gloamxun/modeling_zhaocui
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Train ATI-HO and finalize project outputs

Browse Source
This commit is contained in:
gloamxun
2026-09-26 16:05:44 +08:00
parent a86560da64
commit 9cdd604117
358 changed files with 10540 additions and 173 deletions
Download Patch File Download Diff File Expand all files Collapse all files
final/experiments/q3/ati_ho/models/A0/seed_2026/model_best.pt
BIN
View File
Binary file not shown.
final/experiments/q3/ati_ho/models/A0/seed_2026/training_history.csv
+8
View File
@@ -0,0 +1,8 @@
method,seed,epoch,train_loss,valid_selection_loss,valid_clean_loss,lambda_interaction,lambda_mask
A0,2026,1,1.0065954625606537,0.9363003446833118,0.931940136375008,0.001,0.0
A0,2026,2,0.8319815562831031,0.8692282438278198,0.8585401066056975,0.001,0.0
A0,2026,3,0.7417113295307866,0.8601733071135951,0.8481150000959963,0.001,0.0
A0,2026,4,0.6949168112542894,0.8574190991265433,0.8474381146850166,0.001,0.0
A0,2026,5,0.6569236450725131,0.8673936297277827,0.8559725572774698,0.001,0.0
A0,2026,6,0.6017359914603057,0.9111922525770062,0.9027906433566586,0.001,0.0
A0,2026,7,0.5584396239784029,0.9282659668843826,0.9144766841615949,0.001,0.0
1 method seed epoch train_loss valid_selection_loss valid_clean_loss lambda_interaction lambda_mask
2 A0 2026 1 1.0065954625606537 0.9363003446833118 0.931940136375008 0.001 0.0
3 A0 2026 2 0.8319815562831031 0.8692282438278198 0.8585401066056975 0.001 0.0
4 A0 2026 3 0.7417113295307866 0.8601733071135951 0.8481150000959963 0.001 0.0
5 A0 2026 4 0.6949168112542894 0.8574190991265433 0.8474381146850166 0.001 0.0
6 A0 2026 5 0.6569236450725131 0.8673936297277827 0.8559725572774698 0.001 0.0
7 A0 2026 6 0.6017359914603057 0.9111922525770062 0.9027906433566586 0.001 0.0
8 A0 2026 7 0.5584396239784029 0.9282659668843826 0.9144766841615949 0.001 0.0
final/experiments/q3/ati_ho/models/A0/seed_2026/training_manifest.json
+137
View File
@@ -0,0 +1,137 @@
{
"method": "A0",
"seed": 2026,
"best_epoch": 4,
"best_selection_loss": 0.8574190991265433,
"elapsed_seconds": 6.77071833203081,
"batch_size": 64,
"epoch_limit": 12,
"patience": 3,
"optimizer": "AdamW",
"learning_rate": 0.0003,
"weight_decay": 0.001,
"gradient_clip_norm": 1.0,
"training_mask_rates": [
0.0,
0.1,
0.3,
0.5,
0.7
],
"training_mask_patterns": [
"single",
"sync",
"partial",
"async"
],
"training_mask_seed_base": 20261227,
"same_orders_and_masks_across_methods_for_same_seed": true,
"config": {
"name": "A0_main_effects",
"low_rank": false,
"cross_attention": false,
"anchored": true,
"lambda_interaction": 0.001,
"lambda_mask": 0.0,
"rank": 4,
"hidden": 64,
"gru_hidden_per_direction": 32,
"attention_heads": 4,
"attention_ffn": 128,
"eta_init": 0.1
},
"history": [
{
"method": "A0",
"seed": 2026,
"epoch": 1,
"train_loss": 1.0065954625606537,
"valid_selection_loss": 0.9363003446833118,
"valid_clean_loss": 0.931940136375008,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A0",
"seed": 2026,
"epoch": 2,
"train_loss": 0.8319815562831031,
"valid_selection_loss": 0.8692282438278198,
"valid_clean_loss": 0.8585401066056975,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A0",
"seed": 2026,
"epoch": 3,
"train_loss": 0.7417113295307866,
"valid_selection_loss": 0.8601733071135951,
"valid_clean_loss": 0.8481150000959963,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A0",
"seed": 2026,
"epoch": 4,
"train_loss": 0.6949168112542894,
"valid_selection_loss": 0.8574190991265433,
"valid_clean_loss": 0.8474381146850166,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A0",
"seed": 2026,
"epoch": 5,
"train_loss": 0.6569236450725131,
"valid_selection_loss": 0.8673936297277827,
"valid_clean_loss": 0.8559725572774698,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A0",
"seed": 2026,
"epoch": 6,
"train_loss": 0.6017359914603057,
"valid_selection_loss": 0.9111922525770062,
"valid_clean_loss": 0.9027906433566586,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A0",
"seed": 2026,
"epoch": 7,
"train_loss": 0.5584396239784029,
"valid_selection_loss": 0.9282659668843826,
"valid_clean_loss": 0.9144766841615949,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
}
],
"mask_counts": {
"0.1/async": 1149,
"0.0/async": 1181,
"0.5/sync": 1205,
"0.0/sync": 1177,
"0.5/single": 1154,
"0.0/single": 1223,
"0.7/async": 1223,
"0.0/partial": 1254,
"0.7/sync": 1182,
"0.5/partial": 1182,
"0.3/partial": 1153,
"0.1/partial": 1216,
"0.3/single": 1178,
"0.3/async": 1180,
"0.5/async": 1182,
"0.1/single": 1208,
"0.7/single": 1129,
"0.3/sync": 1220,
"0.1/sync": 1187,
"0.7/partial": 1182
}
}
final/experiments/q3/ati_ho/models/A0/seed_3407/model_best.pt
BIN
View File
Binary file not shown.
final/experiments/q3/ati_ho/models/A0/seed_3407/training_history.csv
+7
View File
@@ -0,0 +1,7 @@
method,seed,epoch,train_loss,valid_selection_loss,valid_clean_loss,lambda_interaction,lambda_mask
A0,3407,1,1.0028652648131053,0.9436100423336029,0.9391928321712619,0.001,0.0
A0,3407,2,0.8238278020311285,0.8820641582811272,0.8730134918139532,0.001,0.0
A0,3407,3,0.74309202587163,0.8756466648735842,0.8646609815922413,0.001,0.0
A0,3407,4,0.6922215135009201,0.8815795684253777,0.8710692741058685,0.001,0.0
A0,3407,5,0.6518548451088093,0.9090294297579881,0.8968785254509895,0.001,0.0
A0,3407,6,0.6016097575150154,0.9234253629878326,0.9113659098908141,0.001,0.0
1 method seed epoch train_loss valid_selection_loss valid_clean_loss lambda_interaction lambda_mask
2 A0 3407 1 1.0028652648131053 0.9436100423336029 0.9391928321712619 0.001 0.0
3 A0 3407 2 0.8238278020311285 0.8820641582811272 0.8730134918139532 0.001 0.0
4 A0 3407 3 0.74309202587163 0.8756466648735842 0.8646609815922413 0.001 0.0
5 A0 3407 4 0.6922215135009201 0.8815795684253777 0.8710692741058685 0.001 0.0
6 A0 3407 5 0.6518548451088093 0.9090294297579881 0.8968785254509895 0.001 0.0
7 A0 3407 6 0.6016097575150154 0.9234253629878326 0.9113659098908141 0.001 0.0
final/experiments/q3/ati_ho/models/A0/seed_3407/training_manifest.json
+127
View File
@@ -0,0 +1,127 @@
{
"method": "A0",
"seed": 3407,
"best_epoch": 3,
"best_selection_loss": 0.8756466648735842,
"elapsed_seconds": 5.952770586998668,
"batch_size": 64,
"epoch_limit": 12,
"patience": 3,
"optimizer": "AdamW",
"learning_rate": 0.0003,
"weight_decay": 0.001,
"gradient_clip_norm": 1.0,
"training_mask_rates": [
0.0,
0.1,
0.3,
0.5,
0.7
],
"training_mask_patterns": [
"single",
"sync",
"partial",
"async"
],
"training_mask_seed_base": 20261227,
"same_orders_and_masks_across_methods_for_same_seed": true,
"config": {
"name": "A0_main_effects",
"low_rank": false,
"cross_attention": false,
"anchored": true,
"lambda_interaction": 0.001,
"lambda_mask": 0.0,
"rank": 4,
"hidden": 64,
"gru_hidden_per_direction": 32,
"attention_heads": 4,
"attention_ffn": 128,
"eta_init": 0.1
},
"history": [
{
"method": "A0",
"seed": 3407,
"epoch": 1,
"train_loss": 1.0028652648131053,
"valid_selection_loss": 0.9436100423336029,
"valid_clean_loss": 0.9391928321712619,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A0",
"seed": 3407,
"epoch": 2,
"train_loss": 0.8238278020311285,
"valid_selection_loss": 0.8820641582811272,
"valid_clean_loss": 0.8730134918139532,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A0",
"seed": 3407,
"epoch": 3,
"train_loss": 0.74309202587163,
"valid_selection_loss": 0.8756466648735842,
"valid_clean_loss": 0.8646609815922413,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A0",
"seed": 3407,
"epoch": 4,
"train_loss": 0.6922215135009201,
"valid_selection_loss": 0.8815795684253777,
"valid_clean_loss": 0.8710692741058685,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A0",
"seed": 3407,
"epoch": 5,
"train_loss": 0.6518548451088093,
"valid_selection_loss": 0.9090294297579881,
"valid_clean_loss": 0.8968785254509895,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A0",
"seed": 3407,
"epoch": 6,
"train_loss": 0.6016097575150154,
"valid_selection_loss": 0.9234253629878326,
"valid_clean_loss": 0.9113659098908141,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
}
],
"mask_counts": {
"0.7/async": 1003,
"0.0/async": 1014,
"0.5/sync": 1033,
"0.1/sync": 1024,
"0.5/partial": 996,
"0.3/partial": 1019,
"0.5/async": 1053,
"0.3/single": 1003,
"0.0/sync": 1042,
"0.7/partial": 1079,
"0.7/sync": 986,
"0.0/single": 1029,
"0.1/async": 959,
"0.3/sync": 1055,
"0.0/partial": 1016,
"0.1/single": 1011,
"0.1/partial": 988,
"0.5/single": 1076,
"0.7/single": 997,
"0.3/async": 987
}
}
final/experiments/q3/ati_ho/models/A0/seed_42/model_best.pt
BIN
View File
Binary file not shown.
final/experiments/q3/ati_ho/models/A0/seed_42/training_history.csv
+8
View File
@@ -0,0 +1,8 @@
method,seed,epoch,train_loss,valid_selection_loss,valid_clean_loss,lambda_interaction,lambda_mask
A0,42,1,0.9974622616061458,0.9360404843157465,0.9310548764008743,0.001,0.0
A0,42,2,0.8197665159349088,0.8663458909307207,0.8535200449136587,0.001,0.0
A0,42,3,0.7383816402267527,0.8669271827726575,0.8542619085573888,0.001,0.0
A0,42,4,0.6841275570569215,0.8643234267339601,0.850496660222064,0.001,0.0
A0,42,5,0.6515898009141287,0.8761576001460736,0.8627192581092918,0.001,0.0
A0,42,6,0.6044590015102316,0.9111036181777388,0.8968614366028335,0.001,0.0
A0,42,7,0.5584419562860772,0.9260857973124955,0.9113180768358838,0.001,0.0
1 method seed epoch train_loss valid_selection_loss valid_clean_loss lambda_interaction lambda_mask
2 A0 42 1 0.9974622616061458 0.9360404843157465 0.9310548764008743 0.001 0.0
3 A0 42 2 0.8197665159349088 0.8663458909307207 0.8535200449136587 0.001 0.0
4 A0 42 3 0.7383816402267527 0.8669271827726575 0.8542619085573888 0.001 0.0
5 A0 42 4 0.6841275570569215 0.8643234267339601 0.850496660222064 0.001 0.0
6 A0 42 5 0.6515898009141287 0.8761576001460736 0.8627192581092918 0.001 0.0
7 A0 42 6 0.6044590015102316 0.9111036181777388 0.8968614366028335 0.001 0.0
8 A0 42 7 0.5584419562860772 0.9260857973124955 0.9113180768358838 0.001 0.0
final/experiments/q3/ati_ho/models/A0/seed_42/training_manifest.json
+137
View File
@@ -0,0 +1,137 @@
{
"method": "A0",
"seed": 42,
"best_epoch": 4,
"best_selection_loss": 0.8643234267339601,
"elapsed_seconds": 12.891749234986492,
"batch_size": 64,
"epoch_limit": 12,
"patience": 3,
"optimizer": "AdamW",
"learning_rate": 0.0003,
"weight_decay": 0.001,
"gradient_clip_norm": 1.0,
"training_mask_rates": [
0.0,
0.1,
0.3,
0.5,
0.7
],
"training_mask_patterns": [
"single",
"sync",
"partial",
"async"
],
"training_mask_seed_base": 20261227,
"same_orders_and_masks_across_methods_for_same_seed": true,
"config": {
"name": "A0_main_effects",
"low_rank": false,
"cross_attention": false,
"anchored": true,
"lambda_interaction": 0.001,
"lambda_mask": 0.0,
"rank": 4,
"hidden": 64,
"gru_hidden_per_direction": 32,
"attention_heads": 4,
"attention_ffn": 128,
"eta_init": 0.1
},
"history": [
{
"method": "A0",
"seed": 42,
"epoch": 1,
"train_loss": 0.9974622616061458,
"valid_selection_loss": 0.9360404843157465,
"valid_clean_loss": 0.9310548764008743,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A0",
"seed": 42,
"epoch": 2,
"train_loss": 0.8197665159349088,
"valid_selection_loss": 0.8663458909307207,
"valid_clean_loss": 0.8535200449136587,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A0",
"seed": 42,
"epoch": 3,
"train_loss": 0.7383816402267527,
"valid_selection_loss": 0.8669271827726575,
"valid_clean_loss": 0.8542619085573888,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A0",
"seed": 42,
"epoch": 4,
"train_loss": 0.6841275570569215,
"valid_selection_loss": 0.8643234267339601,
"valid_clean_loss": 0.850496660222064,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A0",
"seed": 42,
"epoch": 5,
"train_loss": 0.6515898009141287,
"valid_selection_loss": 0.8761576001460736,
"valid_clean_loss": 0.8627192581092918,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A0",
"seed": 42,
"epoch": 6,
"train_loss": 0.6044590015102316,
"valid_selection_loss": 0.9111036181777388,
"valid_clean_loss": 0.8968614366028335,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A0",
"seed": 42,
"epoch": 7,
"train_loss": 0.5584419562860772,
"valid_selection_loss": 0.9260857973124955,
"valid_clean_loss": 0.9113180768358838,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
}
],
"mask_counts": {
"0.0/single": 1169,
"0.3/single": 1187,
"0.5/single": 1158,
"0.7/sync": 1272,
"0.7/partial": 1207,
"0.0/async": 1145,
"0.3/async": 1200,
"0.3/partial": 1209,
"0.3/sync": 1170,
"0.1/single": 1205,
"0.5/async": 1185,
"0.7/single": 1232,
"0.5/sync": 1143,
"0.5/partial": 1236,
"0.0/partial": 1183,
"0.0/sync": 1159,
"0.7/async": 1179,
"0.1/sync": 1202,
"0.1/async": 1192,
"0.1/partial": 1132
}
}
final/experiments/q3/ati_ho/models/A1/seed_2026/model_best.pt
BIN
View File
Binary file not shown.
final/experiments/q3/ati_ho/models/A1/seed_2026/training_history.csv
+8
View File
@@ -0,0 +1,8 @@
method,seed,epoch,train_loss,valid_selection_loss,valid_clean_loss,lambda_interaction,lambda_mask
A1,2026,1,1.0153538673012346,0.9397767204177248,0.9357580549114353,0.001,0.0
A1,2026,2,0.8344417214393616,0.8647092963968004,0.8555307774753361,0.001,0.0
A1,2026,3,0.7413759132226309,0.8561831976358707,0.8470297724336058,0.001,0.0
A1,2026,4,0.6923596881054066,0.8557776766163963,0.8481834858328432,0.001,0.0
A1,2026,5,0.6530446432254933,0.8681209758742825,0.8606502813297313,0.001,0.0
A1,2026,6,0.5979297994463532,0.9125843056283154,0.9086607209928743,0.001,0.0
A1,2026,7,0.5576094913261908,0.9276157835355172,0.9214578185762677,0.001,0.0
1 method seed epoch train_loss valid_selection_loss valid_clean_loss lambda_interaction lambda_mask
2 A1 2026 1 1.0153538673012346 0.9397767204177248 0.9357580549114353 0.001 0.0
3 A1 2026 2 0.8344417214393616 0.8647092963968004 0.8555307774753361 0.001 0.0
4 A1 2026 3 0.7413759132226309 0.8561831976358707 0.8470297724336058 0.001 0.0
5 A1 2026 4 0.6923596881054066 0.8557776766163963 0.8481834858328432 0.001 0.0
6 A1 2026 5 0.6530446432254933 0.8681209758742825 0.8606502813297313 0.001 0.0
7 A1 2026 6 0.5979297994463532 0.9125843056283154 0.9086607209928743 0.001 0.0
8 A1 2026 7 0.5576094913261908 0.9276157835355172 0.9214578185762677 0.001 0.0
final/experiments/q3/ati_ho/models/A1/seed_2026/training_manifest.json
+137
View File
@@ -0,0 +1,137 @@
{
"method": "A1",
"seed": 2026,
"best_epoch": 4,
"best_selection_loss": 0.8557776766163963,
"elapsed_seconds": 7.712336958968081,
"batch_size": 64,
"epoch_limit": 12,
"patience": 3,
"optimizer": "AdamW",
"learning_rate": 0.0003,
"weight_decay": 0.001,
"gradient_clip_norm": 1.0,
"training_mask_rates": [
0.0,
0.1,
0.3,
0.5,
0.7
],
"training_mask_patterns": [
"single",
"sync",
"partial",
"async"
],
"training_mask_seed_base": 20261227,
"same_orders_and_masks_across_methods_for_same_seed": true,
"config": {
"name": "A1_low_rank_pairs",
"low_rank": true,
"cross_attention": false,
"anchored": true,
"lambda_interaction": 0.001,
"lambda_mask": 0.0,
"rank": 4,
"hidden": 64,
"gru_hidden_per_direction": 32,
"attention_heads": 4,
"attention_ffn": 128,
"eta_init": 0.1
},
"history": [
{
"method": "A1",
"seed": 2026,
"epoch": 1,
"train_loss": 1.0153538673012346,
"valid_selection_loss": 0.9397767204177248,
"valid_clean_loss": 0.9357580549114353,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A1",
"seed": 2026,
"epoch": 2,
"train_loss": 0.8344417214393616,
"valid_selection_loss": 0.8647092963968004,
"valid_clean_loss": 0.8555307774753361,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A1",
"seed": 2026,
"epoch": 3,
"train_loss": 0.7413759132226309,
"valid_selection_loss": 0.8561831976358707,
"valid_clean_loss": 0.8470297724336058,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A1",
"seed": 2026,
"epoch": 4,
"train_loss": 0.6923596881054066,
"valid_selection_loss": 0.8557776766163963,
"valid_clean_loss": 0.8481834858328432,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A1",
"seed": 2026,
"epoch": 5,
"train_loss": 0.6530446432254933,
"valid_selection_loss": 0.8681209758742825,
"valid_clean_loss": 0.8606502813297313,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A1",
"seed": 2026,
"epoch": 6,
"train_loss": 0.5979297994463532,
"valid_selection_loss": 0.9125843056283154,
"valid_clean_loss": 0.9086607209928743,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A1",
"seed": 2026,
"epoch": 7,
"train_loss": 0.5576094913261908,
"valid_selection_loss": 0.9276157835355172,
"valid_clean_loss": 0.9214578185762677,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
}
],
"mask_counts": {
"0.1/async": 1149,
"0.0/async": 1181,
"0.5/sync": 1205,
"0.0/sync": 1177,
"0.5/single": 1154,
"0.0/single": 1223,
"0.7/async": 1223,
"0.0/partial": 1254,
"0.7/sync": 1182,
"0.5/partial": 1182,
"0.3/partial": 1153,
"0.1/partial": 1216,
"0.3/single": 1178,
"0.3/async": 1180,
"0.5/async": 1182,
"0.1/single": 1208,
"0.7/single": 1129,
"0.3/sync": 1220,
"0.1/sync": 1187,
"0.7/partial": 1182
}
}
final/experiments/q3/ati_ho/models/A1/seed_3407/model_best.pt
BIN
View File
Binary file not shown.
final/experiments/q3/ati_ho/models/A1/seed_3407/training_history.csv
+7
View File
@@ -0,0 +1,7 @@
method,seed,epoch,train_loss,valid_selection_loss,valid_clean_loss,lambda_interaction,lambda_mask
A1,3407,1,0.993991059285623,0.9350224992076119,0.9297198895569686,0.001,0.0
A1,3407,2,0.8231704179887418,0.8853522272227885,0.8782615517521952,0.001,0.0
A1,3407,3,0.7412442289016865,0.8810990981675766,0.872995504966149,0.001,0.0
A1,3407,4,0.6908998235508248,0.8885695379186462,0.8816041920211289,0.001,0.0
A1,3407,5,0.6549228540173283,0.9160289600655273,0.9081327443594461,0.001,0.0
A1,3407,6,0.6015714597370889,0.9299172356233493,0.9222097593349415,0.001,0.0
1 method seed epoch train_loss valid_selection_loss valid_clean_loss lambda_interaction lambda_mask
2 A1 3407 1 0.993991059285623 0.9350224992076119 0.9297198895569686 0.001 0.0
3 A1 3407 2 0.8231704179887418 0.8853522272227885 0.8782615517521952 0.001 0.0
4 A1 3407 3 0.7412442289016865 0.8810990981675766 0.872995504966149 0.001 0.0
5 A1 3407 4 0.6908998235508248 0.8885695379186462 0.8816041920211289 0.001 0.0
6 A1 3407 5 0.6549228540173283 0.9160289600655273 0.9081327443594461 0.001 0.0
7 A1 3407 6 0.6015714597370889 0.9299172356233493 0.9222097593349415 0.001 0.0
final/experiments/q3/ati_ho/models/A1/seed_3407/training_manifest.json
+127
View File
@@ -0,0 +1,127 @@
{
"method": "A1",
"seed": 3407,
"best_epoch": 3,
"best_selection_loss": 0.8810990981675766,
"elapsed_seconds": 6.539086283999495,
"batch_size": 64,
"epoch_limit": 12,
"patience": 3,
"optimizer": "AdamW",
"learning_rate": 0.0003,
"weight_decay": 0.001,
"gradient_clip_norm": 1.0,
"training_mask_rates": [
0.0,
0.1,
0.3,
0.5,
0.7
],
"training_mask_patterns": [
"single",
"sync",
"partial",
"async"
],
"training_mask_seed_base": 20261227,
"same_orders_and_masks_across_methods_for_same_seed": true,
"config": {
"name": "A1_low_rank_pairs",
"low_rank": true,
"cross_attention": false,
"anchored": true,
"lambda_interaction": 0.001,
"lambda_mask": 0.0,
"rank": 4,
"hidden": 64,
"gru_hidden_per_direction": 32,
"attention_heads": 4,
"attention_ffn": 128,
"eta_init": 0.1
},
"history": [
{
"method": "A1",
"seed": 3407,
"epoch": 1,
"train_loss": 0.993991059285623,
"valid_selection_loss": 0.9350224992076119,
"valid_clean_loss": 0.9297198895569686,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A1",
"seed": 3407,
"epoch": 2,
"train_loss": 0.8231704179887418,
"valid_selection_loss": 0.8853522272227885,
"valid_clean_loss": 0.8782615517521952,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A1",
"seed": 3407,
"epoch": 3,
"train_loss": 0.7412442289016865,
"valid_selection_loss": 0.8810990981675766,
"valid_clean_loss": 0.872995504966149,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A1",
"seed": 3407,
"epoch": 4,
"train_loss": 0.6908998235508248,
"valid_selection_loss": 0.8885695379186462,
"valid_clean_loss": 0.8816041920211289,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A1",
"seed": 3407,
"epoch": 5,
"train_loss": 0.6549228540173283,
"valid_selection_loss": 0.9160289600655273,
"valid_clean_loss": 0.9081327443594461,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A1",
"seed": 3407,
"epoch": 6,
"train_loss": 0.6015714597370889,
"valid_selection_loss": 0.9299172356233493,
"valid_clean_loss": 0.9222097593349415,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
}
],
"mask_counts": {
"0.7/async": 1003,
"0.0/async": 1014,
"0.5/sync": 1033,
"0.1/sync": 1024,
"0.5/partial": 996,
"0.3/partial": 1019,
"0.5/async": 1053,
"0.3/single": 1003,
"0.0/sync": 1042,
"0.7/partial": 1079,
"0.7/sync": 986,
"0.0/single": 1029,
"0.1/async": 959,
"0.3/sync": 1055,
"0.0/partial": 1016,
"0.1/single": 1011,
"0.1/partial": 988,
"0.5/single": 1076,
"0.7/single": 997,
"0.3/async": 987
}
}
final/experiments/q3/ati_ho/models/A1/seed_42/model_best.pt
BIN
View File
Binary file not shown.
final/experiments/q3/ati_ho/models/A1/seed_42/training_history.csv
+8
View File
@@ -0,0 +1,8 @@
method,seed,epoch,train_loss,valid_selection_loss,valid_clean_loss,lambda_interaction,lambda_mask
A1,42,1,0.9995549983448453,0.9426855438358182,0.9384826362788022,0.001,0.0
A1,42,2,0.8219450292763887,0.8703031135456902,0.8586705464583176,0.001,0.0
A1,42,3,0.7375812182823817,0.8694013199963413,0.8590540899025215,0.001,0.0
A1,42,4,0.6825026941520197,0.8649765662439577,0.8543586613057734,0.001,0.0
A1,42,5,0.648727031217681,0.877989024087623,0.8682481325589694,0.001,0.0
A1,42,6,0.6008665031856961,0.9173483715935067,0.9082090225848523,0.001,0.0
A1,42,7,0.5538152815015228,0.9287053037148255,0.9185833616571112,0.001,0.0
1 method seed epoch train_loss valid_selection_loss valid_clean_loss lambda_interaction lambda_mask
2 A1 42 1 0.9995549983448453 0.9426855438358182 0.9384826362788022 0.001 0.0
3 A1 42 2 0.8219450292763887 0.8703031135456902 0.8586705464583176 0.001 0.0
4 A1 42 3 0.7375812182823817 0.8694013199963413 0.8590540899025215 0.001 0.0
5 A1 42 4 0.6825026941520197 0.8649765662439577 0.8543586613057734 0.001 0.0
6 A1 42 5 0.648727031217681 0.877989024087623 0.8682481325589694 0.001 0.0
7 A1 42 6 0.6008665031856961 0.9173483715935067 0.9082090225848523 0.001 0.0
8 A1 42 7 0.5538152815015228 0.9287053037148255 0.9185833616571112 0.001 0.0
final/experiments/q3/ati_ho/models/A1/seed_42/training_manifest.json
+137
View File
@@ -0,0 +1,137 @@
{
"method": "A1",
"seed": 42,
"best_epoch": 4,
"best_selection_loss": 0.8649765662439577,
"elapsed_seconds": 7.483738042006735,
"batch_size": 64,
"epoch_limit": 12,
"patience": 3,
"optimizer": "AdamW",
"learning_rate": 0.0003,
"weight_decay": 0.001,
"gradient_clip_norm": 1.0,
"training_mask_rates": [
0.0,
0.1,
0.3,
0.5,
0.7
],
"training_mask_patterns": [
"single",
"sync",
"partial",
"async"
],
"training_mask_seed_base": 20261227,
"same_orders_and_masks_across_methods_for_same_seed": true,
"config": {
"name": "A1_low_rank_pairs",
"low_rank": true,
"cross_attention": false,
"anchored": true,
"lambda_interaction": 0.001,
"lambda_mask": 0.0,
"rank": 4,
"hidden": 64,
"gru_hidden_per_direction": 32,
"attention_heads": 4,
"attention_ffn": 128,
"eta_init": 0.1
},
"history": [
{
"method": "A1",
"seed": 42,
"epoch": 1,
"train_loss": 0.9995549983448453,
"valid_selection_loss": 0.9426855438358182,
"valid_clean_loss": 0.9384826362788022,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A1",
"seed": 42,
"epoch": 2,
"train_loss": 0.8219450292763887,
"valid_selection_loss": 0.8703031135456902,
"valid_clean_loss": 0.8586705464583176,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A1",
"seed": 42,
"epoch": 3,
"train_loss": 0.7375812182823817,
"valid_selection_loss": 0.8694013199963413,
"valid_clean_loss": 0.8590540899025215,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A1",
"seed": 42,
"epoch": 4,
"train_loss": 0.6825026941520197,
"valid_selection_loss": 0.8649765662439577,
"valid_clean_loss": 0.8543586613057734,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A1",
"seed": 42,
"epoch": 5,
"train_loss": 0.648727031217681,
"valid_selection_loss": 0.877989024087623,
"valid_clean_loss": 0.8682481325589694,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A1",
"seed": 42,
"epoch": 6,
"train_loss": 0.6008665031856961,
"valid_selection_loss": 0.9173483715935067,
"valid_clean_loss": 0.9082090225848523,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A1",
"seed": 42,
"epoch": 7,
"train_loss": 0.5538152815015228,
"valid_selection_loss": 0.9287053037148255,
"valid_clean_loss": 0.9185833616571112,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
}
],
"mask_counts": {
"0.0/single": 1169,
"0.3/single": 1187,
"0.5/single": 1158,
"0.7/sync": 1272,
"0.7/partial": 1207,
"0.0/async": 1145,
"0.3/async": 1200,
"0.3/partial": 1209,
"0.3/sync": 1170,
"0.1/single": 1205,
"0.5/async": 1185,
"0.7/single": 1232,
"0.5/sync": 1143,
"0.5/partial": 1236,
"0.0/partial": 1183,
"0.0/sync": 1159,
"0.7/async": 1179,
"0.1/sync": 1202,
"0.1/async": 1192,
"0.1/partial": 1132
}
}
final/experiments/q3/ati_ho/models/A2/seed_2026/model_best.pt
BIN
View File
Binary file not shown.
final/experiments/q3/ati_ho/models/A2/seed_2026/training_history.csv
+7
View File
@@ -0,0 +1,7 @@
method,seed,epoch,train_loss,valid_selection_loss,valid_clean_loss,lambda_interaction,lambda_mask
A2,2026,1,0.9926471710205078,0.928315669953168,0.9239131356333639,0.001,0.0
A2,2026,2,0.820087315859618,0.8673638532777409,0.8601592170013176,0.001,0.0
A2,2026,3,0.7308831590193289,0.862087192279952,0.8561929890087673,0.001,0.0
A2,2026,4,0.6851603929643277,0.8666458973190287,0.8619590087251349,0.001,0.0
A2,2026,5,0.6444870498445299,0.88101957656525,0.8752215864894154,0.001,0.0
A2,2026,6,0.5834257365376861,0.9354289185542327,0.9353187752294017,0.001,0.0
1 method seed epoch train_loss valid_selection_loss valid_clean_loss lambda_interaction lambda_mask
2 A2 2026 1 0.9926471710205078 0.928315669953168 0.9239131356333639 0.001 0.0
3 A2 2026 2 0.820087315859618 0.8673638532777409 0.8601592170013176 0.001 0.0
4 A2 2026 3 0.7308831590193289 0.862087192279952 0.8561929890087673 0.001 0.0
5 A2 2026 4 0.6851603929643277 0.8666458973190287 0.8619590087251349 0.001 0.0
6 A2 2026 5 0.6444870498445299 0.88101957656525 0.8752215864894154 0.001 0.0
7 A2 2026 6 0.5834257365376861 0.9354289185542327 0.9353187752294017 0.001 0.0
final/experiments/q3/ati_ho/models/A2/seed_2026/training_manifest.json
+127
View File
@@ -0,0 +1,127 @@
{
"method": "A2",
"seed": 2026,
"best_epoch": 3,
"best_selection_loss": 0.862087192279952,
"elapsed_seconds": 9.472242687013932,
"batch_size": 64,
"epoch_limit": 12,
"patience": 3,
"optimizer": "AdamW",
"learning_rate": 0.0003,
"weight_decay": 0.001,
"gradient_clip_norm": 1.0,
"training_mask_rates": [
0.0,
0.1,
0.3,
0.5,
0.7
],
"training_mask_patterns": [
"single",
"sync",
"partial",
"async"
],
"training_mask_seed_base": 20261227,
"same_orders_and_masks_across_methods_for_same_seed": true,
"config": {
"name": "A2_anchored_pairwise",
"low_rank": true,
"cross_attention": true,
"anchored": true,
"lambda_interaction": 0.001,
"lambda_mask": 0.0,
"rank": 4,
"hidden": 64,
"gru_hidden_per_direction": 32,
"attention_heads": 4,
"attention_ffn": 128,
"eta_init": 0.1
},
"history": [
{
"method": "A2",
"seed": 2026,
"epoch": 1,
"train_loss": 0.9926471710205078,
"valid_selection_loss": 0.928315669953168,
"valid_clean_loss": 0.9239131356333639,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A2",
"seed": 2026,
"epoch": 2,
"train_loss": 0.820087315859618,
"valid_selection_loss": 0.8673638532777409,
"valid_clean_loss": 0.8601592170013176,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A2",
"seed": 2026,
"epoch": 3,
"train_loss": 0.7308831590193289,
"valid_selection_loss": 0.862087192279952,
"valid_clean_loss": 0.8561929890087673,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A2",
"seed": 2026,
"epoch": 4,
"train_loss": 0.6851603929643277,
"valid_selection_loss": 0.8666458973190287,
"valid_clean_loss": 0.8619590087251349,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A2",
"seed": 2026,
"epoch": 5,
"train_loss": 0.6444870498445299,
"valid_selection_loss": 0.88101957656525,
"valid_clean_loss": 0.8752215864894154,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A2",
"seed": 2026,
"epoch": 6,
"train_loss": 0.5834257365376861,
"valid_selection_loss": 0.9354289185542327,
"valid_clean_loss": 0.9353187752294017,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
}
],
"mask_counts": {
"0.1/async": 970,
"0.0/async": 995,
"0.5/sync": 1040,
"0.0/sync": 1012,
"0.5/single": 999,
"0.0/single": 1046,
"0.7/async": 1048,
"0.0/partial": 1063,
"0.7/sync": 1020,
"0.5/partial": 1022,
"0.3/partial": 1002,
"0.1/partial": 1022,
"0.3/single": 1004,
"0.3/async": 1026,
"0.5/async": 1028,
"0.1/single": 1048,
"0.7/single": 959,
"0.3/sync": 1030,
"0.1/sync": 1007,
"0.7/partial": 1029
}
}
final/experiments/q3/ati_ho/models/A2/seed_3407/model_best.pt
BIN
View File
Binary file not shown.
final/experiments/q3/ati_ho/models/A2/seed_3407/training_history.csv
+7
View File
@@ -0,0 +1,7 @@
method,seed,epoch,train_loss,valid_selection_loss,valid_clean_loss,lambda_interaction,lambda_mask
A2,3407,1,0.9647035289693762,0.9118595260840197,0.9059003676686969,0.001,0.0
A2,3407,2,0.8049548080673924,0.8821067369573719,0.8777574326965835,0.001,0.0
A2,3407,3,0.7312485785396011,0.880271397449158,0.8748716823347322,0.001,0.0
A2,3407,4,0.6844777796003554,0.88496050123985,0.881083797622513,0.001,0.0
A2,3407,5,0.6395798369690224,0.9139724308317835,0.9094404204861148,0.001,0.0
A2,3407,6,0.5926711849040456,0.9307978958873957,0.9263341426849365,0.001,0.0
1 method seed epoch train_loss valid_selection_loss valid_clean_loss lambda_interaction lambda_mask
2 A2 3407 1 0.9647035289693762 0.9118595260840197 0.9059003676686969 0.001 0.0
3 A2 3407 2 0.8049548080673924 0.8821067369573719 0.8777574326965835 0.001 0.0
4 A2 3407 3 0.7312485785396011 0.880271397449158 0.8748716823347322 0.001 0.0
5 A2 3407 4 0.6844777796003554 0.88496050123985 0.881083797622513 0.001 0.0
6 A2 3407 5 0.6395798369690224 0.9139724308317835 0.9094404204861148 0.001 0.0
7 A2 3407 6 0.5926711849040456 0.9307978958873957 0.9263341426849365 0.001 0.0
final/experiments/q3/ati_ho/models/A2/seed_3407/training_manifest.json
+127
View File
@@ -0,0 +1,127 @@
{
"method": "A2",
"seed": 3407,
"best_epoch": 3,
"best_selection_loss": 0.880271397449158,
"elapsed_seconds": 9.558659436006565,
"batch_size": 64,
"epoch_limit": 12,
"patience": 3,
"optimizer": "AdamW",
"learning_rate": 0.0003,
"weight_decay": 0.001,
"gradient_clip_norm": 1.0,
"training_mask_rates": [
0.0,
0.1,
0.3,
0.5,
0.7
],
"training_mask_patterns": [
"single",
"sync",
"partial",
"async"
],
"training_mask_seed_base": 20261227,
"same_orders_and_masks_across_methods_for_same_seed": true,
"config": {
"name": "A2_anchored_pairwise",
"low_rank": true,
"cross_attention": true,
"anchored": true,
"lambda_interaction": 0.001,
"lambda_mask": 0.0,
"rank": 4,
"hidden": 64,
"gru_hidden_per_direction": 32,
"attention_heads": 4,
"attention_ffn": 128,
"eta_init": 0.1
},
"history": [
{
"method": "A2",
"seed": 3407,
"epoch": 1,
"train_loss": 0.9647035289693762,
"valid_selection_loss": 0.9118595260840197,
"valid_clean_loss": 0.9059003676686969,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A2",
"seed": 3407,
"epoch": 2,
"train_loss": 0.8049548080673924,
"valid_selection_loss": 0.8821067369573719,
"valid_clean_loss": 0.8777574326965835,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A2",
"seed": 3407,
"epoch": 3,
"train_loss": 0.7312485785396011,
"valid_selection_loss": 0.880271397449158,
"valid_clean_loss": 0.8748716823347322,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A2",
"seed": 3407,
"epoch": 4,
"train_loss": 0.6844777796003554,
"valid_selection_loss": 0.88496050123985,
"valid_clean_loss": 0.881083797622513,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A2",
"seed": 3407,
"epoch": 5,
"train_loss": 0.6395798369690224,
"valid_selection_loss": 0.9139724308317835,
"valid_clean_loss": 0.9094404204861148,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A2",
"seed": 3407,
"epoch": 6,
"train_loss": 0.5926711849040456,
"valid_selection_loss": 0.9307978958873957,
"valid_clean_loss": 0.9263341426849365,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
}
],
"mask_counts": {
"0.7/async": 1003,
"0.0/async": 1014,
"0.5/sync": 1033,
"0.1/sync": 1024,
"0.5/partial": 996,
"0.3/partial": 1019,
"0.5/async": 1053,
"0.3/single": 1003,
"0.0/sync": 1042,
"0.7/partial": 1079,
"0.7/sync": 986,
"0.0/single": 1029,
"0.1/async": 959,
"0.3/sync": 1055,
"0.0/partial": 1016,
"0.1/single": 1011,
"0.1/partial": 988,
"0.5/single": 1076,
"0.7/single": 997,
"0.3/async": 987
}
}
final/experiments/q3/ati_ho/models/A2/seed_42/model_best.pt
BIN
View File
Binary file not shown.
final/experiments/q3/ati_ho/models/A2/seed_42/training_history.csv
+6
View File
@@ -0,0 +1,6 @@
method,seed,epoch,train_loss,valid_selection_loss,valid_clean_loss,lambda_interaction,lambda_mask
A2,42,1,0.9741679154060505,0.9116741605512388,0.9052737184933254,0.001,0.0
A2,42,2,0.7997706675970996,0.8634063961741689,0.8530163961452443,0.001,0.0
A2,42,3,0.7264242089456983,0.8717996972602802,0.8635974290606739,0.001,0.0
A2,42,4,0.6684221158976908,0.8713849653581996,0.8629672671412374,0.001,0.0
A2,42,5,0.6371411096166681,0.8887755983805918,0.8809175399633554,0.001,0.0
1 method seed epoch train_loss valid_selection_loss valid_clean_loss lambda_interaction lambda_mask
2 A2 42 1 0.9741679154060505 0.9116741605512388 0.9052737184933254 0.001 0.0
3 A2 42 2 0.7997706675970996 0.8634063961741689 0.8530163961452443 0.001 0.0
4 A2 42 3 0.7264242089456983 0.8717996972602802 0.8635974290606739 0.001 0.0
5 A2 42 4 0.6684221158976908 0.8713849653581996 0.8629672671412374 0.001 0.0
6 A2 42 5 0.6371411096166681 0.8887755983805918 0.8809175399633554 0.001 0.0
final/experiments/q3/ati_ho/models/A2/seed_42/training_manifest.json
+117
View File
@@ -0,0 +1,117 @@
{
"method": "A2",
"seed": 42,
"best_epoch": 2,
"best_selection_loss": 0.8634063961741689,
"elapsed_seconds": 8.171100930019747,
"batch_size": 64,
"epoch_limit": 12,
"patience": 3,
"optimizer": "AdamW",
"learning_rate": 0.0003,
"weight_decay": 0.001,
"gradient_clip_norm": 1.0,
"training_mask_rates": [
0.0,
0.1,
0.3,
0.5,
0.7
],
"training_mask_patterns": [
"single",
"sync",
"partial",
"async"
],
"training_mask_seed_base": 20261227,
"same_orders_and_masks_across_methods_for_same_seed": true,
"config": {
"name": "A2_anchored_pairwise",
"low_rank": true,
"cross_attention": true,
"anchored": true,
"lambda_interaction": 0.001,
"lambda_mask": 0.0,
"rank": 4,
"hidden": 64,
"gru_hidden_per_direction": 32,
"attention_heads": 4,
"attention_ffn": 128,
"eta_init": 0.1
},
"history": [
{
"method": "A2",
"seed": 42,
"epoch": 1,
"train_loss": 0.9741679154060505,
"valid_selection_loss": 0.9116741605512388,
"valid_clean_loss": 0.9052737184933254,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A2",
"seed": 42,
"epoch": 2,
"train_loss": 0.7997706675970996,
"valid_selection_loss": 0.8634063961741689,
"valid_clean_loss": 0.8530163961452443,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A2",
"seed": 42,
"epoch": 3,
"train_loss": 0.7264242089456983,
"valid_selection_loss": 0.8717996972602802,
"valid_clean_loss": 0.8635974290606739,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A2",
"seed": 42,
"epoch": 4,
"train_loss": 0.6684221158976908,
"valid_selection_loss": 0.8713849653581996,
"valid_clean_loss": 0.8629672671412374,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "A2",
"seed": 42,
"epoch": 5,
"train_loss": 0.6371411096166681,
"valid_selection_loss": 0.8887755983805918,
"valid_clean_loss": 0.8809175399633554,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
}
],
"mask_counts": {
"0.0/single": 838,
"0.3/single": 841,
"0.5/single": 832,
"0.7/sync": 931,
"0.7/partial": 854,
"0.0/async": 818,
"0.3/async": 856,
"0.3/partial": 861,
"0.3/sync": 841,
"0.1/single": 857,
"0.5/async": 832,
"0.7/single": 872,
"0.5/sync": 848,
"0.5/partial": 854,
"0.0/partial": 870,
"0.0/sync": 827,
"0.7/async": 831,
"0.1/sync": 839,
"0.1/async": 847,
"0.1/partial": 826
}
}
final/experiments/q3/ati_ho/models/A3/seed_42/model_best.pt
BIN
View File
Binary file not shown.
final/experiments/q3/ati_ho/models/A3/seed_42/training_history.csv
+6
View File
@@ -0,0 +1,6 @@
method,seed,epoch,train_loss,valid_selection_loss,valid_clean_loss,lambda_interaction,lambda_mask
A3,42,1,1.0065002750467371,0.9119112013460515,0.9055310710445865,0.001,0.05
A3,42,2,0.8278417852189806,0.8637567355737581,0.8532818588581714,0.001,0.05
A3,42,3,0.7517132185123585,0.8728943449127805,0.8647457944167839,0.001,0.05
A3,42,4,0.6915535065862868,0.8723347707764133,0.8638496981872307,0.001,0.05
A3,42,5,0.6589032621295364,0.8902343159521019,0.8822925837485345,0.001,0.05
1 method seed epoch train_loss valid_selection_loss valid_clean_loss lambda_interaction lambda_mask
2 A3 42 1 1.0065002750467371 0.9119112013460515 0.9055310710445865 0.001 0.05
3 A3 42 2 0.8278417852189806 0.8637567355737581 0.8532818588581714 0.001 0.05
4 A3 42 3 0.7517132185123585 0.8728943449127805 0.8647457944167839 0.001 0.05
5 A3 42 4 0.6915535065862868 0.8723347707764133 0.8638496981872307 0.001 0.05
6 A3 42 5 0.6589032621295364 0.8902343159521019 0.8822925837485345 0.001 0.05
final/experiments/q3/ati_ho/models/A3/seed_42/training_manifest.json
+117
View File
@@ -0,0 +1,117 @@
{
"method": "A3",
"seed": 42,
"best_epoch": 2,
"best_selection_loss": 0.8637567355737581,
"elapsed_seconds": 8.470124277984723,
"batch_size": 64,
"epoch_limit": 12,
"patience": 3,
"optimizer": "AdamW",
"learning_rate": 0.0003,
"weight_decay": 0.001,
"gradient_clip_norm": 1.0,
"training_mask_rates": [
0.0,
0.1,
0.3,
0.5,
0.7
],
"training_mask_patterns": [
"single",
"sync",
"partial",
"async"
],
"training_mask_seed_base": 20261227,
"same_orders_and_masks_across_methods_for_same_seed": true,
"config": {
"name": "A3_pairwise_mask_aux",
"low_rank": true,
"cross_attention": true,
"anchored": true,
"lambda_interaction": 0.001,
"lambda_mask": 0.05,
"rank": 4,
"hidden": 64,
"gru_hidden_per_direction": 32,
"attention_heads": 4,
"attention_ffn": 128,
"eta_init": 0.1
},
"history": [
{
"method": "A3",
"seed": 42,
"epoch": 1,
"train_loss": 1.0065002750467371,
"valid_selection_loss": 0.9119112013460515,
"valid_clean_loss": 0.9055310710445865,
"lambda_interaction": 0.001,
"lambda_mask": 0.05
},
{
"method": "A3",
"seed": 42,
"epoch": 2,
"train_loss": 0.8278417852189806,
"valid_selection_loss": 0.8637567355737581,
"valid_clean_loss": 0.8532818588581714,
"lambda_interaction": 0.001,
"lambda_mask": 0.05
},
{
"method": "A3",
"seed": 42,
"epoch": 3,
"train_loss": 0.7517132185123585,
"valid_selection_loss": 0.8728943449127805,
"valid_clean_loss": 0.8647457944167839,
"lambda_interaction": 0.001,
"lambda_mask": 0.05
},
{
"method": "A3",
"seed": 42,
"epoch": 4,
"train_loss": 0.6915535065862868,
"valid_selection_loss": 0.8723347707764133,
"valid_clean_loss": 0.8638496981872307,
"lambda_interaction": 0.001,
"lambda_mask": 0.05
},
{
"method": "A3",
"seed": 42,
"epoch": 5,
"train_loss": 0.6589032621295364,
"valid_selection_loss": 0.8902343159521019,
"valid_clean_loss": 0.8822925837485345,
"lambda_interaction": 0.001,
"lambda_mask": 0.05
}
],
"mask_counts": {
"0.0/single": 838,
"0.3/single": 841,
"0.5/single": 832,
"0.7/sync": 931,
"0.7/partial": 854,
"0.0/async": 818,
"0.3/async": 856,
"0.3/partial": 861,
"0.3/sync": 841,
"0.1/single": 857,
"0.5/async": 832,
"0.7/single": 872,
"0.5/sync": 848,
"0.5/partial": 854,
"0.0/partial": 870,
"0.0/sync": 827,
"0.7/async": 831,
"0.1/sync": 839,
"0.1/async": 847,
"0.1/partial": 826
}
}
final/experiments/q3/ati_ho/models/B0_early_concat/seed_2026/model_best.pt
BIN
View File
Binary file not shown.
final/experiments/q3/ati_ho/models/B0_early_concat/seed_2026/training_history.csv
+6
View File
@@ -0,0 +1,6 @@
method,seed,epoch,train_loss,valid_selection_loss,valid_clean_loss,lambda_interaction,lambda_mask
B0_early_concat,2026,1,0.9894467581201483,0.9033395762626941,0.8919235042163304,0.0,0.0
B0_early_concat,2026,2,0.8127173174310613,0.8722499026047006,0.8655281905289535,0.0,0.0
B0_early_concat,2026,3,0.7274277905623118,0.9010504417039537,0.9002949733000535,0.0,0.0
B0_early_concat,2026,4,0.6712073153919644,0.9205121268610378,0.9208705294263232,0.0,0.0
B0_early_concat,2026,5,0.6352586569609465,0.9440761974879672,0.9462754883608975,0.0,0.0
1 method seed epoch train_loss valid_selection_loss valid_clean_loss lambda_interaction lambda_mask
2 B0_early_concat 2026 1 0.9894467581201483 0.9033395762626941 0.8919235042163304 0.0 0.0
3 B0_early_concat 2026 2 0.8127173174310613 0.8722499026047006 0.8655281905289535 0.0 0.0
4 B0_early_concat 2026 3 0.7274277905623118 0.9010504417039537 0.9002949733000535 0.0 0.0
5 B0_early_concat 2026 4 0.6712073153919644 0.9205121268610378 0.9208705294263232 0.0 0.0
6 B0_early_concat 2026 5 0.6352586569609465 0.9440761974879672 0.9462754883608975 0.0 0.0
final/experiments/q3/ati_ho/models/B0_early_concat/seed_2026/training_manifest.json
+118
View File
@@ -0,0 +1,118 @@
{
"method": "B0_early_concat",
"seed": 2026,
"best_epoch": 2,
"best_selection_loss": 0.8722499026047006,
"elapsed_seconds": 2.9758365789894015,
"batch_size": 64,
"epoch_limit": 12,
"patience": 3,
"optimizer": "AdamW",
"learning_rate": 0.0003,
"weight_decay": 0.001,
"gradient_clip_norm": 1.0,
"training_mask_rates": [
0.0,
0.1,
0.3,
0.5,
0.7
],
"training_mask_patterns": [
"single",
"sync",
"partial",
"async"
],
"training_mask_seed_base": 20261227,
"same_orders_and_masks_across_methods_for_same_seed": true,
"config": {
"model_config": {
"router": "mlp",
"expert_names": [
"T",
"A",
"V",
"TA",
"TV",
"AV",
"TAV"
],
"availability_mode": "hard"
}
},
"history": [
{
"method": "B0_early_concat",
"seed": 2026,
"epoch": 1,
"train_loss": 0.9894467581201483,
"valid_selection_loss": 0.9033395762626941,
"valid_clean_loss": 0.8919235042163304,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B0_early_concat",
"seed": 2026,
"epoch": 2,
"train_loss": 0.8127173174310613,
"valid_selection_loss": 0.8722499026047006,
"valid_clean_loss": 0.8655281905289535,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B0_early_concat",
"seed": 2026,
"epoch": 3,
"train_loss": 0.7274277905623118,
"valid_selection_loss": 0.9010504417039537,
"valid_clean_loss": 0.9002949733000535,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B0_early_concat",
"seed": 2026,
"epoch": 4,
"train_loss": 0.6712073153919644,
"valid_selection_loss": 0.9205121268610378,
"valid_clean_loss": 0.9208705294263232,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B0_early_concat",
"seed": 2026,
"epoch": 5,
"train_loss": 0.6352586569609465,
"valid_selection_loss": 0.9440761974879672,
"valid_clean_loss": 0.9462754883608975,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
}
],
"mask_counts": {
"0.1/async": 795,
"0.0/async": 832,
"0.5/sync": 860,
"0.0/sync": 859,
"0.5/single": 824,
"0.0/single": 878,
"0.7/async": 880,
"0.0/partial": 860,
"0.7/sync": 842,
"0.5/partial": 867,
"0.3/partial": 815,
"0.1/partial": 874,
"0.3/single": 837,
"0.3/async": 873,
"0.5/async": 830,
"0.1/single": 886,
"0.7/single": 801,
"0.3/sync": 841,
"0.1/sync": 847,
"0.7/partial": 874
}
}
final/experiments/q3/ati_ho/models/B0_early_concat/seed_3407/model_best.pt
BIN
View File
Binary file not shown.
final/experiments/q3/ati_ho/models/B0_early_concat/seed_3407/training_history.csv
+7
View File
@@ -0,0 +1,7 @@
method,seed,epoch,train_loss,valid_selection_loss,valid_clean_loss,lambda_interaction,lambda_mask
B0_early_concat,3407,1,0.9901614431981687,0.9334849718507829,0.9280904976876228,0.0,0.0
B0_early_concat,3407,2,0.8284304141998291,0.8979402235248587,0.8948127568423093,0.0,0.0
B0_early_concat,3407,3,0.7437613738907708,0.8850730860626306,0.8822627912510882,0.0,0.0
B0_early_concat,3407,4,0.682034013999833,0.8985609801915976,0.8979688529129867,0.0,0.0
B0_early_concat,3407,5,0.6572728455066681,0.9590781547211028,0.9659944576221507,0.0,0.0
B0_early_concat,3407,6,0.5803193593466723,0.9992094658888303,1.013474199798081,0.0,0.0
1 method seed epoch train_loss valid_selection_loss valid_clean_loss lambda_interaction lambda_mask
2 B0_early_concat 3407 1 0.9901614431981687 0.9334849718507829 0.9280904976876228 0.0 0.0
3 B0_early_concat 3407 2 0.8284304141998291 0.8979402235248587 0.8948127568423093 0.0 0.0
4 B0_early_concat 3407 3 0.7437613738907708 0.8850730860626306 0.8822627912510882 0.0 0.0
5 B0_early_concat 3407 4 0.682034013999833 0.8985609801915976 0.8979688529129867 0.0 0.0
6 B0_early_concat 3407 5 0.6572728455066681 0.9590781547211028 0.9659944576221507 0.0 0.0
7 B0_early_concat 3407 6 0.5803193593466723 0.9992094658888303 1.013474199798081 0.0 0.0
final/experiments/q3/ati_ho/models/B0_early_concat/seed_3407/training_manifest.json
+128
View File
@@ -0,0 +1,128 @@
{
"method": "B0_early_concat",
"seed": 3407,
"best_epoch": 3,
"best_selection_loss": 0.8850730860626306,
"elapsed_seconds": 3.5102919560158625,
"batch_size": 64,
"epoch_limit": 12,
"patience": 3,
"optimizer": "AdamW",
"learning_rate": 0.0003,
"weight_decay": 0.001,
"gradient_clip_norm": 1.0,
"training_mask_rates": [
0.0,
0.1,
0.3,
0.5,
0.7
],
"training_mask_patterns": [
"single",
"sync",
"partial",
"async"
],
"training_mask_seed_base": 20261227,
"same_orders_and_masks_across_methods_for_same_seed": true,
"config": {
"model_config": {
"router": "mlp",
"expert_names": [
"T",
"A",
"V",
"TA",
"TV",
"AV",
"TAV"
],
"availability_mode": "hard"
}
},
"history": [
{
"method": "B0_early_concat",
"seed": 3407,
"epoch": 1,
"train_loss": 0.9901614431981687,
"valid_selection_loss": 0.9334849718507829,
"valid_clean_loss": 0.9280904976876228,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B0_early_concat",
"seed": 3407,
"epoch": 2,
"train_loss": 0.8284304141998291,
"valid_selection_loss": 0.8979402235248587,
"valid_clean_loss": 0.8948127568423093,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B0_early_concat",
"seed": 3407,
"epoch": 3,
"train_loss": 0.7437613738907708,
"valid_selection_loss": 0.8850730860626306,
"valid_clean_loss": 0.8822627912510882,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B0_early_concat",
"seed": 3407,
"epoch": 4,
"train_loss": 0.682034013999833,
"valid_selection_loss": 0.8985609801915976,
"valid_clean_loss": 0.8979688529129867,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B0_early_concat",
"seed": 3407,
"epoch": 5,
"train_loss": 0.6572728455066681,
"valid_selection_loss": 0.9590781547211028,
"valid_clean_loss": 0.9659944576221507,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B0_early_concat",
"seed": 3407,
"epoch": 6,
"train_loss": 0.5803193593466723,
"valid_selection_loss": 0.9992094658888303,
"valid_clean_loss": 1.013474199798081,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
}
],
"mask_counts": {
"0.7/async": 1003,
"0.0/async": 1014,
"0.5/sync": 1033,
"0.1/sync": 1024,
"0.5/partial": 996,
"0.3/partial": 1019,
"0.5/async": 1053,
"0.3/single": 1003,
"0.0/sync": 1042,
"0.7/partial": 1079,
"0.7/sync": 986,
"0.0/single": 1029,
"0.1/async": 959,
"0.3/sync": 1055,
"0.0/partial": 1016,
"0.1/single": 1011,
"0.1/partial": 988,
"0.5/single": 1076,
"0.7/single": 997,
"0.3/async": 987
}
}
final/experiments/q3/ati_ho/models/B0_early_concat/seed_42/model_best.pt
BIN
View File
Binary file not shown.
final/experiments/q3/ati_ho/models/B0_early_concat/seed_42/training_history.csv
+6
View File
@@ -0,0 +1,6 @@
method,seed,epoch,train_loss,valid_selection_loss,valid_clean_loss,lambda_interaction,lambda_mask
B0_early_concat,42,1,0.9883709682358636,0.9093860420551929,0.9040598345326853,0.0,0.0
B0_early_concat,42,2,0.8108003448556971,0.8755410200619436,0.8639584504641019,0.0,0.0
B0_early_concat,42,3,0.7378355805520658,0.8942322922604424,0.8897933003666637,0.0,0.0
B0_early_concat,42,4,0.6694756401357828,0.9000292073239337,0.8993391833462558,0.0,0.0
B0_early_concat,42,5,0.6367529040133512,0.9381853456680591,0.9431815828595843,0.0,0.0
1 method seed epoch train_loss valid_selection_loss valid_clean_loss lambda_interaction lambda_mask
2 B0_early_concat 42 1 0.9883709682358636 0.9093860420551929 0.9040598345326853 0.0 0.0
3 B0_early_concat 42 2 0.8108003448556971 0.8755410200619436 0.8639584504641019 0.0 0.0
4 B0_early_concat 42 3 0.7378355805520658 0.8942322922604424 0.8897933003666637 0.0 0.0
5 B0_early_concat 42 4 0.6694756401357828 0.9000292073239337 0.8993391833462558 0.0 0.0
6 B0_early_concat 42 5 0.6367529040133512 0.9381853456680591 0.9431815828595843 0.0 0.0
final/experiments/q3/ati_ho/models/B0_early_concat/seed_42/training_manifest.json
+118
View File
@@ -0,0 +1,118 @@
{
"method": "B0_early_concat",
"seed": 42,
"best_epoch": 2,
"best_selection_loss": 0.8755410200619436,
"elapsed_seconds": 6.863459730986506,
"batch_size": 64,
"epoch_limit": 12,
"patience": 3,
"optimizer": "AdamW",
"learning_rate": 0.0003,
"weight_decay": 0.001,
"gradient_clip_norm": 1.0,
"training_mask_rates": [
0.0,
0.1,
0.3,
0.5,
0.7
],
"training_mask_patterns": [
"single",
"sync",
"partial",
"async"
],
"training_mask_seed_base": 20261227,
"same_orders_and_masks_across_methods_for_same_seed": true,
"config": {
"model_config": {
"router": "mlp",
"expert_names": [
"T",
"A",
"V",
"TA",
"TV",
"AV",
"TAV"
],
"availability_mode": "hard"
}
},
"history": [
{
"method": "B0_early_concat",
"seed": 42,
"epoch": 1,
"train_loss": 0.9883709682358636,
"valid_selection_loss": 0.9093860420551929,
"valid_clean_loss": 0.9040598345326853,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B0_early_concat",
"seed": 42,
"epoch": 2,
"train_loss": 0.8108003448556971,
"valid_selection_loss": 0.8755410200619436,
"valid_clean_loss": 0.8639584504641019,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B0_early_concat",
"seed": 42,
"epoch": 3,
"train_loss": 0.7378355805520658,
"valid_selection_loss": 0.8942322922604424,
"valid_clean_loss": 0.8897933003666637,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B0_early_concat",
"seed": 42,
"epoch": 4,
"train_loss": 0.6694756401357828,
"valid_selection_loss": 0.9000292073239337,
"valid_clean_loss": 0.8993391833462558,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B0_early_concat",
"seed": 42,
"epoch": 5,
"train_loss": 0.6367529040133512,
"valid_selection_loss": 0.9381853456680591,
"valid_clean_loss": 0.9431815828595843,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
}
],
"mask_counts": {
"0.0/single": 838,
"0.3/single": 841,
"0.5/single": 832,
"0.7/sync": 931,
"0.7/partial": 854,
"0.0/async": 818,
"0.3/async": 856,
"0.3/partial": 861,
"0.3/sync": 841,
"0.1/single": 857,
"0.5/async": 832,
"0.7/single": 872,
"0.5/sync": 848,
"0.5/partial": 854,
"0.0/partial": 870,
"0.0/sync": 827,
"0.7/async": 831,
"0.1/sync": 839,
"0.1/async": 847,
"0.1/partial": 826
}
}
final/experiments/q3/ati_ho/models/B5_mofe_mlp/seed_2026/model_best.pt
BIN
View File
Binary file not shown.
final/experiments/q3/ati_ho/models/B5_mofe_mlp/seed_2026/training_history.csv
+7
View File
@@ -0,0 +1,7 @@
method,seed,epoch,train_loss,valid_selection_loss,valid_clean_loss,lambda_interaction,lambda_mask
B5_mofe_mlp,2026,1,1.0356682870123122,0.9561890172106879,0.9502097872587351,0.0,0.0
B5_mofe_mlp,2026,2,0.8621677259604136,0.8906986942658057,0.8774622966954996,0.0,0.0
B5_mofe_mlp,2026,3,0.7639763178648772,0.8864746625934329,0.8795440236290732,0.0,0.0
B5_mofe_mlp,2026,4,0.7062149555594833,0.8968080696496334,0.898658872960688,0.0,0.0
B5_mofe_mlp,2026,5,0.6749658584594727,0.9011682678054977,0.9081905451449719,0.0,0.0
B5_mofe_mlp,2026,6,0.6303896639082167,0.9725698265400562,0.9927981803705405,0.0,0.0
1 method seed epoch train_loss valid_selection_loss valid_clean_loss lambda_interaction lambda_mask
2 B5_mofe_mlp 2026 1 1.0356682870123122 0.9561890172106879 0.9502097872587351 0.0 0.0
3 B5_mofe_mlp 2026 2 0.8621677259604136 0.8906986942658057 0.8774622966954996 0.0 0.0
4 B5_mofe_mlp 2026 3 0.7639763178648772 0.8864746625934329 0.8795440236290732 0.0 0.0
5 B5_mofe_mlp 2026 4 0.7062149555594833 0.8968080696496334 0.898658872960688 0.0 0.0
6 B5_mofe_mlp 2026 5 0.6749658584594727 0.9011682678054977 0.9081905451449719 0.0 0.0
7 B5_mofe_mlp 2026 6 0.6303896639082167 0.9725698265400562 0.9927981803705405 0.0 0.0
final/experiments/q3/ati_ho/models/B5_mofe_mlp/seed_2026/training_manifest.json
+128
View File
@@ -0,0 +1,128 @@
{
"method": "B5_mofe_mlp",
"seed": 2026,
"best_epoch": 3,
"best_selection_loss": 0.8864746625934329,
"elapsed_seconds": 4.902324867958669,
"batch_size": 64,
"epoch_limit": 12,
"patience": 3,
"optimizer": "AdamW",
"learning_rate": 0.0003,
"weight_decay": 0.001,
"gradient_clip_norm": 1.0,
"training_mask_rates": [
0.0,
0.1,
0.3,
0.5,
0.7
],
"training_mask_patterns": [
"single",
"sync",
"partial",
"async"
],
"training_mask_seed_base": 20261227,
"same_orders_and_masks_across_methods_for_same_seed": true,
"config": {
"model_config": {
"router": "mlp",
"expert_names": [
"T",
"A",
"V",
"TA",
"TV",
"AV",
"TAV"
],
"availability_mode": "hard"
}
},
"history": [
{
"method": "B5_mofe_mlp",
"seed": 2026,
"epoch": 1,
"train_loss": 1.0356682870123122,
"valid_selection_loss": 0.9561890172106879,
"valid_clean_loss": 0.9502097872587351,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B5_mofe_mlp",
"seed": 2026,
"epoch": 2,
"train_loss": 0.8621677259604136,
"valid_selection_loss": 0.8906986942658057,
"valid_clean_loss": 0.8774622966954996,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B5_mofe_mlp",
"seed": 2026,
"epoch": 3,
"train_loss": 0.7639763178648772,
"valid_selection_loss": 0.8864746625934329,
"valid_clean_loss": 0.8795440236290732,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B5_mofe_mlp",
"seed": 2026,
"epoch": 4,
"train_loss": 0.7062149555594833,
"valid_selection_loss": 0.8968080696496334,
"valid_clean_loss": 0.898658872960688,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B5_mofe_mlp",
"seed": 2026,
"epoch": 5,
"train_loss": 0.6749658584594727,
"valid_selection_loss": 0.9011682678054977,
"valid_clean_loss": 0.9081905451449719,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B5_mofe_mlp",
"seed": 2026,
"epoch": 6,
"train_loss": 0.6303896639082167,
"valid_selection_loss": 0.9725698265400562,
"valid_clean_loss": 0.9927981803705405,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
}
],
"mask_counts": {
"0.1/async": 970,
"0.0/async": 995,
"0.5/sync": 1040,
"0.0/sync": 1012,
"0.5/single": 999,
"0.0/single": 1046,
"0.7/async": 1048,
"0.0/partial": 1063,
"0.7/sync": 1020,
"0.5/partial": 1022,
"0.3/partial": 1002,
"0.1/partial": 1022,
"0.3/single": 1004,
"0.3/async": 1026,
"0.5/async": 1028,
"0.1/single": 1048,
"0.7/single": 959,
"0.3/sync": 1030,
"0.1/sync": 1007,
"0.7/partial": 1029
}
}
final/experiments/q3/ati_ho/models/B5_mofe_mlp/seed_3407/model_best.pt
BIN
View File
Binary file not shown.
final/experiments/q3/ati_ho/models/B5_mofe_mlp/seed_3407/training_history.csv
+8
View File
@@ -0,0 +1,8 @@
method,seed,epoch,train_loss,valid_selection_loss,valid_clean_loss,lambda_interaction,lambda_mask
B5_mofe_mlp,3407,1,1.0173597953937672,0.9501426021803867,0.9452906283703479,0.0,0.0
B5_mofe_mlp,3407,2,0.8566740729190685,0.922284932254435,0.9245965945851672,0.0,0.0
B5_mofe_mlp,3407,3,0.7812010469260039,0.8953835820103739,0.901839215021867,0.0,0.0
B5_mofe_mlp,3407,4,0.7314516603946686,0.8908331908367492,0.8956135635847574,0.0,0.0
B5_mofe_mlp,3407,5,0.7109767960177528,0.941465525836735,0.9525768114970281,0.0,0.0
B5_mofe_mlp,3407,6,0.6440567677771604,0.9769726883579086,1.0012760489851564,0.0,0.0
B5_mofe_mlp,3407,7,0.609118523421111,1.0052462698339106,1.0282764369314843,0.0,0.0
1 method seed epoch train_loss valid_selection_loss valid_clean_loss lambda_interaction lambda_mask
2 B5_mofe_mlp 3407 1 1.0173597953937672 0.9501426021803867 0.9452906283703479 0.0 0.0
3 B5_mofe_mlp 3407 2 0.8566740729190685 0.922284932254435 0.9245965945851672 0.0 0.0
4 B5_mofe_mlp 3407 3 0.7812010469260039 0.8953835820103739 0.901839215021867 0.0 0.0
5 B5_mofe_mlp 3407 4 0.7314516603946686 0.8908331908367492 0.8956135635847574 0.0 0.0
6 B5_mofe_mlp 3407 5 0.7109767960177528 0.941465525836735 0.9525768114970281 0.0 0.0
7 B5_mofe_mlp 3407 6 0.6440567677771604 0.9769726883579086 1.0012760489851564 0.0 0.0
8 B5_mofe_mlp 3407 7 0.609118523421111 1.0052462698339106 1.0282764369314843 0.0 0.0
final/experiments/q3/ati_ho/models/B5_mofe_mlp/seed_3407/training_manifest.json
+138
View File
@@ -0,0 +1,138 @@
{
"method": "B5_mofe_mlp",
"seed": 3407,
"best_epoch": 4,
"best_selection_loss": 0.8908331908367492,
"elapsed_seconds": 5.8394740910152905,
"batch_size": 64,
"epoch_limit": 12,
"patience": 3,
"optimizer": "AdamW",
"learning_rate": 0.0003,
"weight_decay": 0.001,
"gradient_clip_norm": 1.0,
"training_mask_rates": [
0.0,
0.1,
0.3,
0.5,
0.7
],
"training_mask_patterns": [
"single",
"sync",
"partial",
"async"
],
"training_mask_seed_base": 20261227,
"same_orders_and_masks_across_methods_for_same_seed": true,
"config": {
"model_config": {
"router": "mlp",
"expert_names": [
"T",
"A",
"V",
"TA",
"TV",
"AV",
"TAV"
],
"availability_mode": "hard"
}
},
"history": [
{
"method": "B5_mofe_mlp",
"seed": 3407,
"epoch": 1,
"train_loss": 1.0173597953937672,
"valid_selection_loss": 0.9501426021803867,
"valid_clean_loss": 0.9452906283703479,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B5_mofe_mlp",
"seed": 3407,
"epoch": 2,
"train_loss": 0.8566740729190685,
"valid_selection_loss": 0.922284932254435,
"valid_clean_loss": 0.9245965945851672,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B5_mofe_mlp",
"seed": 3407,
"epoch": 3,
"train_loss": 0.7812010469260039,
"valid_selection_loss": 0.8953835820103739,
"valid_clean_loss": 0.901839215021867,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B5_mofe_mlp",
"seed": 3407,
"epoch": 4,
"train_loss": 0.7314516603946686,
"valid_selection_loss": 0.8908331908367492,
"valid_clean_loss": 0.8956135635847574,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B5_mofe_mlp",
"seed": 3407,
"epoch": 5,
"train_loss": 0.7109767960177528,
"valid_selection_loss": 0.941465525836735,
"valid_clean_loss": 0.9525768114970281,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B5_mofe_mlp",
"seed": 3407,
"epoch": 6,
"train_loss": 0.6440567677771604,
"valid_selection_loss": 0.9769726883579086,
"valid_clean_loss": 1.0012760489851564,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B5_mofe_mlp",
"seed": 3407,
"epoch": 7,
"train_loss": 0.609118523421111,
"valid_selection_loss": 1.0052462698339106,
"valid_clean_loss": 1.0282764369314843,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
}
],
"mask_counts": {
"0.7/async": 1162,
"0.0/async": 1173,
"0.5/sync": 1215,
"0.1/sync": 1215,
"0.5/partial": 1175,
"0.3/partial": 1195,
"0.5/async": 1228,
"0.3/single": 1162,
"0.0/sync": 1197,
"0.7/partial": 1263,
"0.7/sync": 1165,
"0.0/single": 1199,
"0.1/async": 1114,
"0.3/sync": 1218,
"0.0/partial": 1182,
"0.1/single": 1187,
"0.1/partial": 1158,
"0.5/single": 1238,
"0.7/single": 1161,
"0.3/async": 1158
}
}
final/experiments/q3/ati_ho/models/B5_mofe_mlp/seed_42/model_best.pt
BIN
View File
Binary file not shown.
final/experiments/q3/ati_ho/models/B5_mofe_mlp/seed_42/training_history.csv
+8
View File
@@ -0,0 +1,8 @@
method,seed,epoch,train_loss,valid_selection_loss,valid_clean_loss,lambda_interaction,lambda_mask
B5_mofe_mlp,42,1,1.0171989544674203,0.9322710561228322,0.9271787070966029,0.0,0.0
B5_mofe_mlp,42,2,0.8356565082514728,0.9130383135525735,0.9016369266824408,0.0,0.0
B5_mofe_mlp,42,3,0.7714545528093973,0.9085815730658207,0.9123272260466775,0.0,0.0
B5_mofe_mlp,42,4,0.7048188387243836,0.9027628702121776,0.9053291953527011,0.0,0.0
B5_mofe_mlp,42,5,0.6746232868344696,0.9353020928063236,0.9397049303893205,0.0,0.0
B5_mofe_mlp,42,6,0.6332692697092339,1.025483563706115,1.0391453098464798,0.0,0.0
B5_mofe_mlp,42,7,0.5814774891844502,1.0453854943369771,1.060925617322817,0.0,0.0
1 method seed epoch train_loss valid_selection_loss valid_clean_loss lambda_interaction lambda_mask
2 B5_mofe_mlp 42 1 1.0171989544674203 0.9322710561228322 0.9271787070966029 0.0 0.0
3 B5_mofe_mlp 42 2 0.8356565082514728 0.9130383135525735 0.9016369266824408 0.0 0.0
4 B5_mofe_mlp 42 3 0.7714545528093973 0.9085815730658207 0.9123272260466775 0.0 0.0
5 B5_mofe_mlp 42 4 0.7048188387243836 0.9027628702121776 0.9053291953527011 0.0 0.0
6 B5_mofe_mlp 42 5 0.6746232868344696 0.9353020928063236 0.9397049303893205 0.0 0.0
7 B5_mofe_mlp 42 6 0.6332692697092339 1.025483563706115 1.0391453098464798 0.0 0.0
8 B5_mofe_mlp 42 7 0.5814774891844502 1.0453854943369771 1.060925617322817 0.0 0.0
final/experiments/q3/ati_ho/models/B5_mofe_mlp/seed_42/training_manifest.json
+138
View File
@@ -0,0 +1,138 @@
{
"method": "B5_mofe_mlp",
"seed": 42,
"best_epoch": 4,
"best_selection_loss": 0.9027628702121776,
"elapsed_seconds": 6.507645265024621,
"batch_size": 64,
"epoch_limit": 12,
"patience": 3,
"optimizer": "AdamW",
"learning_rate": 0.0003,
"weight_decay": 0.001,
"gradient_clip_norm": 1.0,
"training_mask_rates": [
0.0,
0.1,
0.3,
0.5,
0.7
],
"training_mask_patterns": [
"single",
"sync",
"partial",
"async"
],
"training_mask_seed_base": 20261227,
"same_orders_and_masks_across_methods_for_same_seed": true,
"config": {
"model_config": {
"router": "mlp",
"expert_names": [
"T",
"A",
"V",
"TA",
"TV",
"AV",
"TAV"
],
"availability_mode": "hard"
}
},
"history": [
{
"method": "B5_mofe_mlp",
"seed": 42,
"epoch": 1,
"train_loss": 1.0171989544674203,
"valid_selection_loss": 0.9322710561228322,
"valid_clean_loss": 0.9271787070966029,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B5_mofe_mlp",
"seed": 42,
"epoch": 2,
"train_loss": 0.8356565082514728,
"valid_selection_loss": 0.9130383135525735,
"valid_clean_loss": 0.9016369266824408,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B5_mofe_mlp",
"seed": 42,
"epoch": 3,
"train_loss": 0.7714545528093973,
"valid_selection_loss": 0.9085815730658207,
"valid_clean_loss": 0.9123272260466775,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B5_mofe_mlp",
"seed": 42,
"epoch": 4,
"train_loss": 0.7048188387243836,
"valid_selection_loss": 0.9027628702121776,
"valid_clean_loss": 0.9053291953527011,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B5_mofe_mlp",
"seed": 42,
"epoch": 5,
"train_loss": 0.6746232868344696,
"valid_selection_loss": 0.9353020928063236,
"valid_clean_loss": 0.9397049303893205,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B5_mofe_mlp",
"seed": 42,
"epoch": 6,
"train_loss": 0.6332692697092339,
"valid_selection_loss": 1.025483563706115,
"valid_clean_loss": 1.0391453098464798,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
},
{
"method": "B5_mofe_mlp",
"seed": 42,
"epoch": 7,
"train_loss": 0.5814774891844502,
"valid_selection_loss": 1.0453854943369771,
"valid_clean_loss": 1.060925617322817,
"lambda_interaction": 0.0,
"lambda_mask": 0.0
}
],
"mask_counts": {
"0.0/single": 1169,
"0.3/single": 1187,
"0.5/single": 1158,
"0.7/sync": 1272,
"0.7/partial": 1207,
"0.0/async": 1145,
"0.3/async": 1200,
"0.3/partial": 1209,
"0.3/sync": 1170,
"0.1/single": 1205,
"0.5/async": 1185,
"0.7/single": 1232,
"0.5/sync": 1143,
"0.5/partial": 1236,
"0.0/partial": 1183,
"0.0/sync": 1159,
"0.7/async": 1179,
"0.1/sync": 1202,
"0.1/async": 1192,
"0.1/partial": 1132
}
}
final/experiments/q3/ati_ho/models/D0/seed_42/model_best.pt
BIN
View File
Binary file not shown.
final/experiments/q3/ati_ho/models/D0/seed_42/training_history.csv
+6
View File
@@ -0,0 +1,6 @@
method,seed,epoch,train_loss,valid_selection_loss,valid_clean_loss,lambda_interaction,lambda_mask
D0,42,1,0.9736692993729202,0.9111956561004722,0.9047908566810272,0.001,0.0
D0,42,2,0.7995399965180291,0.8630276628575482,0.8526584676333836,0.001,0.0
D0,42,3,0.7263524250851737,0.8714345292403147,0.863228269985744,0.001,0.0
D0,42,4,0.6683531569110023,0.871083995961881,0.8627031235904484,0.001,0.0
D0,42,5,0.6370693781861553,0.8886248764100966,0.8807592830815159,0.001,0.0
1 method seed epoch train_loss valid_selection_loss valid_clean_loss lambda_interaction lambda_mask
2 D0 42 1 0.9736692993729202 0.9111956561004722 0.9047908566810272 0.001 0.0
3 D0 42 2 0.7995399965180291 0.8630276628575482 0.8526584676333836 0.001 0.0
4 D0 42 3 0.7263524250851737 0.8714345292403147 0.863228269985744 0.001 0.0
5 D0 42 4 0.6683531569110023 0.871083995961881 0.8627031235904484 0.001 0.0
6 D0 42 5 0.6370693781861553 0.8886248764100966 0.8807592830815159 0.001 0.0
final/experiments/q3/ati_ho/models/D0/seed_42/training_manifest.json
+117
View File
@@ -0,0 +1,117 @@
{
"method": "D0",
"seed": 42,
"best_epoch": 2,
"best_selection_loss": 0.8630276628575482,
"elapsed_seconds": 8.120354533020873,
"batch_size": 64,
"epoch_limit": 12,
"patience": 3,
"optimizer": "AdamW",
"learning_rate": 0.0003,
"weight_decay": 0.001,
"gradient_clip_norm": 1.0,
"training_mask_rates": [
0.0,
0.1,
0.3,
0.5,
0.7
],
"training_mask_patterns": [
"single",
"sync",
"partial",
"async"
],
"training_mask_seed_base": 20261227,
"same_orders_and_masks_across_methods_for_same_seed": true,
"config": {
"name": "D0_unanchored_diagnostic",
"low_rank": true,
"cross_attention": true,
"anchored": false,
"lambda_interaction": 0.001,
"lambda_mask": 0.0,
"rank": 4,
"hidden": 64,
"gru_hidden_per_direction": 32,
"attention_heads": 4,
"attention_ffn": 128,
"eta_init": 0.1
},
"history": [
{
"method": "D0",
"seed": 42,
"epoch": 1,
"train_loss": 0.9736692993729202,
"valid_selection_loss": 0.9111956561004722,
"valid_clean_loss": 0.9047908566810272,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "D0",
"seed": 42,
"epoch": 2,
"train_loss": 0.7995399965180291,
"valid_selection_loss": 0.8630276628575482,
"valid_clean_loss": 0.8526584676333836,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "D0",
"seed": 42,
"epoch": 3,
"train_loss": 0.7263524250851737,
"valid_selection_loss": 0.8714345292403147,
"valid_clean_loss": 0.863228269985744,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "D0",
"seed": 42,
"epoch": 4,
"train_loss": 0.6683531569110023,
"valid_selection_loss": 0.871083995961881,
"valid_clean_loss": 0.8627031235904484,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
},
{
"method": "D0",
"seed": 42,
"epoch": 5,
"train_loss": 0.6370693781861553,
"valid_selection_loss": 0.8886248764100966,
"valid_clean_loss": 0.8807592830815159,
"lambda_interaction": 0.001,
"lambda_mask": 0.0
}
],
"mask_counts": {
"0.0/single": 838,
"0.3/single": 841,
"0.5/single": 832,
"0.7/sync": 931,
"0.7/partial": 854,
"0.0/async": 818,
"0.3/async": 856,
"0.3/partial": 861,
"0.3/sync": 841,
"0.1/single": 857,
"0.5/async": 832,
"0.7/single": 872,
"0.5/sync": 848,
"0.5/partial": 854,
"0.0/partial": 870,
"0.0/sync": 827,
"0.7/async": 831,
"0.1/sync": 839,
"0.1/async": 847,
"0.1/partial": 826
}
}