config.json

{
  "model_series": "uce",
  "model_name": "4layer",
  "model_type": "embedding_extractor",
  "nlayers": 4,
  "d_model": 1280,
  "nhead": 20,
  "d_hid": 5120,
  "output_dim": 1280,
  "token_dim": 5120,
  "dropout": 0.05,
  "pad_length": 1536,
  "sample_size": 1024,
  "pad_token_idx": 0,
  "chrom_token_left_idx": 1,
  "chrom_token_right_idx": 2,
  "cls_token_idx": 3,
  "CHROM_TOKEN_OFFSET": 143574,
  "species": "human",
  "embedding_model": "ESM2",
  "spec_chrom_csv_path": "species_chrom.csv",
  "token_file": "tokens.pt",
  "protein_embeddings_dir": "protein_embeddings/",
  "offset_pkl_path": "species_offsets.pkl",
  "batch_first": true
}