tokenizer_info.json

{
  "vocab_size": 10303,
  "pad_token_id": 0,
  "mf_start_token_id": 1,
  "mf_end_token_id": 2,
  "bp_start_token_id": 3,
  "bp_end_token_id": 4,
  "cc_start_token_id": 5,
  "cc_end_token_id": 6,
  "organism_vocab_size": 201,
  "embed_model_type": "esm2",
  "max_protein_length": 1024,
  "vocab_config": {
    "MF": {
      "method": "min_freq",
      "value": 20
    },
    "BP": {
      "method": "min_freq",
      "value": 20
    },
    "CC": {
      "method": "min_freq",
      "value": 20
    }
  },
  "vocab_stats": {
    "MF": {
      "method": "min_freq",
      "threshold": 20,
      "selected": 1537,
      "freq_range": [
        20,
        73296
      ],
      "total_terms": 7183
    },
    "BP": {
      "method": "min_freq",
      "threshold": 20,
      "selected": 7720,
      "freq_range": [
        20,
        81287
      ],
      "total_terms": 21200
    },
    "CC": {
      "method": "min_freq",
      "threshold": 20,
      "selected": 1039,
      "freq_range": [
        20,
        84240
      ],
      "total_terms": 2939
    }
  }
}