Skip to content

Latest commit

 

History

History
36 lines (17 loc) · 682 Bytes

README.md

File metadata and controls

36 lines (17 loc) · 682 Bytes

Tensorflow implementation of Natural Language Processing Algos

1. Seq2Seq (machine translation)

encoder: LSTM

decorder: LSTM with beam search

dataset: Eng-french translation dataset

2. Show attend and tell (Image Captioning)

encoder: Convnet-VGG

attention model: Bahdanua Attention

decoder: LSTM with beam search     

dataset: COCO 

3. Sentiment Analysis

 encoder: multilayer LSTM 
 
 decoder: Classification Dense Layer
 
 dataset: Imdb-dataset

4. Loading pre-trained word embeddings

facebook: multilingual embeddings: MUSE 

stanford: Glove