Hugging Face's logo

Mahdip72
/
prot2token

Model card Files
xet
Community
Mahdip72 commited on
Commit
e381a82
·
verified ·
1 Parent(s): a72c22a

Upload 4 files

Browse files
Files changed (4) hide show
  1. stability/2024-07-05__17-35-31/checkpoints/best_valid_stability_spearman.pth +3 -0
  2. stability/2024-07-05__17-35-31/config.yaml +119 -0
  3. stability/2024-07-05__17-35-31/decoder_tokenizer.yaml +15 -0
  4. stability/2024-07-05__17-35-31/logs.txt +224 -0
stability/2024-07-05__17-35-31/checkpoints/best_valid_stability_spearman.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:838561f19ffb3237af9ccf5fb7c15483e79b401c0e5187a5a010218c71186cda
3
+ size 4045597881
stability/2024-07-05__17-35-31/config.yaml ADDED
@@ -0,0 +1,119 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ fix_seed: 0
2
+ checkpoints_every: 64
3
+ tensorboard_log: True
4
+ tqdm_progress_bar: False
5
+ result_path: ./results/stability/
6
+
7
+ resume:
8
+ resume: False
9
+ resume_path: /path/to/checkpoint.pth
10
+ restart_optimizer: True
11
+ handle_shape_missmatch: True
12
+
13
+ prot2token_model:
14
+ compile_model: True
15
+ positional_encoding_type: learned # absolute, learned
16
+ protein_encoder:
17
+ model_name: facebook/esm2_t33_650M_UR50D # facebook/esm2_t33_650M_UR50D, facebook/esm2_t30_150M_UR50D, facebook/esm2_t12_35M_UR50D, facebook/esm2_t6_8M_UR50D, Rostlab/prot_t5_base_mt_uniref50
18
+ max_len: 1024
19
+ drop_positional_encoding: True
20
+ quantization_4_bit: False # use with tune_embedding enabled
21
+ tune_embedding: False
22
+ fine_tune:
23
+ enable: True
24
+ last_layers_trainable: 6
25
+ lora:
26
+ enable: False
27
+ r: 8
28
+ lora_alpha: 32
29
+ lora_dropout: 0.05
30
+ molecule_encoder:
31
+ enable: False
32
+ model_name: gayane/BARTSmiles
33
+ max_len: 128
34
+ drop_positional_encoding: True
35
+ tune_embedding: False
36
+ fine_tune:
37
+ enable: False
38
+ last_layers_trainable: 2
39
+ decoder:
40
+ dimension: 640
41
+ dim_feedforward: 1280
42
+ num_heads: 8
43
+ num_layers: 8
44
+ max_len: 32
45
+ activation_function: gelu
46
+
47
+ train_settings:
48
+ skip: False
49
+ data_path: /mnt/hdd8/mehdi/datasets/Joint_training/
50
+ num_epochs: 48
51
+ start_metric_epoch: 2
52
+ shuffle: True
53
+ loss: crossentropy # crossentropy or focal
54
+ sample_weight: True
55
+ task_weight: False
56
+ mixed_precision: bf16 # no, fp16, bf16, fp8
57
+ device: cuda
58
+ batch_size: 16
59
+ random_masking: 0.15 # 0 to 1.0
60
+ num_workers: 0
61
+ grad_accumulation: 32
62
+ max_task_samples: 250000
63
+
64
+ valid_settings:
65
+ data_path: /mnt/hdd8/mehdi/datasets/Joint_training/
66
+ do_every: 1
67
+ batch_size: 1
68
+ num_workers: 0
69
+
70
+ test_settings:
71
+ enable: True
72
+ data_path: /mnt/hdd8/mehdi/datasets/Joint_training/
73
+ batch_size: 1
74
+ num_workers: 0
75
+ monitoring_metrics:
76
+ fluorescence: spearman
77
+ stability: spearman
78
+ protein_ligand_affinity: rmse
79
+ human_ppi: f1
80
+ structure_similarity: spearman
81
+ protein_protein_interface: auc
82
+ enzyme_reaction: f1
83
+ fold: f1
84
+ localization_deeploc: macro_f1
85
+ secondary_structure: accuracy
86
+
87
+ optimizer:
88
+ name: adam
89
+ lr: 5e-5
90
+ weight_decouple: True
91
+ weight_decay: 1e-2
92
+ eps: 1e-16
93
+ beta_1: 0.9
94
+ beta_2: 0.999
95
+ use_8bit_adam: False
96
+ grad_clip_norm: 1
97
+ decay:
98
+ warmup: 256
99
+ min_lr: 1e-6
100
+ gamma: 0.2
101
+ num_restarts: 1
102
+
103
+ tasks:
104
+ phosphorylation: False
105
+ localization: False
106
+ localization_deeploc: False
107
+ fold: False
108
+ enzyme_reaction: False
109
+ human_ppi: False
110
+ structure_similarity: False
111
+ protein_protein_interface: False # decoder max length: 210
112
+ protein_ligand_affinity: False
113
+ fluorescence: False
114
+ stability: True
115
+ amino_to_fold_seek: False # decoder max length: max input length + 1
116
+ secondary_structure: False # decoder max length: max input length + 1
117
+ gene_ontology: False # decoder max length: 625
118
+ enzyme_commission: False # decoder max length: 47
119
+ auxiliary: False
stability/2024-07-05__17-35-31/decoder_tokenizer.yaml ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ .: 14
2
+ '0': 4
3
+ '1': 5
4
+ '2': 6
5
+ '3': 7
6
+ '4': 8
7
+ '5': 9
8
+ '6': 10
9
+ '7': 11
10
+ '8': 12
11
+ '9': 13
12
+ <bos>: 1
13
+ <eos>: 2
14
+ <pad>: 0
15
+ <task_stability>: 3
stability/2024-07-05__17-35-31/logs.txt ADDED
@@ -0,0 +1,224 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2024-07-05 17:35:31,450 - <task_stability>: remaining samples: 53614
2
+ 2024-07-05 17:35:31,744 - <task_stability>: remaining samples: 2512
3
+ 2024-07-05 17:35:31,781 - <task_stability>: remaining samples: 12851
4
+ 2024-07-05 17:35:31,782 - preparing dataloaders are done
5
+ 2024-07-05 17:35:33,302 - protein encoder trainable params: 118,887,040 || all params: 653,173,781 || trainable%: 18.20144094240672
6
+ 2024-07-05 17:35:33,396 - decoder trainable params: 40,103,055 || all params: 40,103,055 || trainable%: 100.0
7
+ 2024-07-05 17:35:33,397 - supermodel trainable params: 158,990,095 || all params: 693,276,836 || trainable%: 22.93313244350198
8
+ 2024-07-05 17:35:33,397 - preparing model is done
9
+ 2024-07-05 17:35:33,399 - preparing optimizer is done
10
+ 2024-07-05 17:35:37,350 - compile model is done
11
+ 2024-07-05 17:35:37,352 - number of train steps per epoch: 105
12
+ 2024-07-05 17:35:37,352 - number of stability valid steps per epoch: 2512
13
+ 2024-07-05 18:05:09,177 - epoch 1/48 - steps 105, - time 1771.82s, train loss 0.381185
14
+ 2024-07-05 18:31:11,659 - epoch 2/48 - steps 105, - time 1562.48s, train loss 0.26480179
15
+ 2024-07-05 18:57:15,738 - epoch 3/48 - steps 105, - time 1563.72s, train loss 0.23365109
16
+ 2024-07-05 18:57:15,752 - train stability spearman p 0.0302, mae 0.0882, number of correct prediction 53149
17
+ 2024-07-05 19:01:15,160 - evaluation dataset 1 - steps 2512 - time 239.36s, valid loss 0.33956882
18
+ 2024-07-05 19:01:15,160 - valid stability spearman p 0.6081, mae 0.0689, number of correct prediction 2512
19
+ 2024-07-05 19:01:20,149 - new best spearman for stability: 0.6081
20
+ 2024-07-05 19:01:20,149 - saving the best model in /mnt/hdd8/mehdi/projects/prot2token/results/stability/2024-07-05__17-35-31/checkpoints/best_valid_stability_spearman.pth
21
+ 2024-07-05 19:28:23,082 - epoch 4/48 - steps 105, - time 1622.54s, train loss 0.20023499
22
+ 2024-07-05 19:28:23,082 - train stability spearman p 0.0698, mae 0.086, number of correct prediction 53182
23
+ 2024-07-05 19:31:30,154 - evaluation dataset 1 - steps 2512 - time 187.04s, valid loss 0.40229009
24
+ 2024-07-05 19:31:30,154 - valid stability spearman p 0.6449, mae 0.0649, number of correct prediction 2512
25
+ 2024-07-05 19:33:34,804 - new best spearman for stability: 0.6449
26
+ 2024-07-05 19:33:34,805 - saving the best model in /mnt/hdd8/mehdi/projects/prot2token/results/stability/2024-07-05__17-35-31/checkpoints/best_valid_stability_spearman.pth
27
+ 2024-07-05 20:01:54,924 - epoch 5/48 - steps 105, - time 1699.73s, train loss 0.18428825
28
+ 2024-07-05 20:01:54,935 - train stability spearman p 0.1295, mae 0.0843, number of correct prediction 53204
29
+ 2024-07-05 20:05:01,870 - evaluation dataset 1 - steps 2512 - time 186.91s, valid loss 0.44731443
30
+ 2024-07-05 20:05:01,871 - valid stability spearman p 0.7055, mae 0.0594, number of correct prediction 2512
31
+ 2024-07-05 20:05:22,357 - new best spearman for stability: 0.7055
32
+ 2024-07-05 20:05:22,358 - saving the best model in /mnt/hdd8/mehdi/projects/prot2token/results/stability/2024-07-05__17-35-31/checkpoints/best_valid_stability_spearman.pth
33
+ 2024-07-05 20:32:13,923 - epoch 6/48 - steps 105, - time 1611.15s, train loss 0.17623341
34
+ 2024-07-05 20:32:13,923 - train stability spearman p 0.2035, mae 0.0829, number of correct prediction 53231
35
+ 2024-07-05 20:35:12,223 - evaluation dataset 1 - steps 2512 - time 178.27s, valid loss 0.46854883
36
+ 2024-07-05 20:35:12,224 - valid stability spearman p 0.7019, mae 0.0642, number of correct prediction 2512
37
+ 2024-07-05 21:00:46,165 - epoch 7/48 - steps 105, - time 1533.53s, train loss 0.17145256
38
+ 2024-07-05 21:00:46,165 - train stability spearman p 0.2589, mae 0.0811, number of correct prediction 53241
39
+ 2024-07-05 21:03:44,364 - evaluation dataset 1 - steps 2512 - time 178.17s, valid loss 0.50273507
40
+ 2024-07-05 21:03:44,364 - valid stability spearman p 0.712, mae 0.0584, number of correct prediction 2512
41
+ 2024-07-05 21:04:04,732 - new best spearman for stability: 0.712
42
+ 2024-07-05 21:04:04,732 - saving the best model in /mnt/hdd8/mehdi/projects/prot2token/results/stability/2024-07-05__17-35-31/checkpoints/best_valid_stability_spearman.pth
43
+ 2024-07-05 21:29:37,243 - epoch 8/48 - steps 105, - time 1532.1s, train loss 0.16834861
44
+ 2024-07-05 21:29:37,244 - train stability spearman p 0.2966, mae 0.0798, number of correct prediction 53239
45
+ 2024-07-05 21:32:34,979 - evaluation dataset 1 - steps 2512 - time 177.7s, valid loss 0.49872354
46
+ 2024-07-05 21:32:34,979 - valid stability spearman p 0.7226, mae 0.0581, number of correct prediction 2512
47
+ 2024-07-05 21:32:55,539 - new best spearman for stability: 0.7226
48
+ 2024-07-05 21:32:55,539 - saving the best model in /mnt/hdd8/mehdi/projects/prot2token/results/stability/2024-07-05__17-35-31/checkpoints/best_valid_stability_spearman.pth
49
+ 2024-07-05 21:58:29,621 - epoch 9/48 - steps 105, - time 1533.67s, train loss 0.16531481
50
+ 2024-07-05 21:58:29,621 - train stability spearman p 0.3378, mae 0.0776, number of correct prediction 53276
51
+ 2024-07-05 22:01:26,298 - evaluation dataset 1 - steps 2512 - time 176.65s, valid loss 0.51312769
52
+ 2024-07-05 22:01:26,298 - valid stability spearman p 0.7233, mae 0.0663, number of correct prediction 2512
53
+ 2024-07-05 22:01:46,973 - new best spearman for stability: 0.7233
54
+ 2024-07-05 22:01:46,973 - saving the best model in /mnt/hdd8/mehdi/projects/prot2token/results/stability/2024-07-05__17-35-31/checkpoints/best_valid_stability_spearman.pth
55
+ 2024-07-05 22:27:21,344 - epoch 10/48 - steps 105, - time 1533.94s, train loss 0.16316066
56
+ 2024-07-05 22:27:21,344 - train stability spearman p 0.3724, mae 0.0764, number of correct prediction 53262
57
+ 2024-07-05 22:30:18,295 - evaluation dataset 1 - steps 2512 - time 176.92s, valid loss 0.54175883
58
+ 2024-07-05 22:30:18,295 - valid stability spearman p 0.7437, mae 0.054, number of correct prediction 2512
59
+ 2024-07-05 22:30:38,703 - new best spearman for stability: 0.7437
60
+ 2024-07-05 22:30:38,704 - saving the best model in /mnt/hdd8/mehdi/projects/prot2token/results/stability/2024-07-05__17-35-31/checkpoints/best_valid_stability_spearman.pth
61
+ 2024-07-05 22:56:12,609 - epoch 11/48 - steps 105, - time 1533.48s, train loss 0.16117712
62
+ 2024-07-05 22:56:12,609 - train stability spearman p 0.4078, mae 0.0742, number of correct prediction 53271
63
+ 2024-07-05 22:59:09,976 - evaluation dataset 1 - steps 2512 - time 177.34s, valid loss 0.55451429
64
+ 2024-07-05 22:59:09,977 - valid stability spearman p 0.7354, mae 0.0547, number of correct prediction 2512
65
+ 2024-07-05 23:24:41,255 - epoch 12/48 - steps 105, - time 1530.85s, train loss 0.15981591
66
+ 2024-07-05 23:24:41,255 - train stability spearman p 0.4316, mae 0.0727, number of correct prediction 53269
67
+ 2024-07-05 23:27:38,432 - evaluation dataset 1 - steps 2512 - time 177.14s, valid loss 0.53740785
68
+ 2024-07-05 23:27:38,432 - valid stability spearman p 0.7315, mae 0.0561, number of correct prediction 2512
69
+ 2024-07-05 23:54:00,607 - epoch 13/48 - steps 105, - time 1581.75s, train loss 0.15810851
70
+ 2024-07-05 23:54:00,608 - train stability spearman p 0.4586, mae 0.0709, number of correct prediction 53294
71
+ 2024-07-05 23:56:56,755 - evaluation dataset 1 - steps 2512 - time 176.11s, valid loss 0.57103281
72
+ 2024-07-05 23:56:56,755 - valid stability spearman p 0.7281, mae 0.0544, number of correct prediction 2512
73
+ 2024-07-06 00:22:39,173 - epoch 14/48 - steps 105, - time 1541.97s, train loss 0.15659422
74
+ 2024-07-06 00:22:39,174 - train stability spearman p 0.4792, mae 0.0695, number of correct prediction 53290
75
+ 2024-07-06 00:25:30,633 - evaluation dataset 1 - steps 2512 - time 171.43s, valid loss 0.58697268
76
+ 2024-07-06 00:25:30,633 - valid stability spearman p 0.7403, mae 0.0551, number of correct prediction 2512
77
+ 2024-07-06 00:51:12,854 - epoch 15/48 - steps 105, - time 1541.69s, train loss 0.15534321
78
+ 2024-07-06 00:51:12,854 - train stability spearman p 0.5026, mae 0.0675, number of correct prediction 53310
79
+ 2024-07-06 00:54:10,545 - evaluation dataset 1 - steps 2512 - time 177.66s, valid loss 0.56736496
80
+ 2024-07-06 00:54:10,545 - valid stability spearman p 0.7459, mae 0.0528, number of correct prediction 2512
81
+ 2024-07-06 00:54:29,250 - new best spearman for stability: 0.7459
82
+ 2024-07-06 00:54:29,252 - saving the best model in /mnt/hdd8/mehdi/projects/prot2token/results/stability/2024-07-05__17-35-31/checkpoints/best_valid_stability_spearman.pth
83
+ 2024-07-06 01:20:12,180 - epoch 16/48 - steps 105, - time 1542.49s, train loss 0.15383106
84
+ 2024-07-06 01:20:12,181 - train stability spearman p 0.5228, mae 0.0662, number of correct prediction 53276
85
+ 2024-07-06 01:23:25,122 - evaluation dataset 1 - steps 2512 - time 192.78s, valid loss 0.591676
86
+ 2024-07-06 01:23:25,122 - valid stability spearman p 0.7357, mae 0.053, number of correct prediction 2512
87
+ 2024-07-06 01:49:12,194 - epoch 17/48 - steps 105, - time 1546.63s, train loss 0.15279829
88
+ 2024-07-06 01:49:12,194 - train stability spearman p 0.5419, mae 0.0646, number of correct prediction 53304
89
+ 2024-07-06 01:52:08,951 - evaluation dataset 1 - steps 2512 - time 176.72s, valid loss 0.59653156
90
+ 2024-07-06 01:52:08,952 - valid stability spearman p 0.7379, mae 0.0531, number of correct prediction 2512
91
+ 2024-07-06 02:17:50,682 - epoch 18/48 - steps 105, - time 1541.29s, train loss 0.15158923
92
+ 2024-07-06 02:17:50,682 - train stability spearman p 0.5574, mae 0.0631, number of correct prediction 53302
93
+ 2024-07-06 02:20:48,944 - evaluation dataset 1 - steps 2512 - time 178.23s, valid loss 0.592355
94
+ 2024-07-06 02:20:48,944 - valid stability spearman p 0.7279, mae 0.0552, number of correct prediction 2512
95
+ 2024-07-06 02:46:33,038 - epoch 19/48 - steps 105, - time 1543.65s, train loss 0.15060912
96
+ 2024-07-06 02:46:33,038 - train stability spearman p 0.5698, mae 0.0619, number of correct prediction 53298
97
+ 2024-07-06 02:49:30,836 - evaluation dataset 1 - steps 2512 - time 177.76s, valid loss 0.59448483
98
+ 2024-07-06 02:49:30,837 - valid stability spearman p 0.7466, mae 0.0526, number of correct prediction 2512
99
+ 2024-07-06 02:49:49,121 - new best spearman for stability: 0.7466
100
+ 2024-07-06 02:49:49,122 - saving the best model in /mnt/hdd8/mehdi/projects/prot2token/results/stability/2024-07-05__17-35-31/checkpoints/best_valid_stability_spearman.pth
101
+ 2024-07-06 03:15:31,367 - epoch 20/48 - steps 105, - time 1541.8s, train loss 0.14970614
102
+ 2024-07-06 03:15:31,367 - train stability spearman p 0.5792, mae 0.0615, number of correct prediction 53304
103
+ 2024-07-06 03:18:29,323 - evaluation dataset 1 - steps 2512 - time 177.92s, valid loss 0.59908577
104
+ 2024-07-06 03:18:29,324 - valid stability spearman p 0.7392, mae 0.053, number of correct prediction 2512
105
+ 2024-07-06 03:44:15,622 - epoch 21/48 - steps 105, - time 1545.85s, train loss 0.14841781
106
+ 2024-07-06 03:44:15,622 - train stability spearman p 0.5938, mae 0.0597, number of correct prediction 53301
107
+ 2024-07-06 03:47:11,424 - evaluation dataset 1 - steps 2512 - time 175.77s, valid loss 0.59845865
108
+ 2024-07-06 03:47:11,424 - valid stability spearman p 0.7318, mae 0.0547, number of correct prediction 2512
109
+ 2024-07-06 04:12:55,730 - epoch 22/48 - steps 105, - time 1543.86s, train loss 0.14786205
110
+ 2024-07-06 04:12:55,730 - train stability spearman p 0.5975, mae 0.0592, number of correct prediction 53318
111
+ 2024-07-06 04:15:51,892 - evaluation dataset 1 - steps 2512 - time 176.13s, valid loss 0.61498887
112
+ 2024-07-06 04:15:51,892 - valid stability spearman p 0.7301, mae 0.0541, number of correct prediction 2512
113
+ 2024-07-06 04:41:36,075 - epoch 23/48 - steps 105, - time 1543.73s, train loss 0.14677871
114
+ 2024-07-06 04:41:36,075 - train stability spearman p 0.6106, mae 0.058, number of correct prediction 53307
115
+ 2024-07-06 04:44:28,976 - evaluation dataset 1 - steps 2512 - time 172.87s, valid loss 0.6000444
116
+ 2024-07-06 04:44:28,976 - valid stability spearman p 0.7346, mae 0.0541, number of correct prediction 2512
117
+ 2024-07-06 05:10:14,474 - epoch 24/48 - steps 105, - time 1545.04s, train loss 0.14593586
118
+ 2024-07-06 05:10:14,475 - train stability spearman p 0.6187, mae 0.0572, number of correct prediction 53306
119
+ 2024-07-06 05:13:12,427 - evaluation dataset 1 - steps 2512 - time 177.92s, valid loss 0.61570224
120
+ 2024-07-06 05:13:12,427 - valid stability spearman p 0.7339, mae 0.0533, number of correct prediction 2512
121
+ 2024-07-06 05:38:57,006 - epoch 25/48 - steps 105, - time 1544.13s, train loss 0.14512503
122
+ 2024-07-06 05:38:57,006 - train stability spearman p 0.623, mae 0.0566, number of correct prediction 53299
123
+ 2024-07-06 05:41:52,087 - evaluation dataset 1 - steps 2512 - time 175.05s, valid loss 0.63505717
124
+ 2024-07-06 05:41:52,088 - valid stability spearman p 0.7261, mae 0.056, number of correct prediction 2512
125
+ 2024-07-06 06:07:38,515 - epoch 26/48 - steps 105, - time 1545.97s, train loss 0.144289
126
+ 2024-07-06 06:07:38,515 - train stability spearman p 0.6285, mae 0.0559, number of correct prediction 53289
127
+ 2024-07-06 06:10:33,061 - evaluation dataset 1 - steps 2512 - time 174.51s, valid loss 0.61991238
128
+ 2024-07-06 06:10:33,061 - valid stability spearman p 0.731, mae 0.054, number of correct prediction 2512
129
+ 2024-07-06 06:36:17,122 - epoch 27/48 - steps 105, - time 1543.61s, train loss 0.14366293
130
+ 2024-07-06 06:36:17,122 - train stability spearman p 0.638, mae 0.0552, number of correct prediction 53314
131
+ 2024-07-06 06:39:15,832 - evaluation dataset 1 - steps 2512 - time 178.67s, valid loss 0.61764782
132
+ 2024-07-06 06:39:15,832 - valid stability spearman p 0.7258, mae 0.0547, number of correct prediction 2512
133
+ 2024-07-06 07:05:03,636 - epoch 28/48 - steps 105, - time 1547.35s, train loss 0.14253053
134
+ 2024-07-06 07:05:03,637 - train stability spearman p 0.6508, mae 0.0536, number of correct prediction 53320
135
+ 2024-07-06 07:08:02,321 - evaluation dataset 1 - steps 2512 - time 178.65s, valid loss 0.6193421
136
+ 2024-07-06 07:08:02,321 - valid stability spearman p 0.7204, mae 0.0551, number of correct prediction 2512
137
+ 2024-07-06 07:33:51,510 - epoch 29/48 - steps 105, - time 1548.74s, train loss 0.14204362
138
+ 2024-07-06 07:33:51,511 - train stability spearman p 0.6557, mae 0.0532, number of correct prediction 53301
139
+ 2024-07-06 07:36:47,359 - evaluation dataset 1 - steps 2512 - time 175.81s, valid loss 0.62733793
140
+ 2024-07-06 07:36:47,360 - valid stability spearman p 0.7292, mae 0.054, number of correct prediction 2512
141
+ 2024-07-06 08:02:33,704 - epoch 30/48 - steps 105, - time 1545.89s, train loss 0.14151941
142
+ 2024-07-06 08:02:33,705 - train stability spearman p 0.6556, mae 0.0532, number of correct prediction 53318
143
+ 2024-07-06 08:05:31,549 - evaluation dataset 1 - steps 2512 - time 177.8s, valid loss 0.63140021
144
+ 2024-07-06 08:05:31,550 - valid stability spearman p 0.73, mae 0.0545, number of correct prediction 2512
145
+ 2024-07-06 08:31:12,091 - epoch 31/48 - steps 105, - time 1540.08s, train loss 0.14072536
146
+ 2024-07-06 08:31:12,092 - train stability spearman p 0.6657, mae 0.0521, number of correct prediction 53324
147
+ 2024-07-06 08:34:08,121 - evaluation dataset 1 - steps 2512 - time 175.99s, valid loss 0.62926954
148
+ 2024-07-06 08:34:08,121 - valid stability spearman p 0.7225, mae 0.0551, number of correct prediction 2512
149
+ 2024-07-06 08:59:46,194 - epoch 32/48 - steps 105, - time 1537.62s, train loss 0.14034777
150
+ 2024-07-06 08:59:46,194 - train stability spearman p 0.6695, mae 0.0517, number of correct prediction 53295
151
+ 2024-07-06 09:02:43,284 - evaluation dataset 1 - steps 2512 - time 177.05s, valid loss 0.62579863
152
+ 2024-07-06 09:02:43,285 - valid stability spearman p 0.7206, mae 0.0555, number of correct prediction 2512
153
+ 2024-07-06 09:28:22,049 - epoch 33/48 - steps 105, - time 1538.3s, train loss 0.13980339
154
+ 2024-07-06 09:28:22,050 - train stability spearman p 0.6737, mae 0.0513, number of correct prediction 53305
155
+ 2024-07-06 09:31:15,518 - evaluation dataset 1 - steps 2512 - time 173.43s, valid loss 0.63753542
156
+ 2024-07-06 09:31:15,518 - valid stability spearman p 0.7256, mae 0.0544, number of correct prediction 2512
157
+ 2024-07-06 09:56:52,535 - epoch 34/48 - steps 105, - time 1536.56s, train loss 0.13909172
158
+ 2024-07-06 09:56:52,535 - train stability spearman p 0.6766, mae 0.0506, number of correct prediction 53312
159
+ 2024-07-06 09:59:46,873 - evaluation dataset 1 - steps 2512 - time 174.3s, valid loss 0.63789942
160
+ 2024-07-06 09:59:46,873 - valid stability spearman p 0.7275, mae 0.0545, number of correct prediction 2512
161
+ 2024-07-06 10:25:27,569 - epoch 35/48 - steps 105, - time 1540.24s, train loss 0.13869807
162
+ 2024-07-06 10:25:27,569 - train stability spearman p 0.6791, mae 0.0503, number of correct prediction 53313
163
+ 2024-07-06 10:28:22,042 - evaluation dataset 1 - steps 2512 - time 174.43s, valid loss 0.6326882
164
+ 2024-07-06 10:28:22,043 - valid stability spearman p 0.7234, mae 0.0548, number of correct prediction 2512
165
+ 2024-07-06 10:54:00,461 - epoch 36/48 - steps 105, - time 1537.96s, train loss 0.13812206
166
+ 2024-07-06 10:54:00,462 - train stability spearman p 0.6831, mae 0.0501, number of correct prediction 53320
167
+ 2024-07-06 10:56:55,864 - evaluation dataset 1 - steps 2512 - time 175.36s, valid loss 0.64251012
168
+ 2024-07-06 10:56:55,864 - valid stability spearman p 0.7197, mae 0.0548, number of correct prediction 2512
169
+ 2024-07-06 11:22:35,719 - epoch 37/48 - steps 105, - time 1539.39s, train loss 0.13756861
170
+ 2024-07-06 11:22:35,719 - train stability spearman p 0.6922, mae 0.0491, number of correct prediction 53355
171
+ 2024-07-06 11:25:31,957 - evaluation dataset 1 - steps 2512 - time 176.2s, valid loss 0.64436017
172
+ 2024-07-06 11:25:31,957 - valid stability spearman p 0.7115, mae 0.0561, number of correct prediction 2512
173
+ 2024-07-06 11:51:10,099 - epoch 38/48 - steps 105, - time 1537.69s, train loss 0.13751318
174
+ 2024-07-06 11:51:10,100 - train stability spearman p 0.686, mae 0.0495, number of correct prediction 53306
175
+ 2024-07-06 11:54:04,048 - evaluation dataset 1 - steps 2512 - time 173.91s, valid loss 0.64465082
176
+ 2024-07-06 11:54:04,049 - valid stability spearman p 0.7256, mae 0.0546, number of correct prediction 2512
177
+ 2024-07-06 12:19:41,515 - epoch 39/48 - steps 105, - time 1537.02s, train loss 0.13719218
178
+ 2024-07-06 12:19:41,515 - train stability spearman p 0.6877, mae 0.0492, number of correct prediction 53342
179
+ 2024-07-06 12:22:36,967 - evaluation dataset 1 - steps 2512 - time 175.41s, valid loss 0.65021068
180
+ 2024-07-06 12:22:36,968 - valid stability spearman p 0.7238, mae 0.055, number of correct prediction 2512
181
+ 2024-07-06 12:48:17,724 - epoch 40/48 - steps 105, - time 1540.3s, train loss 0.13661633
182
+ 2024-07-06 12:48:17,724 - train stability spearman p 0.6949, mae 0.0486, number of correct prediction 53314
183
+ 2024-07-06 12:51:12,931 - evaluation dataset 1 - steps 2512 - time 175.17s, valid loss 0.652825
184
+ 2024-07-06 12:51:12,931 - valid stability spearman p 0.7211, mae 0.0547, number of correct prediction 2512
185
+ 2024-07-06 13:16:53,360 - epoch 41/48 - steps 105, - time 1539.97s, train loss 0.13669245
186
+ 2024-07-06 13:16:53,360 - train stability spearman p 0.6922, mae 0.0489, number of correct prediction 53324
187
+ 2024-07-06 13:19:48,674 - evaluation dataset 1 - steps 2512 - time 175.27s, valid loss 0.64914848
188
+ 2024-07-06 13:19:48,675 - valid stability spearman p 0.7211, mae 0.055, number of correct prediction 2512
189
+ 2024-07-06 13:45:29,880 - epoch 42/48 - steps 105, - time 1540.73s, train loss 0.13632903
190
+ 2024-07-06 13:45:29,880 - train stability spearman p 0.6963, mae 0.0485, number of correct prediction 53351
191
+ 2024-07-06 13:48:25,810 - evaluation dataset 1 - steps 2512 - time 175.89s, valid loss 0.64850231
192
+ 2024-07-06 13:48:25,810 - valid stability spearman p 0.7226, mae 0.0547, number of correct prediction 2512
193
+ 2024-07-06 14:14:02,505 - epoch 43/48 - steps 105, - time 1536.22s, train loss 0.13618307
194
+ 2024-07-06 14:14:02,505 - train stability spearman p 0.6991, mae 0.0483, number of correct prediction 53298
195
+ 2024-07-06 14:16:58,703 - evaluation dataset 1 - steps 2512 - time 176.16s, valid loss 0.65442398
196
+ 2024-07-06 14:16:58,704 - valid stability spearman p 0.7183, mae 0.055, number of correct prediction 2512
197
+ 2024-07-06 14:42:39,325 - epoch 44/48 - steps 105, - time 1540.16s, train loss 0.1359691
198
+ 2024-07-06 14:42:39,325 - train stability spearman p 0.6957, mae 0.0484, number of correct prediction 53315
199
+ 2024-07-06 14:45:35,365 - evaluation dataset 1 - steps 2512 - time 176.0s, valid loss 0.65047561
200
+ 2024-07-06 14:45:35,365 - valid stability spearman p 0.7233, mae 0.0547, number of correct prediction 2512
201
+ 2024-07-06 15:11:12,465 - epoch 45/48 - steps 105, - time 1536.65s, train loss 0.13577146
202
+ 2024-07-06 15:11:12,465 - train stability spearman p 0.7031, mae 0.0477, number of correct prediction 53323
203
+ 2024-07-06 15:14:07,831 - evaluation dataset 1 - steps 2512 - time 175.32s, valid loss 0.6487082
204
+ 2024-07-06 15:14:07,832 - valid stability spearman p 0.721, mae 0.0548, number of correct prediction 2512
205
+ 2024-07-06 15:39:46,841 - epoch 46/48 - steps 105, - time 1538.55s, train loss 0.13574993
206
+ 2024-07-06 15:39:46,841 - train stability spearman p 0.7003, mae 0.0479, number of correct prediction 53324
207
+ 2024-07-06 15:42:41,823 - evaluation dataset 1 - steps 2512 - time 174.94s, valid loss 0.64949482
208
+ 2024-07-06 15:42:41,823 - valid stability spearman p 0.7194, mae 0.055, number of correct prediction 2512
209
+ 2024-07-06 16:08:19,528 - epoch 47/48 - steps 105, - time 1537.26s, train loss 0.13570484
210
+ 2024-07-06 16:08:19,528 - train stability spearman p 0.701, mae 0.0478, number of correct prediction 53306
211
+ 2024-07-06 16:11:13,899 - evaluation dataset 1 - steps 2512 - time 174.33s, valid loss 0.6503134
212
+ 2024-07-06 16:11:13,900 - valid stability spearman p 0.7194, mae 0.0549, number of correct prediction 2512
213
+ 2024-07-06 16:36:53,253 - epoch 48/48 - steps 105, - time 1538.82s, train loss 0.13576934
214
+ 2024-07-06 16:36:53,253 - train stability spearman p 0.7006, mae 0.0479, number of correct prediction 53315
215
+ 2024-07-06 16:39:47,753 - evaluation dataset 1 - steps 2512 - time 174.46s, valid loss 0.65165197
216
+ 2024-07-06 16:39:47,753 - valid stability spearman p 0.7164, mae 0.0551, number of correct prediction 2512
217
+ 2024-07-06 16:40:07,820 -
218
+
219
+ start testing the best validation checkpoints
220
+ 2024-07-06 16:40:07,820 -
221
+ testing this checkpoint: /mnt/hdd8/mehdi/projects/prot2token/results/stability/2024-07-05__17-35-31/checkpoints/best_valid_stability_spearman.pth
222
+ 2024-07-06 16:40:25,836 - Loading checkpoint log: <All keys matched successfully>
223
+ 2024-07-06 16:55:53,205 - evaluation dataset 1 - steps 12851 - time 927.32s, valid loss 0.5607418
224
+ 2024-07-06 16:55:53,205 - test stability spearman p 0.7974, mae 0.0572, number of correct prediction 12851