https://github.com/Bakuutin/lstm_text_generator

Neural network powered telegram bot
https://github.com/Bakuutin/lstm_text_generator

lstm machine-learning rnn telegram tensorflow

Last synced: about 2 months ago
JSON representation

Neural network powered telegram bot

• Host: GitHub
• URL: https://github.com/Bakuutin/lstm_text_generator
• Owner: Bakuutin
• Created: 2017-03-02T07:45:34.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2017-03-19T16:06:17.000Z (over 7 years ago)
• Last Synced: 2024-07-04T17:20:24.041Z (3 months ago)
• Topics: lstm, machine-learning, rnn, telegram, tensorflow
• Language: Python
• Homepage:
• Size: 9.77 KB
• Stars: 12
• Watchers: 1
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: readme.md

Awesome Lists containing this project

README

        
# Installation

    pip install -Ur requirements.txt

    mkdir -p data/checkpoints

    cp config.example.py config.py

*Optional:* Add your telegram key to `config.py`.

# Train on the new text

Set proper `TEXT_PATH`.

Delete `checkpoints/*`, `vocab.npy` and `vec_vocab.npy` from `data` directory.

Run `./trainer.py`.

Press `ctrl+c` to stop training.

# Run

Use `./generator.py` to check the texts.

For bot mode you must set `BOT_TOKEN` and `BOT_LOG_CHANNEL_ID`.

Run it via `./bot.py`.

# Pre trained net

Feel free to use [this snapshot of data](https://drive.google.com/file/d/0BwKD7s64TX1KeVZucnp5NzkxZnM/view).

The text is written by Eliezer Yudkowsky on human rationality and irrationality in cognitive science. You can find the original on the [LessWrong](http://lesswrong.com/).