Protein Engineering

ube4b · Fully connected

reference_ube4b_fc_modulof0_20261104-144719

Metrics

Setmsepearsonrspearmanrr2above_wt_nabove_wt_spearmanrbudget_to_maxhit_rate_at_10mean_at_10max_at_10precision_at_10captured_at_10hit_rate_at_50mean_at_50max_at_50precision_at_50captured_at_50hit_rate_at_100mean_at_100max_at_100precision_at_100captured_at_100hit_rate_at_1pctmean_at_1pctmax_at_1pctprecision_at_1pctcaptured_at_1pct
train2.47200.63270.60080.39115520.00000.506423.00000.90004.93297.82590.20000.74190.96004.45078.62460.34000.78240.98004.19728.62460.37000.80040.96733.80268.62460.49800.8270
validation3.0956-0.0832-0.1095-0.0164571.00000.04351580.00000.1000-1.84250.14300.00000.0000-0.06900.1600-1.59213.56230.02000.0000-0.02790.1100-1.61073.56230.04000.0000-0.03720.1935-1.45473.56230.03230.00000.0055
test3.3218-0.0793-0.0905-0.0179645.00000.05371034.00000.2000-2.31563.41600.00000.0000-0.19100.2000-1.92813.41600.02000.0000-0.15400.1400-1.89033.41600.03000.0000-0.16670.2500-1.84993.41600.03570.0000-0.1252

Loss per epoch

training validation

Config

name: reference_ube4b_fc_modulof0
task: dms
data:
  dataset: ube4b
  encoding: one_hot,aa_index
  split_dir: null
  split_scheme: modulo
  split_fold: 0
model:
  name: fcn
  hidden:
  - 100
  - 100
  - 100
  dropout: 0.2
train:
  epochs: 300
  batch_size: 64
  learning_rate: 0.0001
  patience: 15
  min_delta: 1.0e-05
  seed: 7