=2000,  weights=[embedding_matrix], trainable=False))
my_model.add(Conv1D(filters=128, kernel_size=3, activation='relu'))
my_model.add(GlobalMaxPooling1D())
my_model.add(Bidirectional(LSTM(units=128)))
my_model.add(Dense(64, activation='relu'))
my_model.add(Dense(1, activation='sigmoid'))  # first try with sigmoid function

my_model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])

my_model.fit(smiles_vectors, epochs=100, batch_size=32)

This is my model which receives vectors from word2vec but when I try to fit it is raised following error: ValueError: Input 0 of layer "conv1d_7" is incompatible with the layer: expected min_ndim=3, found ndim=2. Full shape received: (None, 2000)

How can I fix it? Number of vectors is 2000, tokens which was produced before vectorization are 1255.

I tried to reshape but it didn't worked.