eval_config.json

{
    "Description": "This is a pre-training configuration for codon optimization model based on BART language model.",
    "project_name": "COnTRA_SCPECBS3_wintest",
    "dataset_path": "/mount/COnTRA/data/datasets/processed_data_SCPECBS3/homologs/SCPECBS3_SCPECBS3_ExprRefined30",
    "tokenizer_path": "/mount/COnTRA/tokenizers/contra_tokenizer_gen_exprrefined",
    "cai_refference_path": "/mount/COnTRA/data/datasets/processed_data_SCPECBS3/S_cerevisiae/S_cerevisiae.0.nt.fasta",
    "checkpoint_flag": true,
    "checkpoint_path": "/mount/COnTRA/models/Finetuned2Steps_homologs_30_ExR/best_model/",
    "special_token_th": 42,
    "mask_all": false,
    "sw_aa_size": 30,
    "eval_type": "model",
    "model_type": "COBaBExRi",
    "cai_query_species": "S_cerevisiae",
    "outdir": "/mount/COnTRA/models/Finetuned2Steps_homologs_30_ExR/best_model/",
    "outfile": "model_eval_mimic.csv",
    "out_dict": [
        "prot_len",
        "num_of_correct_predicted_codons",
        "accuracy",
        "cross_entropy_loss",
        "entropy",
        "query_codons",
        "subject_codons",
        "pred_codons"
    ],
    "orig_dict": [
        "qseqid",
        "sseqid",
        "query_species",
        "subject_species",
        "pident",
        "length",
        "mismatch",
        "gapopen",
        "qstart",
        "qend",
        "sstart",
        "send",
        "evalue",
        "bitscore"
    ],
    "debug": false
}