bgl3 · Sequence CNN

reference_bgl3_cnn_mr5plm_modulof0_20261104-222247

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	0.5316	0.8285	0.8367	0.6789	109.0000	0.0025	516.0000	0.6000	0.0099	0.2595	0.0000	0.7280	0.3600	-0.0635	0.4161	0.2000	0.7975	0.2900	-0.1215	0.7042	0.2800	0.8160	0.3793	-0.0681	0.2792	0.0690	0.7517
validation	0.6741	0.6351	0.6591	0.4027	47.0000	-0.0850	67.0000	0.4000	-0.1117	0.2216	0.0000	0.5721	0.2800	-0.2264	0.2216	0.2800	0.6586	0.2200	-0.2352	1.6908	0.3800	0.7222	0.4286	-0.0743	0.2216	0.0000	0.5547
test	0.8128	0.5534	0.6228	0.2961	49.0000	0.1635	64.0000	0.3000	-0.0299	0.3887	0.2000	0.6903	0.2000	-0.2081	0.3887	0.2000	0.6913	0.1900	-0.3117	0.7630	0.3600	0.6742	0.4286	0.0697	0.3887	0.1429	0.7384

Loss per epoch

training validation

Config

name: reference_bgl3_cnn_mr5plm_modulof0
task: dms
data:
  dataset: bgl3
  encoding: one_hot,aa_index,plm,plm_sum
  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