ldm/modules/encoders/CLAP/bert-base-uncased/tokenizer_config.json

{"do_lower_case": true, "model_max_length": 512}