I need PyTorch expert who have experience in Transfer Learning and Multi-task Learning used for NLP applications

I have a pre-trained model that fine-tuned on Stance Detection task. I would like to further improve the model by implementing a multi-task learning, specifically by combining three tasks: stance, sentiment, sarcasm, in order to improve the performance of the target task (i.e., stance detection).

Upon confirming, the repository including the data and the code used to fine-tune BERT-based models for stance detection (using PyTorch Lightning framework) will be shared with you.