Senda

🚀 Danish BERT fine-tuned for Sentiment Analysis with senda

This model is designed to detect the polarity ('positive', 'neutral', 'negative') of Danish texts, offering valuable insights into the sentiment of such content. It is trained and tested on Tweets annotated by Alexandra Institute, and the training process is carried out using the senda package.

🚀 Quick Start

Here is an example of how to load the model in PyTorch using the 🤗Transformers library:

from transformers import AutoTokenizer, AutoModelForSequenceClassification, pipeline
tokenizer = AutoTokenizer.from_pretrained("pin/senda")
model = AutoModelForSequenceClassification.from_pretrained("pin/senda")

# create 'senda' sentiment analysis pipeline 
senda_pipeline = pipeline('sentiment-analysis', model=model, tokenizer=tokenizer)

text = "Sikke en dejlig dag det er i dag"
# in English: 'what a lovely day'
senda_pipeline(text)

✨ Features

• Sentiment Polarity Detection: The model can accurately detect whether a Danish text expresses a positive, neutral, or negative sentiment.
• Trained on Real - world Data: It is trained and tested on Tweets annotated by the Alexandra Institute, ensuring its practical applicability.

📚 Documentation

Performance

The senda model achieves an accuracy of 0.77 and a macro - averaged F1 - score of 0.73 on a small test data set provided by Alexandra Institute. The model can most certainly be improved, and we encourage all NLP - enthusiasts to give it their best shot. You can use the senda package to do this.

Contact

Feel free to contact author Lars Kjeldgaard on lars.kjeldgaard@eb.dk.

Shout - outs

Props to Malte Højmark - Berthelsen for pretraining Danish BERT and helping out adding a TensorFlow backend for senda.

📄 License

This project is licensed under the cc - by - 4.0 license.