TypeError: Unhashable type: 'numpy.ndarray' in keras

Question

I have a code below

while True:
   
    question = input("你： ")


    question_seq = tokenizer.texts_to_sequences([question])
    
    question_seq_padded = keras.preprocessing.sequence.pad_sequences(question_seq, maxlen=max_len)
    
    answer_seq = model.predict(question_seq_padded).argmax(axis=-1)[0]
    answer = tokenizer.index_word[answer_seq]
    print("機器人：", answer)

And when i ran it, happened some wrong

answer = tokenizer.index_word[answer_seq]

issued error

TypeError: Unhashable type: 'numpy.ndarray'

I try to find the wrong, maybe answer_seq can’t hash, I don’t know.

Asked By: rtyrtyrtyqa

||

Source

Answer 1

You should use tokenizer.sequences_to_texts([answer_seq]) instead.

Answered By: TheEngineerProgrammer

Answer 2

I found the way to fix the code, just add something like below.

    answer_seq_tuple = tuple(answer_seq)
    answer = tokenizer.sequences_to_texts(answer_seq_tuple)

Answered By: rtyrtyrtyqa

TypeError: Unhashable type: 'numpy.ndarray' in keras

Question:

Answers: