[EPIC] Sparsification for KPI extraction question answering task

[Shreyanand]

The current KPI extraction question answering model is huge (1.7gb) and it takes around a total of 7mins to infer for a pdf. We want to find smaller version of the models that get similar performance but using a smaller and faster model.

Overall, we want to investigate model pruning and test the effects of tools such as NeuralMagic to measure the performance impact of different levels of pruning. This EPIC is the first step for this overall goal.

• Create huggingface version of QA KPI extraction task #223
• Use sparsified neural models for KPI extraction task