joint_config.json

{
  "model_type": "joint_nt_esm2",
  "architecture": "joint_dna_protein",
  "dna_model": {
    "type": "nucleotide_transformer",
    "path": "./dna",
    "context_length": 4096,
    "vocab_size": 4107
  },
  "protein_model": {
    "type": "esm2",
    "path": "./protein",
    "variant": "large",
    "vocab_size": 33
  },
  "training": {
    "joint_training": true,
    "batch_size": 8,
    "data_type": "transcript_specific_coding",
    "lowercase": true
  }
}