bgl3 · Sequence CNN

reference_bgl3_cnn_mr5_modulof0_20261104-221645

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	0.6815	0.8238	0.8166	0.5884	109.0000	0.0241	807.0000	0.5000	0.0170	0.3940	0.1000	0.7317	0.3400	-0.0672	0.7042	0.2000	0.7954	0.2900	-0.1586	0.7042	0.2800	0.7926	0.3103	-0.0832	0.3940	0.1724	0.7433
validation	0.9196	0.4758	0.4781	0.1852	47.0000	-0.0407	37.0000	0.3000	-0.0584	0.4856	0.1000	0.6072	0.2600	-0.2518	1.6908	0.2600	0.6364	0.1600	-0.2839	1.6908	0.3300	0.6748	0.1429	-0.1361	0.0294	0.0000	0.5168
test	1.0200	0.3425	0.3140	0.1166	49.0000	0.0664	170.0000	0.5000	-0.3045	0.3887	0.2000	0.5048	0.2400	-0.5998	0.3887	0.2600	0.3701	0.1700	-0.6615	0.3887	0.2900	0.3554	0.4286	-0.5039	0.3347	0.0000	0.3606

Loss per epoch

training validation

Config

name: reference_bgl3_cnn_mr5_modulof0
task: dms
data:
  dataset: bgl3
  encoding: one_hot,aa_index
  split_dir: null
  split_scheme: modulo
  split_fold: 0
  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