Updated default reader model in extractive reader component "[PrimeQA](https://huggingface.co/PrimeQA)/[nq_tydi_sq1-reader-xlmr_large-20221110](https://huggingface.co/PrimeQA/nq_tydi_sq1-reader-xlmr_large-20221110)"
0.9.8alpha
Improve component initialization for services Documentation updates - how to set up service store and drop in index, checkpoint - Install FAISS library from conda
0.9.7alpha
1 Add MRQA preprocessor 2. Simplified run_ir arguments 3. Supporting annotate EM 4. Updates to README to. install faiss from conda 4 Updates to services/README on how to finetune with feedback data
0.9.3alpha
This release adds support for:
1. Dense Passage Retrieval (DPR) 2. Pipelines and components interface 3. gRPC and REST service layer 4. Containerize PrimeQA (cpu and gpu) 5. MRC training with custom data and feedback data 6. Automate documentation build 7. Updated PrimeQA logo
0.8.0alpha
This version includes the following: - IR via primeqa/ir/run_ir.py script supports dense and sparse retrieval - MRC via primeqa/mrc/run_mrc.py supports TyDI, SQUAD, MLQA - TyDIQA support for Boolean Questions - Question Generation over table and text - Continuous Integration with GitHub Actions