Errors and missing instructions

[giusidp]

I've tried running your code, I followed your instructions of preprocessing the data (s2s_preprocess.py) and running the main under ./baselines, but it generates just .pt files, while the main for the flask web application needs a .tfmodel file...which is missing and it's not generated (there are only tfmodel.meta and tfmodel.index, but the file containing the weights is missing, of course). I've seen that some .tfmodel are generated inside s2s_train.py but i get this error that I just can't fix:, even if i'm using the same pre-trained embedding as yours I guess: ValueError: Cannot feed value of shape (33017, 100) for Tensor 'pretrained_decoder_embeddings:0', which has shape '(33090, 100)' I already solved previous bugs and also fixed wiki_dbpedia_train.txt (in some points of the txt there were too many tabs and it caused an error), so please...could you be more clear on what I am missing? I've been struck on this code for way too long