https://github.com/ZeRoGerc/rnn-autocomplete
Bachelor's grad work on code autocompletion with rnn
https://github.com/ZeRoGerc/rnn-autocomplete
Last synced: 5 days ago
JSON representation
Bachelor's grad work on code autocompletion with rnn
- Host: GitHub
- URL: https://github.com/ZeRoGerc/rnn-autocomplete
- Owner: zerogerc
- License: apache-2.0
- Created: 2017-12-17T14:30:36.000Z (almost 8 years ago)
- Default Branch: master
- Last Pushed: 2019-05-19T14:10:25.000Z (over 6 years ago)
- Last Synced: 2025-04-13T14:43:41.833Z (7 months ago)
- Language: Python
- Size: 15.1 MB
- Stars: 10
- Watchers: 3
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.rst
- License: LICENSE.txt
Awesome Lists containing this project
- awesome-machine-learning-on-source-code - rnn-autocomplete - Neural code autocompletion with RNN (bachelor's thesis). (Software)
- awesome-machine-learning-on-source-code - rnn-autocomplete - Neural code autocompletion with RNN (bachelor's thesis). (Software)
README
##################
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``