Autocomplete

Extremely simple(and maybe useless) LSTM model trained on google search history data

Getting started

1. Get your google search data (only google search)
2. Put the MyActivity.html file in the data folder
3. Generate the JSON python src/save_to_json.py
4. cd src
5. Create dataset python process_data.py --past_count 4 --freq_tresh 5 --window 1
6. Train cbow model (for embeddings) python train_cbow.py
7. Train autocomplete model python train.py
8. Run web app python app.py

Demo

noodle_demo_one_word.mp4