- Combination of different NMT architectures for Machine Translation Task. Currently, GCNN (Gated Convolution Neural Network) model is completed for German to English Translation and it has a BLEU score of 32.4.
- BERT-base model is completed.
- English to Hindi CNN Sequence to Sequence Model is trained with MultiIndic Language Corpus. Performance evaluation is left.
- English to Hindi Pre-process part is done in generic for all BERT-base, CNN Seq2Seq and RNN-With Attention Model.
- Training BERT-base is left.
- Web-App Frontend and Backend for Language Translation is currently being done parallelly in separate repo.