About

Pipeline for training Stanford Seq2Seq Neural Machine Translation using PyTorch.
Model trained on IWSLT'15 English-Vietnamese.
State-of-the-art on IWSLT'15 English-Vietnamese reference.

Usage

First, install dependencies:

# clone repo
git clone https://github.com/dayyass/neural_machine_translation.git

# install dependencies
cd neural_machine_translation
pip install -r requirements.txt

Data Format

Parallel corpora for Machine Translation.
More about it here.

Vocabulary

Before train any models, you need to create vocabularies for two languages.
More about it here.

Training

Train Neural Machine Translation:

python train.py

At the beginning of the script there is a list of parameters (written in uppercase) for training that can be changed.
Validation performed on every epoch, testing performed after the last epoch.

Validation

NotImplementedError: opened issue.

Inference

NotImplementedError: opened issue.

Models

List of implemented models:

• Seq2SeqModel
• Seq2SeqAttnModel