hi, I'm trying to implement a seq2seq model. when I use the actual sequence length as TrainingHelper's parameter, I get an error for the loss calculation:
InvalidArgumentError (see above for traceback): logits and labels must have the same first dimension, got logits shape [960,15038] and labels shape [1280]
But if I use all the same sequence length, the code works fine. what's the effects of using actual sequence length instead of using max sequence length?

Korean Is below

Hi, I'm beginner in tensorflow. After your tutorial(https://github.com/j-min/tf_tutorial_plus/blob/master/RNN_seq2seq/contrib_seq2seq/04_AttentionWrapper.ipynb), I have a question, So I write this.

After running your tutorial, I changed data for Korean tokenizing(white space).
I think there is no difference code. But in prediction, there are big difference between train and inference.

So if you have any idea about this(I think this may be hyper-parameter or difference between TrainingHelper and GreedyEmbeddingHelper), Any Comment will be thanks

These are my hyper-parameter and examples

Hyper-parameter:
enc_vocab_size : 2126
dec_vocab_size : 105
enc_sentence_length : 115
dec_sentence_length : 116
hidden_size : 64 or 128 or enc_vocab_size
attn_size = hidden_size

example :
In Train,prediction
Input(title): 마 이 수 야 여 성 용 아 디 오 재 킷
Prediction: _GO 4 7 10 _GO _GO _GO _GO _GO _GO _GO _GO _GO _GO _GO
Target(white space index):, 4 7 10

In Inference,prediction:
Input: 마 이 수 야 여 성 용 아 디 오 재 킷
Prediction: _GO 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9
Target: 4 7 10

안녕하세요 텐서플로우 초보자라 해당 tutorial 을 따라하다 궁금한점이 있어서 문의 드립니다.

해당 tutorial을 작동하고 이후 데이터를 변경하여 한글 띄워쓰기를 시도해 봤는데 코드 관련 해서는 차이점이 없는데 train 과정과 inference 과정에서 차이가 너무 나서 혹시 Hyper Parameter나 혹은 TrainingHelper 와 GreedyEmbeddingHelper 관련 차이가 있는지 문의 드립니다.

하이퍼 파라미터:
enc_vocab_size : 2126
dec_vocab_size : 105
enc_sentence_length : 115
dec_sentence_length : 116
hidden_size : 64 or 128 or enc_vocab_size
attn_size = hidden_size

Train 과정에서의 추측 :
Input: 마 이 수 야 여 성 용 아 디 오 재 킷
Prediction: _GO 4 7 10 _GO _GO _GO _GO _GO _GO _GO _GO _GO _GO _GO
Target:, 4 7 10

Inference 과정에서의 추측 :
Input: 마 이 수 야 여 성 용 아 디 오 재 킷
Prediction: _GO 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9
Target: 4 7 10

Thanks A Lot for your tutorial.

To support TF 1.2
I think you need to update below line
"from tensorflow.contrib.rnn.python.ops.core_rnn_cell_impl import * "
To
"from tensorflow.contrib.rnn import *"

Also to use "embedding_attention_seq2seq" set include path or use:
tf.contrib.legacy_seq2seq.embedding_attention_seq2seq

Thanks Again for your tutorial.

Hi, I run the code and exactly same dataset, I cannot achieve similar results as you, the predictions just make no sense like "hi" and "bye bye'.

I'm trying to train a seq2seq model to predict sequence point coordinates (like [0.1,0.3,0.2]), so I do not have embedding matrix in my model. According to your wonderful tutorial, I know that TrainingHelper is helpful when training my model, but I do not know how to do the inference after training. It seems InferenceHelper may help, but I do not know how to use it.

To simply my question, I need the output of step t-1 as input of step t, and output looks like this [0.1,0.3,0.2].

Thanks!