# Need:
This is a balanced dataset around 106k sentence with their respective emotions. This dataset was created by fusing 3 different smaller datasets. The dataset was primarily created to cater to NLP tasks which use the **transformer** architecture. Since transformers are very data hungry I couldn't use the smaller datasets without over fitting, hence I created this dataset with 11 different emotions. I personally trained DistilBERT from the BERT family.
# Technical details:
The datasets has no NaNs but do double check just in case. I've removed all the usernames and URLs from the texts. I've also removed just the '#' symbol from hashtags instead of the entire thing as hashtags can carry valuable information about the emotion of the sentence.
# Here are the 3 datasets I used to create mine:
1. [emotion analysis based on text](https://www.kaggle.com/datasets/simaanjali/emotion-analysis-based-on-text)
2. [Text with Sentiment](https://www.kaggle.com/datasets/divu2001/text-…
