zerogerc / rnn-autocomplete

Bachelor's grad work on code autocompletion with rnn

Bachelor's grad work in neural code completion

Initial set up

Create virtual environment: ./venv.sh

Activate virtual environment: source env/bin/activate

Proposed models are working with AST so there is a possibility to complete any language. For now there is possibility to test model on two datasets:

1. Javascript (js150 dataset link)
2. Python (py150 dataset link)

Javascript

To train model on Javascript dataset:

1. Download data: ./scripts/ast/data_download.sh
2. Process data: ./scripts/ast/data_process.sh
3. Train model: ./scripts/ast/run.sh

To change model parameters edit file: scripts/ast/train.sh

Python

To train model on Python dataset:

1. Download data: ./scripts/pyast/data_download.sh
2. Process data: ./scripts/pyast/data_process.sh
3. Train model: ./scripts/pyast/run.sh

To change model parameters edit file: scripts/pyast/train.sh

Results

For accuracy visualization tensorboard is used. To run it use: ./scripts/tensorboard.sh