This is machine learning model that is trained to predict next word in the sequence. Model is defined in keras and then converted to tensorflowjs model using tfjs_converter.
- keras
- tensorflow
- tensorflowjs_converter (for web deployment)
To install all dependencies at once use:
pip3 install -r requirements.txtFirst create necessary directories or else you will get an error.
mkdir model/checkpoint
mkdir web/dist/web_modelOnce you have your dependencies installed via pip, train the model by running :
python3 train.pyAfter training you can either use the model by running :
python3 test.pyAfter running test.py your output should be :
Enter atleast 3 words:
welcome to new
indexes of last 3 words : [4402, 5, 35]
prediction indexes : [92, 1, 2645, 1491, 189, 1762, 207, 362, 2261, 1727]
1 - welcome to new york
2 - welcome to new <unk>
3 - welcome to new york-based
4 - welcome to new england
5 - welcome to new products
6 - welcome to new jersey
7 - welcome to new debt
8 - welcome to new loans
9 - welcome to new hampshire
10 - welcome to new venturesor by converting the model to tensorflow js model by running :
tensorflowjs_converter --input_format=keras /model/model.h5 /web/dist/web_modelThen Read the README.md in /web directory
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent