bgl3 · Sequence CNN

reference_bgl3_cnn_mr5_randomf0_20261104-221613

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	0.5734	0.8360	0.8453	0.6946	134.0000	-0.0217	125.0000	0.4000	-0.0806	0.3940	0.1000	0.6833	0.2800	-0.1341	0.7042	0.1200	0.7715	0.2200	-0.1648	0.7042	0.1700	0.8047	0.2941	-0.1420	0.7042	0.1471	0.7398
validation	0.8339	0.7499	0.7667	0.5612	37.0000	0.1408	362.0000	0.3000	-0.1970	0.4856	0.3000	0.7067	0.1400	-0.2806	0.4856	0.2200	0.7733	0.1200	-0.3039	0.5578	0.3600	0.8123	0.3636	-0.1487	0.4856	0.3636	0.7335
test	0.9231	0.7027	0.7177	0.4916	46.0000	0.0087	359.0000	0.2000	-0.6130	0.1063	0.0000	0.4627	0.1000	-0.5523	0.2388	0.1000	0.5903	0.0800	-0.5188	0.4991	0.2600	0.6538	0.1818	-0.7981	0.1063	0.0000	0.3809

Loss per epoch

training validation

Config

name: reference_bgl3_cnn_mr5_randomf0
task: dms
data:
  dataset: bgl3
  encoding: one_hot,aa_index
  split_dir: null
  split_scheme: random
  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