Blame for tokenizer_config.json

{"model_max_length": 1024}