Add files using upload-large-folder tool

logs/brainfm_frozen_mlp.log

@@ -0,0 +1,389 @@
+device=cuda backbone=brainfm encoder_scope=none contrastive_weight=0.2 regression_weight=1.0 train=710 val=152
+epoch=1 step=20/710 loss=0.1219
+epoch=1 step=40/710 loss=0.0105
+epoch=1 step=60/710 loss=0.0214
+epoch=1 step=80/710 loss=0.0091
+epoch=1 step=100/710 loss=0.1297
+epoch=1 step=120/710 loss=0.0441
+epoch=1 step=140/710 loss=0.0076
+epoch=1 step=160/710 loss=0.0079
+epoch=1 step=180/710 loss=0.0164
+epoch=1 step=200/710 loss=0.0376
+epoch=1 step=220/710 loss=0.0214
+epoch=1 step=240/710 loss=0.0293
+epoch=1 step=260/710 loss=0.0251
+epoch=1 step=280/710 loss=0.0192
+epoch=1 step=300/710 loss=0.0085
+epoch=1 step=320/710 loss=0.0298
+epoch=1 step=340/710 loss=0.0086
+epoch=1 step=360/710 loss=0.0206
+epoch=1 step=380/710 loss=0.0124
+epoch=1 step=400/710 loss=0.0095
+epoch=1 step=420/710 loss=0.0078
+epoch=1 step=440/710 loss=0.0333
+epoch=1 step=460/710 loss=0.0349
+epoch=1 step=480/710 loss=0.0124
+epoch=1 step=500/710 loss=0.0074
+epoch=1 step=520/710 loss=0.0168
+epoch=1 step=540/710 loss=0.0059
+epoch=1 step=560/710 loss=0.0101
+epoch=1 step=580/710 loss=0.0308
+epoch=1 step=600/710 loss=0.0074
+epoch=1 step=620/710 loss=0.0307
+epoch=1 step=640/710 loss=0.0057
+epoch=1 step=660/710 loss=0.0178
+epoch=1 step=680/710 loss=0.0072
+epoch=1 step=700/710 loss=0.0187
+epoch=1 train_loss=0.0362 train_contrastive=0.0000 train_regression=0.0362 val_loss=0.0207 val_contrastive=0.0000 val_regression=0.0207
+saved_best runs/foundation/brainfm_frozen_mlp_best.pt val_loss=0.0207 epoch=1
+epoch=2 step=20/710 loss=0.0268
+epoch=2 step=40/710 loss=0.0042
+epoch=2 step=60/710 loss=0.0127
+epoch=2 step=80/710 loss=0.0408
+epoch=2 step=100/710 loss=0.0159
+epoch=2 step=120/710 loss=0.0482
+epoch=2 step=140/710 loss=0.0084
+epoch=2 step=160/710 loss=0.0065
+epoch=2 step=180/710 loss=0.0040
+epoch=2 step=200/710 loss=0.0069
+epoch=2 step=220/710 loss=0.0254
+epoch=2 step=240/710 loss=0.0086
+epoch=2 step=260/710 loss=0.0254
+epoch=2 step=280/710 loss=0.0143
+epoch=2 step=300/710 loss=0.0052
+epoch=2 step=320/710 loss=0.0273
+epoch=2 step=340/710 loss=0.0073
+epoch=2 step=360/710 loss=0.0121
+epoch=2 step=380/710 loss=0.0200
+epoch=2 step=400/710 loss=0.0241
+epoch=2 step=420/710 loss=0.0365
+epoch=2 step=440/710 loss=0.0057
+epoch=2 step=460/710 loss=0.0172
+epoch=2 step=480/710 loss=0.0309
+epoch=2 step=500/710 loss=0.0059
+epoch=2 step=520/710 loss=0.0088
+epoch=2 step=540/710 loss=0.0226
+epoch=2 step=560/710 loss=0.0319
+epoch=2 step=580/710 loss=0.0676
+epoch=2 step=600/710 loss=0.0083
+epoch=2 step=620/710 loss=0.0048
+epoch=2 step=640/710 loss=0.0119
+epoch=2 step=660/710 loss=0.1012
+epoch=2 step=680/710 loss=0.0222
+epoch=2 step=700/710 loss=0.0062
+epoch=2 train_loss=0.0200 train_contrastive=0.0000 train_regression=0.0200 val_loss=0.0267 val_contrastive=0.0000 val_regression=0.0267
+epoch=3 step=20/710 loss=0.0168
+epoch=3 step=40/710 loss=0.0225
+epoch=3 step=60/710 loss=0.0235
+epoch=3 step=80/710 loss=0.0067
+epoch=3 step=100/710 loss=0.0077
+epoch=3 step=120/710 loss=0.0419
+epoch=3 step=140/710 loss=0.0065
+epoch=3 step=160/710 loss=0.0114
+epoch=3 step=180/710 loss=0.0102
+epoch=3 step=200/710 loss=0.0121
+epoch=3 step=220/710 loss=0.0349
+epoch=3 step=240/710 loss=0.0239
+epoch=3 step=260/710 loss=0.0103
+epoch=3 step=280/710 loss=0.0055
+epoch=3 step=300/710 loss=0.0082
+epoch=3 step=320/710 loss=0.0095
+epoch=3 step=340/710 loss=0.0063
+epoch=3 step=360/710 loss=0.0067
+epoch=3 step=380/710 loss=0.0055
+epoch=3 step=400/710 loss=0.0123
+epoch=3 step=420/710 loss=0.0194
+epoch=3 step=440/710 loss=0.0148
+epoch=3 step=460/710 loss=0.0194
+epoch=3 step=480/710 loss=0.0156
+epoch=3 step=500/710 loss=0.0071
+epoch=3 step=520/710 loss=0.0170
+epoch=3 step=540/710 loss=0.0062
+epoch=3 step=560/710 loss=0.0065
+epoch=3 step=580/710 loss=0.0554
+epoch=3 step=600/710 loss=0.0200
+epoch=3 step=620/710 loss=0.0050
+epoch=3 step=640/710 loss=0.0064
+epoch=3 step=660/710 loss=0.0871
+epoch=3 step=680/710 loss=0.0047
+epoch=3 step=700/710 loss=0.0385
+epoch=3 train_loss=0.0180 train_contrastive=0.0000 train_regression=0.0180 val_loss=0.0201 val_contrastive=0.0000 val_regression=0.0201
+saved_best runs/foundation/brainfm_frozen_mlp_best.pt val_loss=0.0201 epoch=3
+epoch=4 step=20/710 loss=0.0103
+epoch=4 step=40/710 loss=0.0239
+epoch=4 step=60/710 loss=0.0224
+epoch=4 step=80/710 loss=0.0101
+epoch=4 step=100/710 loss=0.0346
+epoch=4 step=120/710 loss=0.0075
+epoch=4 step=140/710 loss=0.0067
+epoch=4 step=160/710 loss=0.0056
+epoch=4 step=180/710 loss=0.0055
+epoch=4 step=200/710 loss=0.0044
+epoch=4 step=220/710 loss=0.0418
+epoch=4 step=240/710 loss=0.0084
+epoch=4 step=260/710 loss=0.0121
+epoch=4 step=280/710 loss=0.0048
+epoch=4 step=300/710 loss=0.0195
+epoch=4 step=320/710 loss=0.0150
+epoch=4 step=340/710 loss=0.0311
+epoch=4 step=360/710 loss=0.0201
+epoch=4 step=380/710 loss=0.0056
+epoch=4 step=400/710 loss=0.0142
+epoch=4 step=420/710 loss=0.0063
+epoch=4 step=440/710 loss=0.0326
+epoch=4 step=460/710 loss=0.0045
+epoch=4 step=480/710 loss=0.0062
+epoch=4 step=500/710 loss=0.0091
+epoch=4 step=520/710 loss=0.0232
+epoch=4 step=540/710 loss=0.0154
+epoch=4 step=560/710 loss=0.0092
+epoch=4 step=580/710 loss=0.0074
+epoch=4 step=600/710 loss=0.0124
+epoch=4 step=620/710 loss=0.0061
+epoch=4 step=640/710 loss=0.0121
+epoch=4 step=660/710 loss=0.0105
+epoch=4 step=680/710 loss=0.0165
+epoch=4 step=700/710 loss=0.0198
+epoch=4 train_loss=0.0166 train_contrastive=0.0000 train_regression=0.0166 val_loss=0.0182 val_contrastive=0.0000 val_regression=0.0182
+saved_best runs/foundation/brainfm_frozen_mlp_best.pt val_loss=0.0182 epoch=4
+epoch=5 step=20/710 loss=0.0120
+epoch=5 step=40/710 loss=0.0027
+epoch=5 step=60/710 loss=0.0222
+epoch=5 step=80/710 loss=0.0112
+epoch=5 step=100/710 loss=0.0080
+epoch=5 step=120/710 loss=0.0057
+epoch=5 step=140/710 loss=0.0154
+epoch=5 step=160/710 loss=0.0086
+epoch=5 step=180/710 loss=0.0128
+epoch=5 step=200/710 loss=0.0237
+epoch=5 step=220/710 loss=0.0049
+epoch=5 step=240/710 loss=0.0189
+epoch=5 step=260/710 loss=0.0063
+epoch=5 step=280/710 loss=0.0235
+epoch=5 step=300/710 loss=0.0079
+epoch=5 step=320/710 loss=0.0108
+epoch=5 step=340/710 loss=0.0053
+epoch=5 step=360/710 loss=0.0155
+epoch=5 step=380/710 loss=0.0082
+epoch=5 step=400/710 loss=0.0072
+epoch=5 step=420/710 loss=0.0065
+epoch=5 step=440/710 loss=0.0081
+epoch=5 step=460/710 loss=0.0178
+epoch=5 step=480/710 loss=0.0062
+epoch=5 step=500/710 loss=0.0152
+epoch=5 step=520/710 loss=0.0227
+epoch=5 step=540/710 loss=0.0071
+epoch=5 step=560/710 loss=0.0093
+epoch=5 step=580/710 loss=0.0076
+epoch=5 step=600/710 loss=0.0073
+epoch=5 step=620/710 loss=0.0051
+epoch=5 step=640/710 loss=0.0212
+epoch=5 step=660/710 loss=0.0051
+epoch=5 step=680/710 loss=0.0145
+epoch=5 step=700/710 loss=0.0069
+epoch=5 train_loss=0.0155 train_contrastive=0.0000 train_regression=0.0155 val_loss=0.0169 val_contrastive=0.0000 val_regression=0.0169
+saved_best runs/foundation/brainfm_frozen_mlp_best.pt val_loss=0.0169 epoch=5
+epoch=6 step=20/710 loss=0.0088
+epoch=6 step=40/710 loss=0.0089
+epoch=6 step=60/710 loss=0.0162
+epoch=6 step=80/710 loss=0.0102
+epoch=6 step=100/710 loss=0.0137
+epoch=6 step=120/710 loss=0.0063
+epoch=6 step=140/710 loss=0.0151
+epoch=6 step=160/710 loss=0.0039
+epoch=6 step=180/710 loss=0.0113
+epoch=6 step=200/710 loss=0.0086
+epoch=6 step=220/710 loss=0.0206
+epoch=6 step=240/710 loss=0.0211
+epoch=6 step=260/710 loss=0.0146
+epoch=6 step=280/710 loss=0.0186
+epoch=6 step=300/710 loss=0.0052
+epoch=6 step=320/710 loss=0.0126
+epoch=6 step=340/710 loss=0.0056
+epoch=6 step=360/710 loss=0.0315
+epoch=6 step=380/710 loss=0.0119
+epoch=6 step=400/710 loss=0.0245
+epoch=6 step=420/710 loss=0.0093
+epoch=6 step=440/710 loss=0.0310
+epoch=6 step=460/710 loss=0.0203
+epoch=6 step=480/710 loss=0.0084
+epoch=6 step=500/710 loss=0.0056
+epoch=6 step=520/710 loss=0.0063
+epoch=6 step=540/710 loss=0.0105
+epoch=6 step=560/710 loss=0.0072
+epoch=6 step=580/710 loss=0.0088
+epoch=6 step=600/710 loss=0.0059
+epoch=6 step=620/710 loss=0.0081
+epoch=6 step=640/710 loss=0.0082
+epoch=6 step=660/710 loss=0.0127
+epoch=6 step=680/710 loss=0.0043
+epoch=6 step=700/710 loss=0.0075
+epoch=6 train_loss=0.0154 train_contrastive=0.0000 train_regression=0.0154 val_loss=0.0173 val_contrastive=0.0000 val_regression=0.0173
+epoch=7 step=20/710 loss=0.0127
+epoch=7 step=40/710 loss=0.0141
+epoch=7 step=60/710 loss=0.0073
+epoch=7 step=80/710 loss=0.0079
+epoch=7 step=100/710 loss=0.0120
+epoch=7 step=120/710 loss=0.0172
+epoch=7 step=140/710 loss=0.0076
+epoch=7 step=160/710 loss=0.0078
+epoch=7 step=180/710 loss=0.0029
+epoch=7 step=200/710 loss=0.0069
+epoch=7 step=220/710 loss=0.0238
+epoch=7 step=240/710 loss=0.0118
+epoch=7 step=260/710 loss=0.0088
+epoch=7 step=280/710 loss=0.0085
+epoch=7 step=300/710 loss=0.0064
+epoch=7 step=320/710 loss=0.0052
+epoch=7 step=340/710 loss=0.0061
+epoch=7 step=360/710 loss=0.0048
+epoch=7 step=380/710 loss=0.0123
+epoch=7 step=400/710 loss=0.0213
+epoch=7 step=420/710 loss=0.0098
+epoch=7 step=440/710 loss=0.0138
+epoch=7 step=460/710 loss=0.0101
+epoch=7 step=480/710 loss=0.0150
+epoch=7 step=500/710 loss=0.0046
+epoch=7 step=520/710 loss=0.0207
+epoch=7 step=540/710 loss=0.0088
+epoch=7 step=560/710 loss=0.0096
+epoch=7 step=580/710 loss=0.0073
+epoch=7 step=600/710 loss=0.0033
+epoch=7 step=620/710 loss=0.0079
+epoch=7 step=640/710 loss=0.0125
+epoch=7 step=660/710 loss=0.0141
+epoch=7 step=680/710 loss=0.0101
+epoch=7 step=700/710 loss=0.0089
+epoch=7 train_loss=0.0143 train_contrastive=0.0000 train_regression=0.0143 val_loss=0.0187 val_contrastive=0.0000 val_regression=0.0187
+epoch=8 step=20/710 loss=0.0059
+epoch=8 step=40/710 loss=0.0389
+epoch=8 step=60/710 loss=0.0223
+epoch=8 step=80/710 loss=0.0327
+epoch=8 step=100/710 loss=0.0103
+epoch=8 step=120/710 loss=0.0075
+epoch=8 step=140/710 loss=0.0137
+epoch=8 step=160/710 loss=0.0063
+epoch=8 step=180/710 loss=0.0099
+epoch=8 step=200/710 loss=0.0038
+epoch=8 step=220/710 loss=0.0069
+epoch=8 step=240/710 loss=0.0157
+epoch=8 step=260/710 loss=0.0141
+epoch=8 step=280/710 loss=0.0135
+epoch=8 step=300/710 loss=0.0070
+epoch=8 step=320/710 loss=0.0144
+epoch=8 step=340/710 loss=0.0088
+epoch=8 step=360/710 loss=0.0069
+epoch=8 step=380/710 loss=0.0044
+epoch=8 step=400/710 loss=0.0052
+epoch=8 step=420/710 loss=0.0035
+epoch=8 step=440/710 loss=0.0550
+epoch=8 step=460/710 loss=0.0039
+epoch=8 step=480/710 loss=0.0037
+epoch=8 step=500/710 loss=0.0040
+epoch=8 step=520/710 loss=0.0036
+epoch=8 step=540/710 loss=0.0170
+epoch=8 step=560/710 loss=0.0138
+epoch=8 step=580/710 loss=0.0246
+epoch=8 step=600/710 loss=0.0232
+epoch=8 step=620/710 loss=0.0479
+epoch=8 step=640/710 loss=0.0084
+epoch=8 step=660/710 loss=0.0039
+epoch=8 step=680/710 loss=0.0188
+epoch=8 step=700/710 loss=0.0520
+epoch=8 train_loss=0.0140 train_contrastive=0.0000 train_regression=0.0140 val_loss=0.0155 val_contrastive=0.0000 val_regression=0.0155
+saved_best runs/foundation/brainfm_frozen_mlp_best.pt val_loss=0.0155 epoch=8
+epoch=9 step=20/710 loss=0.0117
+epoch=9 step=40/710 loss=0.0385
+epoch=9 step=60/710 loss=0.0077
+epoch=9 step=80/710 loss=0.0072
+epoch=9 step=100/710 loss=0.0178
+epoch=9 step=120/710 loss=0.0186
+epoch=9 step=140/710 loss=0.0075
+epoch=9 step=160/710 loss=0.0041
+epoch=9 step=180/710 loss=0.0037
+epoch=9 step=200/710 loss=0.0098
+epoch=9 step=220/710 loss=0.0084
+epoch=9 step=240/710 loss=0.0133
+epoch=9 step=260/710 loss=0.0237
+epoch=9 step=280/710 loss=0.0057
+epoch=9 step=300/710 loss=0.0067
+epoch=9 step=320/710 loss=0.0058
+epoch=9 step=340/710 loss=0.0666
+epoch=9 step=360/710 loss=0.0100
+epoch=9 step=380/710 loss=0.0162
+epoch=9 step=400/710 loss=0.0024
+epoch=9 step=420/710 loss=0.0077
+epoch=9 step=440/710 loss=0.0044
+epoch=9 step=460/710 loss=0.0045
+epoch=9 step=480/710 loss=0.0074
+epoch=9 step=500/710 loss=0.0217
+epoch=9 step=520/710 loss=0.0067
+epoch=9 step=540/710 loss=0.0263
+epoch=9 step=560/710 loss=0.0084
+epoch=9 step=580/710 loss=0.0042
+epoch=9 step=600/710 loss=0.0061
+epoch=9 step=620/710 loss=0.0166
+epoch=9 step=640/710 loss=0.0041
+epoch=9 step=660/710 loss=0.0254
+epoch=9 step=680/710 loss=0.0078
+epoch=9 step=700/710 loss=0.0041
+epoch=9 train_loss=0.0137 train_contrastive=0.0000 train_regression=0.0137 val_loss=0.0150 val_contrastive=0.0000 val_regression=0.0150
+saved_best runs/foundation/brainfm_frozen_mlp_best.pt val_loss=0.0150 epoch=9
+epoch=10 step=20/710 loss=0.0161
+epoch=10 step=40/710 loss=0.0123
+epoch=10 step=60/710 loss=0.0135
+epoch=10 step=80/710 loss=0.0058
+epoch=10 step=100/710 loss=0.0207
+epoch=10 step=120/710 loss=0.0062
+epoch=10 step=140/710 loss=0.0327
+epoch=10 step=160/710 loss=0.0029
+epoch=10 step=180/710 loss=0.0175
+epoch=10 step=200/710 loss=0.0064
+epoch=10 step=220/710 loss=0.0062
+epoch=10 step=240/710 loss=0.0056
+epoch=10 step=260/710 loss=0.0211
+epoch=10 step=280/710 loss=0.0185
+epoch=10 step=300/710 loss=0.0078
+epoch=10 step=320/710 loss=0.0039
+epoch=10 step=340/710 loss=0.0022
+epoch=10 step=360/710 loss=0.0210
+epoch=10 step=380/710 loss=0.0113
+epoch=10 step=400/710 loss=0.0101
+epoch=10 step=420/710 loss=0.0084
+epoch=10 step=440/710 loss=0.0097
+epoch=10 step=460/710 loss=0.0229
+epoch=10 step=480/710 loss=0.0234
+epoch=10 step=500/710 loss=0.0068
+epoch=10 step=520/710 loss=0.0048
+epoch=10 step=540/710 loss=0.0029
+epoch=10 step=560/710 loss=0.0054
+epoch=10 step=580/710 loss=0.0073
+epoch=10 step=600/710 loss=0.0099
+epoch=10 step=620/710 loss=0.0110
+epoch=10 step=640/710 loss=0.0131
+epoch=10 step=660/710 loss=0.0205
+epoch=10 step=680/710 loss=0.0080
+epoch=10 step=700/710 loss=0.0163
+epoch=10 train_loss=0.0135 train_contrastive=0.0000 train_regression=0.0135 val_loss=0.0164 val_contrastive=0.0000 val_regression=0.0164
+epoch=11 step=20/710 loss=0.0078
+epoch=11 step=40/710 loss=0.0063
+epoch=11 step=60/710 loss=0.0066
+epoch=11 step=80/710 loss=0.0074
+epoch=11 step=100/710 loss=0.0039
+epoch=11 step=120/710 loss=0.0139
+epoch=11 step=140/710 loss=0.0134
+epoch=11 step=160/710 loss=0.0287
+epoch=11 step=180/710 loss=0.0073
+epoch=11 step=200/710 loss=0.0041
+epoch=11 step=220/710 loss=0.0115
+epoch=11 step=240/710 loss=0.0216
+epoch=11 step=260/710 loss=0.0357
+epoch=11 step=280/710 loss=0.0081
+epoch=11 step=300/710 loss=0.0081
+epoch=11 step=320/710 loss=0.0057
+epoch=11 step=340/710 loss=0.0040
+epoch=11 step=360/710 loss=0.0312
+epoch=11 step=380/710 loss=0.0057
+epoch=11 step=400/710 loss=0.0125
+epoch=11 step=420/710 loss=0.0045
+epoch=11 step=440/710 loss=0.0056

logs/brainfm_frozen_mlp_eval_after_train.log

@@ -0,0 +1,19 @@
+checkpoint=runs/foundation/brainfm_frozen_mlp_best.pt
+manifest=metadata/splits/test.csv
+samples=153.000000
+mae=0.091787
+rmse=0.119520
+pearson=0.851365
+spearman=0.893663
+top5_high_overlap=0.552941
+top5_low_overlap=0.747712
+pet_to_suvr_recall@1=0.006536
+pet_to_suvr_recall@5=0.039216
+pet_to_suvr_recall@10=0.071895
+pet_to_suvr_mrr=0.036891
+pet_to_suvr_median_rank=77.000000
+suvr_to_pet_recall@1=0.019608
+suvr_to_pet_recall@5=0.052288
+suvr_to_pet_recall@10=0.071895
+suvr_to_pet_mrr=0.050539
+suvr_to_pet_median_rank=66.000000

logs/brainiac_e2e_mlp_b4_20260514_123841.log

@@ -0,0 +1,184 @@
+<frozen importlib._bootstrap_external>:1325: FutureWarning: The cuda.cudart module is deprecated and will be removed in a future release, please switch to use the cuda.bindings.runtime module instead.
+brainiac_missing_keys=['blocks.0.norm_cross_attn.weight', 'blocks.0.norm_cross_attn.bias', 'blocks.0.cross_attn.out_proj.weight', 'blocks.0.cross_attn.out_proj.bias', 'blocks.0.cross_attn.to_q.weight', 'blocks.0.cross_attn.to_k.weight', 'blocks.0.cross_attn.to_v.weight', 'blocks.1.norm_cross_attn.weight']
+device=cuda backbone=brainiac freeze=False train=710 val=152
+epoch=1 step=20/178 loss=2.7477
+epoch=1 step=40/178 loss=2.4494
+epoch=1 step=60/178 loss=1.9561
+epoch=1 step=80/178 loss=1.7050
+epoch=1 step=100/178 loss=1.7074
+epoch=1 step=120/178 loss=1.6398
+epoch=1 step=140/178 loss=1.5386
+epoch=1 step=160/178 loss=1.4806
+epoch=1 train_loss=1.9187 val_loss=1.4929
+epoch=2 step=20/178 loss=1.4866
+epoch=2 step=40/178 loss=1.4530
+epoch=2 step=60/178 loss=1.4581
+epoch=2 step=80/178 loss=1.4286
+epoch=2 step=100/178 loss=1.4201
+epoch=2 step=120/178 loss=1.4039
+epoch=2 step=140/178 loss=1.4323
+epoch=2 step=160/178 loss=1.4008
+epoch=2 train_loss=1.4351 val_loss=1.4147
+epoch=3 step=20/178 loss=1.4215
+epoch=3 step=40/178 loss=1.4142
+epoch=3 step=60/178 loss=1.3929
+epoch=3 step=80/178 loss=1.4224
+epoch=3 step=100/178 loss=1.4120
+epoch=3 step=120/178 loss=1.4124
+epoch=3 step=140/178 loss=1.4064
+epoch=3 step=160/178 loss=1.3747
+epoch=3 train_loss=1.4011 val_loss=1.3977
+epoch=4 step=20/178 loss=1.4019
+epoch=4 step=40/178 loss=1.4002
+epoch=4 step=60/178 loss=1.3850
+epoch=4 step=80/178 loss=1.3895
+epoch=4 step=100/178 loss=1.3851
+epoch=4 step=120/178 loss=1.3855
+epoch=4 step=140/178 loss=1.4152
+epoch=4 step=160/178 loss=1.3623
+epoch=4 train_loss=1.3869 val_loss=1.3813
+epoch=5 step=20/178 loss=1.3909
+epoch=5 step=40/178 loss=1.3857
+epoch=5 step=60/178 loss=1.2669
+epoch=5 step=80/178 loss=1.3531
+epoch=5 step=100/178 loss=1.2557
+epoch=5 step=120/178 loss=1.1314
+epoch=5 step=140/178 loss=1.3942
+epoch=5 step=160/178 loss=1.3280
+epoch=5 train_loss=1.3739 val_loss=1.3734
+epoch=6 step=20/178 loss=1.4298
+epoch=6 step=40/178 loss=1.3505
+epoch=6 step=60/178 loss=1.3278
+epoch=6 step=80/178 loss=1.0664
+epoch=6 step=100/178 loss=1.4560
+epoch=6 step=120/178 loss=1.5205
+epoch=6 step=140/178 loss=1.0904
+epoch=6 step=160/178 loss=1.4003
+epoch=6 train_loss=1.3465 val_loss=1.3487
+epoch=7 step=20/178 loss=1.4617
+epoch=7 step=40/178 loss=1.3708
+epoch=7 step=60/178 loss=1.3128
+epoch=7 step=80/178 loss=1.1522
+epoch=7 step=100/178 loss=1.1868
+epoch=7 step=120/178 loss=1.2577
+epoch=7 step=140/178 loss=1.1712
+epoch=7 step=160/178 loss=1.1340
+epoch=7 train_loss=1.2330 val_loss=1.2656
+epoch=8 step=20/178 loss=0.8030
+epoch=8 step=40/178 loss=1.2733
+epoch=8 step=60/178 loss=1.2478
+epoch=8 step=80/178 loss=1.4158
+epoch=8 step=100/178 loss=1.8038
+epoch=8 step=120/178 loss=1.4911
+epoch=8 step=140/178 loss=1.4074
+epoch=8 step=160/178 loss=1.3697
+epoch=8 train_loss=1.2964 val_loss=1.3946
+epoch=9 step=20/178 loss=1.3679
+epoch=9 step=40/178 loss=1.4070
+epoch=9 step=60/178 loss=1.4304
+epoch=9 step=80/178 loss=1.3337
+epoch=9 step=100/178 loss=1.3835
+epoch=9 step=120/178 loss=1.5095
+epoch=9 step=140/178 loss=1.3403
+epoch=9 step=160/178 loss=1.7522
+epoch=9 train_loss=1.3829 val_loss=1.4120
+epoch=10 step=20/178 loss=1.3916
+epoch=10 step=40/178 loss=1.2785
+epoch=10 step=60/178 loss=1.3909
+epoch=10 step=80/178 loss=1.2816
+epoch=10 step=100/178 loss=1.2579
+epoch=10 step=120/178 loss=1.2597
+epoch=10 step=140/178 loss=1.3562
+epoch=10 step=160/178 loss=1.5344
+epoch=10 train_loss=1.3722 val_loss=1.4236
+epoch=11 step=20/178 loss=1.3389
+epoch=11 step=40/178 loss=1.4463
+epoch=11 step=60/178 loss=1.3295
+epoch=11 step=80/178 loss=1.3310
+epoch=11 step=100/178 loss=1.2883
+epoch=11 step=120/178 loss=1.3063
+epoch=11 step=140/178 loss=1.2969
+epoch=11 step=160/178 loss=1.4154
+epoch=11 train_loss=1.3414 val_loss=1.2965
+epoch=12 step=20/178 loss=1.4114
+epoch=12 step=40/178 loss=1.0259
+epoch=12 step=60/178 loss=1.4089
+epoch=12 step=80/178 loss=1.1336
+epoch=12 step=100/178 loss=1.3582
+epoch=12 step=120/178 loss=1.1584
+epoch=12 step=140/178 loss=1.3623
+epoch=12 step=160/178 loss=1.2000
+epoch=12 train_loss=1.3072 val_loss=1.3726
+epoch=13 step=20/178 loss=1.4152
+epoch=13 step=40/178 loss=1.2959
+epoch=13 step=60/178 loss=1.2332
+epoch=13 step=80/178 loss=1.3008
+epoch=13 step=100/178 loss=1.5912
+epoch=13 step=120/178 loss=1.3502
+epoch=13 step=140/178 loss=1.3887
+epoch=13 step=160/178 loss=1.5290
+epoch=13 train_loss=1.3243 val_loss=1.3560
+epoch=14 step=20/178 loss=1.1668
+epoch=14 step=40/178 loss=1.1744
+epoch=14 step=60/178 loss=1.1625
+epoch=14 step=80/178 loss=1.4453
+epoch=14 step=100/178 loss=1.2376
+epoch=14 step=120/178 loss=1.2605
+epoch=14 step=140/178 loss=1.2995
+epoch=14 step=160/178 loss=1.2790
+epoch=14 train_loss=1.2965 val_loss=1.3678
+epoch=15 step=20/178 loss=1.6269
+epoch=15 step=40/178 loss=1.2135
+epoch=15 step=60/178 loss=1.1822
+epoch=15 step=80/178 loss=1.3781
+epoch=15 step=100/178 loss=1.1750
+epoch=15 step=120/178 loss=1.0385
+epoch=15 step=140/178 loss=1.0788
+epoch=15 step=160/178 loss=1.2553
+epoch=15 train_loss=1.2844 val_loss=1.3587
+epoch=16 step=20/178 loss=1.3235
+epoch=16 step=40/178 loss=1.2870
+epoch=16 step=60/178 loss=1.6865
+epoch=16 step=80/178 loss=1.5873
+epoch=16 step=100/178 loss=1.5687
+epoch=16 step=120/178 loss=1.3438
+epoch=16 step=140/178 loss=1.4837
+epoch=16 step=160/178 loss=1.3507
+epoch=16 train_loss=1.2437 val_loss=1.3241
+epoch=17 step=20/178 loss=1.0311
+epoch=17 step=40/178 loss=0.9715
+epoch=17 step=60/178 loss=1.5381
+epoch=17 step=80/178 loss=1.0434
+epoch=17 step=100/178 loss=1.2054
+epoch=17 step=120/178 loss=0.9097
+epoch=17 step=140/178 loss=1.5504
+epoch=17 step=160/178 loss=0.9485
+epoch=17 train_loss=1.1533 val_loss=1.3083
+epoch=18 step=20/178 loss=1.0471
+epoch=18 step=40/178 loss=0.9124
+epoch=18 step=60/178 loss=1.0259
+epoch=18 step=80/178 loss=0.8158
+epoch=18 step=100/178 loss=1.6901
+epoch=18 step=120/178 loss=1.1825
+epoch=18 step=140/178 loss=1.3904
+epoch=18 step=160/178 loss=1.2119
+epoch=18 train_loss=1.2641 val_loss=1.3522
+epoch=19 step=20/178 loss=1.4824
+epoch=19 step=40/178 loss=1.1201
+epoch=19 step=60/178 loss=1.4100
+epoch=19 step=80/178 loss=1.3447
+epoch=19 step=100/178 loss=1.3165
+epoch=19 step=120/178 loss=1.4345
+epoch=19 step=140/178 loss=1.1947
+epoch=19 step=160/178 loss=1.1696
+epoch=19 train_loss=1.2681 val_loss=1.2997
+epoch=20 step=20/178 loss=1.0460
+epoch=20 step=40/178 loss=1.4625
+epoch=20 step=60/178 loss=1.0141
+epoch=20 step=80/178 loss=0.9798
+epoch=20 step=100/178 loss=1.2182
+epoch=20 step=120/178 loss=1.0026
+epoch=20 step=140/178 loss=0.8596
+epoch=20 step=160/178 loss=1.1291
+epoch=20 train_loss=1.1818 val_loss=1.2860
+saved runs/foundation/brainiac_e2e_mlp_b4.pt

logs/brainiac_frozen_mlp_20260514_114857.log

@@ -0,0 +1,544 @@
+<frozen importlib._bootstrap_external>:1325: FutureWarning: The cuda.cudart module is deprecated and will be removed in a future release, please switch to use the cuda.bindings.runtime module instead.
+brainiac_missing_keys=['blocks.0.norm_cross_attn.weight', 'blocks.0.norm_cross_attn.bias', 'blocks.0.cross_attn.out_proj.weight', 'blocks.0.cross_attn.out_proj.bias', 'blocks.0.cross_attn.to_q.weight', 'blocks.0.cross_attn.to_k.weight', 'blocks.0.cross_attn.to_v.weight', 'blocks.1.norm_cross_attn.weight']
+device=cuda backbone=brainiac freeze=True train=710 val=152
+epoch=1 step=10/178 loss=1.7412
+epoch=1 step=20/178 loss=1.5394
+epoch=1 step=30/178 loss=1.4918
+epoch=1 step=40/178 loss=1.4163
+epoch=1 step=50/178 loss=1.4050
+epoch=1 step=60/178 loss=1.4727
+epoch=1 step=70/178 loss=1.4039
+epoch=1 step=80/178 loss=1.3965
+epoch=1 step=90/178 loss=1.4145
+epoch=1 step=100/178 loss=1.3965
+epoch=1 step=110/178 loss=1.4023
+epoch=1 step=120/178 loss=1.4071
+epoch=1 step=130/178 loss=1.3955
+epoch=1 step=140/178 loss=1.4086
+epoch=1 step=150/178 loss=1.4037
+epoch=1 step=160/178 loss=1.4115
+epoch=1 step=170/178 loss=1.4074
+epoch=1 train_loss=1.4813 val_loss=1.4110
+epoch=2 step=10/178 loss=1.4050
+epoch=2 step=20/178 loss=1.4006
+epoch=2 step=30/178 loss=1.4292
+epoch=2 step=40/178 loss=1.4041
+epoch=2 step=50/178 loss=1.4146
+epoch=2 step=60/178 loss=1.4040
+epoch=2 step=70/178 loss=1.4127
+epoch=2 step=80/178 loss=1.4075
+epoch=2 step=90/178 loss=1.3968
+epoch=2 step=100/178 loss=1.4169
+epoch=2 step=110/178 loss=1.4006
+epoch=2 step=120/178 loss=1.4011
+epoch=2 step=130/178 loss=1.4007
+epoch=2 step=140/178 loss=1.4242
+epoch=2 step=150/178 loss=1.3959
+epoch=2 step=160/178 loss=1.4170
+epoch=2 step=170/178 loss=1.3971
+epoch=2 train_loss=1.4078 val_loss=1.4102
+epoch=3 step=10/178 loss=1.4196
+epoch=3 step=20/178 loss=1.4093
+epoch=3 step=30/178 loss=1.4023
+epoch=3 step=40/178 loss=1.4007
+epoch=3 step=50/178 loss=1.4008
+epoch=3 step=60/178 loss=1.4155
+epoch=3 step=70/178 loss=1.4404
+epoch=3 step=80/178 loss=1.4186
+epoch=3 step=90/178 loss=1.4290
+epoch=3 step=100/178 loss=1.4092
+epoch=3 step=110/178 loss=1.4048
+epoch=3 step=120/178 loss=1.4133
+epoch=3 step=130/178 loss=1.4754
+epoch=3 step=140/178 loss=1.4135
+epoch=3 step=150/178 loss=1.4070
+epoch=3 step=160/178 loss=1.4201
+epoch=3 step=170/178 loss=1.4145
+epoch=3 train_loss=1.4074 val_loss=1.4102
+epoch=4 step=10/178 loss=1.4137
+epoch=4 step=20/178 loss=1.4097
+epoch=4 step=30/178 loss=1.4081
+epoch=4 step=40/178 loss=1.4061
+epoch=4 step=50/178 loss=1.4272
+epoch=4 step=60/178 loss=1.4002
+epoch=4 step=70/178 loss=1.4001
+epoch=4 step=80/178 loss=1.4594
+epoch=4 step=90/178 loss=1.3974
+epoch=4 step=100/178 loss=1.4332
+epoch=4 step=110/178 loss=1.3973
+epoch=4 step=120/178 loss=1.4190
+epoch=4 step=130/178 loss=1.4114
+epoch=4 step=140/178 loss=1.4071
+epoch=4 step=150/178 loss=1.4020
+epoch=4 step=160/178 loss=1.4123
+epoch=4 step=170/178 loss=1.4045
+epoch=4 train_loss=1.4074 val_loss=1.4121
+epoch=5 step=10/178 loss=1.4308
+epoch=5 step=20/178 loss=1.4115
+epoch=5 step=30/178 loss=1.4121
+epoch=5 step=40/178 loss=1.4004
+epoch=5 step=50/178 loss=1.3975
+epoch=5 step=60/178 loss=1.4094
+epoch=5 step=70/178 loss=1.4561
+epoch=5 step=80/178 loss=1.4357
+epoch=5 step=90/178 loss=1.4047
+epoch=5 step=100/178 loss=1.4065
+epoch=5 step=110/178 loss=1.4014
+epoch=5 step=120/178 loss=1.4031
+epoch=5 step=130/178 loss=1.4041
+epoch=5 step=140/178 loss=1.4031
+epoch=5 step=150/178 loss=1.4077
+epoch=5 step=160/178 loss=1.4866
+epoch=5 step=170/178 loss=1.4046
+epoch=5 train_loss=1.4074 val_loss=1.4096
+epoch=6 step=10/178 loss=1.4026
+epoch=6 step=20/178 loss=1.4172
+epoch=6 step=30/178 loss=1.4001
+epoch=6 step=40/178 loss=1.3986
+epoch=6 step=50/178 loss=1.4018
+epoch=6 step=60/178 loss=1.4054
+epoch=6 step=70/178 loss=1.4005
+epoch=6 step=80/178 loss=1.4122
+epoch=6 step=90/178 loss=1.4119
+epoch=6 step=100/178 loss=1.4014
+epoch=6 step=110/178 loss=1.4444
+epoch=6 step=120/178 loss=1.4055
+epoch=6 step=130/178 loss=1.4038
+epoch=6 step=140/178 loss=1.4002
+epoch=6 step=150/178 loss=1.4023
+epoch=6 step=160/178 loss=1.4097
+epoch=6 step=170/178 loss=1.4045
+epoch=6 train_loss=1.4075 val_loss=1.4106
+epoch=7 step=10/178 loss=1.4010
+epoch=7 step=20/178 loss=1.4018
+epoch=7 step=30/178 loss=1.4234
+epoch=7 step=40/178 loss=1.4074
+epoch=7 step=50/178 loss=1.3937
+epoch=7 step=60/178 loss=1.3964
+epoch=7 step=70/178 loss=1.4124
+epoch=7 step=80/178 loss=1.3977
+epoch=7 step=90/178 loss=1.3938
+epoch=7 step=100/178 loss=1.4039
+epoch=7 step=110/178 loss=1.4027
+epoch=7 step=120/178 loss=1.4015
+epoch=7 step=130/178 loss=1.3954
+epoch=7 step=140/178 loss=1.4153
+epoch=7 step=150/178 loss=1.4096
+epoch=7 step=160/178 loss=1.3928
+epoch=7 step=170/178 loss=1.4001
+epoch=7 train_loss=1.4074 val_loss=1.4109
+epoch=8 step=10/178 loss=1.4042
+epoch=8 step=20/178 loss=1.4013
+epoch=8 step=30/178 loss=1.3988
+epoch=8 step=40/178 loss=1.4044
+epoch=8 step=50/178 loss=1.4065
+epoch=8 step=60/178 loss=1.4284
+epoch=8 step=70/178 loss=1.4223
+epoch=8 step=80/178 loss=1.4002
+epoch=8 step=90/178 loss=1.4076
+epoch=8 step=100/178 loss=1.4011
+epoch=8 step=110/178 loss=1.4157
+epoch=8 step=120/178 loss=1.4006
+epoch=8 step=130/178 loss=1.4202
+epoch=8 step=140/178 loss=1.4230
+epoch=8 step=150/178 loss=1.4033
+epoch=8 step=160/178 loss=1.3991
+epoch=8 step=170/178 loss=1.4126
+epoch=8 train_loss=1.4074 val_loss=1.4107
+epoch=9 step=10/178 loss=1.4034
+epoch=9 step=20/178 loss=1.4031
+epoch=9 step=30/178 loss=1.3971
+epoch=9 step=40/178 loss=1.3998
+epoch=9 step=50/178 loss=1.4057
+epoch=9 step=60/178 loss=1.3979
+epoch=9 step=70/178 loss=1.4105
+epoch=9 step=80/178 loss=1.3993
+epoch=9 step=90/178 loss=1.3970
+epoch=9 step=100/178 loss=1.4170
+epoch=9 step=110/178 loss=1.4083
+epoch=9 step=120/178 loss=1.4081
+epoch=9 step=130/178 loss=1.4148
+epoch=9 step=140/178 loss=1.4647
+epoch=9 step=150/178 loss=1.3988
+epoch=9 step=160/178 loss=1.3996
+epoch=9 step=170/178 loss=1.3991
+epoch=9 train_loss=1.4069 val_loss=1.4100
+epoch=10 step=10/178 loss=1.4077
+epoch=10 step=20/178 loss=1.3982
+epoch=10 step=30/178 loss=1.4060
+epoch=10 step=40/178 loss=1.4075
+epoch=10 step=50/178 loss=1.4105
+epoch=10 step=60/178 loss=1.4021
+epoch=10 step=70/178 loss=1.4209
+epoch=10 step=80/178 loss=1.4120
+epoch=10 step=90/178 loss=1.4026
+epoch=10 step=100/178 loss=1.4090
+epoch=10 step=110/178 loss=1.4030
+epoch=10 step=120/178 loss=1.3989
+epoch=10 step=130/178 loss=1.4017
+epoch=10 step=140/178 loss=1.4015
+epoch=10 step=150/178 loss=1.3981
+epoch=10 step=160/178 loss=1.4036
+epoch=10 step=170/178 loss=1.3987
+epoch=10 train_loss=1.4070 val_loss=1.4115
+epoch=11 step=10/178 loss=1.3992
+epoch=11 step=20/178 loss=1.4072
+epoch=11 step=30/178 loss=1.4022
+epoch=11 step=40/178 loss=1.3991
+epoch=11 step=50/178 loss=1.3965
+epoch=11 step=60/178 loss=1.4044
+epoch=11 step=70/178 loss=1.4107
+epoch=11 step=80/178 loss=1.4088
+epoch=11 step=90/178 loss=1.4455
+epoch=11 step=100/178 loss=1.4391
+epoch=11 step=110/178 loss=1.3965
+epoch=11 step=120/178 loss=1.4012
+epoch=11 step=130/178 loss=1.4182
+epoch=11 step=140/178 loss=1.4028
+epoch=11 step=150/178 loss=1.4045
+epoch=11 step=160/178 loss=1.4042
+epoch=11 step=170/178 loss=1.4808
+epoch=11 train_loss=1.4070 val_loss=1.4101
+epoch=12 step=10/178 loss=1.4007
+epoch=12 step=20/178 loss=1.4055
+epoch=12 step=30/178 loss=1.3959
+epoch=12 step=40/178 loss=1.4064
+epoch=12 step=50/178 loss=1.4052
+epoch=12 step=60/178 loss=1.3931
+epoch=12 step=70/178 loss=1.3993
+epoch=12 step=80/178 loss=1.4290
+epoch=12 step=90/178 loss=1.4015
+epoch=12 step=100/178 loss=1.4002
+epoch=12 step=110/178 loss=1.4077
+epoch=12 step=120/178 loss=1.4044
+epoch=12 step=130/178 loss=1.3979
+epoch=12 step=140/178 loss=1.4118
+epoch=12 step=150/178 loss=1.4018
+epoch=12 step=160/178 loss=1.4037
+epoch=12 step=170/178 loss=1.4160
+epoch=12 train_loss=1.4071 val_loss=1.4101
+epoch=13 step=10/178 loss=1.4066
+epoch=13 step=20/178 loss=1.4114
+epoch=13 step=30/178 loss=1.4106
+epoch=13 step=40/178 loss=1.4105
+epoch=13 step=50/178 loss=1.4013
+epoch=13 step=60/178 loss=1.4442
+epoch=13 step=70/178 loss=1.4108
+epoch=13 step=80/178 loss=1.4006
+epoch=13 step=90/178 loss=1.4092
+epoch=13 step=100/178 loss=1.4239
+epoch=13 step=110/178 loss=1.4526
+epoch=13 step=120/178 loss=1.4087
+epoch=13 step=130/178 loss=1.4118
+epoch=13 step=140/178 loss=1.3999
+epoch=13 step=150/178 loss=1.3994
+epoch=13 step=160/178 loss=1.4105
+epoch=13 step=170/178 loss=1.3955
+epoch=13 train_loss=1.4069 val_loss=1.4093
+epoch=14 step=10/178 loss=1.4392
+epoch=14 step=20/178 loss=1.4009
+epoch=14 step=30/178 loss=1.4051
+epoch=14 step=40/178 loss=1.4029
+epoch=14 step=50/178 loss=1.4043
+epoch=14 step=60/178 loss=1.4110
+epoch=14 step=70/178 loss=1.4038
+epoch=14 step=80/178 loss=1.3936
+epoch=14 step=90/178 loss=1.4002
+epoch=14 step=100/178 loss=1.3979
+epoch=14 step=110/178 loss=1.4053
+epoch=14 step=120/178 loss=1.3964
+epoch=14 step=130/178 loss=1.4091
+epoch=14 step=140/178 loss=1.4474
+epoch=14 step=150/178 loss=1.4037
+epoch=14 step=160/178 loss=1.3981
+epoch=14 step=170/178 loss=1.3945
+epoch=14 train_loss=1.4070 val_loss=1.4104
+epoch=15 step=10/178 loss=1.4397
+epoch=15 step=20/178 loss=1.4022
+epoch=15 step=30/178 loss=1.3984
+epoch=15 step=40/178 loss=1.4052
+epoch=15 step=50/178 loss=1.4090
+epoch=15 step=60/178 loss=1.4058
+epoch=15 step=70/178 loss=1.4180
+epoch=15 step=80/178 loss=1.4317
+epoch=15 step=90/178 loss=1.4104
+epoch=15 step=100/178 loss=1.4011
+epoch=15 step=110/178 loss=1.4140
+epoch=15 step=120/178 loss=1.4191
+epoch=15 step=130/178 loss=1.4036
+epoch=15 step=140/178 loss=1.4004
+epoch=15 step=150/178 loss=1.4217
+epoch=15 step=160/178 loss=1.4019
+epoch=15 step=170/178 loss=1.4065
+epoch=15 train_loss=1.4067 val_loss=1.4093
+epoch=16 step=10/178 loss=1.4028
+epoch=16 step=20/178 loss=1.4194
+epoch=16 step=30/178 loss=1.4056
+epoch=16 step=40/178 loss=1.4014
+epoch=16 step=50/178 loss=1.3944
+epoch=16 step=60/178 loss=1.4060
+epoch=16 step=70/178 loss=1.3977
+epoch=16 step=80/178 loss=1.3999
+epoch=16 step=90/178 loss=1.4083
+epoch=16 step=100/178 loss=1.3969
+epoch=16 step=110/178 loss=1.3986
+epoch=16 step=120/178 loss=1.4153
+epoch=16 step=130/178 loss=1.4328
+epoch=16 step=140/178 loss=1.4079
+epoch=16 step=150/178 loss=1.4248
+epoch=16 step=160/178 loss=1.3984
+epoch=16 step=170/178 loss=1.4070
+epoch=16 train_loss=1.4069 val_loss=1.4093
+epoch=17 step=10/178 loss=1.4054
+epoch=17 step=20/178 loss=1.4010
+epoch=17 step=30/178 loss=1.3962
+epoch=17 step=40/178 loss=1.4053
+epoch=17 step=50/178 loss=1.4052
+epoch=17 step=60/178 loss=1.4064
+epoch=17 step=70/178 loss=1.4093
+epoch=17 step=80/178 loss=1.3995
+epoch=17 step=90/178 loss=1.4309
+epoch=17 step=100/178 loss=1.4126
+epoch=17 step=110/178 loss=1.4210
+epoch=17 step=120/178 loss=1.3952
+epoch=17 step=130/178 loss=1.4154
+epoch=17 step=140/178 loss=1.4032
+epoch=17 step=150/178 loss=1.3952
+epoch=17 step=160/178 loss=1.4199
+epoch=17 step=170/178 loss=1.4024
+epoch=17 train_loss=1.4069 val_loss=1.4116
+epoch=18 step=10/178 loss=1.4109
+epoch=18 step=20/178 loss=1.4094
+epoch=18 step=30/178 loss=1.3982
+epoch=18 step=40/178 loss=1.4056
+epoch=18 step=50/178 loss=1.4313
+epoch=18 step=60/178 loss=1.3941
+epoch=18 step=70/178 loss=1.4091
+epoch=18 step=80/178 loss=1.4086
+epoch=18 step=90/178 loss=1.4005
+epoch=18 step=100/178 loss=1.4044
+epoch=18 step=110/178 loss=1.3942
+epoch=18 step=120/178 loss=1.4129
+epoch=18 step=130/178 loss=1.3942
+epoch=18 step=140/178 loss=1.4101
+epoch=18 step=150/178 loss=1.4436
+epoch=18 step=160/178 loss=1.3957
+epoch=18 step=170/178 loss=1.3951
+epoch=18 train_loss=1.4066 val_loss=1.4087
+epoch=19 step=10/178 loss=1.3939
+epoch=19 step=20/178 loss=1.4060
+epoch=19 step=30/178 loss=1.3955
+epoch=19 step=40/178 loss=1.3994
+epoch=19 step=50/178 loss=1.4160
+epoch=19 step=60/178 loss=1.4051
+epoch=19 step=70/178 loss=1.3939
+epoch=19 step=80/178 loss=1.4132
+epoch=19 step=90/178 loss=1.3992
+epoch=19 step=100/178 loss=1.4011
+epoch=19 step=110/178 loss=1.4085
+epoch=19 step=120/178 loss=1.4010
+epoch=19 step=130/178 loss=1.3988
+epoch=19 step=140/178 loss=1.4120
+epoch=19 step=150/178 loss=1.3934
+epoch=19 step=160/178 loss=1.3938
+epoch=19 step=170/178 loss=1.3935
+epoch=19 train_loss=1.4054 val_loss=1.4087
+epoch=20 step=10/178 loss=1.3936
+epoch=20 step=20/178 loss=1.4030
+epoch=20 step=30/178 loss=1.4003
+epoch=20 step=40/178 loss=1.4149
+epoch=20 step=50/178 loss=1.4052
+epoch=20 step=60/178 loss=1.4008
+epoch=20 step=70/178 loss=1.4061
+epoch=20 step=80/178 loss=1.4038
+epoch=20 step=90/178 loss=1.4124
+epoch=20 step=100/178 loss=1.3991
+epoch=20 step=110/178 loss=1.4195
+epoch=20 step=120/178 loss=1.3980
+epoch=20 step=130/178 loss=1.4086
+epoch=20 step=140/178 loss=1.3954
+epoch=20 step=150/178 loss=1.3995
+epoch=20 step=160/178 loss=1.4066
+epoch=20 step=170/178 loss=1.4004
+epoch=20 train_loss=1.4066 val_loss=1.4087
+epoch=21 step=10/178 loss=1.4126
+epoch=21 step=20/178 loss=1.4022
+epoch=21 step=30/178 loss=1.4004
+epoch=21 step=40/178 loss=1.4097
+epoch=21 step=50/178 loss=1.3986
+epoch=21 step=60/178 loss=1.4111
+epoch=21 step=70/178 loss=1.4202
+epoch=21 step=80/178 loss=1.3904
+epoch=21 step=90/178 loss=1.4045
+epoch=21 step=100/178 loss=1.4789
+epoch=21 step=110/178 loss=1.4109
+epoch=21 step=120/178 loss=1.3992
+epoch=21 step=130/178 loss=1.3994
+epoch=21 step=140/178 loss=1.4139
+epoch=21 step=150/178 loss=1.4225
+epoch=21 step=160/178 loss=1.3954
+epoch=21 step=170/178 loss=1.4105
+epoch=21 train_loss=1.4066 val_loss=1.4092
+epoch=22 step=10/178 loss=1.4025
+epoch=22 step=20/178 loss=1.4025
+epoch=22 step=30/178 loss=1.4112
+epoch=22 step=40/178 loss=1.4087
+epoch=22 step=50/178 loss=1.4046
+epoch=22 step=60/178 loss=1.3956
+epoch=22 step=70/178 loss=1.4043
+epoch=22 step=80/178 loss=1.3944
+epoch=22 step=90/178 loss=1.4152
+epoch=22 step=100/178 loss=1.3996
+epoch=22 step=110/178 loss=1.3996
+epoch=22 step=120/178 loss=1.4075
+epoch=22 step=130/178 loss=1.4014
+epoch=22 step=140/178 loss=1.3985
+epoch=22 step=150/178 loss=1.4151
+epoch=22 step=160/178 loss=1.3973
+epoch=22 step=170/178 loss=1.4106
+epoch=22 train_loss=1.4048 val_loss=1.4072
+epoch=23 step=10/178 loss=1.3926
+epoch=23 step=20/178 loss=1.3909
+epoch=23 step=30/178 loss=1.4244
+epoch=23 step=40/178 loss=1.3922
+epoch=23 step=50/178 loss=1.3914
+epoch=23 step=60/178 loss=1.4157
+epoch=23 step=70/178 loss=1.3931
+epoch=23 step=80/178 loss=1.4285
+epoch=23 step=90/178 loss=1.4011
+epoch=23 step=100/178 loss=1.4057
+epoch=23 step=110/178 loss=1.3943
+epoch=23 step=120/178 loss=1.4225
+epoch=23 step=130/178 loss=1.3979
+epoch=23 step=140/178 loss=1.4347
+epoch=23 step=150/178 loss=1.4415
+epoch=23 step=160/178 loss=1.4137
+epoch=23 step=170/178 loss=1.4273
+epoch=23 train_loss=1.4059 val_loss=1.4105
+epoch=24 step=10/178 loss=1.4006
+epoch=24 step=20/178 loss=1.3969
+epoch=24 step=30/178 loss=1.3991
+epoch=24 step=40/178 loss=1.3955
+epoch=24 step=50/178 loss=1.4109
+epoch=24 step=60/178 loss=1.3979
+epoch=24 step=70/178 loss=1.3897
+epoch=24 step=80/178 loss=1.3986
+epoch=24 step=90/178 loss=1.4038
+epoch=24 step=100/178 loss=1.3914
+epoch=24 step=110/178 loss=1.4083
+epoch=24 step=120/178 loss=1.4099
+epoch=24 step=130/178 loss=1.4106
+epoch=24 step=140/178 loss=1.3985
+epoch=24 step=150/178 loss=1.4156
+epoch=24 step=160/178 loss=1.3986
+epoch=24 step=170/178 loss=1.4236
+epoch=24 train_loss=1.4030 val_loss=1.4073
+epoch=25 step=10/178 loss=1.3976
+epoch=25 step=20/178 loss=1.4244
+epoch=25 step=30/178 loss=1.4076
+epoch=25 step=40/178 loss=1.4106
+epoch=25 step=50/178 loss=1.3955
+epoch=25 step=60/178 loss=1.4146
+epoch=25 step=70/178 loss=1.4056
+epoch=25 step=80/178 loss=1.3958
+epoch=25 step=90/178 loss=1.4091
+epoch=25 step=100/178 loss=1.4004
+epoch=25 step=110/178 loss=1.4882
+epoch=25 step=120/178 loss=1.4493
+epoch=25 step=130/178 loss=1.4258
+epoch=25 step=140/178 loss=1.4073
+epoch=25 step=150/178 loss=1.4034
+epoch=25 step=160/178 loss=1.4851
+epoch=25 step=170/178 loss=1.4175
+epoch=25 train_loss=1.4077 val_loss=1.4078
+epoch=26 step=10/178 loss=1.3901
+epoch=26 step=20/178 loss=1.4334
+epoch=26 step=30/178 loss=1.3922
+epoch=26 step=40/178 loss=1.4193
+epoch=26 step=50/178 loss=1.3938
+epoch=26 step=60/178 loss=1.3984
+epoch=26 step=70/178 loss=1.4003
+epoch=26 step=80/178 loss=1.4021
+epoch=26 step=90/178 loss=1.3914
+epoch=26 step=100/178 loss=1.4060
+epoch=26 step=110/178 loss=1.4133
+epoch=26 step=120/178 loss=1.4651
+epoch=26 step=130/178 loss=1.3937
+epoch=26 step=140/178 loss=1.4019
+epoch=26 step=150/178 loss=1.3944
+epoch=26 step=160/178 loss=1.3942
+epoch=26 step=170/178 loss=1.3946
+epoch=26 train_loss=1.4029 val_loss=1.4057
+epoch=27 step=10/178 loss=1.4021
+epoch=27 step=20/178 loss=1.3956
+epoch=27 step=30/178 loss=1.3927
+epoch=27 step=40/178 loss=1.3996
+epoch=27 step=50/178 loss=1.4078
+epoch=27 step=60/178 loss=1.4003
+epoch=27 step=70/178 loss=1.4162
+epoch=27 step=80/178 loss=1.3919
+epoch=27 step=90/178 loss=1.3973
+epoch=27 step=100/178 loss=1.3781
+epoch=27 step=110/178 loss=1.4024
+epoch=27 step=120/178 loss=1.4094
+epoch=27 step=130/178 loss=1.4180
+epoch=27 step=140/178 loss=1.4057
+epoch=27 step=150/178 loss=1.3870
+epoch=27 step=160/178 loss=1.3921
+epoch=27 step=170/178 loss=1.4165
+epoch=27 train_loss=1.4016 val_loss=1.4069
+epoch=28 step=10/178 loss=1.3999
+epoch=28 step=20/178 loss=1.4015
+epoch=28 step=30/178 loss=1.4064
+epoch=28 step=40/178 loss=1.4020
+epoch=28 step=50/178 loss=1.4017
+epoch=28 step=60/178 loss=1.3969
+epoch=28 step=70/178 loss=1.3915
+epoch=28 step=80/178 loss=1.3987
+epoch=28 step=90/178 loss=1.4064
+epoch=28 step=100/178 loss=1.3928
+epoch=28 step=110/178 loss=1.3866
+epoch=28 step=120/178 loss=1.4135
+epoch=28 step=130/178 loss=1.4124
+epoch=28 step=140/178 loss=1.4072
+epoch=28 step=150/178 loss=1.4261
+epoch=28 step=160/178 loss=1.3925
+epoch=28 step=170/178 loss=1.3908
+epoch=28 train_loss=1.4040 val_loss=1.4032
+epoch=29 step=10/178 loss=1.3964
+epoch=29 step=20/178 loss=1.4010
+epoch=29 step=30/178 loss=1.4329
+epoch=29 step=40/178 loss=1.3845
+epoch=29 step=50/178 loss=1.3873
+epoch=29 step=60/178 loss=1.4000
+epoch=29 step=70/178 loss=1.4046
+epoch=29 step=80/178 loss=1.3934
+epoch=29 step=90/178 loss=1.3787
+epoch=29 step=100/178 loss=1.4171
+epoch=29 step=110/178 loss=1.4354
+epoch=29 step=120/178 loss=1.5154
+epoch=29 step=130/178 loss=1.4443
+epoch=29 step=140/178 loss=1.4090
+epoch=29 step=150/178 loss=1.3928
+epoch=29 step=160/178 loss=1.4139
+epoch=29 step=170/178 loss=1.4023
+epoch=29 train_loss=1.4041 val_loss=1.4059
+epoch=30 step=10/178 loss=1.4047
+epoch=30 step=20/178 loss=1.3992
+epoch=30 step=30/178 loss=1.3817
+epoch=30 step=40/178 loss=1.4073
+epoch=30 step=50/178 loss=1.3849
+epoch=30 step=60/178 loss=1.3913
+epoch=30 step=70/178 loss=1.4040
+epoch=30 step=80/178 loss=1.3952
+epoch=30 step=90/178 loss=1.3887
+epoch=30 step=100/178 loss=1.4015
+epoch=30 step=110/178 loss=1.3782
+epoch=30 step=120/178 loss=1.3872
+epoch=30 step=130/178 loss=1.3902
+epoch=30 step=140/178 loss=1.4019
+epoch=30 step=150/178 loss=1.3940
+epoch=30 step=160/178 loss=1.4044
+epoch=30 step=170/178 loss=1.3948
+epoch=30 train_loss=1.3969 val_loss=1.4019
+saved runs/foundation/brainiac_frozen_mlp.pt

logs/brainiac_lastblock_regalign.log

@@ -0,0 +1,307 @@
+<frozen importlib._bootstrap_external>:1325: FutureWarning: The cuda.cudart module is deprecated and will be removed in a future release, please switch to use the cuda.bindings.runtime module instead.
+brainiac_missing_keys=['blocks.0.norm_cross_attn.weight', 'blocks.0.norm_cross_attn.bias', 'blocks.0.cross_attn.out_proj.weight', 'blocks.0.cross_attn.out_proj.bias', 'blocks.0.cross_attn.to_q.weight', 'blocks.0.cross_attn.to_k.weight', 'blocks.0.cross_attn.to_v.weight', 'blocks.1.norm_cross_attn.weight']
+device=cuda backbone=brainiac encoder_scope=last_block contrastive_weight=0.2 regression_weight=1.0 train=710 val=152
+epoch=1 step=20/178 loss=1.3135
+epoch=1 step=40/178 loss=1.1481
+epoch=1 step=60/178 loss=0.9881
+epoch=1 step=80/178 loss=0.6453
+epoch=1 step=100/178 loss=0.6136
+epoch=1 step=120/178 loss=0.5926
+epoch=1 step=140/178 loss=0.4593
+epoch=1 step=160/178 loss=0.4182
+epoch=1 train_loss=0.7994 train_contrastive=1.3975 train_regression=0.5200 val_loss=0.4029 val_contrastive=1.3925 val_regression=0.1244
+saved_best runs/foundation/brainiac_lastblock_regalign_best.pt val_loss=0.4029 epoch=1
+epoch=2 step=20/178 loss=0.3766
+epoch=2 step=40/178 loss=0.3224
+epoch=2 step=60/178 loss=0.3400
+epoch=2 step=80/178 loss=0.3211
+epoch=2 step=100/178 loss=0.3303
+epoch=2 step=120/178 loss=0.3137
+epoch=2 step=140/178 loss=0.3098
+epoch=2 step=160/178 loss=0.3635
+epoch=2 train_loss=0.3381 train_contrastive=1.3902 train_regression=0.0601 val_loss=0.3124 val_contrastive=1.3904 val_regression=0.0343
+saved_best runs/foundation/brainiac_lastblock_regalign_best.pt val_loss=0.3124 epoch=2
+epoch=3 step=20/178 loss=0.3125
+epoch=3 step=40/178 loss=0.3013
+epoch=3 step=60/178 loss=0.2963
+epoch=3 step=80/178 loss=0.2870
+epoch=3 step=100/178 loss=0.2936
+epoch=3 step=120/178 loss=0.3187
+epoch=3 step=140/178 loss=0.2929
+epoch=3 step=160/178 loss=0.2962
+epoch=3 train_loss=0.3032 train_contrastive=1.3879 train_regression=0.0256 val_loss=0.3018 val_contrastive=1.3891 val_regression=0.0240
+saved_best runs/foundation/brainiac_lastblock_regalign_best.pt val_loss=0.3018 epoch=3
+epoch=4 step=20/178 loss=0.2934
+epoch=4 step=40/178 loss=0.3002
+epoch=4 step=60/178 loss=0.2875
+epoch=4 step=80/178 loss=0.2889
+epoch=4 step=100/178 loss=0.2978
+epoch=4 step=120/178 loss=0.2913
+epoch=4 step=140/178 loss=0.2888
+epoch=4 step=160/178 loss=0.3075
+epoch=4 train_loss=0.2996 train_contrastive=1.3870 train_regression=0.0222 val_loss=0.3009 val_contrastive=1.3886 val_regression=0.0232
+saved_best runs/foundation/brainiac_lastblock_regalign_best.pt val_loss=0.3009 epoch=4
+epoch=5 step=20/178 loss=0.2854
+epoch=5 step=40/178 loss=0.3089
+epoch=5 step=60/178 loss=0.3024
+epoch=5 step=80/178 loss=0.2931
+epoch=5 step=100/178 loss=0.2938
+epoch=5 step=120/178 loss=0.2961
+epoch=5 step=140/178 loss=0.3159
+epoch=5 step=160/178 loss=0.3064
+epoch=5 train_loss=0.2993 train_contrastive=1.3864 train_regression=0.0220 val_loss=0.3006 val_contrastive=1.3880 val_regression=0.0230
+saved_best runs/foundation/brainiac_lastblock_regalign_best.pt val_loss=0.3006 epoch=5
+epoch=6 step=20/178 loss=0.2958
+epoch=6 step=40/178 loss=0.3081
+epoch=6 step=60/178 loss=0.2924
+epoch=6 step=80/178 loss=0.3638
+epoch=6 step=100/178 loss=0.2960
+epoch=6 step=120/178 loss=0.2918
+epoch=6 step=140/178 loss=0.2917
+epoch=6 step=160/178 loss=0.2997
+epoch=6 train_loss=0.2991 train_contrastive=1.3859 train_regression=0.0220 val_loss=0.3005 val_contrastive=1.3877 val_regression=0.0230
+saved_best runs/foundation/brainiac_lastblock_regalign_best.pt val_loss=0.3005 epoch=6
+epoch=7 step=20/178 loss=0.2920
+epoch=7 step=40/178 loss=0.2868
+epoch=7 step=60/178 loss=0.2941
+epoch=7 step=80/178 loss=0.2995
+epoch=7 step=100/178 loss=0.2930
+epoch=7 step=120/178 loss=0.3059
+epoch=7 step=140/178 loss=0.2936
+epoch=7 step=160/178 loss=0.2934
+epoch=7 train_loss=0.2992 train_contrastive=1.3857 train_regression=0.0221 val_loss=0.3007 val_contrastive=1.3876 val_regression=0.0232
+epoch=8 step=20/178 loss=0.2942
+epoch=8 step=40/178 loss=0.2973
+epoch=8 step=60/178 loss=0.3111
+epoch=8 step=80/178 loss=0.3367
+epoch=8 step=100/178 loss=0.2973
+epoch=8 step=120/178 loss=0.2931
+epoch=8 step=140/178 loss=0.3194
+epoch=8 step=160/178 loss=0.2991
+epoch=8 train_loss=0.2991 train_contrastive=1.3854 train_regression=0.0220 val_loss=0.3007 val_contrastive=1.3874 val_regression=0.0232
+epoch=9 step=20/178 loss=0.2950
+epoch=9 step=40/178 loss=0.2951
+epoch=9 step=60/178 loss=0.2903
+epoch=9 step=80/178 loss=0.3156
+epoch=9 step=100/178 loss=0.2995
+epoch=9 step=120/178 loss=0.2939
+epoch=9 step=140/178 loss=0.2993
+epoch=9 step=160/178 loss=0.2921
+epoch=9 train_loss=0.2991 train_contrastive=1.3852 train_regression=0.0220 val_loss=0.3003 val_contrastive=1.3872 val_regression=0.0229
+saved_best runs/foundation/brainiac_lastblock_regalign_best.pt val_loss=0.3003 epoch=9
+epoch=10 step=20/178 loss=0.2927
+epoch=10 step=40/178 loss=0.2887
+epoch=10 step=60/178 loss=0.2869
+epoch=10 step=80/178 loss=0.2943
+epoch=10 step=100/178 loss=0.3080
+epoch=10 step=120/178 loss=0.3022
+epoch=10 step=140/178 loss=0.2857
+epoch=10 step=160/178 loss=0.2934
+epoch=10 train_loss=0.2991 train_contrastive=1.3851 train_regression=0.0220 val_loss=0.3002 val_contrastive=1.3871 val_regression=0.0228
+saved_best runs/foundation/brainiac_lastblock_regalign_best.pt val_loss=0.3002 epoch=10
+epoch=11 step=20/178 loss=0.2902
+epoch=11 step=40/178 loss=0.3082
+epoch=11 step=60/178 loss=0.2948
+epoch=11 step=80/178 loss=0.2951
+epoch=11 step=100/178 loss=0.2915
+epoch=11 step=120/178 loss=0.2929
+epoch=11 step=140/178 loss=0.2997
+epoch=11 step=160/178 loss=0.2947
+epoch=11 train_loss=0.2991 train_contrastive=1.3850 train_regression=0.0221 val_loss=0.3004 val_contrastive=1.3870 val_regression=0.0230
+epoch=12 step=20/178 loss=0.2896
+epoch=12 step=40/178 loss=0.2924
+epoch=12 step=60/178 loss=0.2900
+epoch=12 step=80/178 loss=0.2966
+epoch=12 step=100/178 loss=0.2960
+epoch=12 step=120/178 loss=0.3021
+epoch=12 step=140/178 loss=0.3353
+epoch=12 step=160/178 loss=0.2962
+epoch=12 train_loss=0.2991 train_contrastive=1.3849 train_regression=0.0221 val_loss=0.3002 val_contrastive=1.3869 val_regression=0.0228
+epoch=13 step=20/178 loss=0.3080
+epoch=13 step=40/178 loss=0.2851
+epoch=13 step=60/178 loss=0.2933
+epoch=13 step=80/178 loss=0.2949
+epoch=13 step=100/178 loss=0.3000
+epoch=13 step=120/178 loss=0.2871
+epoch=13 step=140/178 loss=0.3029
+epoch=13 step=160/178 loss=0.3076
+epoch=13 train_loss=0.2990 train_contrastive=1.3848 train_regression=0.0221 val_loss=0.3002 val_contrastive=1.3868 val_regression=0.0228
+saved_best runs/foundation/brainiac_lastblock_regalign_best.pt val_loss=0.3002 epoch=13
+epoch=14 step=20/178 loss=0.2904
+epoch=14 step=40/178 loss=0.2894
+epoch=14 step=60/178 loss=0.3143
+epoch=14 step=80/178 loss=0.2886
+epoch=14 step=100/178 loss=0.2931
+epoch=14 step=120/178 loss=0.3158
+epoch=14 step=140/178 loss=0.2933
+epoch=14 step=160/178 loss=0.2960
+epoch=14 train_loss=0.2990 train_contrastive=1.3847 train_regression=0.0221 val_loss=0.3005 val_contrastive=1.3868 val_regression=0.0231
+epoch=15 step=20/178 loss=0.3098
+epoch=15 step=40/178 loss=0.2963
+epoch=15 step=60/178 loss=0.3313
+epoch=15 step=80/178 loss=0.2891
+epoch=15 step=100/178 loss=0.2960
+epoch=15 step=120/178 loss=0.2947
+epoch=15 step=140/178 loss=0.2993
+epoch=15 step=160/178 loss=0.2978
+epoch=15 train_loss=0.2990 train_contrastive=1.3846 train_regression=0.0221 val_loss=0.3003 val_contrastive=1.3867 val_regression=0.0229
+epoch=16 step=20/178 loss=0.2958
+epoch=16 step=40/178 loss=0.2984
+epoch=16 step=60/178 loss=0.3018
+epoch=16 step=80/178 loss=0.2898
+epoch=16 step=100/178 loss=0.3599
+epoch=16 step=120/178 loss=0.2969
+epoch=16 step=140/178 loss=0.2897
+epoch=16 step=160/178 loss=0.2875
+epoch=16 train_loss=0.2990 train_contrastive=1.3846 train_regression=0.0221 val_loss=0.3002 val_contrastive=1.3866 val_regression=0.0229
+epoch=17 step=20/178 loss=0.3663
+epoch=17 step=40/178 loss=0.2939
+epoch=17 step=60/178 loss=0.2962
+epoch=17 step=80/178 loss=0.2969
+epoch=17 step=100/178 loss=0.2951
+epoch=17 step=120/178 loss=0.2991
+epoch=17 step=140/178 loss=0.2861
+epoch=17 step=160/178 loss=0.3057
+epoch=17 train_loss=0.2990 train_contrastive=1.3845 train_regression=0.0221 val_loss=0.3003 val_contrastive=1.3866 val_regression=0.0230
+epoch=18 step=20/178 loss=0.2965
+epoch=18 step=40/178 loss=0.2888
+epoch=18 step=60/178 loss=0.2933
+epoch=18 step=80/178 loss=0.2875
+epoch=18 step=100/178 loss=0.2910
+epoch=18 step=120/178 loss=0.2997
+epoch=18 step=140/178 loss=0.2868
+epoch=18 step=160/178 loss=0.2980
+epoch=18 train_loss=0.2991 train_contrastive=1.3845 train_regression=0.0223 val_loss=0.3001 val_contrastive=1.3866 val_regression=0.0228
+saved_best runs/foundation/brainiac_lastblock_regalign_best.pt val_loss=0.3001 epoch=18
+epoch=19 step=20/178 loss=0.2900
+epoch=19 step=40/178 loss=0.2905
+epoch=19 step=60/178 loss=0.3097
+epoch=19 step=80/178 loss=0.2934
+epoch=19 step=100/178 loss=0.2947
+epoch=19 step=120/178 loss=0.3004
+epoch=19 step=140/178 loss=0.2983
+epoch=19 step=160/178 loss=0.2917
+epoch=19 train_loss=0.2990 train_contrastive=1.3845 train_regression=0.0221 val_loss=0.3002 val_contrastive=1.3865 val_regression=0.0229
+epoch=20 step=20/178 loss=0.2833
+epoch=20 step=40/178 loss=0.2920
+epoch=20 step=60/178 loss=0.2929
+epoch=20 step=80/178 loss=0.2956
+epoch=20 step=100/178 loss=0.2856
+epoch=20 step=120/178 loss=0.2936
+epoch=20 step=140/178 loss=0.2885
+epoch=20 step=160/178 loss=0.2976
+epoch=20 train_loss=0.2990 train_contrastive=1.3844 train_regression=0.0221 val_loss=0.3003 val_contrastive=1.3865 val_regression=0.0230
+epoch=21 step=20/178 loss=0.2908
+epoch=21 step=40/178 loss=0.3267
+epoch=21 step=60/178 loss=0.2907
+epoch=21 step=80/178 loss=0.2930
+epoch=21 step=100/178 loss=0.3173
+epoch=21 step=120/178 loss=0.2888
+epoch=21 step=140/178 loss=0.2938
+epoch=21 step=160/178 loss=0.2872
+epoch=21 train_loss=0.2990 train_contrastive=1.3843 train_regression=0.0222 val_loss=0.3002 val_contrastive=1.3864 val_regression=0.0230
+epoch=22 step=20/178 loss=0.3011
+epoch=22 step=40/178 loss=0.3080
+epoch=22 step=60/178 loss=0.2881
+epoch=22 step=80/178 loss=0.2961
+epoch=22 step=100/178 loss=0.3398
+epoch=22 step=120/178 loss=0.3354
+epoch=22 step=140/178 loss=0.2921
+epoch=22 step=160/178 loss=0.2875
+epoch=22 train_loss=0.2989 train_contrastive=1.3843 train_regression=0.0221 val_loss=0.3004 val_contrastive=1.3864 val_regression=0.0231
+epoch=23 step=20/178 loss=0.2996
+epoch=23 step=40/178 loss=0.3122
+epoch=23 step=60/178 loss=0.2859
+epoch=23 step=80/178 loss=0.2985
+epoch=23 step=100/178 loss=0.3003
+epoch=23 step=120/178 loss=0.2920
+epoch=23 step=140/178 loss=0.3060
+epoch=23 step=160/178 loss=0.2859
+epoch=23 train_loss=0.2990 train_contrastive=1.3843 train_regression=0.0221 val_loss=0.3003 val_contrastive=1.3864 val_regression=0.0231
+epoch=24 step=20/178 loss=0.3156
+epoch=24 step=40/178 loss=0.2938
+epoch=24 step=60/178 loss=0.2846
+epoch=24 step=80/178 loss=0.3004
+epoch=24 step=100/178 loss=0.3019
+epoch=24 step=120/178 loss=0.2974
+epoch=24 step=140/178 loss=0.2851
+epoch=24 step=160/178 loss=0.2910
+epoch=24 train_loss=0.2990 train_contrastive=1.3842 train_regression=0.0222 val_loss=0.3005 val_contrastive=1.3863 val_regression=0.0232
+epoch=25 step=20/178 loss=0.2903
+epoch=25 step=40/178 loss=0.3081
+epoch=25 step=60/178 loss=0.3006
+epoch=25 step=80/178 loss=0.2898
+epoch=25 step=100/178 loss=0.2920
+epoch=25 step=120/178 loss=0.2952
+epoch=25 step=140/178 loss=0.3087
+epoch=25 step=160/178 loss=0.3051
+epoch=25 train_loss=0.2989 train_contrastive=1.3842 train_regression=0.0221 val_loss=0.3001 val_contrastive=1.3862 val_regression=0.0229
+saved_best runs/foundation/brainiac_lastblock_regalign_best.pt val_loss=0.3001 epoch=25
+epoch=26 step=20/178 loss=0.2983
+epoch=26 step=40/178 loss=0.3193
+epoch=26 step=60/178 loss=0.2922
+epoch=26 step=80/178 loss=0.2926
+epoch=26 step=100/178 loss=0.2925
+epoch=26 step=120/178 loss=0.2968
+epoch=26 step=140/178 loss=0.3209
+epoch=26 step=160/178 loss=0.2963
+epoch=26 train_loss=0.2989 train_contrastive=1.3841 train_regression=0.0221 val_loss=0.3005 val_contrastive=1.3862 val_regression=0.0232
+epoch=27 step=20/178 loss=0.2955
+epoch=27 step=40/178 loss=0.3051
+epoch=27 step=60/178 loss=0.3094
+epoch=27 step=80/178 loss=0.3118
+epoch=27 step=100/178 loss=0.3105
+epoch=27 step=120/178 loss=0.2979
+epoch=27 step=140/178 loss=0.2948
+epoch=27 step=160/178 loss=0.3281
+epoch=27 train_loss=0.2990 train_contrastive=1.3840 train_regression=0.0222 val_loss=0.3002 val_contrastive=1.3862 val_regression=0.0229
+epoch=28 step=20/178 loss=0.2948
+epoch=28 step=40/178 loss=0.2883
+epoch=28 step=60/178 loss=0.2887
+epoch=28 step=80/178 loss=0.2938
+epoch=28 step=100/178 loss=0.3075
+epoch=28 step=120/178 loss=0.2869
+epoch=28 step=140/178 loss=0.2949
+epoch=28 step=160/178 loss=0.3073
+epoch=28 train_loss=0.2988 train_contrastive=1.3840 train_regression=0.0220 val_loss=0.3001 val_contrastive=1.3861 val_regression=0.0229
+saved_best runs/foundation/brainiac_lastblock_regalign_best.pt val_loss=0.3001 epoch=28
+epoch=29 step=20/178 loss=0.2884
+epoch=29 step=40/178 loss=0.2845
+epoch=29 step=60/178 loss=0.2910
+epoch=29 step=80/178 loss=0.3802
+epoch=29 step=100/178 loss=0.2892
+epoch=29 step=120/178 loss=0.2976
+epoch=29 step=140/178 loss=0.3071
+epoch=29 step=160/178 loss=0.2939
+epoch=29 train_loss=0.2989 train_contrastive=1.3840 train_regression=0.0221 val_loss=0.3006 val_contrastive=1.3861 val_regression=0.0234
+epoch=30 step=20/178 loss=0.3272
+epoch=30 step=40/178 loss=0.3204
+epoch=30 step=60/178 loss=0.2958
+epoch=30 step=80/178 loss=0.3011
+epoch=30 step=100/178 loss=0.3031
+epoch=30 step=120/178 loss=0.3244
+epoch=30 step=140/178 loss=0.2923
+epoch=30 step=160/178 loss=0.3000
+epoch=30 train_loss=0.2991 train_contrastive=1.3839 train_regression=0.0224 val_loss=0.3006 val_contrastive=1.3860 val_regression=0.0234
+saved runs/foundation/brainiac_lastblock_regalign.pt
+<frozen importlib._bootstrap_external>:1325: FutureWarning: The cuda.cudart module is deprecated and will be removed in a future release, please switch to use the cuda.bindings.runtime module instead.
+brainiac_missing_keys=['blocks.0.norm_cross_attn.weight', 'blocks.0.norm_cross_attn.bias', 'blocks.0.cross_attn.out_proj.weight', 'blocks.0.cross_attn.out_proj.bias', 'blocks.0.cross_attn.to_q.weight', 'blocks.0.cross_attn.to_k.weight', 'blocks.0.cross_attn.to_v.weight', 'blocks.1.norm_cross_attn.weight']
+checkpoint=runs/foundation/brainiac_lastblock_regalign_best.pt
+manifest=metadata/splits/test.csv
+samples=153.000000
+mae=0.119062
+rmse=0.155278
+pearson=0.731406
+spearman=0.847326
+top5_high_overlap=0.401307
+top5_low_overlap=0.729412
+pet_to_suvr_recall@1=0.006536
+pet_to_suvr_recall@5=0.039216
+pet_to_suvr_recall@10=0.071895
+pet_to_suvr_mrr=0.037240
+pet_to_suvr_median_rank=75.000000
+suvr_to_pet_recall@1=0.026144
+suvr_to_pet_recall@5=0.058824
+suvr_to_pet_recall@10=0.130719
+suvr_to_pet_mrr=0.066782
+suvr_to_pet_median_rank=50.000000

logs/clinical_brainiac_frozen_v2.log

@@ -0,0 +1,23 @@
+<frozen importlib._bootstrap_external>:1325: FutureWarning: The cuda.cudart module is deprecated and will be removed in a future release, please switch to use the cuda.bindings.runtime module instead.
+brainiac_missing_keys=['blocks.0.norm_cross_attn.weight', 'blocks.0.norm_cross_attn.bias', 'blocks.0.cross_attn.out_proj.weight', 'blocks.0.cross_attn.out_proj.bias', 'blocks.0.cross_attn.to_q.weight', 'blocks.0.cross_attn.to_k.weight', 'blocks.0.cross_attn.to_v.weight', 'blocks.1.norm_cross_attn.weight']
+/data/Albus/miniconda3/lib/python3.13/site-packages/sklearn/linear_model/_ridge.py:228: LinAlgWarning: An ill-conditioned matrix detected: slice 0 has rcond = 8.86297328861474e-08.
+  return linalg.solve(A, Xy, assume_a="pos", overwrite_a=True).T
+/data/Albus/miniconda3/lib/python3.13/site-packages/sklearn/linear_model/_ridge.py:228: LinAlgWarning: An ill-conditioned matrix detected: slice 0 has rcond = 8.896465431007528e-08.
+  return linalg.solve(A, Xy, assume_a="pos", overwrite_a=True).T
+/data/Albus/miniconda3/lib/python3.13/site-packages/sklearn/linear_model/_ridge.py:228: LinAlgWarning: An ill-conditioned matrix detected: slice 0 has rcond = 1.182330819915478e-07.
+  return linalg.solve(A, Xy, assume_a="pos", overwrite_a=True).T
+/data/Albus/miniconda3/lib/python3.13/site-packages/sklearn/linear_model/_ridge.py:228: LinAlgWarning: An ill-conditioned matrix detected: slice 0 has rcond = 8.917045590806083e-08.
+  return linalg.solve(A, Xy, assume_a="pos", overwrite_a=True).T
+/data/Albus/miniconda3/lib/python3.13/site-packages/sklearn/linear_model/_ridge.py:228: LinAlgWarning: An ill-conditioned matrix detected: slice 0 has rcond = 8.896466141550263e-08.
+  return linalg.solve(A, Xy, assume_a="pos", overwrite_a=True).T
+/data/Albus/miniconda3/lib/python3.13/site-packages/sklearn/linear_model/_ridge.py:228: LinAlgWarning: An ill-conditioned matrix detected: slice 0 has rcond = 1.182330819915478e-07.
+  return linalg.solve(A, Xy, assume_a="pos", overwrite_a=True).T
+checkpoint=runs/foundation/brainiac_frozen_mlp.pt
+wrote=runs/clinical/brainiac_frozen_clinical_probe.csv
+{'task': 'clinical_label_3way', 'type': 'classification', 'n_train': 710, 'n_val': 152, 'n_test': 153, 'selected_param': 0.3, 'accuracy': 0.43790849673202614, 'balanced_accuracy': 0.4543898263634141, 'macro_f1': 0.4279793700497329, 'auroc': 0.6520980101618948}
+{'task': 'ad_vs_cn', 'type': 'classification', 'n_train': 332, 'n_val': 69, 'n_test': 59, 'selected_param': 1.0, 'accuracy': 0.7627118644067796, 'balanced_accuracy': 0.7632183908045977, 'macro_f1': 0.7626436781609196, 'auroc': 0.8091954022988506}
+{'task': 'pmci_vs_smci', 'type': 'classification', 'n_train': 269, 'n_val': 51, 'n_test': 69, 'selected_param': 10.0, 'accuracy': 0.6956521739130435, 'balanced_accuracy': 0.6491745283018868, 'macro_f1': 0.6247086247086246, 'auroc': 0.660377358490566}
+{'task': 'adas11', 'type': 'regression', 'n_train': 709, 'n_val': 151, 'n_test': 153, 'selected_param': 100.0, 'mae': 4.597776591606389, 'rmse': 5.857034542659296, 'r2': 0.23026390620228632, 'pearson': 0.49249220671669164}
+{'task': 'mmse', 'type': 'regression', 'n_train': 710, 'n_val': 152, 'n_test': 153, 'selected_param': 100.0, 'mae': 2.0437283983417585, 'rmse': 2.466731047611196, 'r2': 0.14414524544080032, 'pearson': 0.3802017119312998}
+{'task': 'ravlt_immediate', 'type': 'regression', 'n_train': 708, 'n_val': 152, 'n_test': 153, 'selected_param': 30.0, 'mae': 9.761533724716287, 'rmse': 11.842277526038245, 'r2': 0.13641071111813963, 'pearson': 0.38600975120126757}
+{'task': 'ldeltotal', 'type': 'regression', 'n_train': 710, 'n_val': 152, 'n_test': 153, 'selected_param': 100.0, 'mae': 3.5811307040694493, 'rmse': 4.512365236674753, 'r2': 0.11637267545449115, 'pearson': 0.3422350912199664}

logs/clinical_sam_med3d_frozen.log

@@ -0,0 +1,29 @@
+[2026-05-20 21:43:33] start sam_med3d_frozen gpu=1 ckpt=runs/foundation/sam_med3d_frozen_mlp_best.pt
+creating model SAM-Med3D
+try to load pretrained weights from pretrained/sam-med3d/sam_med3d_turbo.pth
+/data/Albus/miniconda3/lib/python3.13/site-packages/sklearn/linear_model/_ridge.py:228: LinAlgWarning: An ill-conditioned matrix detected: slice 0 has rcond = 4.511748485924727e-08.
+  return linalg.solve(A, Xy, assume_a="pos", overwrite_a=True).T
+/data/Albus/miniconda3/lib/python3.13/site-packages/sklearn/linear_model/_ridge.py:228: LinAlgWarning: An ill-conditioned matrix detected: slice 0 has rcond = 9.895542518734146e-08.
+  return linalg.solve(A, Xy, assume_a="pos", overwrite_a=True).T
+/data/Albus/miniconda3/lib/python3.13/site-packages/sklearn/linear_model/_ridge.py:228: LinAlgWarning: An ill-conditioned matrix detected: slice 0 has rcond = 4.511748485924727e-08.
+  return linalg.solve(A, Xy, assume_a="pos", overwrite_a=True).T
+/data/Albus/miniconda3/lib/python3.13/site-packages/sklearn/linear_model/_ridge.py:228: LinAlgWarning: An ill-conditioned matrix detected: slice 0 has rcond = 9.895542518734146e-08.
+  return linalg.solve(A, Xy, assume_a="pos", overwrite_a=True).T
+/data/Albus/miniconda3/lib/python3.13/site-packages/sklearn/linear_model/_ridge.py:228: LinAlgWarning: An ill-conditioned matrix detected: slice 0 has rcond = 4.470589942684455e-08.
+  return linalg.solve(A, Xy, assume_a="pos", overwrite_a=True).T
+/data/Albus/miniconda3/lib/python3.13/site-packages/sklearn/linear_model/_ridge.py:228: LinAlgWarning: An ill-conditioned matrix detected: slice 0 has rcond = 9.901283704039088e-08.
+  return linalg.solve(A, Xy, assume_a="pos", overwrite_a=True).T
+/data/Albus/miniconda3/lib/python3.13/site-packages/sklearn/linear_model/_ridge.py:228: LinAlgWarning: An ill-conditioned matrix detected: slice 0 has rcond = 4.475874959553039e-08.
+  return linalg.solve(A, Xy, assume_a="pos", overwrite_a=True).T
+/data/Albus/miniconda3/lib/python3.13/site-packages/sklearn/linear_model/_ridge.py:228: LinAlgWarning: An ill-conditioned matrix detected: slice 0 has rcond = 9.906475639809287e-08.
+  return linalg.solve(A, Xy, assume_a="pos", overwrite_a=True).T
+checkpoint=runs/foundation/sam_med3d_frozen_mlp_best.pt
+wrote=runs/clinical/sam_med3d_frozen_clinical_probe.csv
+{'task': 'clinical_label_3way', 'type': 'classification', 'n_train': 710, 'n_val': 152, 'n_test': 153, 'selected_param': 10.0, 'accuracy': 0.47058823529411764, 'balanced_accuracy': 0.4868672046955245, 'macro_f1': 0.4509157509157509, 'auroc': 0.6999575462312609}
+{'task': 'ad_vs_cn', 'type': 'classification', 'n_train': 332, 'n_val': 69, 'n_test': 59, 'selected_param': 10.0, 'accuracy': 0.8135593220338984, 'balanced_accuracy': 0.8132183908045978, 'macro_f1': 0.813344837503595, 'auroc': 0.8505747126436782}
+{'task': 'pmci_vs_smci', 'type': 'classification', 'n_train': 269, 'n_val': 51, 'n_test': 69, 'selected_param': 0.01, 'accuracy': 0.7681159420289855, 'balanced_accuracy': 0.6963443396226415, 'macro_f1': 0.6877828054298643, 'auroc': 0.7594339622641509}
+{'task': 'mmse', 'type': 'regression', 'n_train': 710, 'n_val': 152, 'n_test': 153, 'selected_param': 30.0, 'mae': 1.84983027838414, 'rmse': 2.2609941849077426, 'r2': 0.28095617966745545, 'pearson': 0.5310391758366578}
+{'task': 'cdrsb', 'type': 'regression', 'n_train': 710, 'n_val': 152, 'n_test': 153, 'selected_param': 0.03, 'mae': 1.050908666810179, 'rmse': 1.353166386070657, 'r2': 0.35818541107926016, 'pearson': 0.6331574717010728}
+{'task': 'adas13', 'type': 'regression', 'n_train': 707, 'n_val': 151, 'n_test': 148, 'selected_param': 0.1, 'mae': 5.581859120549383, 'rmse': 7.179592316072405, 'r2': 0.412714357130716, 'pearson': 0.6661076432108629}
+{'task': 'faq', 'type': 'regression', 'n_train': 704, 'n_val': 152, 'n_test': 153, 'selected_param': 1.0, 'mae': 4.035065822351992, 'rmse': 5.581804945349844, 'r2': 0.3191912593232088, 'pearson': 0.5671677685306808}
+[2026-05-20 21:44:56] done sam_med3d_frozen

logs/clinical_swinunetr_frozen.log

@@ -0,0 +1,12 @@
+[2026-05-20 21:44:56] start swinunetr_frozen gpu=1 ckpt=runs/foundation/swinunetr_frozen_mlp_best.pt
+<frozen importlib._bootstrap_external>:1325: FutureWarning: The cuda.cudart module is deprecated and will be removed in a future release, please switch to use the cuda.bindings.runtime module instead.
+checkpoint=runs/foundation/swinunetr_frozen_mlp_best.pt
+wrote=runs/clinical/swinunetr_frozen_clinical_probe.csv
+{'task': 'clinical_label_3way', 'type': 'classification', 'n_train': 710, 'n_val': 152, 'n_test': 153, 'selected_param': 10.0, 'accuracy': 0.5294117647058824, 'balanced_accuracy': 0.5414771337735388, 'macro_f1': 0.514267736353626, 'auroc': 0.6837291845130701}
+{'task': 'ad_vs_cn', 'type': 'classification', 'n_train': 332, 'n_val': 69, 'n_test': 59, 'selected_param': 10.0, 'accuracy': 0.847457627118644, 'balanced_accuracy': 0.8482758620689655, 'macro_f1': 0.8472821397756687, 'auroc': 0.9114942528735632}
+{'task': 'pmci_vs_smci', 'type': 'classification', 'n_train': 269, 'n_val': 51, 'n_test': 69, 'selected_param': 3.0, 'accuracy': 0.7101449275362319, 'balanced_accuracy': 0.6367924528301887, 'macro_f1': 0.6241830065359477, 'auroc': 0.7370283018867925}
+{'task': 'mmse', 'type': 'regression', 'n_train': 710, 'n_val': 152, 'n_test': 153, 'selected_param': 30.0, 'mae': 1.8839712828592536, 'rmse': 2.297914219287171, 'r2': 0.25728175515439977, 'pearson': 0.5132029557195673}
+{'task': 'cdrsb', 'type': 'regression', 'n_train': 710, 'n_val': 152, 'n_test': 153, 'selected_param': 30.0, 'mae': 1.0241044716897354, 'rmse': 1.322133316801039, 'r2': 0.3872861726655592, 'pearson': 0.623195591701983}
+{'task': 'adas13', 'type': 'regression', 'n_train': 707, 'n_val': 151, 'n_test': 148, 'selected_param': 10.0, 'mae': 6.121648664989986, 'rmse': 7.4727311331796855, 'r2': 0.3637782245701815, 'pearson': 0.6073704834299607}
+{'task': 'faq', 'type': 'regression', 'n_train': 704, 'n_val': 152, 'n_test': 153, 'selected_param': 30.0, 'mae': 3.966766862308278, 'rmse': 5.595969716827848, 'r2': 0.3157315410669784, 'pearson': 0.5623856224956587}
+[2026-05-20 21:45:49] done swinunetr_frozen

logs/download_neurovfm_20260518_005116.log

@@ -0,0 +1,102 @@
+START neurovfm download
+target /data/Albus/Brain/pretrained/neurovfm-encoder
+Traceback (most recent call last):
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/urllib3/connection.py", line 204, in _new_conn
+    sock = connection.create_connection(
+        (self._dns_host, self.port),
+    ...<2 lines>...
+        socket_options=self.socket_options,
+    )
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/urllib3/util/connection.py", line 85, in create_connection
+    raise err
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/urllib3/util/connection.py", line 73, in create_connection
+    sock.connect(sa)
+    ~~~~~~~~~~~~^^^^
+OSError: [Errno 101] Network is unreachable
+
+The above exception was the direct cause of the following exception:
+
+Traceback (most recent call last):
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/urllib3/connectionpool.py", line 787, in urlopen
+    response = self._make_request(
+        conn,
+    ...<10 lines>...
+        **response_kw,
+    )
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/urllib3/connectionpool.py", line 488, in _make_request
+    raise new_e
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/urllib3/connectionpool.py", line 464, in _make_request
+    self._validate_conn(conn)
+    ~~~~~~~~~~~~~~~~~~~^^^^^^
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/urllib3/connectionpool.py", line 1093, in _validate_conn
+    conn.connect()
+    ~~~~~~~~~~~~^^
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/urllib3/connection.py", line 759, in connect
+    self.sock = sock = self._new_conn()
+                       ~~~~~~~~~~~~~~^^
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/urllib3/connection.py", line 219, in _new_conn
+    raise NewConnectionError(
+        self, f"Failed to establish a new connection: {e}"
+    ) from e
+urllib3.exceptions.NewConnectionError: HTTPSConnection(host='huggingface.co', port=443): Failed to establish a new connection: [Errno 101] Network is unreachable
+
+The above exception was the direct cause of the following exception:
+
+Traceback (most recent call last):
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/requests/adapters.py", line 645, in send
+    resp = conn.urlopen(
+        method=request.method,
+    ...<9 lines>...
+        chunked=chunked,
+    )
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/urllib3/connectionpool.py", line 841, in urlopen
+    retries = retries.increment(
+        method, url, error=new_e, _pool=self, _stacktrace=sys.exc_info()[2]
+    )
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/urllib3/util/retry.py", line 535, in increment
+    raise MaxRetryError(_pool, url, reason) from reason  # type: ignore[arg-type]
+    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+urllib3.exceptions.MaxRetryError: HTTPSConnectionPool(host='huggingface.co', port=443): Max retries exceeded with url: /api/models/mlinslab/neurovfm-encoder/revision/main (Caused by NewConnectionError("HTTPSConnection(host='huggingface.co', port=443): Failed to establish a new connection: [Errno 101] Network is unreachable"))
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/tmp/download_neurovfm.py", line 10, in <module>
+    snapshot_download(
+    ~~~~~~~~~~~~~~~~~^
+        repo_id=repo,
+        ^^^^^^^^^^^^^
+    ...<4 lines>...
+        max_workers=1,
+        ^^^^^^^^^^^^^^
+    )
+    ^
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/huggingface_hub/utils/_validators.py", line 118, in _inner_fn
+    return fn(*args, **kwargs)
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/huggingface_hub/_snapshot_download.py", line 187, in snapshot_download
+    repo_info = api.repo_info(repo_id=repo_id, repo_type=repo_type, revision=revision, token=token)
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/huggingface_hub/utils/_validators.py", line 118, in _inner_fn
+    return fn(*args, **kwargs)
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/huggingface_hub/hf_api.py", line 2112, in repo_info
+    return method(
+        repo_id,
+    ...<3 lines>...
+        files_metadata=files_metadata,
+    )
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/huggingface_hub/utils/_validators.py", line 118, in _inner_fn
+    return fn(*args, **kwargs)
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/huggingface_hub/hf_api.py", line 1921, in model_info
+    r = get_session().get(path, headers=headers, timeout=timeout, params=params)
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/requests/sessions.py", line 605, in get
+    return self.request("GET", url, **kwargs)
+           ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/requests/sessions.py", line 592, in request
+    resp = self.send(prep, **send_kwargs)
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/requests/sessions.py", line 706, in send
+    r = adapter.send(request, **kwargs)
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/huggingface_hub/utils/_http.py", line 63, in send
+    return super().send(request, *args, **kwargs)
+           ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/data/Albus/miniconda3/lib/python3.13/site-packages/requests/adapters.py", line 678, in send
+    raise ConnectionError(e, request=request)
+requests.exceptions.ConnectionError: (MaxRetryError('HTTPSConnectionPool(host=\'huggingface.co\', port=443): Max retries exceeded with url: /api/models/mlinslab/neurovfm-encoder/revision/main (Caused by NewConnectionError("HTTPSConnection(host=\'huggingface.co\', port=443): Failed to establish a new connection: [Errno 101] Network is unreachable"))'), '(Request ID: c105d0ad-85cc-41dd-8958-1a7833b0d24d)')

logs/download_neurovfm_official_20260518_005602.log

@@ -0,0 +1 @@
+repo= mlinslab/neurovfm-encoder

logs/eval_remap_pet_clinicalbert_text_alignment_test.log

@@ -0,0 +1,13 @@
+samples=153.000000
+pet_to_text_recall@1=0.078431
+pet_to_text_recall@5=0.333333
+pet_to_text_recall@10=0.490196
+pet_to_text_mrr=0.213128
+pet_to_text_median_rank=11.000000
+text_to_pet_recall@1=0.098039
+text_to_pet_recall@5=0.294118
+text_to_pet_recall@10=0.470588
+text_to_pet_mrr=0.222179
+text_to_pet_median_rank=12.000000
+retrieved_text_low_overlap=0.698039
+retrieved_text_high_overlap=0.563399

logs/eval_sam_med3d_frozen_clinicalbert_text_alignment_test.log

@@ -0,0 +1,15 @@
+creating model SAM-Med3D
+try to load pretrained weights from pretrained/sam-med3d/sam_med3d_turbo.pth
+samples=153.000000
+pet_to_text_recall@1=0.052288
+pet_to_text_recall@5=0.169935
+pet_to_text_recall@10=0.254902
+pet_to_text_mrr=0.131163
+pet_to_text_median_rank=26.000000
+text_to_pet_recall@1=0.026144
+text_to_pet_recall@5=0.163399
+text_to_pet_recall@10=0.209150
+text_to_pet_mrr=0.105663
+text_to_pet_median_rank=30.000000
+retrieved_text_low_overlap=0.662745
+retrieved_text_high_overlap=0.465359

logs/eval_sam_med3d_frozen_mlp_test.log

@@ -0,0 +1,21 @@
+creating model SAM-Med3D
+try to load pretrained weights from pretrained/sam-med3d/sam_med3d_turbo.pth
+checkpoint=runs/foundation/sam_med3d_frozen_mlp_best.pt
+manifest=metadata/splits/test.csv
+samples=153.000000
+mae=0.115097
+rmse=0.146874
+pearson=0.828614
+spearman=0.858010
+top5_high_overlap=0.462745
+top5_low_overlap=0.724183
+pet_to_suvr_recall@1=0.163399
+pet_to_suvr_recall@5=0.418301
+pet_to_suvr_recall@10=0.620915
+pet_to_suvr_mrr=0.299050
+pet_to_suvr_median_rank=7.000000
+suvr_to_pet_recall@1=0.196078
+suvr_to_pet_recall@5=0.562092
+suvr_to_pet_recall@10=0.777778
+suvr_to_pet_mrr=0.362741
+suvr_to_pet_median_rank=4.000000

logs/eval_swinunetr_frozen_clinicalbert_text_alignment_test.log

@@ -0,0 +1,14 @@
+<frozen importlib._bootstrap_external>:1325: FutureWarning: The cuda.cudart module is deprecated and will be removed in a future release, please switch to use the cuda.bindings.runtime module instead.
+samples=153.000000
+pet_to_text_recall@1=0.045752
+pet_to_text_recall@5=0.196078
+pet_to_text_recall@10=0.281046
+pet_to_text_mrr=0.133791
+pet_to_text_median_rank=21.000000
+text_to_pet_recall@1=0.052288
+text_to_pet_recall@5=0.143791
+text_to_pet_recall@10=0.261438
+text_to_pet_mrr=0.122184
+text_to_pet_median_rank=24.000000
+retrieved_text_low_overlap=0.694118
+retrieved_text_high_overlap=0.458824

logs/remap_pet_clinicalbert_text_alignment_b16.log

@@ -0,0 +1,34 @@
+epoch=1 train_loss=2.753767 val_loss=2.710755
+saved_best runs/vlm/remap_pet_clinicalbert_text_alignment_b16_best.pt val_loss=2.710755
+epoch=2 train_loss=2.638914 val_loss=2.664012
+saved_best runs/vlm/remap_pet_clinicalbert_text_alignment_b16_best.pt val_loss=2.664012
+epoch=3 train_loss=2.408463 val_loss=2.619754
+saved_best runs/vlm/remap_pet_clinicalbert_text_alignment_b16_best.pt val_loss=2.619754
+epoch=4 train_loss=2.353396 val_loss=2.296605
+saved_best runs/vlm/remap_pet_clinicalbert_text_alignment_b16_best.pt val_loss=2.296605
+epoch=5 train_loss=2.114704 val_loss=2.224501
+saved_best runs/vlm/remap_pet_clinicalbert_text_alignment_b16_best.pt val_loss=2.224501
+epoch=6 train_loss=1.979567 val_loss=2.115859
+saved_best runs/vlm/remap_pet_clinicalbert_text_alignment_b16_best.pt val_loss=2.115859
+epoch=7 train_loss=1.857410 val_loss=2.103239
+saved_best runs/vlm/remap_pet_clinicalbert_text_alignment_b16_best.pt val_loss=2.103239
+epoch=8 train_loss=1.787650 val_loss=2.099148
+saved_best runs/vlm/remap_pet_clinicalbert_text_alignment_b16_best.pt val_loss=2.099148
+epoch=9 train_loss=1.698219 val_loss=2.055482
+saved_best runs/vlm/remap_pet_clinicalbert_text_alignment_b16_best.pt val_loss=2.055482
+epoch=10 train_loss=1.686719 val_loss=1.947024
+saved_best runs/vlm/remap_pet_clinicalbert_text_alignment_b16_best.pt val_loss=1.947024
+epoch=11 train_loss=1.716927 val_loss=1.970050
+epoch=12 train_loss=1.557844 val_loss=2.024685
+epoch=13 train_loss=1.508503 val_loss=2.026529
+epoch=14 train_loss=1.468290 val_loss=1.844181
+saved_best runs/vlm/remap_pet_clinicalbert_text_alignment_b16_best.pt val_loss=1.844181
+epoch=15 train_loss=1.499628 val_loss=1.832888
+saved_best runs/vlm/remap_pet_clinicalbert_text_alignment_b16_best.pt val_loss=1.832888
+epoch=16 train_loss=1.481773 val_loss=1.839653
+epoch=17 train_loss=1.587373 val_loss=1.683570
+saved_best runs/vlm/remap_pet_clinicalbert_text_alignment_b16_best.pt val_loss=1.683570
+epoch=18 train_loss=1.382078 val_loss=1.853615
+epoch=19 train_loss=1.415400 val_loss=1.759831
+epoch=20 train_loss=1.371480 val_loss=1.730412
+saved runs/vlm/remap_pet_clinicalbert_text_alignment_b16.pt

logs/sam_med3d_frozen_clinicalbert_text_alignment.log

@@ -0,0 +1,32 @@
+creating model SAM-Med3D
+try to load pretrained weights from pretrained/sam-med3d/sam_med3d_turbo.pth
+epoch=1 train_loss=2.751899 val_loss=2.714804
+saved_best runs/vlm/sam_med3d_frozen_clinicalbert_text_alignment_best.pt val_loss=2.714804
+epoch=2 train_loss=2.679979 val_loss=2.669987
+saved_best runs/vlm/sam_med3d_frozen_clinicalbert_text_alignment_best.pt val_loss=2.669987
+epoch=3 train_loss=2.591764 val_loss=2.632059
+saved_best runs/vlm/sam_med3d_frozen_clinicalbert_text_alignment_best.pt val_loss=2.632059
+epoch=4 train_loss=2.490655 val_loss=2.662694
+epoch=5 train_loss=2.555299 val_loss=2.517349
+saved_best runs/vlm/sam_med3d_frozen_clinicalbert_text_alignment_best.pt val_loss=2.517349
+epoch=6 train_loss=2.368108 val_loss=2.718388
+epoch=7 train_loss=2.428306 val_loss=2.610473
+epoch=8 train_loss=2.343087 val_loss=2.551209
+epoch=9 train_loss=2.291833 val_loss=2.511172
+saved_best runs/vlm/sam_med3d_frozen_clinicalbert_text_alignment_best.pt val_loss=2.511172
+epoch=10 train_loss=2.251726 val_loss=2.665643
+epoch=11 train_loss=2.279050 val_loss=2.418705
+saved_best runs/vlm/sam_med3d_frozen_clinicalbert_text_alignment_best.pt val_loss=2.418705
+epoch=12 train_loss=2.185663 val_loss=2.415046
+saved_best runs/vlm/sam_med3d_frozen_clinicalbert_text_alignment_best.pt val_loss=2.415046
+epoch=13 train_loss=2.161236 val_loss=2.496397
+epoch=14 train_loss=2.234160 val_loss=2.441421
+epoch=15 train_loss=2.121753 val_loss=2.380541
+saved_best runs/vlm/sam_med3d_frozen_clinicalbert_text_alignment_best.pt val_loss=2.380541
+epoch=16 train_loss=2.106255 val_loss=2.360190
+saved_best runs/vlm/sam_med3d_frozen_clinicalbert_text_alignment_best.pt val_loss=2.360190
+epoch=17 train_loss=2.087960 val_loss=2.430537
+epoch=18 train_loss=2.112058 val_loss=2.482278
+epoch=19 train_loss=2.021039 val_loss=2.430172
+epoch=20 train_loss=2.116062 val_loss=2.402798
+saved runs/vlm/sam_med3d_frozen_clinicalbert_text_alignment.pt

logs/sam_med3d_frozen_mlp.log

@@ -0,0 +1,1095 @@
+creating model SAM-Med3D
+try to load pretrained weights from pretrained/sam-med3d/sam_med3d_turbo.pth
+device=cuda backbone=sam_med3d encoder_scope=none contrastive_weight=0.2 regression_weight=1.0 train=710 val=152
+epoch=1 step=10/355 loss=0.7136
+epoch=1 step=20/355 loss=0.4104
+epoch=1 step=30/355 loss=0.2083
+epoch=1 step=40/355 loss=0.1626
+epoch=1 step=50/355 loss=0.2020
+epoch=1 step=60/355 loss=0.1439
+epoch=1 step=70/355 loss=0.1541
+epoch=1 step=80/355 loss=0.1885
+epoch=1 step=90/355 loss=0.1540
+epoch=1 step=100/355 loss=0.1724
+epoch=1 step=110/355 loss=0.1469
+epoch=1 step=120/355 loss=0.1501
+epoch=1 step=130/355 loss=0.1580
+epoch=1 step=140/355 loss=0.1602
+epoch=1 step=150/355 loss=0.1466
+epoch=1 step=160/355 loss=0.1405
+epoch=1 step=170/355 loss=0.1486
+epoch=1 step=180/355 loss=0.1685
+epoch=1 step=190/355 loss=0.1415
+epoch=1 step=200/355 loss=0.1461
+epoch=1 step=210/355 loss=0.1108
+epoch=1 step=220/355 loss=0.1542
+epoch=1 step=230/355 loss=0.1561
+epoch=1 step=240/355 loss=0.1687
+epoch=1 step=250/355 loss=0.1459
+epoch=1 step=260/355 loss=0.1140
+epoch=1 step=270/355 loss=0.1003
+epoch=1 step=280/355 loss=0.1400
+epoch=1 step=290/355 loss=0.1727
+epoch=1 step=300/355 loss=0.1309
+epoch=1 step=310/355 loss=0.1253
+epoch=1 step=320/355 loss=0.1999
+epoch=1 step=330/355 loss=0.1294
+epoch=1 step=340/355 loss=0.1837
+epoch=1 step=350/355 loss=0.1632
+epoch=1 train_loss=0.1912 train_contrastive=0.6505 train_regression=0.0611 val_loss=0.1459 val_contrastive=0.6030 val_regression=0.0253
+saved_best runs/foundation/sam_med3d_frozen_mlp_best.pt val_loss=0.1459 epoch=1
+epoch=2 step=10/355 loss=0.1704
+epoch=2 step=20/355 loss=0.1268
+epoch=2 step=30/355 loss=0.1205
+epoch=2 step=40/355 loss=0.1296
+epoch=2 step=50/355 loss=0.1414
+epoch=2 step=60/355 loss=0.1324
+epoch=2 step=70/355 loss=0.1052
+epoch=2 step=80/355 loss=0.0982
+epoch=2 step=90/355 loss=0.0761
+epoch=2 step=100/355 loss=0.1271
+epoch=2 step=110/355 loss=0.1347
+epoch=2 step=120/355 loss=0.1730
+epoch=2 step=130/355 loss=0.2264
+epoch=2 step=140/355 loss=0.1205
+epoch=2 step=150/355 loss=0.1483
+epoch=2 step=160/355 loss=0.1480
+epoch=2 step=170/355 loss=0.2486
+epoch=2 step=180/355 loss=0.0675
+epoch=2 step=190/355 loss=0.1136
+epoch=2 step=200/355 loss=0.1336
+epoch=2 step=210/355 loss=0.1884
+epoch=2 step=220/355 loss=0.0832
+epoch=2 step=230/355 loss=0.1647
+epoch=2 step=240/355 loss=0.0909
+epoch=2 step=250/355 loss=0.0721
+epoch=2 step=260/355 loss=0.2330
+epoch=2 step=270/355 loss=0.0511
+epoch=2 step=280/355 loss=0.0204
+epoch=2 step=290/355 loss=0.1151
+epoch=2 step=300/355 loss=0.0905
+epoch=2 step=310/355 loss=0.1791
+epoch=2 step=320/355 loss=0.0393
+epoch=2 step=330/355 loss=0.0813
+epoch=2 step=340/355 loss=0.1563
+epoch=2 step=350/355 loss=0.0546
+epoch=2 train_loss=0.1200 train_contrastive=0.4719 train_regression=0.0257 val_loss=0.1278 val_contrastive=0.4888 val_regression=0.0300
+saved_best runs/foundation/sam_med3d_frozen_mlp_best.pt val_loss=0.1278 epoch=2
+epoch=3 step=10/355 loss=0.3154
+epoch=3 step=20/355 loss=0.0740
+epoch=3 step=30/355 loss=0.1483
+epoch=3 step=40/355 loss=0.1533
+epoch=3 step=50/355 loss=0.1379
+epoch=3 step=60/355 loss=0.0796
+epoch=3 step=70/355 loss=0.0612
+epoch=3 step=80/355 loss=0.2111
+epoch=3 step=90/355 loss=0.0591
+epoch=3 step=100/355 loss=0.1420
+epoch=3 step=110/355 loss=0.0458
+epoch=3 step=120/355 loss=0.0914
+epoch=3 step=130/355 loss=0.0870
+epoch=3 step=140/355 loss=0.0275
+epoch=3 step=150/355 loss=0.0459
+epoch=3 step=160/355 loss=0.1222
+epoch=3 step=170/355 loss=0.0577
+epoch=3 step=180/355 loss=0.0475
+epoch=3 step=190/355 loss=0.0825
+epoch=3 step=200/355 loss=0.1491
+epoch=3 step=210/355 loss=0.1630
+epoch=3 step=220/355 loss=0.4348
+epoch=3 step=230/355 loss=0.1223
+epoch=3 step=240/355 loss=0.1312
+epoch=3 step=250/355 loss=0.0548
+epoch=3 step=260/355 loss=0.0746
+epoch=3 step=270/355 loss=0.0623
+epoch=3 step=280/355 loss=0.0193
+epoch=3 step=290/355 loss=0.0555
+epoch=3 step=300/355 loss=0.0980
+epoch=3 step=310/355 loss=0.0411
+epoch=3 step=320/355 loss=0.0286
+epoch=3 step=330/355 loss=0.0877
+epoch=3 step=340/355 loss=0.1958
+epoch=3 step=350/355 loss=0.0814
+epoch=3 train_loss=0.0998 train_contrastive=0.3702 train_regression=0.0258 val_loss=0.1133 val_contrastive=0.4312 val_regression=0.0271
+saved_best runs/foundation/sam_med3d_frozen_mlp_best.pt val_loss=0.1133 epoch=3
+epoch=4 step=10/355 loss=0.0700
+epoch=4 step=20/355 loss=0.0654
+epoch=4 step=30/355 loss=0.1634
+epoch=4 step=40/355 loss=0.0260
+epoch=4 step=50/355 loss=0.2388
+epoch=4 step=60/355 loss=0.2101
+epoch=4 step=70/355 loss=0.2292
+epoch=4 step=80/355 loss=0.1180
+epoch=4 step=90/355 loss=0.1188
+epoch=4 step=100/355 loss=0.0467
+epoch=4 step=110/355 loss=0.1484
+epoch=4 step=120/355 loss=0.1815
+epoch=4 step=130/355 loss=0.0549
+epoch=4 step=140/355 loss=0.0351
+epoch=4 step=150/355 loss=0.1020
+epoch=4 step=160/355 loss=0.0360
+epoch=4 step=170/355 loss=0.1094
+epoch=4 step=180/355 loss=0.0667
+epoch=4 step=190/355 loss=0.1368
+epoch=4 step=200/355 loss=0.0180
+epoch=4 step=210/355 loss=0.0204
+epoch=4 step=220/355 loss=0.0333
+epoch=4 step=230/355 loss=0.0324
+epoch=4 step=240/355 loss=0.0674
+epoch=4 step=250/355 loss=0.2402
+epoch=4 step=260/355 loss=0.0721
+epoch=4 step=270/355 loss=0.0897
+epoch=4 step=280/355 loss=0.3145
+epoch=4 step=290/355 loss=0.0214
+epoch=4 step=300/355 loss=0.1652
+epoch=4 step=310/355 loss=0.0291
+epoch=4 step=320/355 loss=0.0362
+epoch=4 step=330/355 loss=0.0962
+epoch=4 step=340/355 loss=0.0191
+epoch=4 step=350/355 loss=0.0408
+epoch=4 train_loss=0.0852 train_contrastive=0.3077 train_regression=0.0236 val_loss=0.0964 val_contrastive=0.3600 val_regression=0.0244
+saved_best runs/foundation/sam_med3d_frozen_mlp_best.pt val_loss=0.0964 epoch=4
+epoch=5 step=10/355 loss=0.0842
+epoch=5 step=20/355 loss=0.0698
+epoch=5 step=30/355 loss=0.1150
+epoch=5 step=40/355 loss=0.0564
+epoch=5 step=50/355 loss=0.0873
+epoch=5 step=60/355 loss=0.2761
+epoch=5 step=70/355 loss=0.1216
+epoch=5 step=80/355 loss=0.0296
+epoch=5 step=90/355 loss=0.0304
+epoch=5 step=100/355 loss=0.0237
+epoch=5 step=110/355 loss=0.0152
+epoch=5 step=120/355 loss=0.0549
+epoch=5 step=130/355 loss=0.0403
+epoch=5 step=140/355 loss=0.0401
+epoch=5 step=150/355 loss=0.1060
+epoch=5 step=160/355 loss=0.1013
+epoch=5 step=170/355 loss=0.0686
+epoch=5 step=180/355 loss=0.0508
+epoch=5 step=190/355 loss=0.0271
+epoch=5 step=200/355 loss=0.0550
+epoch=5 step=210/355 loss=0.0699
+epoch=5 step=220/355 loss=0.0765
+epoch=5 step=230/355 loss=0.2985
+epoch=5 step=240/355 loss=0.0353
+epoch=5 step=250/355 loss=0.0265
+epoch=5 step=260/355 loss=0.0422
+epoch=5 step=270/355 loss=0.0155
+epoch=5 step=280/355 loss=0.0345
+epoch=5 step=290/355 loss=0.0273
+epoch=5 step=300/355 loss=0.0236
+epoch=5 step=310/355 loss=0.1952
+epoch=5 step=320/355 loss=0.0231
+epoch=5 step=330/355 loss=0.0234
+epoch=5 step=340/355 loss=0.0437
+epoch=5 step=350/355 loss=0.0539
+epoch=5 train_loss=0.0781 train_contrastive=0.2764 train_regression=0.0229 val_loss=0.1013 val_contrastive=0.3744 val_regression=0.0264
+epoch=6 step=10/355 loss=0.0955
+epoch=6 step=20/355 loss=0.0408
+epoch=6 step=30/355 loss=0.1542
+epoch=6 step=40/355 loss=0.0517
+epoch=6 step=50/355 loss=0.0464
+epoch=6 step=60/355 loss=0.0237
+epoch=6 step=70/355 loss=0.0350
+epoch=6 step=80/355 loss=0.0187
+epoch=6 step=90/355 loss=0.1097
+epoch=6 step=100/355 loss=0.0536
+epoch=6 step=110/355 loss=0.0454
+epoch=6 step=120/355 loss=0.0275
+epoch=6 step=130/355 loss=0.0463
+epoch=6 step=140/355 loss=0.0457
+epoch=6 step=150/355 loss=0.0245
+epoch=6 step=160/355 loss=0.0460
+epoch=6 step=170/355 loss=0.0370
+epoch=6 step=180/355 loss=0.2482
+epoch=6 step=190/355 loss=0.0492
+epoch=6 step=200/355 loss=0.0393
+epoch=6 step=210/355 loss=0.0682
+epoch=6 step=220/355 loss=0.2146
+epoch=6 step=230/355 loss=0.1287
+epoch=6 step=240/355 loss=0.1588
+epoch=6 step=250/355 loss=0.0961
+epoch=6 step=260/355 loss=0.0663
+epoch=6 step=270/355 loss=0.0363
+epoch=6 step=280/355 loss=0.0208
+epoch=6 step=290/355 loss=0.0195
+epoch=6 step=300/355 loss=0.0534
+epoch=6 step=310/355 loss=0.0400
+epoch=6 step=320/355 loss=0.0514
+epoch=6 step=330/355 loss=0.0387
+epoch=6 step=340/355 loss=0.2138
+epoch=6 step=350/355 loss=0.1313
+epoch=6 train_loss=0.0730 train_contrastive=0.2555 train_regression=0.0219 val_loss=0.1012 val_contrastive=0.3912 val_regression=0.0229
+epoch=7 step=10/355 loss=0.0645
+epoch=7 step=20/355 loss=0.0255
+epoch=7 step=30/355 loss=0.2188
+epoch=7 step=40/355 loss=0.0633
+epoch=7 step=50/355 loss=0.0133
+epoch=7 step=60/355 loss=0.0110
+epoch=7 step=70/355 loss=0.0347
+epoch=7 step=80/355 loss=0.0759
+epoch=7 step=90/355 loss=0.0191
+epoch=7 step=100/355 loss=0.0144
+epoch=7 step=110/355 loss=0.0132
+epoch=7 step=120/355 loss=0.0072
+epoch=7 step=130/355 loss=0.0443
+epoch=7 step=140/355 loss=0.0345
+epoch=7 step=150/355 loss=0.0445
+epoch=7 step=160/355 loss=0.0221
+epoch=7 step=170/355 loss=0.0675
+epoch=7 step=180/355 loss=0.1014
+epoch=7 step=190/355 loss=0.0262
+epoch=7 step=200/355 loss=0.2089
+epoch=7 step=210/355 loss=0.0223
+epoch=7 step=220/355 loss=0.0083
+epoch=7 step=230/355 loss=0.0985
+epoch=7 step=240/355 loss=0.0156
+epoch=7 step=250/355 loss=0.1346
+epoch=7 step=260/355 loss=0.0437
+epoch=7 step=270/355 loss=0.1303
+epoch=7 step=280/355 loss=0.0162
+epoch=7 step=290/355 loss=0.0606
+epoch=7 step=300/355 loss=0.0378
+epoch=7 step=310/355 loss=0.1308
+epoch=7 step=320/355 loss=0.0173
+epoch=7 step=330/355 loss=0.0703
+epoch=7 step=340/355 loss=0.0183
+epoch=7 step=350/355 loss=0.1999
+epoch=7 train_loss=0.0630 train_contrastive=0.2072 train_regression=0.0216 val_loss=0.0910 val_contrastive=0.3400 val_regression=0.0230
+saved_best runs/foundation/sam_med3d_frozen_mlp_best.pt val_loss=0.0910 epoch=7
+epoch=8 step=10/355 loss=0.1262
+epoch=8 step=20/355 loss=0.0322
+epoch=8 step=30/355 loss=0.0546
+epoch=8 step=40/355 loss=0.0166
+epoch=8 step=50/355 loss=0.0154
+epoch=8 step=60/355 loss=0.0886
+epoch=8 step=70/355 loss=0.0627
+epoch=8 step=80/355 loss=0.0452
+epoch=8 step=90/355 loss=0.0408
+epoch=8 step=100/355 loss=0.0354
+epoch=8 step=110/355 loss=0.0388
+epoch=8 step=120/355 loss=0.1588
+epoch=8 step=130/355 loss=0.0950
+epoch=8 step=140/355 loss=0.0918
+epoch=8 step=150/355 loss=0.0122
+epoch=8 step=160/355 loss=0.1754
+epoch=8 step=170/355 loss=0.0470
+epoch=8 step=180/355 loss=0.0164
+epoch=8 step=190/355 loss=0.0317
+epoch=8 step=200/355 loss=0.0755
+epoch=8 step=210/355 loss=0.0310
+epoch=8 step=220/355 loss=0.2704
+epoch=8 step=230/355 loss=0.1094
+epoch=8 step=240/355 loss=0.0279
+epoch=8 step=250/355 loss=0.0267
+epoch=8 step=260/355 loss=0.0102
+epoch=8 step=270/355 loss=0.0625
+epoch=8 step=280/355 loss=0.2066
+epoch=8 step=290/355 loss=0.0105
+epoch=8 step=300/355 loss=0.1889
+epoch=8 step=310/355 loss=0.0132
+epoch=8 step=320/355 loss=0.0166
+epoch=8 step=330/355 loss=0.0101
+epoch=8 step=340/355 loss=0.1069
+epoch=8 step=350/355 loss=0.0197
+epoch=8 train_loss=0.0606 train_contrastive=0.2021 train_regression=0.0202 val_loss=0.0906 val_contrastive=0.3428 val_regression=0.0221
+saved_best runs/foundation/sam_med3d_frozen_mlp_best.pt val_loss=0.0906 epoch=8
+epoch=9 step=10/355 loss=0.0320
+epoch=9 step=20/355 loss=0.2058
+epoch=9 step=30/355 loss=0.0116
+epoch=9 step=40/355 loss=0.0331
+epoch=9 step=50/355 loss=0.0768
+epoch=9 step=60/355 loss=0.0158
+epoch=9 step=70/355 loss=0.0162
+epoch=9 step=80/355 loss=0.0444
+epoch=9 step=90/355 loss=0.0473
+epoch=9 step=100/355 loss=0.0136
+epoch=9 step=110/355 loss=0.0167
+epoch=9 step=120/355 loss=0.2044
+epoch=9 step=130/355 loss=0.0521
+epoch=9 step=140/355 loss=0.0232
+epoch=9 step=150/355 loss=0.0268
+epoch=9 step=160/355 loss=0.0143
+epoch=9 step=170/355 loss=0.1851
+epoch=9 step=180/355 loss=0.0658
+epoch=9 step=190/355 loss=0.0595
+epoch=9 step=200/355 loss=0.2091
+epoch=9 step=210/355 loss=0.2153
+epoch=9 step=220/355 loss=0.1182
+epoch=9 step=230/355 loss=0.1359
+epoch=9 step=240/355 loss=0.0474
+epoch=9 step=250/355 loss=0.0084
+epoch=9 step=260/355 loss=0.0287
+epoch=9 step=270/355 loss=0.0172
+epoch=9 step=280/355 loss=0.0097
+epoch=9 step=290/355 loss=0.0486
+epoch=9 step=300/355 loss=0.0486
+epoch=9 step=310/355 loss=0.0228
+epoch=9 step=320/355 loss=0.0738
+epoch=9 step=330/355 loss=0.0586
+epoch=9 step=340/355 loss=0.0673
+epoch=9 step=350/355 loss=0.0684
+epoch=9 train_loss=0.0615 train_contrastive=0.2066 train_regression=0.0202 val_loss=0.0789 val_contrastive=0.2942 val_regression=0.0201
+saved_best runs/foundation/sam_med3d_frozen_mlp_best.pt val_loss=0.0789 epoch=9
+epoch=10 step=10/355 loss=0.0205
+epoch=10 step=20/355 loss=0.0270
+epoch=10 step=30/355 loss=0.0445
+epoch=10 step=40/355 loss=0.1162
+epoch=10 step=50/355 loss=0.0212
+epoch=10 step=60/355 loss=0.0604
+epoch=10 step=70/355 loss=0.0245
+epoch=10 step=80/355 loss=0.0243
+epoch=10 step=90/355 loss=0.1620
+epoch=10 step=100/355 loss=0.0325
+epoch=10 step=110/355 loss=0.0102
+epoch=10 step=120/355 loss=0.0372
+epoch=10 step=130/355 loss=0.0159
+epoch=10 step=140/355 loss=0.0965
+epoch=10 step=150/355 loss=0.1286
+epoch=10 step=160/355 loss=0.0703
+epoch=10 step=170/355 loss=0.1340
+epoch=10 step=180/355 loss=0.0582
+epoch=10 step=190/355 loss=0.0308
+epoch=10 step=200/355 loss=0.3056
+epoch=10 step=210/355 loss=0.1556
+epoch=10 step=220/355 loss=0.0743
+epoch=10 step=230/355 loss=0.0194
+epoch=10 step=240/355 loss=0.0110
+epoch=10 step=250/355 loss=0.0145
+epoch=10 step=260/355 loss=0.0203
+epoch=10 step=270/355 loss=0.0195
+epoch=10 step=280/355 loss=0.0921
+epoch=10 step=290/355 loss=0.0534
+epoch=10 step=300/355 loss=0.0652
+epoch=10 step=310/355 loss=0.1474
+epoch=10 step=320/355 loss=0.0398
+epoch=10 step=330/355 loss=0.0290
+epoch=10 step=340/355 loss=0.0200
+epoch=10 step=350/355 loss=0.0361
+epoch=10 train_loss=0.0542 train_contrastive=0.1724 train_regression=0.0198 val_loss=0.0949 val_contrastive=0.3738 val_regression=0.0201
+epoch=11 step=10/355 loss=0.0269
+epoch=11 step=20/355 loss=0.0161
+epoch=11 step=30/355 loss=0.0848
+epoch=11 step=40/355 loss=0.0107
+epoch=11 step=50/355 loss=0.0283
+epoch=11 step=60/355 loss=0.0358
+epoch=11 step=70/355 loss=0.0782
+epoch=11 step=80/355 loss=0.0270
+epoch=11 step=90/355 loss=0.0149
+epoch=11 step=100/355 loss=0.0356
+epoch=11 step=110/355 loss=0.1782
+epoch=11 step=120/355 loss=0.0479
+epoch=11 step=130/355 loss=0.0118
+epoch=11 step=140/355 loss=0.0934
+epoch=11 step=150/355 loss=0.0370
+epoch=11 step=160/355 loss=0.0253
+epoch=11 step=170/355 loss=0.0407
+epoch=11 step=180/355 loss=0.0266
+epoch=11 step=190/355 loss=0.0167
+epoch=11 step=200/355 loss=0.0842
+epoch=11 step=210/355 loss=0.0256
+epoch=11 step=220/355 loss=0.0137
+epoch=11 step=230/355 loss=0.0277
+epoch=11 step=240/355 loss=0.0385
+epoch=11 step=250/355 loss=0.0663
+epoch=11 step=260/355 loss=0.0681
+epoch=11 step=270/355 loss=0.0176
+epoch=11 step=280/355 loss=0.0218
+epoch=11 step=290/355 loss=0.0288
+epoch=11 step=300/355 loss=0.0103
+epoch=11 step=310/355 loss=0.0188
+epoch=11 step=320/355 loss=0.0120
+epoch=11 step=330/355 loss=0.0677
+epoch=11 step=340/355 loss=0.0785
+epoch=11 step=350/355 loss=0.0736
+epoch=11 train_loss=0.0509 train_contrastive=0.1561 train_regression=0.0197 val_loss=0.0785 val_contrastive=0.2771 val_regression=0.0230
+saved_best runs/foundation/sam_med3d_frozen_mlp_best.pt val_loss=0.0785 epoch=11
+epoch=12 step=10/355 loss=0.1544
+epoch=12 step=20/355 loss=0.0933
+epoch=12 step=30/355 loss=0.0103
+epoch=12 step=40/355 loss=0.0168
+epoch=12 step=50/355 loss=0.0697
+epoch=12 step=60/355 loss=0.0470
+epoch=12 step=70/355 loss=0.0129
+epoch=12 step=80/355 loss=0.0309
+epoch=12 step=90/355 loss=0.0111
+epoch=12 step=100/355 loss=0.0113
+epoch=12 step=110/355 loss=0.0913
+epoch=12 step=120/355 loss=0.0269
+epoch=12 step=130/355 loss=0.0175
+epoch=12 step=140/355 loss=0.0197
+epoch=12 step=150/355 loss=0.0237
+epoch=12 step=160/355 loss=0.1036
+epoch=12 step=170/355 loss=0.0387
+epoch=12 step=180/355 loss=0.1888
+epoch=12 step=190/355 loss=0.0337
+epoch=12 step=200/355 loss=0.0838
+epoch=12 step=210/355 loss=0.0254
+epoch=12 step=220/355 loss=0.0105
+epoch=12 step=230/355 loss=0.0289
+epoch=12 step=240/355 loss=0.0151
+epoch=12 step=250/355 loss=0.0212
+epoch=12 step=260/355 loss=0.0729
+epoch=12 step=270/355 loss=0.0132
+epoch=12 step=280/355 loss=0.0220
+epoch=12 step=290/355 loss=0.0123
+epoch=12 step=300/355 loss=0.0118
+epoch=12 step=310/355 loss=0.0681
+epoch=12 step=320/355 loss=0.0905
+epoch=12 step=330/355 loss=0.0380
+epoch=12 step=340/355 loss=0.0085
+epoch=12 step=350/355 loss=0.2230
+epoch=12 train_loss=0.0489 train_contrastive=0.1483 train_regression=0.0193 val_loss=0.0719 val_contrastive=0.2578 val_regression=0.0204
+saved_best runs/foundation/sam_med3d_frozen_mlp_best.pt val_loss=0.0719 epoch=12
+epoch=13 step=10/355 loss=0.0163
+epoch=13 step=20/355 loss=0.0281
+epoch=13 step=30/355 loss=0.0343
+epoch=13 step=40/355 loss=0.0470
+epoch=13 step=50/355 loss=0.0936
+epoch=13 step=60/355 loss=0.0558
+epoch=13 step=70/355 loss=0.0242
+epoch=13 step=80/355 loss=0.0896
+epoch=13 step=90/355 loss=0.0114
+epoch=13 step=100/355 loss=0.0829
+epoch=13 step=110/355 loss=0.0501
+epoch=13 step=120/355 loss=0.0139
+epoch=13 step=130/355 loss=0.1523
+epoch=13 step=140/355 loss=0.0133
+epoch=13 step=150/355 loss=0.0454
+epoch=13 step=160/355 loss=0.0296
+epoch=13 step=170/355 loss=0.0534
+epoch=13 step=180/355 loss=0.1737
+epoch=13 step=190/355 loss=0.0248
+epoch=13 step=200/355 loss=0.0176
+epoch=13 step=210/355 loss=0.1679
+epoch=13 step=220/355 loss=0.0786
+epoch=13 step=230/355 loss=0.0609
+epoch=13 step=240/355 loss=0.0721
+epoch=13 step=250/355 loss=0.2285
+epoch=13 step=260/355 loss=0.0195
+epoch=13 step=270/355 loss=0.0341
+epoch=13 step=280/355 loss=0.0093
+epoch=13 step=290/355 loss=0.0716
+epoch=13 step=300/355 loss=0.0359
+epoch=13 step=310/355 loss=0.0058
+epoch=13 step=320/355 loss=0.0170
+epoch=13 step=330/355 loss=0.0389
+epoch=13 step=340/355 loss=0.0149
+epoch=13 step=350/355 loss=0.0202
+epoch=13 train_loss=0.0486 train_contrastive=0.1505 train_regression=0.0185 val_loss=0.0832 val_contrastive=0.3113 val_regression=0.0210
+epoch=14 step=10/355 loss=0.0653
+epoch=14 step=20/355 loss=0.1521
+epoch=14 step=30/355 loss=0.0192
+epoch=14 step=40/355 loss=0.0230
+epoch=14 step=50/355 loss=0.0207
+epoch=14 step=60/355 loss=0.0166
+epoch=14 step=70/355 loss=0.0146
+epoch=14 step=80/355 loss=0.0290
+epoch=14 step=90/355 loss=0.0147
+epoch=14 step=100/355 loss=0.1621
+epoch=14 step=110/355 loss=0.0987
+epoch=14 step=120/355 loss=0.0237
+epoch=14 step=130/355 loss=0.0150
+epoch=14 step=140/355 loss=0.0086
+epoch=14 step=150/355 loss=0.0398
+epoch=14 step=160/355 loss=0.0264
+epoch=14 step=170/355 loss=0.0508
+epoch=14 step=180/355 loss=0.0144
+epoch=14 step=190/355 loss=0.0421
+epoch=14 step=200/355 loss=0.0079
+epoch=14 step=210/355 loss=0.0187
+epoch=14 step=220/355 loss=0.0511
+epoch=14 step=230/355 loss=0.0146
+epoch=14 step=240/355 loss=0.0603
+epoch=14 step=250/355 loss=0.0151
+epoch=14 step=260/355 loss=0.0777
+epoch=14 step=270/355 loss=0.1329
+epoch=14 step=280/355 loss=0.0384
+epoch=14 step=290/355 loss=0.0175
+epoch=14 step=300/355 loss=0.0108
+epoch=14 step=310/355 loss=0.0046
+epoch=14 step=320/355 loss=0.0341
+epoch=14 step=330/355 loss=0.0146
+epoch=14 step=340/355 loss=0.1239
+epoch=14 step=350/355 loss=0.0233
+epoch=14 train_loss=0.0444 train_contrastive=0.1305 train_regression=0.0183 val_loss=0.0862 val_contrastive=0.3372 val_regression=0.0187
+epoch=15 step=10/355 loss=0.0171
+epoch=15 step=20/355 loss=0.1622
+epoch=15 step=30/355 loss=0.1619
+epoch=15 step=40/355 loss=0.0223
+epoch=15 step=50/355 loss=0.0326
+epoch=15 step=60/355 loss=0.0094
+epoch=15 step=70/355 loss=0.0239
+epoch=15 step=80/355 loss=0.0374
+epoch=15 step=90/355 loss=0.2170
+epoch=15 step=100/355 loss=0.0641
+epoch=15 step=110/355 loss=0.0487
+epoch=15 step=120/355 loss=0.2126
+epoch=15 step=130/355 loss=0.0997
+epoch=15 step=140/355 loss=0.1125
+epoch=15 step=150/355 loss=0.0083
+epoch=15 step=160/355 loss=0.0251
+epoch=15 step=170/355 loss=0.0350
+epoch=15 step=180/355 loss=0.0164
+epoch=15 step=190/355 loss=0.0708
+epoch=15 step=200/355 loss=0.0423
+epoch=15 step=210/355 loss=0.0123
+epoch=15 step=220/355 loss=0.0553
+epoch=15 step=230/355 loss=0.0061
+epoch=15 step=240/355 loss=0.0276
+epoch=15 step=250/355 loss=0.0322
+epoch=15 step=260/355 loss=0.1579
+epoch=15 step=270/355 loss=0.0095
+epoch=15 step=280/355 loss=0.0362
+epoch=15 step=290/355 loss=0.0119
+epoch=15 step=300/355 loss=0.0303
+epoch=15 step=310/355 loss=0.0579
+epoch=15 step=320/355 loss=0.0662
+epoch=15 step=330/355 loss=0.0117
+epoch=15 step=340/355 loss=0.0126
+epoch=15 step=350/355 loss=0.0246
+epoch=15 train_loss=0.0469 train_contrastive=0.1427 train_regression=0.0183 val_loss=0.0889 val_contrastive=0.3274 val_regression=0.0234
+epoch=16 step=10/355 loss=0.0088
+epoch=16 step=20/355 loss=0.0198
+epoch=16 step=30/355 loss=0.0114
+epoch=16 step=40/355 loss=0.0121
+epoch=16 step=50/355 loss=0.0454
+epoch=16 step=60/355 loss=0.0201
+epoch=16 step=70/355 loss=0.0289
+epoch=16 step=80/355 loss=0.0316
+epoch=16 step=90/355 loss=0.0454
+epoch=16 step=100/355 loss=0.0651
+epoch=16 step=110/355 loss=0.0226
+epoch=16 step=120/355 loss=0.0733
+epoch=16 step=130/355 loss=0.1161
+epoch=16 step=140/355 loss=0.0131
+epoch=16 step=150/355 loss=0.0197
+epoch=16 step=160/355 loss=0.0117
+epoch=16 step=170/355 loss=0.0223
+epoch=16 step=180/355 loss=0.0369
+epoch=16 step=190/355 loss=0.0244
+epoch=16 step=200/355 loss=0.0367
+epoch=16 step=210/355 loss=0.0126
+epoch=16 step=220/355 loss=0.1387
+epoch=16 step=230/355 loss=0.0214
+epoch=16 step=240/355 loss=0.0156
+epoch=16 step=250/355 loss=0.1735
+epoch=16 step=260/355 loss=0.0168
+epoch=16 step=270/355 loss=0.0268
+epoch=16 step=280/355 loss=0.0221
+epoch=16 step=290/355 loss=0.0100
+epoch=16 step=300/355 loss=0.0313
+epoch=16 step=310/355 loss=0.0381
+epoch=16 step=320/355 loss=0.0596
+epoch=16 step=330/355 loss=0.0385
+epoch=16 step=340/355 loss=0.0163
+epoch=16 step=350/355 loss=0.0233
+epoch=16 train_loss=0.0457 train_contrastive=0.1315 train_regression=0.0194 val_loss=0.0788 val_contrastive=0.2845 val_regression=0.0219
+epoch=17 step=10/355 loss=0.0135
+epoch=17 step=20/355 loss=0.0191
+epoch=17 step=30/355 loss=0.0310
+epoch=17 step=40/355 loss=0.0086
+epoch=17 step=50/355 loss=0.0208
+epoch=17 step=60/355 loss=0.0121
+epoch=17 step=70/355 loss=0.0141
+epoch=17 step=80/355 loss=0.0400
+epoch=17 step=90/355 loss=0.0140
+epoch=17 step=100/355 loss=0.0807
+epoch=17 step=110/355 loss=0.0240
+epoch=17 step=120/355 loss=0.0250
+epoch=17 step=130/355 loss=0.1727
+epoch=17 step=140/355 loss=0.0095
+epoch=17 step=150/355 loss=0.1619
+epoch=17 step=160/355 loss=0.0118
+epoch=17 step=170/355 loss=0.0234
+epoch=17 step=180/355 loss=0.0143
+epoch=17 step=190/355 loss=0.0149
+epoch=17 step=200/355 loss=0.0089
+epoch=17 step=210/355 loss=0.0083
+epoch=17 step=220/355 loss=0.0232
+epoch=17 step=230/355 loss=0.0268
+epoch=17 step=240/355 loss=0.0173
+epoch=17 step=250/355 loss=0.0399
+epoch=17 step=260/355 loss=0.0419
+epoch=17 step=270/355 loss=0.0135
+epoch=17 step=280/355 loss=0.0401
+epoch=17 step=290/355 loss=0.0083
+epoch=17 step=300/355 loss=0.1468
+epoch=17 step=310/355 loss=0.0088
+epoch=17 step=320/355 loss=0.0119
+epoch=17 step=330/355 loss=0.0152
+epoch=17 step=340/355 loss=0.0374
+epoch=17 step=350/355 loss=0.0256
+epoch=17 train_loss=0.0398 train_contrastive=0.1108 train_regression=0.0177 val_loss=0.0817 val_contrastive=0.3148 val_regression=0.0188
+epoch=18 step=10/355 loss=0.0259
+epoch=18 step=20/355 loss=0.0193
+epoch=18 step=30/355 loss=0.1263
+epoch=18 step=40/355 loss=0.0273
+epoch=18 step=50/355 loss=0.0095
+epoch=18 step=60/355 loss=0.0087
+epoch=18 step=70/355 loss=0.0111
+epoch=18 step=80/355 loss=0.0134
+epoch=18 step=90/355 loss=0.0408
+epoch=18 step=100/355 loss=0.0451
+epoch=18 step=110/355 loss=0.0310
+epoch=18 step=120/355 loss=0.0221
+epoch=18 step=130/355 loss=0.1454
+epoch=18 step=140/355 loss=0.0102
+epoch=18 step=150/355 loss=0.0107
+epoch=18 step=160/355 loss=0.0139
+epoch=18 step=170/355 loss=0.0202
+epoch=18 step=180/355 loss=0.1266
+epoch=18 step=190/355 loss=0.0239
+epoch=18 step=200/355 loss=0.1558
+epoch=18 step=210/355 loss=0.0736
+epoch=18 step=220/355 loss=0.0647
+epoch=18 step=230/355 loss=0.0134
+epoch=18 step=240/355 loss=0.0334
+epoch=18 step=250/355 loss=0.0173
+epoch=18 step=260/355 loss=0.0287
+epoch=18 step=270/355 loss=0.0209
+epoch=18 step=280/355 loss=0.0127
+epoch=18 step=290/355 loss=0.0582
+epoch=18 step=300/355 loss=0.0119
+epoch=18 step=310/355 loss=0.0191
+epoch=18 step=320/355 loss=0.0139
+epoch=18 step=330/355 loss=0.0139
+epoch=18 step=340/355 loss=0.0375
+epoch=18 step=350/355 loss=0.1457
+epoch=18 train_loss=0.0445 train_contrastive=0.1323 train_regression=0.0180 val_loss=0.0797 val_contrastive=0.2930 val_regression=0.0211
+epoch=19 step=10/355 loss=0.1000
+epoch=19 step=20/355 loss=0.1062
+epoch=19 step=30/355 loss=0.0773
+epoch=19 step=40/355 loss=0.0112
+epoch=19 step=50/355 loss=0.0281
+epoch=19 step=60/355 loss=0.0875
+epoch=19 step=70/355 loss=0.0609
+epoch=19 step=80/355 loss=0.0071
+epoch=19 step=90/355 loss=0.0095
+epoch=19 step=100/355 loss=0.0135
+epoch=19 step=110/355 loss=0.0111
+epoch=19 step=120/355 loss=0.0261
+epoch=19 step=130/355 loss=0.0394
+epoch=19 step=140/355 loss=0.0138
+epoch=19 step=150/355 loss=0.0995
+epoch=19 step=160/355 loss=0.0204
+epoch=19 step=170/355 loss=0.0421
+epoch=19 step=180/355 loss=0.0313
+epoch=19 step=190/355 loss=0.0119
+epoch=19 step=200/355 loss=0.0300
+epoch=19 step=210/355 loss=0.0080
+epoch=19 step=220/355 loss=0.0519
+epoch=19 step=230/355 loss=0.0350
+epoch=19 step=240/355 loss=0.0193
+epoch=19 step=250/355 loss=0.0169
+epoch=19 step=260/355 loss=0.0152
+epoch=19 step=270/355 loss=0.0390
+epoch=19 step=280/355 loss=0.0360
+epoch=19 step=290/355 loss=0.0527
+epoch=19 step=300/355 loss=0.0134
+epoch=19 step=310/355 loss=0.0359
+epoch=19 step=320/355 loss=0.0132
+epoch=19 step=330/355 loss=0.0340
+epoch=19 step=340/355 loss=0.0185
+epoch=19 step=350/355 loss=0.0398
+epoch=19 train_loss=0.0366 train_contrastive=0.0947 train_regression=0.0176 val_loss=0.0673 val_contrastive=0.2460 val_regression=0.0181
+saved_best runs/foundation/sam_med3d_frozen_mlp_best.pt val_loss=0.0673 epoch=19
+epoch=20 step=10/355 loss=0.0238
+epoch=20 step=20/355 loss=0.1086
+epoch=20 step=30/355 loss=0.0076
+epoch=20 step=40/355 loss=0.0216
+epoch=20 step=50/355 loss=0.0191
+epoch=20 step=60/355 loss=0.0133
+epoch=20 step=70/355 loss=0.0360
+epoch=20 step=80/355 loss=0.0194
+epoch=20 step=90/355 loss=0.3886
+epoch=20 step=100/355 loss=0.0125
+epoch=20 step=110/355 loss=0.0440
+epoch=20 step=120/355 loss=0.0146
+epoch=20 step=130/355 loss=0.0954
+epoch=20 step=140/355 loss=0.1797
+epoch=20 step=150/355 loss=0.0808
+epoch=20 step=160/355 loss=0.0142
+epoch=20 step=170/355 loss=0.0873
+epoch=20 step=180/355 loss=0.0632
+epoch=20 step=190/355 loss=0.0118
+epoch=20 step=200/355 loss=0.0247
+epoch=20 step=210/355 loss=0.1230
+epoch=20 step=220/355 loss=0.0939
+epoch=20 step=230/355 loss=0.1001
+epoch=20 step=240/355 loss=0.0389
+epoch=20 step=250/355 loss=0.0421
+epoch=20 step=260/355 loss=0.0087
+epoch=20 step=270/355 loss=0.0370
+epoch=20 step=280/355 loss=0.0256
+epoch=20 step=290/355 loss=0.0548
+epoch=20 step=300/355 loss=0.0178
+epoch=20 step=310/355 loss=0.0056
+epoch=20 step=320/355 loss=0.0100
+epoch=20 step=330/355 loss=0.0241
+epoch=20 step=340/355 loss=0.0531
+epoch=20 step=350/355 loss=0.0088
+epoch=20 train_loss=0.0396 train_contrastive=0.1127 train_regression=0.0171 val_loss=0.0679 val_contrastive=0.2494 val_regression=0.0180
+epoch=21 step=10/355 loss=0.0202
+epoch=21 step=20/355 loss=0.0186
+epoch=21 step=30/355 loss=0.0150
+epoch=21 step=40/355 loss=0.0307
+epoch=21 step=50/355 loss=0.0308
+epoch=21 step=60/355 loss=0.0127
+epoch=21 step=70/355 loss=0.0870
+epoch=21 step=80/355 loss=0.0124
+epoch=21 step=90/355 loss=0.0063
+epoch=21 step=100/355 loss=0.0290
+epoch=21 step=110/355 loss=0.0734
+epoch=21 step=120/355 loss=0.0078
+epoch=21 step=130/355 loss=0.0194
+epoch=21 step=140/355 loss=0.0309
+epoch=21 step=150/355 loss=0.0469
+epoch=21 step=160/355 loss=0.0127
+epoch=21 step=170/355 loss=0.0254
+epoch=21 step=180/355 loss=0.0470
+epoch=21 step=190/355 loss=0.0106
+epoch=21 step=200/355 loss=0.0211
+epoch=21 step=210/355 loss=0.0243
+epoch=21 step=220/355 loss=0.0068
+epoch=21 step=230/355 loss=0.0236
+epoch=21 step=240/355 loss=0.0160
+epoch=21 step=250/355 loss=0.0474
+epoch=21 step=260/355 loss=0.0100
+epoch=21 step=270/355 loss=0.0381
+epoch=21 step=280/355 loss=0.0245
+epoch=21 step=290/355 loss=0.0161
+epoch=21 step=300/355 loss=0.0068
+epoch=21 step=310/355 loss=0.0120
+epoch=21 step=320/355 loss=0.0111
+epoch=21 step=330/355 loss=0.0107
+epoch=21 step=340/355 loss=0.0246
+epoch=21 step=350/355 loss=0.0108
+epoch=21 train_loss=0.0390 train_contrastive=0.1080 train_regression=0.0174 val_loss=0.0725 val_contrastive=0.2564 val_regression=0.0212
+epoch=22 step=10/355 loss=0.0314
+epoch=22 step=20/355 loss=0.0250
+epoch=22 step=30/355 loss=0.0202
+epoch=22 step=40/355 loss=0.0175
+epoch=22 step=50/355 loss=0.0186
+epoch=22 step=60/355 loss=0.0213
+epoch=22 step=70/355 loss=0.0168
+epoch=22 step=80/355 loss=0.0231
+epoch=22 step=90/355 loss=0.0138
+epoch=22 step=100/355 loss=0.0205
+epoch=22 step=110/355 loss=0.1136
+epoch=22 step=120/355 loss=0.0654
+epoch=22 step=130/355 loss=0.0244
+epoch=22 step=140/355 loss=0.0285
+epoch=22 step=150/355 loss=0.0323
+epoch=22 step=160/355 loss=0.0166
+epoch=22 step=170/355 loss=0.0154
+epoch=22 step=180/355 loss=0.0195
+epoch=22 step=190/355 loss=0.0126
+epoch=22 step=200/355 loss=0.0294
+epoch=22 step=210/355 loss=0.0118
+epoch=22 step=220/355 loss=0.0141
+epoch=22 step=230/355 loss=0.0179
+epoch=22 step=240/355 loss=0.0084
+epoch=22 step=250/355 loss=0.0094
+epoch=22 step=260/355 loss=0.1236
+epoch=22 step=270/355 loss=0.0116
+epoch=22 step=280/355 loss=0.0168
+epoch=22 step=290/355 loss=0.0157
+epoch=22 step=300/355 loss=0.0142
+epoch=22 step=310/355 loss=0.0263
+epoch=22 step=320/355 loss=0.0189
+epoch=22 step=330/355 loss=0.0083
+epoch=22 step=340/355 loss=0.0158
+epoch=22 step=350/355 loss=0.0108
+epoch=22 train_loss=0.0391 train_contrastive=0.1088 train_regression=0.0173 val_loss=0.0724 val_contrastive=0.2589 val_regression=0.0206
+epoch=23 step=10/355 loss=0.1374
+epoch=23 step=20/355 loss=0.0380
+epoch=23 step=30/355 loss=0.0819
+epoch=23 step=40/355 loss=0.0836
+epoch=23 step=50/355 loss=0.0209
+epoch=23 step=60/355 loss=0.0660
+epoch=23 step=70/355 loss=0.0079
+epoch=23 step=80/355 loss=0.0124
+epoch=23 step=90/355 loss=0.0143
+epoch=23 step=100/355 loss=0.0367
+epoch=23 step=110/355 loss=0.0379
+epoch=23 step=120/355 loss=0.0195
+epoch=23 step=130/355 loss=0.0189
+epoch=23 step=140/355 loss=0.0167
+epoch=23 step=150/355 loss=0.0554
+epoch=23 step=160/355 loss=0.0166
+epoch=23 step=170/355 loss=0.0410
+epoch=23 step=180/355 loss=0.0088
+epoch=23 step=190/355 loss=0.0084
+epoch=23 step=200/355 loss=0.0309
+epoch=23 step=210/355 loss=0.0091
+epoch=23 step=220/355 loss=0.0215
+epoch=23 step=230/355 loss=0.0103
+epoch=23 step=240/355 loss=0.0861
+epoch=23 step=250/355 loss=0.0089
+epoch=23 step=260/355 loss=0.0082
+epoch=23 step=270/355 loss=0.0150
+epoch=23 step=280/355 loss=0.0104
+epoch=23 step=290/355 loss=0.0993
+epoch=23 step=300/355 loss=0.0375
+epoch=23 step=310/355 loss=0.0496
+epoch=23 step=320/355 loss=0.0157
+epoch=23 step=330/355 loss=0.0099
+epoch=23 step=340/355 loss=0.0077
+epoch=23 step=350/355 loss=0.0142
+epoch=23 train_loss=0.0356 train_contrastive=0.0928 train_regression=0.0170 val_loss=0.1148 val_contrastive=0.4567 val_regression=0.0234
+epoch=24 step=10/355 loss=0.0135
+epoch=24 step=20/355 loss=0.1625
+epoch=24 step=30/355 loss=0.0546
+epoch=24 step=40/355 loss=0.0102
+epoch=24 step=50/355 loss=0.1708
+epoch=24 step=60/355 loss=0.0133
+epoch=24 step=70/355 loss=0.0131
+epoch=24 step=80/355 loss=0.0124
+epoch=24 step=90/355 loss=0.0371
+epoch=24 step=100/355 loss=0.1932
+epoch=24 step=110/355 loss=0.0723
+epoch=24 step=120/355 loss=0.0216
+epoch=24 step=130/355 loss=0.0403
+epoch=24 step=140/355 loss=0.0324
+epoch=24 step=150/355 loss=0.0239
+epoch=24 step=160/355 loss=0.0904
+epoch=24 step=170/355 loss=0.0398
+epoch=24 step=180/355 loss=0.1156
+epoch=24 step=190/355 loss=0.0117
+epoch=24 step=200/355 loss=0.0700
+epoch=24 step=210/355 loss=0.0125
+epoch=24 step=220/355 loss=0.0076
+epoch=24 step=230/355 loss=0.0088
+epoch=24 step=240/355 loss=0.0092
+epoch=24 step=250/355 loss=0.0129
+epoch=24 step=260/355 loss=0.0201
+epoch=24 step=270/355 loss=0.0489
+epoch=24 step=280/355 loss=0.0186
+epoch=24 step=290/355 loss=0.0076
+epoch=24 step=300/355 loss=0.0245
+epoch=24 step=310/355 loss=0.0116
+epoch=24 step=320/355 loss=0.0312
+epoch=24 step=330/355 loss=0.0169
+epoch=24 step=340/355 loss=0.0627
+epoch=24 step=350/355 loss=0.1326
+epoch=24 train_loss=0.0385 train_contrastive=0.1062 train_regression=0.0172 val_loss=0.0743 val_contrastive=0.2622 val_regression=0.0218
+epoch=25 step=10/355 loss=0.0225
+epoch=25 step=20/355 loss=0.0200
+epoch=25 step=30/355 loss=0.0083
+epoch=25 step=40/355 loss=0.0227
+epoch=25 step=50/355 loss=0.0681
+epoch=25 step=60/355 loss=0.0187
+epoch=25 step=70/355 loss=0.0986
+epoch=25 step=80/355 loss=0.0107
+epoch=25 step=90/355 loss=0.0243
+epoch=25 step=100/355 loss=0.0736
+epoch=25 step=110/355 loss=0.0110
+epoch=25 step=120/355 loss=0.0090
+epoch=25 step=130/355 loss=0.1282
+epoch=25 step=140/355 loss=0.0085
+epoch=25 step=150/355 loss=0.0210
+epoch=25 step=160/355 loss=0.0224
+epoch=25 step=170/355 loss=0.0151
+epoch=25 step=180/355 loss=0.0105
+epoch=25 step=190/355 loss=0.0224
+epoch=25 step=200/355 loss=0.0355
+epoch=25 step=210/355 loss=0.0080
+epoch=25 step=220/355 loss=0.0427
+epoch=25 step=230/355 loss=0.0130
+epoch=25 step=240/355 loss=0.0327
+epoch=25 step=250/355 loss=0.0076
+epoch=25 step=260/355 loss=0.0210
+epoch=25 step=270/355 loss=0.0110
+epoch=25 step=280/355 loss=0.0156
+epoch=25 step=290/355 loss=0.0193
+epoch=25 step=300/355 loss=0.1079
+epoch=25 step=310/355 loss=0.0330
+epoch=25 step=320/355 loss=0.0125
+epoch=25 step=330/355 loss=0.0112
+epoch=25 step=340/355 loss=0.0194
+epoch=25 step=350/355 loss=0.0137
+epoch=25 train_loss=0.0385 train_contrastive=0.1074 train_regression=0.0170 val_loss=0.0785 val_contrastive=0.2807 val_regression=0.0224
+epoch=26 step=10/355 loss=0.2313
+epoch=26 step=20/355 loss=0.0127
+epoch=26 step=30/355 loss=0.0147
+epoch=26 step=40/355 loss=0.0990
+epoch=26 step=50/355 loss=0.0355
+epoch=26 step=60/355 loss=0.0463
+epoch=26 step=70/355 loss=0.0080
+epoch=26 step=80/355 loss=0.0106
+epoch=26 step=90/355 loss=0.0203
+epoch=26 step=100/355 loss=0.0130
+epoch=26 step=110/355 loss=0.0232
+epoch=26 step=120/355 loss=0.0267
+epoch=26 step=130/355 loss=0.1194
+epoch=26 step=140/355 loss=0.0098
+epoch=26 step=150/355 loss=0.0102
+epoch=26 step=160/355 loss=0.0179
+epoch=26 step=170/355 loss=0.0452
+epoch=26 step=180/355 loss=0.0076
+epoch=26 step=190/355 loss=0.0085
+epoch=26 step=200/355 loss=0.0099
+epoch=26 step=210/355 loss=0.0057
+epoch=26 step=220/355 loss=0.0155
+epoch=26 step=230/355 loss=0.0145
+epoch=26 step=240/355 loss=0.0349
+epoch=26 step=250/355 loss=0.0177
+epoch=26 step=260/355 loss=0.0115
+epoch=26 step=270/355 loss=0.1920
+epoch=26 step=280/355 loss=0.0430
+epoch=26 step=290/355 loss=0.0135
+epoch=26 step=300/355 loss=0.0207
+epoch=26 step=310/355 loss=0.0123
+epoch=26 step=320/355 loss=0.0397
+epoch=26 step=330/355 loss=0.0349
+epoch=26 step=340/355 loss=0.0060
+epoch=26 step=350/355 loss=0.0102
+epoch=26 train_loss=0.0383 train_contrastive=0.1025 train_regression=0.0178 val_loss=0.0808 val_contrastive=0.2957 val_regression=0.0217
+epoch=27 step=10/355 loss=0.0877
+epoch=27 step=20/355 loss=0.0187
+epoch=27 step=30/355 loss=0.0068
+epoch=27 step=40/355 loss=0.0203
+epoch=27 step=50/355 loss=0.0244
+epoch=27 step=60/355 loss=0.0094
+epoch=27 step=70/355 loss=0.0403
+epoch=27 step=80/355 loss=0.0211
+epoch=27 step=90/355 loss=0.0282
+epoch=27 step=100/355 loss=0.0287
+epoch=27 step=110/355 loss=0.0088
+epoch=27 step=120/355 loss=0.0214
+epoch=27 step=130/355 loss=0.0148
+epoch=27 step=140/355 loss=0.0213
+epoch=27 step=150/355 loss=0.0527
+epoch=27 step=160/355 loss=0.0597
+epoch=27 step=170/355 loss=0.0200
+epoch=27 step=180/355 loss=0.0316
+epoch=27 step=190/355 loss=0.0290
+epoch=27 step=200/355 loss=0.0155
+epoch=27 step=210/355 loss=0.0206
+epoch=27 step=220/355 loss=0.0261
+epoch=27 step=230/355 loss=0.0321
+epoch=27 step=240/355 loss=0.0315
+epoch=27 step=250/355 loss=0.0258
+epoch=27 step=260/355 loss=0.0288
+epoch=27 step=270/355 loss=0.0151
+epoch=27 step=280/355 loss=0.0087
+epoch=27 step=290/355 loss=0.0148
+epoch=27 step=300/355 loss=0.0142
+epoch=27 step=310/355 loss=0.0487
+epoch=27 step=320/355 loss=0.0112
+epoch=27 step=330/355 loss=0.0043
+epoch=27 step=340/355 loss=0.0109
+epoch=27 step=350/355 loss=0.0048
+epoch=27 train_loss=0.0382 train_contrastive=0.1054 train_regression=0.0171 val_loss=0.0638 val_contrastive=0.1853 val_regression=0.0267
+saved_best runs/foundation/sam_med3d_frozen_mlp_best.pt val_loss=0.0638 epoch=27
+epoch=28 step=10/355 loss=0.0209
+epoch=28 step=20/355 loss=0.0070
+epoch=28 step=30/355 loss=0.0107
+epoch=28 step=40/355 loss=0.0105
+epoch=28 step=50/355 loss=0.2132
+epoch=28 step=60/355 loss=0.0159
+epoch=28 step=70/355 loss=0.0228
+epoch=28 step=80/355 loss=0.0097
+epoch=28 step=90/355 loss=0.0457
+epoch=28 step=100/355 loss=0.0273
+epoch=28 step=110/355 loss=0.0631
+epoch=28 step=120/355 loss=0.0117
+epoch=28 step=130/355 loss=0.0162
+epoch=28 step=140/355 loss=0.0135
+epoch=28 step=150/355 loss=0.0168
+epoch=28 step=160/355 loss=0.0158
+epoch=28 step=170/355 loss=0.1547
+epoch=28 step=180/355 loss=0.0189
+epoch=28 step=190/355 loss=0.0967
+epoch=28 step=200/355 loss=0.0134
+epoch=28 step=210/355 loss=0.0220
+epoch=28 step=220/355 loss=0.0070
+epoch=28 step=230/355 loss=0.0135
+epoch=28 step=240/355 loss=0.0574
+epoch=28 step=250/355 loss=0.0057
+epoch=28 step=260/355 loss=0.0186
+epoch=28 step=270/355 loss=0.0580
+epoch=28 step=280/355 loss=0.0147
+epoch=28 step=290/355 loss=0.0162
+epoch=28 step=300/355 loss=0.0078
+epoch=28 step=310/355 loss=0.0181
+epoch=28 step=320/355 loss=0.0221
+epoch=28 step=330/355 loss=0.0138
+epoch=28 step=340/355 loss=0.0249
+epoch=28 step=350/355 loss=0.0222
+epoch=28 train_loss=0.0360 train_contrastive=0.0963 train_regression=0.0167 val_loss=0.0693 val_contrastive=0.2520 val_regression=0.0189
+epoch=29 step=10/355 loss=0.0638
+epoch=29 step=20/355 loss=0.0283
+epoch=29 step=30/355 loss=0.3332
+epoch=29 step=40/355 loss=0.0103
+epoch=29 step=50/355 loss=0.0152
+epoch=29 step=60/355 loss=0.0327
+epoch=29 step=70/355 loss=0.0788
+epoch=29 step=80/355 loss=0.0221
+epoch=29 step=90/355 loss=0.0390
+epoch=29 step=100/355 loss=0.0174
+epoch=29 step=110/355 loss=0.0068
+epoch=29 step=120/355 loss=0.0084
+epoch=29 step=130/355 loss=0.1244
+epoch=29 step=140/355 loss=0.0288
+epoch=29 step=150/355 loss=0.0274
+epoch=29 step=160/355 loss=0.0155
+epoch=29 step=170/355 loss=0.0868
+epoch=29 step=180/355 loss=0.0086
+epoch=29 step=190/355 loss=0.0260
+epoch=29 step=200/355 loss=0.0100
+epoch=29 step=210/355 loss=0.0831
+epoch=29 step=220/355 loss=0.0357
+epoch=29 step=230/355 loss=0.0117
+epoch=29 step=240/355 loss=0.0201
+epoch=29 step=250/355 loss=0.0111
+epoch=29 step=260/355 loss=0.0332
+epoch=29 step=270/355 loss=0.0531
+epoch=29 step=280/355 loss=0.0094
+epoch=29 step=290/355 loss=0.1234
+epoch=29 step=300/355 loss=0.0258
+epoch=29 step=310/355 loss=0.0142
+epoch=29 step=320/355 loss=0.1096
+epoch=29 step=330/355 loss=0.0108
+epoch=29 step=340/355 loss=0.0223
+epoch=29 step=350/355 loss=0.0127
+epoch=29 train_loss=0.0379 train_contrastive=0.1061 train_regression=0.0167 val_loss=0.0681 val_contrastive=0.2455 val_regression=0.0190
+epoch=30 step=10/355 loss=0.0108
+epoch=30 step=20/355 loss=0.0124
+epoch=30 step=30/355 loss=0.0908
+epoch=30 step=40/355 loss=0.0139
+epoch=30 step=50/355 loss=0.0153
+epoch=30 step=60/355 loss=0.0127
+epoch=30 step=70/355 loss=0.0088
+epoch=30 step=80/355 loss=0.0114
+epoch=30 step=90/355 loss=0.0083
+epoch=30 step=100/355 loss=0.0178
+epoch=30 step=110/355 loss=0.0165
+epoch=30 step=120/355 loss=0.0097
+epoch=30 step=130/355 loss=0.0142
+epoch=30 step=140/355 loss=0.1856
+epoch=30 step=150/355 loss=0.0057
+epoch=30 step=160/355 loss=0.0217
+epoch=30 step=170/355 loss=0.1389
+epoch=30 step=180/355 loss=0.0097
+epoch=30 step=190/355 loss=0.0236
+epoch=30 step=200/355 loss=0.0131
+epoch=30 step=210/355 loss=0.0413
+epoch=30 step=220/355 loss=0.0206
+epoch=30 step=230/355 loss=0.0099
+epoch=30 step=240/355 loss=0.1195
+epoch=30 step=250/355 loss=0.0058
+epoch=30 step=260/355 loss=0.0090
+epoch=30 step=270/355 loss=0.0535
+epoch=30 step=280/355 loss=0.0209
+epoch=30 step=290/355 loss=0.0141
+epoch=30 step=300/355 loss=0.0180
+epoch=30 step=310/355 loss=0.0381
+epoch=30 step=320/355 loss=0.0377
+epoch=30 step=330/355 loss=0.0247
+epoch=30 step=340/355 loss=0.0092
+epoch=30 step=350/355 loss=0.0116
+epoch=30 train_loss=0.0349 train_contrastive=0.0932 train_regression=0.0163 val_loss=0.0664 val_contrastive=0.2439 val_regression=0.0176
+saved runs/foundation/sam_med3d_frozen_mlp.pt

logs/swinunetr_frozen_clinicalbert_text_alignment.log

@@ -0,0 +1,31 @@
+<frozen importlib._bootstrap_external>:1325: FutureWarning: The cuda.cudart module is deprecated and will be removed in a future release, please switch to use the cuda.bindings.runtime module instead.
+epoch=1 train_loss=2.762019 val_loss=2.730054
+saved_best runs/vlm/swinunetr_frozen_clinicalbert_text_alignment_best.pt val_loss=2.730054
+epoch=2 train_loss=2.723667 val_loss=2.679274
+saved_best runs/vlm/swinunetr_frozen_clinicalbert_text_alignment_best.pt val_loss=2.679274
+epoch=3 train_loss=2.646813 val_loss=2.629463
+saved_best runs/vlm/swinunetr_frozen_clinicalbert_text_alignment_best.pt val_loss=2.629463
+epoch=4 train_loss=2.514582 val_loss=2.587226
+saved_best runs/vlm/swinunetr_frozen_clinicalbert_text_alignment_best.pt val_loss=2.587226
+epoch=5 train_loss=2.429484 val_loss=2.611542
+epoch=6 train_loss=2.393816 val_loss=2.477448
+saved_best runs/vlm/swinunetr_frozen_clinicalbert_text_alignment_best.pt val_loss=2.477448
+epoch=7 train_loss=2.374461 val_loss=2.493945
+epoch=8 train_loss=2.304779 val_loss=2.451789
+saved_best runs/vlm/swinunetr_frozen_clinicalbert_text_alignment_best.pt val_loss=2.451789
+epoch=9 train_loss=2.241300 val_loss=2.393683
+saved_best runs/vlm/swinunetr_frozen_clinicalbert_text_alignment_best.pt val_loss=2.393683
+epoch=10 train_loss=2.178645 val_loss=2.426386
+epoch=11 train_loss=2.141102 val_loss=2.432791
+epoch=12 train_loss=2.082751 val_loss=2.443903
+epoch=13 train_loss=2.165531 val_loss=2.626746
+epoch=14 train_loss=2.118653 val_loss=2.448756
+epoch=15 train_loss=2.045155 val_loss=2.443122
+epoch=16 train_loss=2.022221 val_loss=2.387125
+saved_best runs/vlm/swinunetr_frozen_clinicalbert_text_alignment_best.pt val_loss=2.387125
+epoch=17 train_loss=1.984116 val_loss=2.283988
+saved_best runs/vlm/swinunetr_frozen_clinicalbert_text_alignment_best.pt val_loss=2.283988
+epoch=18 train_loss=2.038574 val_loss=2.391189
+epoch=19 train_loss=2.054186 val_loss=2.394509
+epoch=20 train_loss=2.015756 val_loss=2.396744
+saved runs/vlm/swinunetr_frozen_clinicalbert_text_alignment.pt