WikiHow QA Dataset Project
- Analysed a WikiHow FAQ QA dataset for the Cooking and DIY category.
- Experimented on downstream tasks to check how pre-training LM on the custom dataset before fine-tuning on task.
- Used RoBERTa-large for Question Type Classification, Re-Ranking, MRC and Intent Recognition tasks.
- Used T5-large and bloom LLMs for Abstractive Closed Book-ODQA on the WikiHow FAQ dataset.