pab1 · Fully connected

reference_pab1_fc_plm_contiguousf4_20261105-043801

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	captured_at_10	hit_rate_at_50	mean_at_50	max_at_50	precision_at_50	captured_at_50	hit_rate_at_100	mean_at_100	max_at_100	precision_at_100	captured_at_100	hit_rate_at_1pct	mean_at_1pct	max_at_1pct	precision_at_1pct	captured_at_1pct
train	0.4641	0.9441	0.9342	0.8883	1652.0000	0.0432	1175.0000	0.4000	-0.1473	0.4177	0.6629	0.5400	-0.0091	0.5644	0.0200	0.7620	0.6300	0.0300	0.5644	0.0400	0.8034	0.6333	0.0301	0.5644	0.0778	0.8282
validation	1.8542	0.7239	0.7280	0.4998	222.0000	-0.0307	1927.0000	0.2000	-0.2295	0.0634	0.7698	0.2000	-0.2974	0.3723	0.0400	0.7843	0.1800	-0.3779	0.3723	0.0600	0.7814	0.2041	-0.2995	0.3723	0.0408	0.7829
test	1.2885	0.7484	0.6188	0.3092	931.0000	0.0269	1199.0000	0.1000	-0.4591	0.0979	0.2682	0.2000	-0.3565	0.8601	0.0400	0.3680	0.2700	-0.2959	0.8601	0.0300	0.4346	0.1778	-0.3595	0.7491	0.0222	0.3624

Loss per epoch

training validation

Config

name: reference_pab1_fc_plm_contiguousf4
task: dms
data:
  dataset: pab1
  encoding: one_hot,aa_index,plm,plm_sum
  split_dir: null
  split_scheme: contiguous
  split_fold: 4
model:
  name: fcn
  hidden:
  - 100
  - 100
  - 100
  dropout: 0.2
train:
  epochs: 300
  batch_size: 128
  learning_rate: 0.001
  patience: 15
  min_delta: 1.0e-05
  seed: 7