End-to-end fine-tuned Transformer classifiers for financial sentiment (3-class) and sarcasm (2-class). Built with PyTorch + 🤗 Transformers, trained/evaluated on multiple datasets, and packaged for reuse and continued fine-tuning.

Base: microsoft/deberta-v3-base

Labels: {0: negative, 1: neutral, 2: positive}

Kaggle Dataset for training: ankurzing/sentiment-analysis-for-financial-news, cosmos98/twitter-and-reddit-sentimental-analysis-dataset

Base: distilbert-base-uncased

Labels: {0: not sarcastic, 1: sarcastic}

Kaggle Dataset for training: danofer/sarcasm, rmisra/news-headlines-dataset-for-sarcasm-detection

https://github.com/rick-mingyu-liu/wsb-sentiment-analysis https://huggingface.co/imnotrick/sentiment_sarcasm