add gitignore, add requirements, correct format, add main, add modes …

[diazcelsa]

…for show text with sentiment, pickle sentiment of a text, pickle sentiment of a list of texts

The changes are mainly:

• Add requirements with package versions
• Gitignore to avoid editor configuration and cache files
• Add a main in visualize.py with 3 modes: current one "show" and two more for storing the sentiment of one text or a list of texts
• pep-8 formating of the code to be more readable

[diazcelsa]

I evaluated your pre-trained model with my data set and with several sentences from Sutskever et al. 2017 and the sentiment looks coherent. You can have a look:
https://github.com/diazcelsa/NLP_politics_spiegel/blob/master/notebooks/sentiment/Evaluate%20Sentiment%20of%20migration-related%20selected%20articles.ipynb