bgl3 · Graph CNN

reference_bgl3_gcn_mr5plm_randomf1_20261104-222421

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	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	0.5141	0.8533	0.8616	0.7228	127.0000	0.0415	5.0000	0.3000	0.0328	0.9971	0.1000	0.7309	0.2000	-0.0577	0.9971	0.1000	0.8083	0.2100	-0.0925	0.9971	0.1700	0.8420	0.1765	-0.0582	0.9971	0.0588	0.7790
validation	0.9347	0.7124	0.7370	0.5054	53.0000	-0.1498	95.0000	0.2000	-0.3971	0.0327	0.0000	0.5766	0.1400	-0.3917	0.1391	0.1400	0.6894	0.1500	-0.3469	1.6908	0.2600	0.7614	0.1818	-0.3831	0.0327	0.0000	0.5894
test	0.8741	0.7349	0.7534	0.5400	37.0000	0.0389	538.0000	0.1000	-0.1593	0.3347	0.1000	0.7245	0.0800	-0.3754	0.3347	0.1200	0.7214	0.0600	-0.3678	0.4161	0.2600	0.7750	0.0909	-0.1688	0.3347	0.0909	0.7239

Loss per epoch

training validation

Config

name: reference_bgl3_gcn_mr5plm_randomf1
task: dms
data:
  dataset: bgl3
  encoding: one_hot,aa_index,plm,plm_sum
  split_dir: null
  graph_threshold: 6
  split_scheme: random
  split_fold: 1
  min_reads: 5
model:
  name: gcn
  filters:
  - 32
  hidden:
  - 100
  dropout: 0.2
train:
  epochs: 300
  batch_size: 128
  learning_rate: 0.0001
  patience: 15
  min_delta: 1.0e-05
  seed: 7