chunml / seq2seq Goto Github PK

I have used same model just changes dataset with Hindi and English corpus.
Test -
डेली हिन्दी न्यूज - बुंदेलखंड का प्रथम अन्तरजालीय स्थल
मैं भारत से प्यार करता हूँ
ज़बूर जो कि दाउद को प्रदान की गयी
प्रशासनिक विभाजन
वे इस देश के प्रथम UNK -LRB- अफ्रीकी UNK -RRB-
नेपाली विदेश
वेल्श खिलाड़ी इंग्लैंड के लिए खेलने के लिए पात्र हैं
उनके बड़े भाई अजीत तेंडुलकर ने उन्हें खेलने के लिये प्रोत्साहित किया था ।
फिर एक मिनट के बाद किताब छत की ओर उछालकर उन्होंने कहा - चलो ।

Output -
of of of
of of of
the of of of

lrb lrb rrb rrb

the the of of of
the the of of of of
the the the of of of of

Any possible reason behind this, any help much appreciate.

I'm concerned with the test results of the model. I'm trying to implement seq2seq models using word2vec and to before trying to implement on my own , I want to verify the architecture for it. So it would be very helpful if I can get some insights into the test results for this model.

I suppose your sequential model has one-layer encoder and possible multi-layer decoder. however, in most cases encoder has also many layers. To me,
for _ in range(num_layers):
model.add(LSTM(hidden_size, return_sequences=True))
may work, not test yet.
moreover, could bidirectional encoder can be implemented in Keras?