Beam Search For keras Character-level recurrent sequence-to-sequence model

[VallabhMahajan1]

trafficstars

I was trying to implement tensorflow add-ons beam search decoder in my existing char level seq2seq model.

TensorFlow addons Beam search decoder. https://www.tensorflow.org/addons/api_docs/python/tfa/seq2seq/BeamSearchDecoder

keras Character-level recurrent sequence-to-sequence model https://keras.io/examples/nlp/lstm_seq2seq/

There was a tensorflow notebook to implement seq2seq model with beam search, but that model focused on word-level architecture. I tried to change parameter, 'char_level = True' in keras tokenizer, but it didn't worked.

Can someone please help me to solve this issue.

Thank you in advance