I am new to hugginface and I just tried to fine-tune a model(facebook/dpr-ctx_encoder-single-nq-base) with multiple indexes from here, following the tutorial here using TensorFlow , but I am not sure if what I am doing is correct or not and I got several problems.
(1) Since the data I am using is squad_v2, there are multiple indexes and the content of ‘answers’ index is a list of objects, how should I tokenize it? I tried to tokenize them one by one. but get error:
raise ValueError(
ValueError: text input must of type `str` (single example), `List[str]` (batch or single pretokenized example) or `List[List[str]]` (batch of pretokenized examples).
(2) When I tried to remove the ‘answers’ line, another error occurred
raise AttributeError("'{}' object has no attribute '{}'".format(
AttributeError: 'DPRContextEncoder' object has no attribute 'compile'
AttributeError: 'DPRContextEncoder' object has no attribute 'fit'