Using the method of prompt tuning to finetune for various NLP tasks.

summarization.py

question-answering.py

machine_translation.py

1. batch_size: 2 or 4 or 8. CUDA goes out of memory after this. Gradient accumulation has been used.
2. num_epochs: 10. However, early stopping has also been used.
3. gradient clipping: Gradients have been clipped with a norm of 1.0 to prevent exploding gradients.
4. learning_rate: 0.01
5. optimiser: AdamW ,
6. metric: Bleu score

• Best Loss achieved: 3.2

• Best Loss achieved: 1.32
• Bleu score on test set: 0.03

• Best Loss achieved: 4.56
• Bleu score on test set: