bgl3 · Fully connected

reference_bgl3_fc_mr5plm_randomf2_20261104-222605

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	0.5092	0.8659	0.8712	0.7265	124.0000	0.0411	8.0000	0.4000	-0.0060	0.9971	0.1000	0.7102	0.3200	-0.0909	0.9971	0.1400	0.7828	0.3000	-0.1035	0.9971	0.2400	0.8306	0.3235	-0.0753	0.9971	0.1471	0.7602
validation	0.9124	0.7192	0.7375	0.5143	40.0000	-0.0579	170.0000	0.2000	-0.1478	0.3302	0.1000	0.7582	0.2400	-0.2183	0.3302	0.2800	0.8152	0.1900	-0.2979	0.3302	0.4400	0.8166	0.2727	-0.1316	0.3302	0.0909	0.7704
test	0.9071	0.7238	0.7449	0.5200	53.0000	0.0333	73.0000	0.2000	-0.5170	0.1164	0.0000	0.5228	0.2400	-0.4012	0.7042	0.2400	0.6842	0.2500	-0.3258	1.6908	0.3800	0.7735	0.1818	-0.5745	0.1164	0.0000	0.5025

Loss per epoch

training validation

Config

name: reference_bgl3_fc_mr5plm_randomf2
task: dms
data:
  dataset: bgl3
  encoding: one_hot,aa_index,plm,plm_sum
  split_dir: null
  split_scheme: random
  split_fold: 2
  min_reads: 5
model:
  name: fcn
  hidden:
  - 100
  - 100
  dropout: 0.2
train:
  epochs: 300
  batch_size: 32
  learning_rate: 0.0001
  patience: 15
  min_delta: 1.0e-05
  seed: 7