def get_dataset()

in run_language_modeling.py [0:0]

• 8 lines of code
• 3 McCabe index (conditional complexity)

def get_dataset(args: DataTrainingArguments, tokenizer: PreTrainedTokenizer, evaluate=False):
    file_path = args.eval_data_file if evaluate else args.train_data_file
    if args.line_by_line:
        return LineByLineTextDataset(tokenizer=tokenizer, file_path=file_path, block_size=args.block_size)
    else:
        return TextDataset(
            tokenizer=tokenizer, file_path=file_path, block_size=args.block_size, overwrite_cache=args.overwrite_cache
        )