Missing model_max_length for tokeniser

[orkenstein]

I'm trying to use Tokenizer for ModernBERT like this:

model_name = "answerdotai/ModernBERT-large"
tokenizer=AutoTokenizer.from_pretrained(model_name)
print(tokeniser.model_max_length) 
# 1000000000000000019884624838656

Which gives model_max_length == int(1e30)

Shouldn't the tokenizer contain a correct max value of 8192?