Week 2
Milestones
- Generated synthetic data required for finetuning BART.
- Validate the quality of generated samples.
- Finetune and evaluate BART on test cases.
Contributions
- Updated issue.
- Commit: Added training pipeline (from generating data to training BART).
- Commit: Reported results on finetuned BART model.
Learnings
- Learnt more about prompt engineering methods and using Hugging Face tools.
- Read and understood DSP paper.