bgl3 · Sequence CNN

reference_bgl3_cnn_mr5_modulof1_20261104-221850

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	0.6965	0.7925	0.7838	0.5618	136.0000	0.0374	1295.0000	0.3000	-0.2870	0.3347	0.0000	0.5472	0.3800	-0.1001	0.7042	0.2200	0.7416	0.2900	-0.1421	0.7042	0.2000	0.7719	0.4444	-0.1091	0.4161	0.1111	0.6916
validation	0.9800	0.4489	0.4500	0.1658	23.0000	-0.2901	4.0000	0.1000	-0.7732	0.3940	0.1000	0.2421	0.1200	-0.5004	0.3940	0.3200	0.5220	0.0800	-0.5036	0.3940	0.3600	0.5603	0.1250	-0.9458	0.3940	0.1250	0.1138
test	0.9670	0.4714	0.4686	0.1432	47.0000	-0.1013	164.0000	0.4000	-0.0338	0.4856	0.1000	0.6235	0.2600	-0.3045	0.9971	0.2600	0.5903	0.1500	-0.4417	0.9971	0.2700	0.5216	0.2857	-0.0770	0.1045	0.0000	0.5530

Loss per epoch

training validation

Config

name: reference_bgl3_cnn_mr5_modulof1
task: dms
data:
  dataset: bgl3
  encoding: one_hot,aa_index
  split_dir: null
  split_scheme: modulo
  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