irm_enhanced_diffusion_training.json

{
  "VOCAB_SIZE": 33,
  "NUM_LETTERS": 33,
  "PARSE_PROTEIN": 1,
  "PARSE_DNA": 1,
  "PARSE_RNA": 1,
  "PARSE_RNA_AS_DNA": 0,
  "NA_SHARED_TOKENS": 1,
  "NA_REF_ATOM": "C1'",
  "INCLUDE_PRED_NA_N": 1,
  "PROTEIN_BACKBONE_OCC_CUTOFF": 0.8,
  "PROTEIN_SIDE_CHAIN_OCC_CUTOFF": 0.5,
  "DNA_BACKBONE_OCC_CUTOFF": 0.8,
  "DNA_SIDE_CHAIN_OCC_CUTOFF": 0.5,
  "RNA_BACKBONE_OCC_CUTOFF": 0.8,
  "RNA_SIDE_CHAIN_OCC_CUTOFF": 0.5,
  "EXCLUDED_ELEMENTS": [1],
  "DATE_CUTOFF": "2030-01-01",
  "MAX_NUMBER_OF_PDBS_TRAIN": 50000,
  "MAX_NUMBER_OF_PDBS_VALID": 500,
  "BATCH_TOKENS": 6000,
  "LOSS_TOKENS": 10000,
  "LOSS_WEIGHT_STRATEGY": "snr",
  "LOSS_WEIGHT_EPSILON": 0.1,
  "LABEL_SMOOTHING": 0.0,
  "DIFFUSION_LABEL_SMOOTHING": 0.0,
  "EXCLUDE_RES" : ["HOH", "NA", "CL", "K", "BR"],
  "MIN_PROTEIN_LENGTH_CUTOFF": 1,
  "NUM_WORKERS": 4,
  "TOTAL_STEPS": 1800,
  "LR_SCHEDULER": "cosine_warmup",
  "LR_BASE": 1e-05,
  "LR_MIN": 1e-07,
  "LR_WARMUP_STEPS": 100,
  "RANDOMIZE_NMR_MODEL": 0,
  "CROP_LARGE_STRUCTURES": 1,
  "MIN_OVERLAP_LENGTH": 5,
  "DF_PATH_TRAIN": "data/naiad_dataset/train_dna_enhanced.csv",
  "DF_PATH_VALID": "data/naiad_dataset/valid_dna_enhanced.csv",
  "BASE_FOLDER": "models",
  "PREV_CHECKPOINT": "weight_candidates/s1804.pt",
  "BASELINE_CHECKPOINT": null,
  "RESET_TRAINING_STATE": true,
  "COMPARE_WITH_BASELINE": false,
  "HIDDEN_DIM": 128,
  "NUM_ENCODER_LAYERS": 3,
  "NUM_DECODER_LAYERS": 3,
  "NUM_NEIGHBORS": 32,
  "DROPOUT": 0.1,
  "USE_SEQUENCE_CONTEXT": true,
  "PROTEIN_BACKBONE_NOISE": 0.05,
  "DNA_BACKBONE_NOISE": 0.1,
  "RNA_BACKBONE_NOISE": 0.1,
  "MASK_MODE": "na_only",
  "MASK_NA_ONLY": true,
  "MASK_RATIO_MIN": 0.4,
  "MASK_RATIO_MAX": 1.0,
  "FULL_MASK_RATIO": 0.75,
  "MASK_RATIO_SCHEDULE": "uniform",
  "PARSE_PPMS": 0,
  "NA_ONLY_AS_UNIFORM_PPM": 0,
  "DROP_PROTEIN_PROBABILITY": 0,
  "PROTEIN_INTERFACE_RESIDUE_MUTATION_PROBABILITY": 0,
  "MUTATE_BASE_PAIR_TOGETHER": 0,
  "MUTATE_ENTIRE_SIDE_CHAIN_INTERFACE_PROBABILITY": 0,
  "NA_NON_INTERFACE_AS_UNIFORM_PPM": 0,
  "USE_IRM_PENALTY": false,
  "IRM_PENALTY_WEIGHT": 0.0,
  "IRM_MIN_TOKENS_PER_ENV": 8,
  "GRADIENT_NORM": 1.0,
  "MIXED_PRECISION": 1,
  "SAVE_EVERY_N_STEPS": 500,
  "ATOMS_TO_LOAD": "backbone",
  "METRICS_TO_COMPUTE": "basic",
  "USE_WANDB": false,
  "WANDB_ENTITY": null,
  "WANDB_PROJECT": "naiad_irm",
  "WANDB_RUN_NAME": "",
  "_comment": "",
  "_description": ""
}