Alignment Training task using minibatch

Summary:
1. Define a EpochMinibatchIterator which extends the EpochBatchIterator. It has same functionality as EpochBatchIterator except two major changes: use static batching and use MiniBatchIterator for getting the indices.

2. SplitSeqCollater is used instead of Seq2SeqCollater.
3. LSTM_subsample started storing the previous states and reset it once the sample is over.

Reviewed By: jay-mahadeokar

Differential Revision: D15209023

fbshipit-source-id: 900b8bd1f25159ffc77f8106e26729a3e7422a1f

fairseq/models/fairseq_model.py

@@ -325,7 +325,7 @@ class FairseqEncoderModel(BaseFairseqModel):
         Returns:
             the encoder's output, typically of shape `(batch, seq_len, vocab)`
         """
-        return self.encoder(src_tokens, src_lengths)
+        return self.encoder(src_tokens, src_lengths, **kwargs)
 
     def get_normalized_probs(self, net_output, log_probs, sample=None):
         """Get normalized probabilities (or log probs) from a net's output."""