The three arguments that Keras embedding layer specifies are
input_dim: The size of the vocabulary within the text data. For instance, if your information is integer encoded toward values among 0-10, then that size of the vocabulary would comprise 11 words.
output_dim: The size of the vector space within which information will be inserted. It determines the size of each output vectors of this layer for each word. For instance, it could be 32 or else 100 or still larger. Test distinct values of your problem.
input_length: The length of the input series, as you would determine for any input layer from a Keras model. For instance, if all of your input records are composed of 1000 words, this will be 1000.
Keras