config.json

{
  "blacklist": null,
  "ckpt_interval": 100000,
  "column_remapping": null,
  "compile_model": true,
  "console_logging_steps": 2000,
  "conv_peak_encoder": false,
  "dim_feedforward": 1024,
  "dim_model": 768,
  "dropout": 0.1,
  "epochs": 30,
  "fp16": true,
  "grad_accumulation": 1,
  "gradient_clip_val": 10.0,
  "isotope_error_range": [
    0,
    1
  ],
  "lazy_loading": true,
  "learning_rate": 5e-05,
  "logger": null,
  "max_charge": 10,
  "max_iters": 3000000,
  "max_length": 40,
  "max_mz": 2500.0,
  "max_shard_size": 1000000,
  "min_intensity": 0.01,
  "min_mz": 50.0,
  "model_save_folder_path": "checkpoints/instanovo-extended-massivekb",
  "n_beams": 5,
  "n_gpu": 1,
  "n_head": 16,
  "n_layers": 9,
  "n_peaks": 200,
  "num_sanity_val_steps": 10,
  "perform_data_checks": true,
  "precursor_mass_tol": 50,
  "predict_batch_size": 64,
  "preshuffle_shards": true,
  "remove_precursor_tol": 2.0,
  "report_to": "neptune",
  "residue_remapping": null,
  "residues": {
    "A": 71.037114,
    "C": 103.009185,
    "C[UNIMOD:4]": 160.030649,
    "D": 115.026943,
    "E": 129.042593,
    "F": 147.068414,
    "G": 57.021464,
    "H": 137.058912,
    "I": 113.084064,
    "K": 128.094963,
    "L": 113.084064,
    "M": 131.040485,
    "M[UNIMOD:35]": 147.0354,
    "N": 114.042927,
    "N[UNIMOD:7]": 115.026943,
    "P": 97.052764,
    "Q": 128.058578,
    "Q[UNIMOD:7]": 129.042594,
    "R": 156.101111,
    "S": 87.032028,
    "S[UNIMOD:21]": 166.998028,
    "T": 101.04767,
    "T[UNIMOD:21]": 181.01367,
    "V": 99.068414,
    "W": 186.079313,
    "Y": 163.063329,
    "Y[UNIMOD:21]": 243.029329,
    "[UNIMOD:1]": 42.010565,
    "[UNIMOD:385]": -17.026549,
    "[UNIMOD:5]": 43.005814
  },
  "resume_checkpoint": null,
  "run_name": "instanovo_acpt_base",
  "save_model": true,
  "save_weights_only": false,
  "seed": 101,
  "tb_summarywriter": "s3://dtu-denovo-s-2e6da747d6d34f62-outputs/output/8cbd543a-56cb-4866-a257-731cc32e822b/tensorboard/",
  "tensorboard_logging_steps": 500,
  "train_batch_size": 128,
  "train_from_scratch": true,
  "train_partition": null,
  "train_path": "/mnt/instanovo-data-kyber/identity_splits_parquet/*/*-train-*.parquet",
  "train_subset": 1.0,
  "use_flash_attention": false,
  "use_shards": true,
  "val_check_interval": 1.0,
  "valid_partition": null,
  "valid_path": "/mnt/instanovo-data-kyber/identity_splits_parquet/*/*-valid-*.parquet",
  "valid_subset": 0.01,
  "valid_subset_of_train": null,
  "validate_precursor_mass": false,
  "verbose_loading": true,
  "warmup_iters": 100000,
  "weight_decay": 1e-05
}