max length of hfl/chinese-electra-180g-large-discriminator tokenizer is incorrect

[AngledLuffa]

Hey, I've had good luck with using the embeddings of hfl/chinese-electra-180g-large-discriminator in downstream tasks, so thank you for posting that model.

However, I find that the model_max_length of this model is incorrectly set. It appears to be 512 based on what happens when I run it, but instead, the value in the model itself is set to some arbitrary large number (1000000000000000019884624838656)