bgl3 · Fully connected

reference_bgl3_fc_s5_20261104-140535

Metrics

Set	mse	pearsonr	spearmanr	r2	above_wt_n	above_wt_spearmanr	budget_to_max	hit_rate_at_10	mean_at_10	max_at_10	precision_at_10	tie_share_at_10	captured_at_10	hit_rate_at_50	mean_at_50	max_at_50	precision_at_50	tie_share_at_50	captured_at_50	hit_rate_at_100	mean_at_100	max_at_100	precision_at_100	tie_share_at_100	captured_at_100	hit_rate_at_1pct	mean_at_1pct	max_at_1pct	precision_at_1pct	tie_share_at_1pct	captured_at_1pct
test	1.3521	0.5395	0.4731	0.2885	1036.0000	0.1386	502.0000	0.7000	0.7856	1.6946	0.0000	0.0000	0.3854	0.7200	0.6900	2.3136	0.0200	0.0000	0.4046	0.6600	0.5896	2.6204	0.0700	0.0000	0.3936	0.7037	0.6948	1.8690	0.0000	0.0000	0.3831
train	1.2394	0.6135	0.5781	0.3697	8299.0000	0.1885	7209.0000	1.0000	1.4868	3.1160	0.1000	0.5195	0.8800	1.2110	3.1160	0.0200	0.4962	0.8700	1.1419	3.1160	0.0200	0.5023	0.8657	1.1016	3.1160	0.0417	0.5221
validation	1.4149	0.5355	0.4821	0.2842	933.0000	0.0103	1539.0000	0.9000	1.0312	1.6946	0.0000	0.0000	0.4685	0.7400	0.7139	2.0174	0.0200	0.0000	0.4145	0.7200	0.6035	2.8548	0.0800	0.0000	0.3998	0.8750	1.0250	2.0174	0.0000	0.0000	0.5012

Loss per epoch

training validation

Config

name: reference_bgl3_fc_s5
task: dms
data:
  dataset: bgl3
  encoding: one_hot,aa_index
  split_dir: bgl3/splits/main
model:
  name: fcn
  hidden:
  - 100
  - 100
  dropout: 0.2
train:
  epochs: 300
  batch_size: 32
  learning_rate: 0.0001
  patience: 15
  min_delta: 1.0e-05
  seed: 5