arukemre / TextGenerativeModel_Scratch_with_Pytorch

Generating text files processed from a resource using Natural language processing

Text generative model scratch with python

Hi. I want to introduce this repository towards enhance our nlp skill with python. This repository includes Some preprocessing steps and model scratch using torch library. well alraigth what did ı do. I found a text from web alice.txt and with this text that have generated a new text as a continuaton.

|─ alice.txt
|─ TextGenerativeModel.py

    |─ Data Preview
    |─ Data preprocessing
    |─ Data preprocessing and Feature Engineering steps
    |─ textGenerator class
    |─ Model optimization with `Ray Tune` from `Pytorch` library

Requaried libraries

      > pandas 
      > numpy 
      > nltk
      > re
      > torch
      > sklearn
      > collections 
      > ray

LSTM arcitecture

What does LSTM ?

Please look at LSTM link

All rigth!,Which steps were placed?

Then I created some functions for preprocessing steps these functions facilitated pre-processing steps.
After this section. we are in preprocessing step.
The Preprocessing step purpose that how can train computers with these sentences. In order to train the model, we must. Firstly we must label words in the text.
We can do this. We should find all unique words from passed preprocessing steps then we should create a dictionary for encoding unique words.
After all of these steps, we have a text that consists of labeled words.After the process of labeling. 
we created representative tensors that include all labeled words in tex. This process is a little complex for the beginner stage.
For more details please look at the codes... 

Model tuned with RayTune with Pytorch library

RayTune link