Update README.md

update original repo to allow auto infer without manual input labels in code

update readme and requirments

update trainer with eval on the test dataset

training validation works with p-tuning-v2 now; see mrc4ner/scripts/mrc_ner/reproduce/conll03.sh as example

integrate the p-tuning-v2 into MRC-NER

update MRCNERDatasetPtuning and collate_to_max_length_ptuning

init mrc_ner with p-tuning-v2 integration

Initial commit