--- license: apache-2.0 --- Hey ! Well, you can use this model for the Q&A model- given a context and question corresponding from the context, this model can answer the question. Beware that this model can only answer closed-end questions. If you have given context based on the history of the USA from a Wikipedia page and asked the question 'Why the sky is Blue ?', chances will be that you wouldn't get the answer you expected. The exceptional thing is I fine-tuned this model with only 50 data points (out of 10000) due to RAM issues. Even though with such a small dataset, this is working exceptionally well with 98% accuracy! # Context You can input up to 30k words (That's equivalent to 100-page book content). But, it takes time to pre-process (expected!). # Question Ask the question which can be answered from the context only. This field can take upto 1000 words as input. # Answer This section provides you the answer from the above question. Note: Sometimes you may not get the right answer! # Score This number lies between 0 and 1, 1 being best possible answer, 0 being least possible answer. I used BERT model for this task. I've explored different methods and finally arrived at this model. This is pretty good one! PS: I'm working on solving Q&A for physics problems. If you are a big physics buff, please reach out to me to make this possible! (mail:manideepmaruthi.234@gmail.com)