Chainerで学習した対話用のボットをSlackで使用+Twitterから学習データを取得してファインチューニング ref: http://qiita.com/GushiSnow/items/79ca7deeb976f50126d7

file0.py

            if dec_flag:
                if dst["weight_jy"] and child.name == "weight_xi" and self.word2vecFlag:
                    for a, b in zip(child.namedparams(), dst["weight_jy"].namedparams()):
                        b[1].data = a[1].data
                    print('Copy weight_jy')

file0.txt

pip install chainer=="1.5.1"

file1.py

            if dec_flag:
                if dst["weight_jy"] and child.name == "weight_xi" and self.word2vecFlag:
                    for a, b in zip(child.namedparams(), dst["weight_jy"].namedparams()):
                        b[1].data = a[1].data
                    print('Copy weight_jy')

file2.py

    def __input_sentence(self):
        #1
        text = self.__mecab_method(self.data[0]["text"].replace("chainer:", ""))
        #2
        data = [text]
        src_batch = [x + ["</s>"] * (self.generation_limit - len(x) + 1) for x in data]
        return src_batch

    def __predict_sentence(self, src_batch):
        #1
        dialogue = EncoderDecoderModelForwardSlack(self.parameter)
        #2
        src_vocab = Vocabulary.load(self.model_name + '.srcvocab')
        trg_vocab = Vocabulary.load(self.model_name + '.trgvocab')
        #3
        model = EncoderDecoder.load_spec(self.model_name + '.spec')
        serializers.load_hdf5(dialogue.model + '.weights', model)
        #4
        hyp_batch = dialogue.forward(src_batch, None, src_vocab, trg_vocab, model, False, self.generation_limit)
        return hyp_batch

file3.py

                #1
                self.__setting_parameter()
                #2
                model = EncoderDecoder.load_spec(self.model_name + '.spec')
                dialogue = EncoderDecoderModelForwardSlack(self.parameter)
                serializers.load_hdf5(dialogue.model + '.weights', model)
                #3
                dialogue.encdec = model
                dialogue.word2vecFlag = False
                dialogue.train()