This repository is a template to create a Text Generation Artificial Inteligence with Tensorflow. This repository includes a pretrained model called aflmtg.h5
inside the models
folder and a dataset /data/1.artidata
This project is inspired from NeuralNine's Text Generation AI - Next Word Prediction in Python youtube video.
using this template is easy:
$ git clone https://github.com/ArtificialOSS/TextGenTensorflow.git
$ pip install --upgrade pip
$ pip install tensorflow
$ pip install nltk
$ python3 setup.py
or run build.sh
Then you can use the AI
$ python3 main.py
Achiving perfect results requires high computing power (RAM)
The bigger the charsize
the better are the results but takes more time to train
+ charsize = 100000
- charsize = 30000
Lowering the creativity=
parameter brings better results if the charsize is over 50000
+ print(generateResponse(msg, 120, 3))
- print(generateResponse(msg, 120, 5))
The more epochs the more accurate and better is the model
+ Train(next_tokens, unique_tokens, train_x, train_y, epochs=30, batch_size=128)
- Train(next_tokens, unique_tokens, train_x, train_y, epochs=12, batch_size=128)
You're welcome to contribute to this template! open a pull request and do your stuff