NLP 📜 Huggingface Transformers Demo

Demo of different huggingface NLP transformers pipelines :

• Question Answering.
• Text Translation.
• Text Polarity Analysis.

Question Answering

Text Translation

Text Polarity Analysis

Usage

Installing necessary requirements :

python install -r requirements.txt

Running the application :

python app.py

Note : The first time that app is started is going to take some time since the pre-trained models will be downloaded.

Models

1. Question Answering : The default model in use is the distilbert-base-cased-distilled-squad. It also uses the same distilbert-base-cased-distilled-squad tokenizer.

2. Translation : Performs English to French text translation, which can be changed, to any pre-trained language pair available. Example :

translator = pipeline("translation_en_to_fr")
# Or
translator = pipeline("translation_en_to_de")

Note Prediction with sequence length variation can be explored woth the sequence length slide (default=40).

3. Text Polarity Classification : Analyses the input to text to determine if it's a 'POSTITIVE' or 'NEGATIVE' text contxt, and returns the score.