bgl3 · Sequence CNN

reference_bgl3_cnn_mr5plm_contiguousf1_20261104-222521

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	1.2135	0.6590	0.6930	0.3434	151.0000	0.1007	120.0000	0.4000	-0.1892	0.4991	0.1000	0.6154	0.3000	-0.1619	0.4991	0.1600	0.7328	0.3000	-0.1755	0.4991	0.2500	0.7773	0.3529	-0.1714	0.4991	0.1471	0.7027
validation	1.6808	0.3820	0.3558	0.1182	26.0000	-0.2178	670.0000	0.0000	-2.3461	-0.0342	0.0000	-0.2866	0.0200	-1.7271	0.0557	0.0400	0.0141	0.0300	-1.4870	0.1631	0.1000	0.1565	0.0000	-2.2406	-0.0342	0.0000	-0.2386
test	1.4189	0.6346	0.6666	0.0111	40.0000	0.0903	251.0000	0.1000	-0.5843	0.2007	0.1000	0.3883	0.1200	-0.4989	0.4008	0.1400	0.5430	0.1000	-0.4028	0.4008	0.2300	0.6678	0.1111	-0.6154	0.2007	0.0000	0.3639

Loss per epoch

training validation

Config

name: reference_bgl3_cnn_mr5plm_contiguousf1
task: dms
data:
  dataset: bgl3
  encoding: one_hot,aa_index,plm,plm_sum
  split_dir: null
  split_scheme: contiguous
  split_fold: 1
  min_reads: 5
model:
  name: cnn
  conv:
  - - 32
    - 17
  hidden:
  - 100
  dropout: 0.2
train:
  epochs: 300
  batch_size: 64
  learning_rate: 0.0001
  patience: 15
  min_delta: 1.0e-05
  seed: 7