Train a neural network optimized for generating tweets based off of any number of Twitter users! Blend wildly different Twitter users together for hilarity!
Tweet Generator is based off of textgenrnn, and trains the network using context labels for better tweet synthesis.
Open config.yml
and at the top, add the Twitter API keys from a Twitter App you own. Then configure the list of twitter_users
you wish to train the network on. You can also configure the num_epochs
and whether to use the pretrained model or train a new_model
. Then simply run:
python3 tweet_generator.py
The script will automatically save the weights (+ config and vocab for if new_model
) for the trained model, which can then be loaded into textgenrnn and used anywhere.
The included realDonaldTrump_dril_twitter_weights.hdf5
was trained on @realDonaldTrump and @dril's tweets for 200 epochs. You can load it and generate text from it simply with:
from textgenrnn import textgenrnn
textgen=textgenrnn('realDonaldTrump_dril_twitter_weights.hdf5')
textgen.generate_samples()
- textgenrnn
- tensorflow (either CPU or GPU flavors)
- tweepy
Max Woolf (@minimaxir)
Max's open-source projects are supported by his Patreon. If you found this project helpful, any monetary contributions to the Patreon are appreciated and will be put to good creative use.
MIT