A Sequence to Sequence model trained on a dataset of 137861 English and French sentences that can translate new sentences from English to French, built with TensorFlow.
- Number of Epochs: 3
- Batch Size: 128
- LSTM Size: 256
- Number of LSTM Layers: 2
- Embedding Size: 256
- Learning Rate: 0.001
- Dropout Keep Probability: 0.50
Achieved Training Accuracy of 97.2% and Validation Accuracy of 96.8%.