How to use additional input features for NER?

Beginners
14

Thanks for the reply, I somehow managed to follow your method. Now I am not getting desired accuracy, and was setting up optimisers for pos_tag_embedding layer.

optimizer1 = torch.optim.Adam([ 
                                {'params': model.bert.embeddings.pos_tag_embeddings.parameters(), 'lr': 5e-5},
                                {'params': model.bert.parameters()},
                              ], lr=2e-5)

optimizer=torch.optim.Adam(model.parameters(),lr=0.01)
model.base_model.embeddings.parameters

This is throwing me error, any idea how to train this embedding layer accurately?