diff --git a/.gitattributes b/.gitattributes index a4b55454e2d2ec94d71b7eaa6f025d5c8e2cdf8e..47b49c16bf4c73a7b75149a4180693f875cc3e00 100644 --- a/.gitattributes +++ b/.gitattributes @@ -48,3 +48,18 @@ Dataset544/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPla Dataset544/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/progress.png filter=lfs diff=lfs merge=lfs -text Dataset544/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/progress.png filter=lfs diff=lfs merge=lfs -text Dataset544/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/progress.png filter=lfs diff=lfs merge=lfs -text +Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/progress.png filter=lfs diff=lfs merge=lfs -text +Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/progress.png filter=lfs diff=lfs merge=lfs -text +Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/progress.png filter=lfs diff=lfs merge=lfs -text +Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/progress.png filter=lfs diff=lfs merge=lfs -text +Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/progress.png filter=lfs diff=lfs merge=lfs -text +Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/progress.png filter=lfs diff=lfs merge=lfs -text +Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/progress.png filter=lfs diff=lfs merge=lfs -text +Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/progress.png filter=lfs diff=lfs merge=lfs -text +Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/progress.png filter=lfs diff=lfs merge=lfs -text +Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/progress.png filter=lfs diff=lfs merge=lfs -text +Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/progress.png filter=lfs diff=lfs merge=lfs -text +Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/progress.png filter=lfs diff=lfs merge=lfs -text +Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/progress.png filter=lfs diff=lfs merge=lfs -text +Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/progress.png filter=lfs diff=lfs merge=lfs -text +Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/progress.png filter=lfs diff=lfs merge=lfs -text diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/dataset.json b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/dataset.json new file mode 100644 index 0000000000000000000000000000000000000000..f393c13efdc27fdedc5d8ad9dd0cff1b4394a75a --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/dataset.json @@ -0,0 +1,14 @@ +{ + "labels": { + "label_001": "1", + "background": 0 + }, + "channel_names": { + "0": "MR" + }, + "numTraining": 175, + "file_ending": ".nii.gz", + "test_set": [ + "case_0000" + ] +} \ No newline at end of file diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/dataset_fingerprint.json b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/dataset_fingerprint.json new file mode 100644 index 0000000000000000000000000000000000000000..fa251d89e2fc9373f07afa327084d7f0076a4125 --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/dataset_fingerprint.json @@ -0,0 +1,1768 @@ +{ + "foreground_intensity_properties_per_channel": { + "0": { + "max": 4142.0, + "mean": 166.43568420410156, + "median": 4.0, + "min": 0.0, + "percentile_00_5": 0.0, + "percentile_99_5": 1590.0, + "std": 346.18096923828125 + } + }, + "median_relative_size_after_cropping": 1.0, + "shapes_after_crop": [ + [ + 91, + 367, + 465 + ], + [ + 110, + 389, + 494 + ], + [ + 101, + 341, + 459 + ], + [ + 116, + 443, + 503 + ], + [ + 103, + 345, + 471 + ], + [ + 49, + 348, + 465 + ], + [ + 100, + 355, + 469 + ], + [ + 81, + 355, + 462 + ], + [ + 73, + 342, + 462 + ], + [ + 96, + 390, + 534 + ], + [ + 75, + 348, + 467 + ], + [ + 106, + 347, + 463 + ], + [ + 73, + 342, + 459 + ], + [ + 103, + 356, + 465 + ], + [ + 79, + 320, + 344 + ], + [ + 108, + 340, + 464 + ], + [ + 105, + 340, + 460 + ], + [ + 76, + 355, + 470 + ], + [ + 97, + 342, + 462 + ], + [ + 101, + 350, + 470 + ], + [ + 67, + 350, + 460 + ], + [ + 78, + 343, + 466 + ], + [ + 94, + 363, + 564 + ], + [ + 94, + 341, + 425 + ], + [ + 52, + 347, + 461 + ], + [ + 93, + 346, + 465 + ], + [ + 56, + 344, + 462 + ], + [ + 99, + 352, + 469 + ], + [ + 60, + 341, + 461 + ], + [ + 106, + 342, + 462 + ], + [ + 71, + 436, + 465 + ], + [ + 65, + 490, + 499 + ], + [ + 79, + 483, + 497 + ], + [ + 67, + 458, + 462 + ], + [ + 102, + 496, + 511 + ], + [ + 101, + 484, + 485 + ], + [ + 95, + 468, + 478 + ], + [ + 64, + 444, + 482 + ], + [ + 109, + 463, + 473 + ], + [ + 90, + 483, + 487 + ], + [ + 96, + 454, + 487 + ], + [ + 98, + 448, + 492 + ], + [ + 107, + 482, + 484 + ], + [ + 104, + 465, + 480 + ], + [ + 97, + 465, + 481 + ], + [ + 99, + 336, + 481 + ], + [ + 104, + 441, + 481 + ], + [ + 88, + 422, + 484 + ], + [ + 105, + 437, + 485 + ], + [ + 84, + 426, + 483 + ], + [ + 85, + 393, + 458 + ], + [ + 75, + 431, + 486 + ], + [ + 73, + 427, + 488 + ], + [ + 79, + 352, + 379 + ], + [ + 85, + 407, + 496 + ], + [ + 89, + 418, + 469 + ], + [ + 85, + 366, + 488 + ], + [ + 104, + 345, + 446 + ], + [ + 69, + 465, + 483 + ], + [ + 87, + 441, + 488 + ], + [ + 79, + 443, + 477 + ], + [ + 87, + 369, + 484 + ], + [ + 83, + 380, + 481 + ], + [ + 54, + 408, + 494 + ], + [ + 90, + 390, + 497 + ], + [ + 140, + 439, + 447 + ], + [ + 138, + 439, + 440 + ], + [ + 94, + 491, + 494 + ], + [ + 164, + 493, + 498 + ], + [ + 108, + 439, + 448 + ], + [ + 116, + 455, + 449 + ], + [ + 115, + 440, + 437 + ], + [ + 94, + 493, + 495 + ], + [ + 105, + 455, + 447 + ], + [ + 92, + 490, + 486 + ], + [ + 109, + 444, + 454 + ], + [ + 97, + 504, + 500 + ], + [ + 110, + 500, + 540 + ], + [ + 116, + 449, + 445 + ], + [ + 114, + 390, + 387 + ], + [ + 157, + 453, + 446 + ], + [ + 121, + 444, + 441 + ], + [ + 101, + 507, + 540 + ], + [ + 106, + 450, + 448 + ], + [ + 120, + 447, + 449 + ], + [ + 134, + 527, + 587 + ], + [ + 146, + 444, + 443 + ], + [ + 90, + 441, + 446 + ], + [ + 93, + 489, + 489 + ], + [ + 120, + 446, + 446 + ], + [ + 133, + 505, + 549 + ], + [ + 148, + 448, + 453 + ], + [ + 151, + 463, + 464 + ], + [ + 109, + 440, + 441 + ], + [ + 81, + 446, + 445 + ], + [ + 103, + 442, + 452 + ], + [ + 110, + 438, + 437 + ], + [ + 126, + 526, + 608 + ], + [ + 123, + 460, + 452 + ], + [ + 90, + 450, + 461 + ], + [ + 126, + 410, + 404 + ], + [ + 100, + 465, + 460 + ], + [ + 118, + 447, + 444 + ], + [ + 144, + 439, + 443 + ], + [ + 100, + 439, + 442 + ], + [ + 91, + 439, + 448 + ], + [ + 81, + 439, + 440 + ], + [ + 141, + 447, + 438 + ], + [ + 92, + 490, + 489 + ], + [ + 116, + 450, + 459 + ], + [ + 125, + 520, + 584 + ], + [ + 92, + 493, + 491 + ], + [ + 151, + 495, + 556 + ], + [ + 129, + 445, + 446 + ], + [ + 100, + 443, + 441 + ], + [ + 134, + 438, + 446 + ], + [ + 156, + 421, + 440 + ], + [ + 118, + 449, + 439 + ], + [ + 103, + 449, + 439 + ], + [ + 93, + 442, + 442 + ], + [ + 92, + 480, + 488 + ], + [ + 120, + 442, + 440 + ], + [ + 126, + 362, + 583 + ], + [ + 138, + 402, + 398 + ], + [ + 127, + 447, + 445 + ], + [ + 120, + 389, + 393 + ], + [ + 156, + 399, + 413 + ], + [ + 83, + 452, + 444 + ], + [ + 100, + 438, + 444 + ], + [ + 103, + 471, + 499 + ], + [ + 84, + 490, + 486 + ], + [ + 140, + 448, + 446 + ], + [ + 92, + 489, + 487 + ], + [ + 91, + 489, + 486 + ], + [ + 138, + 493, + 555 + ], + [ + 150, + 452, + 449 + ], + [ + 108, + 459, + 449 + ], + [ + 94, + 493, + 493 + ], + [ + 86, + 504, + 551 + ], + [ + 95, + 497, + 493 + ], + [ + 72, + 486, + 490 + ], + [ + 127, + 440, + 441 + ], + [ + 123, + 510, + 586 + ], + [ + 152, + 507, + 568 + ], + [ + 116, + 512, + 584 + ], + [ + 140, + 451, + 453 + ], + [ + 128, + 438, + 438 + ], + [ + 100, + 392, + 400 + ], + [ + 164, + 430, + 438 + ], + [ + 154, + 395, + 399 + ], + [ + 106, + 389, + 392 + ], + [ + 122, + 437, + 437 + ], + [ + 163, + 407, + 420 + ], + [ + 98, + 391, + 399 + ], + [ + 92, + 489, + 489 + ], + [ + 78, + 492, + 491 + ], + [ + 92, + 449, + 495 + ], + [ + 82, + 441, + 495 + ], + [ + 73, + 443, + 505 + ], + [ + 90, + 448, + 505 + ], + [ + 75, + 443, + 474 + ], + [ + 91, + 447, + 492 + ], + [ + 84, + 438, + 489 + ], + [ + 88, + 444, + 491 + ], + [ + 91, + 447, + 495 + ], + [ + 88, + 446, + 494 + ], + [ + 79, + 447, + 501 + ], + [ + 61, + 431, + 477 + ], + [ + 90, + 434, + 441 + ], + [ + 49, + 444, + 443 + ], + [ + 109, + 362, + 439 + ], + [ + 81, + 339, + 514 + ], + [ + 66, + 393, + 436 + ], + [ + 108, + 404, + 536 + ], + [ + 96, + 350, + 426 + ] + ], + "spacings": [ + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ] + ] +} \ No newline at end of file diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/checkpoint_final.pth b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/checkpoint_final.pth new file mode 100644 index 0000000000000000000000000000000000000000..851e50d685c2ed19345f124d6f5de60cf6e4f10e --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/checkpoint_final.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:afc313304f320f9802588aec309eeb15257e4099be8399cb97809756004469e6 +size 815342317 diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/debug.json b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/debug.json new file mode 100644 index 0000000000000000000000000000000000000000..c24b67031a1441a7241273cbf78bd8191af1cc36 --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/debug.json @@ -0,0 +1,57 @@ +{ + "_best_ema": "None", + "aim_run": ">", + "batch_size": "2", + "configuration_manager": "{'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [99.0, 442.0, 465.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}, 'deep_supervision': False}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}", + "configuration_name": "3d_fullres", + "cudnn_version": 90501, + "current_epoch": "0", + "dataloader_train": "", + "dataloader_train.generator": "", + "dataloader_train.num_processes": "12", + "dataloader_train.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataloader_val": "", + "dataloader_val.generator": "", + "dataloader_val.num_processes": "6", + "dataloader_val.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataset_json": "{'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 175, 'file_ending': '.mha'}", + "debug_custom": "{'run_hash': '26699708800b4c02a09dcce7'}", + "decoder_type": "standard", + "device": "cuda:0", + "disable_checkpointing": "False", + "enable_deep_supervision": "False", + "fold": "0", + "folder_with_segs_from_previous_stage": "None", + "gpu_name": "NVIDIA H100", + "grad_scaler": "", + "hostname": "g038", + "inference_allowed_mirroring_axes": "None", + "initial_lr": "0.01", + "is_cascaded": "False", + "is_ddp": "False", + "label_manager": "", + "local_rank": "0", + "log_file": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/training_log_2025_8_11_11_13_37.txt", + "logger": "", + "loss": "SynPerceptionLoss(\n (seg_model): ResidualEncoderUNet(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (decoder): UNetDecoder(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (stages): ModuleList(\n (0): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (1): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (2): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (3): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (4): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (transpconvs): ModuleList(\n (0): ConvTranspose3d(320, 320, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n (1): ConvTranspose3d(320, 256, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (2): ConvTranspose3d(256, 128, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (3): ConvTranspose3d(128, 64, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (4): ConvTranspose3d(64, 32, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n )\n (seg_layers): ModuleList(\n (0): Conv3d(320, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (1): Conv3d(256, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (2): Conv3d(128, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (3): Conv3d(64, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (4): Conv3d(32, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n )\n )\n )\n (L1): L1Loss()\n (image_loss): myMaskedMSE(\n (mse): myMSE()\n )\n)", + "lr_scheduler": "", + "my_init_kwargs": "{'plans': {'dataset_name': 'Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [99, 442, 465], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [442.0, 465.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [99, 292, 307], 'spacing': [3.0, 1.512589724855112, 1.512589724855112], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [99.0, 442.0, 465.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 166.43568420410156, 'median': 4.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1590.0, 'std': 346.18096923828125}}}, 'configuration': '3d_fullres', 'fold': 0, 'dataset_json': {'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 175, 'file_ending': '.mha'}, 'unpack_dataset': True, 'device': device(type='cuda')}", + "network": "ResidualEncoderUNet", + "num_epochs": "1000", + "num_input_channels": "1", + "num_iterations_per_epoch": "250", + "num_val_iterations_per_epoch": "50", + "optimizer": "SGD (\nParameter Group 0\n dampening: 0\n differentiable: False\n foreach: None\n fused: None\n initial_lr: 0.01\n lr: 0.01\n maximize: False\n momentum: 0.99\n nesterov: True\n weight_decay: 3e-05\n)", + "output_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0", + "output_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres", + "oversample_foreground_percent": "0.33", + "perception_masked": "True", + "plans_manager": "{'dataset_name': 'Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [99, 442, 465], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [442.0, 465.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [99, 292, 307], 'spacing': [3.0, 1.512589724855112, 1.512589724855112], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [99.0, 442.0, 465.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 166.43568420410156, 'median': 4.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1590.0, 'std': 346.18096923828125}}}", + "preprocessed_dataset_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetPlans_3d_fullres", + "preprocessed_dataset_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked", + "save_every": "50", + "torch_version": "2.7.1+cu126", + "unpack_dataset": "True", + "was_initialized": "True", + "weight_decay": "3e-05" +} \ No newline at end of file diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/progress.png b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/progress.png new file mode 100644 index 0000000000000000000000000000000000000000..d0c585c441880a3d29c33b7f7cf51f0d2eadac5e --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/progress.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1d6197f5447a11c9a9cb35acae93cf9689e2b6cacae7ce5b2eb40ace14ef54e2 +size 573522 diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/validation/summary.json b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/validation/summary.json new file mode 100644 index 0000000000000000000000000000000000000000..594f016e406dd4ceb3a5ad5e63c8c66a61fd2d92 --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/validation/summary.json @@ -0,0 +1,52 @@ +{ + "mae": { + "mean": 76.8092616512012, + "std": 18.625195811936642, + "max": 140.77994912518722, + "min": 41.62071026364149, + "25pc": 64.72538071310042, + "50pc": 73.85862496516182, + "75pc": 85.8839198540596, + "count": 35 + }, + "psnr": { + "mean": 27.311648936398207, + "std": 1.8531108596131218, + "max": 31.268102989356766, + "min": 22.552571644935476, + "25pc": 26.215671375693184, + "50pc": 27.182330724105693, + "75pc": 28.4978144951083, + "count": 35 + }, + "ms_ssim": { + "mean": 0.8395574636043638, + "std": 0.0858952793371136, + "max": 0.9528723843099408, + "min": 0.5451397112930043, + "25pc": 0.8207952198391668, + "50pc": 0.8547906024860593, + "75pc": 0.8931221833272822, + "count": 35 + }, + "DICE": { + "mean": 0.6560486103807177, + "std": 0.10584852031326245, + "max": 0.8644771575927734, + "min": 0.3993914723396301, + "25pc": 0.5967317223548889, + "50pc": 0.6867832541465759, + "75pc": 0.72818523645401, + "count": 35 + }, + "HD95": { + "mean": 7.068772772380284, + "std": 3.426572914801885, + "max": 14.236841201782227, + "min": 2.4678704738616943, + "25pc": 4.391140460968018, + "50pc": 5.837563991546631, + "75pc": 9.409066677093506, + "count": 35 + } +} \ No newline at end of file diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/checkpoint_final.pth b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/checkpoint_final.pth new file mode 100644 index 0000000000000000000000000000000000000000..7a363f88e853fd66ee204bacc384a4b61d80ffe6 --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/checkpoint_final.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9dbf0c5ea4505fb68d82b28218dc5a55be065c910fc11ee26eab07c95f7e6357 +size 815581869 diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/debug.json b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/debug.json new file mode 100644 index 0000000000000000000000000000000000000000..e6e1415d1dd4cc2355fb338770ff65f58aa637fd --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/debug.json @@ -0,0 +1,58 @@ +{ + "_best_ema": "None", + "aim_run": ">", + "batch_size": "2", + "configuration_manager": "{'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [99.0, 442.0, 465.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}, 'deep_supervision': False}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}", + "configuration_name": "3d_fullres", + "cudnn_version": 90501, + "current_epoch": "0", + "dataloader_train": "", + "dataloader_train.generator": "", + "dataloader_train.num_processes": "12", + "dataloader_train.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataloader_val": "", + "dataloader_val.generator": "", + "dataloader_val.num_processes": "6", + "dataloader_val.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataset_json": "{'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 175, 'file_ending': '.mha'}", + "debug_custom": "{'run_hash': '3dd09716401f402b986ac624'}", + "decoder_type": "standard", + "device": "cuda:0", + "disable_checkpointing": "False", + "enable_deep_supervision": "False", + "fold": "1", + "folder_with_segs_from_previous_stage": "None", + "gpu_name": "NVIDIA H100", + "grad_scaler": "", + "hostname": "g018", + "image_loss_weight": "0.5", + "inference_allowed_mirroring_axes": "None", + "initial_lr": "0.01", + "is_cascaded": "False", + "is_ddp": "False", + "label_manager": "", + "local_rank": "0", + "log_file": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/training_log_2025_8_13_01_35_04.txt", + "logger": "", + "loss": "SynPerceptionLoss(\n (seg_model): ResidualEncoderUNet(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (decoder): UNetDecoder(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (stages): ModuleList(\n (0): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (1): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (2): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (3): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (4): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (transpconvs): ModuleList(\n (0): ConvTranspose3d(320, 320, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n (1): ConvTranspose3d(320, 256, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (2): ConvTranspose3d(256, 128, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (3): ConvTranspose3d(128, 64, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (4): ConvTranspose3d(64, 32, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n )\n (seg_layers): ModuleList(\n (0): Conv3d(320, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (1): Conv3d(256, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (2): Conv3d(128, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (3): Conv3d(64, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (4): Conv3d(32, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n )\n )\n )\n (L1): L1Loss()\n (image_loss): myMaskedMSE(\n (mse): myMSE()\n )\n)", + "lr_scheduler": "", + "my_init_kwargs": "{'plans': {'dataset_name': 'Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [99, 442, 465], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [442.0, 465.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [99, 292, 307], 'spacing': [3.0, 1.512589724855112, 1.512589724855112], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [99.0, 442.0, 465.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 166.43568420410156, 'median': 4.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1590.0, 'std': 346.18096923828125}}}, 'configuration': '3d_fullres', 'fold': 1, 'dataset_json': {'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 175, 'file_ending': '.mha'}, 'unpack_dataset': True, 'device': device(type='cuda')}", + "network": "ResidualEncoderUNet", + "num_epochs": "1000", + "num_input_channels": "1", + "num_iterations_per_epoch": "250", + "num_val_iterations_per_epoch": "50", + "optimizer": "SGD (\nParameter Group 0\n dampening: 0\n differentiable: False\n foreach: None\n fused: None\n initial_lr: 0.01\n lr: 0.01\n maximize: False\n momentum: 0.99\n nesterov: True\n weight_decay: 3e-05\n)", + "output_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1", + "output_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres", + "oversample_foreground_percent": "0.33", + "perception_masked": "True", + "plans_manager": "{'dataset_name': 'Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [99, 442, 465], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [442.0, 465.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [99, 292, 307], 'spacing': [3.0, 1.512589724855112, 1.512589724855112], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [99.0, 442.0, 465.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 166.43568420410156, 'median': 4.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1590.0, 'std': 346.18096923828125}}}", + "preprocessed_dataset_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetPlans_3d_fullres", + "preprocessed_dataset_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked", + "save_every": "50", + "torch_version": "2.7.1+cu126", + "unpack_dataset": "True", + "was_initialized": "True", + "weight_decay": "3e-05" +} \ No newline at end of file diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/progress.png b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/progress.png new file mode 100644 index 0000000000000000000000000000000000000000..47a5e78bfd521e54a15c66ebb6f3b11fce7956af --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/progress.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4431ccbb4775e96c4b70dd5ecc3b449e325d29d57298fc3083a34d3edb85a2c7 +size 494924 diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/validation/summary.json b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/validation/summary.json new file mode 100644 index 0000000000000000000000000000000000000000..0842b8e4ed05142c298c45920334a2731b4f49fc --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/validation/summary.json @@ -0,0 +1,52 @@ +{ + "mae": { + "mean": 70.60732618860474, + "std": 19.041055775358945, + "max": 119.73782460503409, + "min": 35.66236947824612, + "25pc": 57.76567743163339, + "50pc": 68.28826150425549, + "75pc": 81.36687088479742, + "count": 35 + }, + "psnr": { + "mean": 28.059140951732164, + "std": 2.118147886039748, + "max": 32.37409849034857, + "min": 23.868303967056576, + "25pc": 26.495957087833208, + "50pc": 28.056564963775173, + "75pc": 29.147963445553728, + "count": 35 + }, + "ms_ssim": { + "mean": 0.8616194353348334, + "std": 0.08357887966105612, + "max": 0.955569077318667, + "min": 0.5221353830368347, + "25pc": 0.8244010109953923, + "50pc": 0.8854787401313363, + "75pc": 0.9107694131192781, + "count": 35 + }, + "DICE": { + "mean": 0.6700404260839735, + "std": 0.11891196030988815, + "max": 0.8276325464248657, + "min": 0.2527022659778595, + "25pc": 0.6079962253570557, + "50pc": 0.6913292407989502, + "75pc": 0.7501238882541656, + "count": 35 + }, + "HD95": { + "mean": 6.818869611195155, + "std": 3.607450019780616, + "max": 19.41144561767578, + "min": 2.4186062812805176, + "25pc": 4.318592309951782, + "50pc": 6.376376628875732, + "75pc": 7.8472580909729, + "count": 35 + } +} \ No newline at end of file diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/checkpoint_final.pth b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/checkpoint_final.pth new file mode 100644 index 0000000000000000000000000000000000000000..c0ebfc1c7034e9880ac299dfabf6c326b43cb98a --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/checkpoint_final.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ddbbcfe44a62d106d0243a899198adb832469613c676788d033ea0946338631 +size 815581549 diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/debug.json b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/debug.json new file mode 100644 index 0000000000000000000000000000000000000000..2a62db09d1239c6902462cdb8eec0dd7f69c0e8c --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/debug.json @@ -0,0 +1,58 @@ +{ + "_best_ema": "None", + "aim_run": ">", + "batch_size": "2", + "configuration_manager": "{'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [99.0, 442.0, 465.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}, 'deep_supervision': False}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}", + "configuration_name": "3d_fullres", + "cudnn_version": 90501, + "current_epoch": "0", + "dataloader_train": "", + "dataloader_train.generator": "", + "dataloader_train.num_processes": "12", + "dataloader_train.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataloader_val": "", + "dataloader_val.generator": "", + "dataloader_val.num_processes": "6", + "dataloader_val.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataset_json": "{'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 175, 'file_ending': '.mha'}", + "debug_custom": "{'run_hash': 'df86de6bad9d4af88cd79bf4'}", + "decoder_type": "standard", + "device": "cuda:0", + "disable_checkpointing": "False", + "enable_deep_supervision": "False", + "fold": "2", + "folder_with_segs_from_previous_stage": "None", + "gpu_name": "NVIDIA H100", + "grad_scaler": "", + "hostname": "g100", + "image_loss_weight": "0.5", + "inference_allowed_mirroring_axes": "None", + "initial_lr": "0.01", + "is_cascaded": "False", + "is_ddp": "False", + "label_manager": "", + "local_rank": "0", + "log_file": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/training_log_2025_8_13_01_35_04.txt", + "logger": "", + "loss": "SynPerceptionLoss(\n (seg_model): ResidualEncoderUNet(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (decoder): UNetDecoder(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (stages): ModuleList(\n (0): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (1): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (2): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (3): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (4): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (transpconvs): ModuleList(\n (0): ConvTranspose3d(320, 320, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n (1): ConvTranspose3d(320, 256, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (2): ConvTranspose3d(256, 128, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (3): ConvTranspose3d(128, 64, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (4): ConvTranspose3d(64, 32, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n )\n (seg_layers): ModuleList(\n (0): Conv3d(320, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (1): Conv3d(256, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (2): Conv3d(128, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (3): Conv3d(64, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (4): Conv3d(32, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n )\n )\n )\n (L1): L1Loss()\n (image_loss): myMaskedMSE(\n (mse): myMSE()\n )\n)", + "lr_scheduler": "", + "my_init_kwargs": "{'plans': {'dataset_name': 'Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [99, 442, 465], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [442.0, 465.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [99, 292, 307], 'spacing': [3.0, 1.512589724855112, 1.512589724855112], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [99.0, 442.0, 465.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 166.43568420410156, 'median': 4.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1590.0, 'std': 346.18096923828125}}}, 'configuration': '3d_fullres', 'fold': 2, 'dataset_json': {'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 175, 'file_ending': '.mha'}, 'unpack_dataset': True, 'device': device(type='cuda')}", + "network": "ResidualEncoderUNet", + "num_epochs": "1000", + "num_input_channels": "1", + "num_iterations_per_epoch": "250", + "num_val_iterations_per_epoch": "50", + "optimizer": "SGD (\nParameter Group 0\n dampening: 0\n differentiable: False\n foreach: None\n fused: None\n initial_lr: 0.01\n lr: 0.01\n maximize: False\n momentum: 0.99\n nesterov: True\n weight_decay: 3e-05\n)", + "output_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2", + "output_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres", + "oversample_foreground_percent": "0.33", + "perception_masked": "True", + "plans_manager": "{'dataset_name': 'Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [99, 442, 465], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [442.0, 465.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [99, 292, 307], 'spacing': [3.0, 1.512589724855112, 1.512589724855112], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [99.0, 442.0, 465.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 166.43568420410156, 'median': 4.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1590.0, 'std': 346.18096923828125}}}", + "preprocessed_dataset_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetPlans_3d_fullres", + "preprocessed_dataset_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked", + "save_every": "50", + "torch_version": "2.7.1+cu126", + "unpack_dataset": "True", + "was_initialized": "True", + "weight_decay": "3e-05" +} \ No newline at end of file diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/progress.png b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/progress.png new file mode 100644 index 0000000000000000000000000000000000000000..b3f8ec70170edf03bc326b1725a460099fb58a37 --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/progress.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4bb5e2f9910d354333c544dd25b038a9c6d58f2ad2ace3e32d972e9b95936c57 +size 530372 diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/validation/summary.json b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/validation/summary.json new file mode 100644 index 0000000000000000000000000000000000000000..724e73f3eb8e90850640484f70eae9ab703f60bc --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/validation/summary.json @@ -0,0 +1,52 @@ +{ + "mae": { + "mean": 84.74803688363573, + "std": 40.047402284729955, + "max": 263.04195372841815, + "min": 40.32802880932887, + "25pc": 63.28146353775634, + "50pc": 72.06488595910446, + "75pc": 89.78496781277786, + "count": 35 + }, + "psnr": { + "mean": 26.885433281824522, + "std": 2.6544000856675587, + "max": 31.854206682238477, + "min": 19.393644306850135, + "25pc": 25.196885045914968, + "50pc": 27.44324508788327, + "75pc": 28.4630928707804, + "count": 35 + }, + "ms_ssim": { + "mean": 0.8115448857014917, + "std": 0.12441964379373294, + "max": 0.9412200398725382, + "min": 0.42626283493743744, + "25pc": 0.7605181879936237, + "50pc": 0.8653771175752596, + "75pc": 0.8976873502779543, + "count": 35 + }, + "DICE": { + "mean": 0.6612528319869723, + "std": 0.12250841573151652, + "max": 0.808124840259552, + "min": 0.14218591153621674, + "25pc": 0.6211503744125366, + "50pc": 0.68855881690979, + "75pc": 0.740378350019455, + "count": 35 + }, + "HD95": { + "mean": 7.289406333650861, + "std": 5.135283095630513, + "max": 25.59170150756836, + "min": 2.4629297256469727, + "25pc": 4.024229168891907, + "50pc": 5.817408561706543, + "75pc": 8.499962329864502, + "count": 35 + } +} \ No newline at end of file diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/checkpoint_final.pth b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/checkpoint_final.pth new file mode 100644 index 0000000000000000000000000000000000000000..f76c7feafb508dc2fc4b3bea50d4a110662389b1 --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/checkpoint_final.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8231c68e5d0561b2676aa338d8e6fbd9bcf4df58e96255999ca28a5c0338136b +size 815581869 diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/debug.json b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/debug.json new file mode 100644 index 0000000000000000000000000000000000000000..f265d6cb88489b9924ce1d754685e14c0f6da88c --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/debug.json @@ -0,0 +1,58 @@ +{ + "_best_ema": "None", + "aim_run": ">", + "batch_size": "2", + "configuration_manager": "{'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [99.0, 442.0, 465.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}, 'deep_supervision': False}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}", + "configuration_name": "3d_fullres", + "cudnn_version": 90501, + "current_epoch": "0", + "dataloader_train": "", + "dataloader_train.generator": "", + "dataloader_train.num_processes": "12", + "dataloader_train.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataloader_val": "", + "dataloader_val.generator": "", + "dataloader_val.num_processes": "6", + "dataloader_val.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataset_json": "{'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 175, 'file_ending': '.mha'}", + "debug_custom": "{'run_hash': '79dddf6a6f8f4baa8254425b'}", + "decoder_type": "standard", + "device": "cuda:0", + "disable_checkpointing": "False", + "enable_deep_supervision": "False", + "fold": "3", + "folder_with_segs_from_previous_stage": "None", + "gpu_name": "NVIDIA H100", + "grad_scaler": "", + "hostname": "g030", + "image_loss_weight": "0.5", + "inference_allowed_mirroring_axes": "None", + "initial_lr": "0.01", + "is_cascaded": "False", + "is_ddp": "False", + "label_manager": "", + "local_rank": "0", + "log_file": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/training_log_2025_8_13_01_35_04.txt", + "logger": "", + "loss": "SynPerceptionLoss(\n (seg_model): ResidualEncoderUNet(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (decoder): UNetDecoder(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (stages): ModuleList(\n (0): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (1): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (2): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (3): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (4): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (transpconvs): ModuleList(\n (0): ConvTranspose3d(320, 320, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n (1): ConvTranspose3d(320, 256, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (2): ConvTranspose3d(256, 128, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (3): ConvTranspose3d(128, 64, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (4): ConvTranspose3d(64, 32, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n )\n (seg_layers): ModuleList(\n (0): Conv3d(320, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (1): Conv3d(256, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (2): Conv3d(128, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (3): Conv3d(64, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (4): Conv3d(32, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n )\n )\n )\n (L1): L1Loss()\n (image_loss): myMaskedMSE(\n (mse): myMSE()\n )\n)", + "lr_scheduler": "", + "my_init_kwargs": "{'plans': {'dataset_name': 'Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [99, 442, 465], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [442.0, 465.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [99, 292, 307], 'spacing': [3.0, 1.512589724855112, 1.512589724855112], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [99.0, 442.0, 465.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 166.43568420410156, 'median': 4.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1590.0, 'std': 346.18096923828125}}}, 'configuration': '3d_fullres', 'fold': 3, 'dataset_json': {'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 175, 'file_ending': '.mha'}, 'unpack_dataset': True, 'device': device(type='cuda')}", + "network": "ResidualEncoderUNet", + "num_epochs": "1000", + "num_input_channels": "1", + "num_iterations_per_epoch": "250", + "num_val_iterations_per_epoch": "50", + "optimizer": "SGD (\nParameter Group 0\n dampening: 0\n differentiable: False\n foreach: None\n fused: None\n initial_lr: 0.01\n lr: 0.01\n maximize: False\n momentum: 0.99\n nesterov: True\n weight_decay: 3e-05\n)", + "output_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3", + "output_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres", + "oversample_foreground_percent": "0.33", + "perception_masked": "True", + "plans_manager": "{'dataset_name': 'Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [99, 442, 465], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [442.0, 465.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [99, 292, 307], 'spacing': [3.0, 1.512589724855112, 1.512589724855112], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [99.0, 442.0, 465.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 166.43568420410156, 'median': 4.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1590.0, 'std': 346.18096923828125}}}", + "preprocessed_dataset_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetPlans_3d_fullres", + "preprocessed_dataset_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked", + "save_every": "50", + "torch_version": "2.7.1+cu126", + "unpack_dataset": "True", + "was_initialized": "True", + "weight_decay": "3e-05" +} \ No newline at end of file diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/progress.png b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/progress.png new file mode 100644 index 0000000000000000000000000000000000000000..c4e8b7492158a082c0403121b6d7f16862cfac6f --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/progress.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:300bb0c07313d0920e9eb4f1229eebfa66ac075e8277fa193383cd682144fc0d +size 527759 diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/validation/summary.json b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/validation/summary.json new file mode 100644 index 0000000000000000000000000000000000000000..14a2942bb3e2910579672e315db0c729f989f578 --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/validation/summary.json @@ -0,0 +1,52 @@ +{ + "mae": { + "mean": 73.12688302867724, + "std": 19.95995885348866, + "max": 132.248609145247, + "min": 46.339182105789455, + "25pc": 56.624091108605086, + "50pc": 70.36532161707972, + "75pc": 83.9632419496316, + "count": 35 + }, + "psnr": { + "mean": 27.862940087248717, + "std": 2.4213046826808795, + "max": 31.58417669988859, + "min": 20.920674517301556, + "25pc": 26.07172573532648, + "50pc": 27.621538376627477, + "75pc": 29.706829472334555, + "count": 35 + }, + "ms_ssim": { + "mean": 0.8685602022186172, + "std": 0.0669377346891808, + "max": 0.9643831631206568, + "min": 0.7071124984778592, + "25pc": 0.8319130424166855, + "50pc": 0.8801320031447727, + "75pc": 0.9225499086164202, + "count": 35 + }, + "DICE": { + "mean": 0.6758987750325884, + "std": 0.0783866493999751, + "max": 0.7813335657119751, + "min": 0.4357028007507324, + "25pc": 0.6144459247589111, + "50pc": 0.6911219358444214, + "75pc": 0.7368215620517731, + "count": 35 + }, + "HD95": { + "mean": 6.489197506223406, + "std": 3.3116679957565873, + "max": 15.67831039428711, + "min": 2.5454208850860596, + "25pc": 4.271042585372925, + "50pc": 5.40036678314209, + "75pc": 8.04506540298462, + "count": 35 + } +} \ No newline at end of file diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/checkpoint_final.pth b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/checkpoint_final.pth new file mode 100644 index 0000000000000000000000000000000000000000..634582496060a20e55d266ca368baf9ac77db870 --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/checkpoint_final.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:618a99dd20532d8eda6510194973c0d48c4914122d8f2a32aae4d177b9a7e586 +size 815581869 diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/debug.json b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/debug.json new file mode 100644 index 0000000000000000000000000000000000000000..72d2abcf7dce070bdf4f2785a68d23c93a0b73ba --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/debug.json @@ -0,0 +1,58 @@ +{ + "_best_ema": "None", + "aim_run": ">", + "batch_size": "2", + "configuration_manager": "{'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [99.0, 442.0, 465.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}, 'deep_supervision': False}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}", + "configuration_name": "3d_fullres", + "cudnn_version": 90501, + "current_epoch": "0", + "dataloader_train": "", + "dataloader_train.generator": "", + "dataloader_train.num_processes": "12", + "dataloader_train.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataloader_val": "", + "dataloader_val.generator": "", + "dataloader_val.num_processes": "6", + "dataloader_val.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataset_json": "{'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 175, 'file_ending': '.mha'}", + "debug_custom": "{'run_hash': '4747960d6f034e4f83f7649d'}", + "decoder_type": "standard", + "device": "cuda:0", + "disable_checkpointing": "False", + "enable_deep_supervision": "False", + "fold": "4", + "folder_with_segs_from_previous_stage": "None", + "gpu_name": "NVIDIA H100", + "grad_scaler": "", + "hostname": "g031", + "image_loss_weight": "0.5", + "inference_allowed_mirroring_axes": "None", + "initial_lr": "0.01", + "is_cascaded": "False", + "is_ddp": "False", + "label_manager": "", + "local_rank": "0", + "log_file": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/training_log_2025_8_13_01_35_07.txt", + "logger": "", + "loss": "SynPerceptionLoss(\n (seg_model): ResidualEncoderUNet(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (decoder): UNetDecoder(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (stages): ModuleList(\n (0): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (1): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (2): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (3): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (4): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (transpconvs): ModuleList(\n (0): ConvTranspose3d(320, 320, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n (1): ConvTranspose3d(320, 256, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (2): ConvTranspose3d(256, 128, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (3): ConvTranspose3d(128, 64, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (4): ConvTranspose3d(64, 32, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n )\n (seg_layers): ModuleList(\n (0): Conv3d(320, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (1): Conv3d(256, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (2): Conv3d(128, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (3): Conv3d(64, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (4): Conv3d(32, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n )\n )\n )\n (L1): L1Loss()\n (image_loss): myMaskedMSE(\n (mse): myMSE()\n )\n)", + "lr_scheduler": "", + "my_init_kwargs": "{'plans': {'dataset_name': 'Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [99, 442, 465], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [442.0, 465.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [99, 292, 307], 'spacing': [3.0, 1.512589724855112, 1.512589724855112], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [99.0, 442.0, 465.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 166.43568420410156, 'median': 4.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1590.0, 'std': 346.18096923828125}}}, 'configuration': '3d_fullres', 'fold': 4, 'dataset_json': {'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 175, 'file_ending': '.mha'}, 'unpack_dataset': True, 'device': device(type='cuda')}", + "network": "ResidualEncoderUNet", + "num_epochs": "1000", + "num_input_channels": "1", + "num_iterations_per_epoch": "250", + "num_val_iterations_per_epoch": "50", + "optimizer": "SGD (\nParameter Group 0\n dampening: 0\n differentiable: False\n foreach: None\n fused: None\n initial_lr: 0.01\n lr: 0.01\n maximize: False\n momentum: 0.99\n nesterov: True\n weight_decay: 3e-05\n)", + "output_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4", + "output_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres", + "oversample_foreground_percent": "0.33", + "perception_masked": "True", + "plans_manager": "{'dataset_name': 'Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [99, 442, 465], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [442.0, 465.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [99, 292, 307], 'spacing': [3.0, 1.512589724855112, 1.512589724855112], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [99.0, 442.0, 465.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 166.43568420410156, 'median': 4.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1590.0, 'std': 346.18096923828125}}}", + "preprocessed_dataset_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked/nnUNetPlans_3d_fullres", + "preprocessed_dataset_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked", + "save_every": "50", + "torch_version": "2.7.1+cu126", + "unpack_dataset": "True", + "was_initialized": "True", + "weight_decay": "3e-05" +} \ No newline at end of file diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/progress.png b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/progress.png new file mode 100644 index 0000000000000000000000000000000000000000..99f105466ba5b3a564eb29d2830b888628eb26df --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/progress.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:65c3c0810363b741a4823b9ea4a5d4832b5dd4740aab80218a3c38a1f18b9f1e +size 511035 diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/validation/summary.json b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/validation/summary.json new file mode 100644 index 0000000000000000000000000000000000000000..8864523f88c3870d3edc5eb12c233751c2a6c85a --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/validation/summary.json @@ -0,0 +1,52 @@ +{ + "mae": { + "mean": 73.76014493166267, + "std": 26.56964177751206, + "max": 173.77206468173551, + "min": 40.90840481847065, + "25pc": 55.13015548579263, + "50pc": 67.3127463280125, + "75pc": 86.52359464422477, + "count": 35 + }, + "psnr": { + "mean": 27.94676304680717, + "std": 2.3759362559653723, + "max": 32.15384763215193, + "min": 21.935099247792856, + "25pc": 26.33251731594776, + "50pc": 28.44191171626884, + "75pc": 29.779852377601863, + "count": 35 + }, + "ms_ssim": { + "mean": 0.8611710306800014, + "std": 0.10120441133426866, + "max": 0.9523523094969428, + "min": 0.5571973158646525, + "25pc": 0.8318360531504236, + "50pc": 0.8986434315569816, + "75pc": 0.9212279087281977, + "count": 35 + }, + "DICE": { + "mean": 0.655138806785856, + "std": 0.1160120367096364, + "max": 0.8359733819961548, + "min": 0.3324865400791168, + "25pc": 0.594952791929245, + "50pc": 0.6854947805404663, + "75pc": 0.7328222692012787, + "count": 35 + }, + "HD95": { + "mean": 6.714818007605416, + "std": 3.4390077736514857, + "max": 16.03200912475586, + "min": 2.5568649768829346, + "25pc": 4.054888606071472, + "50pc": 5.722603797912598, + "75pc": 9.357222080230713, + "count": 35 + } +} \ No newline at end of file diff --git a/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/plans.json b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/plans.json new file mode 100644 index 0000000000000000000000000000000000000000..430dd05f44371108d500a48a9008ffe638c5ed9e --- /dev/null +++ b/Dataset260/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/plans.json @@ -0,0 +1,521 @@ +{ + "dataset_name": "Dataset260_synthrad2025_task1_MR_AB_pre_v2r_stitched_masked", + "plans_name": "nnUNetResEncUNetLPlans", + "original_median_spacing_after_transp": [ + 3.0, + 1.0, + 1.0 + ], + "original_median_shape_after_transp": [ + 99, + 442, + 465 + ], + "image_reader_writer": "SimpleITKIO", + "transpose_forward": [ + 0, + 1, + 2 + ], + "transpose_backward": [ + 0, + 1, + 2 + ], + "configurations": { + "2d": { + "data_identifier": "nnUNetPlans_2d", + "preprocessor_name": "DefaultPreprocessor", + "batch_size": 13, + "patch_size": [ + 448, + 512 + ], + "median_image_size_in_voxels": [ + 442.0, + 465.0 + ], + "spacing": [ + 1.0, + 1.0 + ], + "normalization_schemes": [ + "ZScoreNormalization" + ], + "use_mask_for_norm": [ + false + ], + "resampling_fn_data": "resample_data_or_seg_to_shape", + "resampling_fn_seg": "resample_data_or_seg_to_shape", + "resampling_fn_data_kwargs": { + "is_seg": false, + "order": 3, + "order_z": 0, + "force_separate_z": null + }, + "resampling_fn_seg_kwargs": { + "is_seg": true, + "order": 1, + "order_z": 0, + "force_separate_z": null + }, + "resampling_fn_probabilities": "resample_data_or_seg_to_shape", + "resampling_fn_probabilities_kwargs": { + "is_seg": false, + "order": 1, + "order_z": 0, + "force_separate_z": null + }, + "architecture": { + "network_class_name": "dynamic_network_architectures.architectures.unet.ResidualEncoderUNet", + "arch_kwargs": { + "n_stages": 7, + "features_per_stage": [ + 32, + 64, + 128, + 256, + 512, + 512, + 512 + ], + "conv_op": "torch.nn.modules.conv.Conv2d", + "kernel_sizes": [ + [ + 3, + 3 + ], + [ + 3, + 3 + ], + [ + 3, + 3 + ], + [ + 3, + 3 + ], + [ + 3, + 3 + ], + [ + 3, + 3 + ], + [ + 3, + 3 + ] + ], + "strides": [ + [ + 1, + 1 + ], + [ + 2, + 2 + ], + [ + 2, + 2 + ], + [ + 2, + 2 + ], + [ + 2, + 2 + ], + [ + 2, + 2 + ], + [ + 2, + 2 + ] + ], + "n_blocks_per_stage": [ + 1, + 3, + 4, + 6, + 6, + 6, + 6 + ], + "n_conv_per_stage_decoder": [ + 1, + 1, + 1, + 1, + 1, + 1 + ], + "conv_bias": true, + "norm_op": "torch.nn.modules.instancenorm.InstanceNorm2d", + "norm_op_kwargs": { + "eps": 1e-05, + "affine": true + }, + "dropout_op": null, + "dropout_op_kwargs": null, + "nonlin": "torch.nn.LeakyReLU", + "nonlin_kwargs": { + "inplace": true + } + }, + "_kw_requires_import": [ + "conv_op", + "norm_op", + "dropout_op", + "nonlin" + ] + }, + "batch_dice": true + }, + "3d_lowres": { + "data_identifier": "nnUNetResEncUNetLPlans_3d_lowres", + "preprocessor_name": "DefaultPreprocessor", + "batch_size": 2, + "patch_size": [ + 64, + 192, + 192 + ], + "median_image_size_in_voxels": [ + 99, + 292, + 307 + ], + "spacing": [ + 3.0, + 1.512589724855112, + 1.512589724855112 + ], + "normalization_schemes": [ + "ZScoreNormalization" + ], + "use_mask_for_norm": [ + false + ], + "resampling_fn_data": "resample_data_or_seg_to_shape", + "resampling_fn_seg": "resample_data_or_seg_to_shape", + "resampling_fn_data_kwargs": { + "is_seg": false, + "order": 3, + "order_z": 0, + "force_separate_z": null + }, + "resampling_fn_seg_kwargs": { + "is_seg": true, + "order": 1, + "order_z": 0, + "force_separate_z": null + }, + "resampling_fn_probabilities": "resample_data_or_seg_to_shape", + "resampling_fn_probabilities_kwargs": { + "is_seg": false, + "order": 1, + "order_z": 0, + "force_separate_z": null + }, + "architecture": { + "network_class_name": "dynamic_network_architectures.architectures.unet.ResidualEncoderUNet", + "arch_kwargs": { + "n_stages": 6, + "features_per_stage": [ + 32, + 64, + 128, + 256, + 320, + 320 + ], + "conv_op": "torch.nn.modules.conv.Conv3d", + "kernel_sizes": [ + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ] + ], + "strides": [ + [ + 1, + 1, + 1 + ], + [ + 2, + 2, + 2 + ], + [ + 2, + 2, + 2 + ], + [ + 2, + 2, + 2 + ], + [ + 2, + 2, + 2 + ], + [ + 1, + 2, + 2 + ] + ], + "n_blocks_per_stage": [ + 1, + 3, + 4, + 6, + 6, + 6 + ], + "n_conv_per_stage_decoder": [ + 1, + 1, + 1, + 1, + 1 + ], + "conv_bias": true, + "norm_op": "torch.nn.modules.instancenorm.InstanceNorm3d", + "norm_op_kwargs": { + "eps": 1e-05, + "affine": true + }, + "dropout_op": null, + "dropout_op_kwargs": null, + "nonlin": "torch.nn.LeakyReLU", + "nonlin_kwargs": { + "inplace": true + } + }, + "_kw_requires_import": [ + "conv_op", + "norm_op", + "dropout_op", + "nonlin" + ] + }, + "batch_dice": false, + "next_stage": "3d_cascade_fullres" + }, + "3d_fullres": { + "data_identifier": "nnUNetPlans_3d_fullres", + "preprocessor_name": "DefaultPreprocessor", + "batch_size": 2, + "patch_size": [ + 40, + 192, + 224 + ], + "median_image_size_in_voxels": [ + 99.0, + 442.0, + 465.0 + ], + "spacing": [ + 3.0, + 1.0, + 1.0 + ], + "normalization_schemes": [ + "ZScoreNormalization" + ], + "use_mask_for_norm": [ + false + ], + "resampling_fn_data": "resample_data_or_seg_to_shape", + "resampling_fn_seg": "resample_data_or_seg_to_shape", + "resampling_fn_data_kwargs": { + "is_seg": false, + "order": 3, + "order_z": 0, + "force_separate_z": null + }, + "resampling_fn_seg_kwargs": { + "is_seg": true, + "order": 1, + "order_z": 0, + "force_separate_z": null + }, + "resampling_fn_probabilities": "resample_data_or_seg_to_shape", + "resampling_fn_probabilities_kwargs": { + "is_seg": false, + "order": 1, + "order_z": 0, + "force_separate_z": null + }, + "architecture": { + "network_class_name": "dynamic_network_architectures.architectures.unet.ResidualEncoderUNet", + "arch_kwargs": { + "n_stages": 6, + "features_per_stage": [ + 32, + 64, + 128, + 256, + 320, + 320 + ], + "conv_op": "torch.nn.modules.conv.Conv3d", + "kernel_sizes": [ + [ + 1, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ] + ], + "strides": [ + [ + 1, + 1, + 1 + ], + [ + 1, + 2, + 2 + ], + [ + 2, + 2, + 2 + ], + [ + 2, + 2, + 2 + ], + [ + 2, + 2, + 2 + ], + [ + 1, + 2, + 2 + ] + ], + "n_blocks_per_stage": [ + 1, + 3, + 4, + 6, + 6, + 6 + ], + "n_conv_per_stage_decoder": [ + 1, + 1, + 1, + 1, + 1 + ], + "conv_bias": true, + "norm_op": "torch.nn.modules.instancenorm.InstanceNorm3d", + "norm_op_kwargs": { + "eps": 1e-05, + "affine": true + }, + "dropout_op": null, + "dropout_op_kwargs": null, + "nonlin": "torch.nn.LeakyReLU", + "nonlin_kwargs": { + "inplace": true + } + }, + "_kw_requires_import": [ + "conv_op", + "norm_op", + "dropout_op", + "nonlin" + ] + }, + "batch_dice": true + }, + "3d_cascade_fullres": { + "inherits_from": "3d_fullres", + "previous_stage": "3d_lowres" + } + }, + "experiment_planner_used": "nnUNetPlannerResEncL", + "label_manager": "LabelManager", + "foreground_intensity_properties_per_channel": { + "0": { + "max": 4142.0, + "mean": 166.43568420410156, + "median": 4.0, + "min": 0.0, + "percentile_00_5": 0.0, + "percentile_99_5": 1590.0, + "std": 346.18096923828125 + } + } +} \ No newline at end of file diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/dataset.json b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/dataset.json new file mode 100644 index 0000000000000000000000000000000000000000..1ddfb413389de2919a2a2e60416cb24432abec72 --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/dataset.json @@ -0,0 +1,14 @@ +{ + "labels": { + "label_001": "1", + "background": 0 + }, + "channel_names": { + "0": "MR" + }, + "numTraining": 221, + "file_ending": ".nii.gz", + "test_set": [ + "case_0000" + ] +} \ No newline at end of file diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/dataset_fingerprint.json b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/dataset_fingerprint.json new file mode 100644 index 0000000000000000000000000000000000000000..ef09421ba98a15dd7ce28a2832a0eeeef13c44b6 --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/dataset_fingerprint.json @@ -0,0 +1,2228 @@ +{ + "foreground_intensity_properties_per_channel": { + "0": { + "max": 4142.0, + "mean": 75.25260162353516, + "median": 0.0, + "min": 0.0, + "percentile_00_5": 0.0, + "percentile_99_5": 1222.0, + "std": 197.03660583496094 + } + }, + "median_relative_size_after_cropping": 1.0, + "shapes_after_crop": [ + [ + 118, + 390, + 533 + ], + [ + 109, + 382, + 499 + ], + [ + 112, + 356, + 529 + ], + [ + 131, + 392, + 590 + ], + [ + 116, + 384, + 560 + ], + [ + 111, + 379, + 499 + ], + [ + 134, + 378, + 585 + ], + [ + 123, + 378, + 573 + ], + [ + 115, + 379, + 499 + ], + [ + 135, + 381, + 542 + ], + [ + 117, + 356, + 516 + ], + [ + 125, + 355, + 577 + ], + [ + 114, + 381, + 482 + ], + [ + 130, + 371, + 593 + ], + [ + 115, + 383, + 499 + ], + [ + 127, + 364, + 588 + ], + [ + 99, + 359, + 563 + ], + [ + 123, + 375, + 535 + ], + [ + 110, + 379, + 499 + ], + [ + 131, + 383, + 552 + ], + [ + 127, + 386, + 523 + ], + [ + 113, + 314, + 541 + ], + [ + 126, + 378, + 551 + ], + [ + 115, + 387, + 499 + ], + [ + 125, + 382, + 588 + ], + [ + 134, + 389, + 584 + ], + [ + 117, + 376, + 509 + ], + [ + 127, + 364, + 541 + ], + [ + 117, + 382, + 587 + ], + [ + 118, + 370, + 563 + ], + [ + 129, + 342, + 547 + ], + [ + 116, + 368, + 474 + ], + [ + 126, + 380, + 562 + ], + [ + 115, + 376, + 499 + ], + [ + 118, + 383, + 545 + ], + [ + 136, + 341, + 542 + ], + [ + 126, + 380, + 555 + ], + [ + 116, + 375, + 558 + ], + [ + 115, + 346, + 484 + ], + [ + 118, + 379, + 499 + ], + [ + 115, + 334, + 499 + ], + [ + 110, + 392, + 499 + ], + [ + 128, + 385, + 592 + ], + [ + 128, + 391, + 516 + ], + [ + 133, + 389, + 523 + ], + [ + 121, + 355, + 519 + ], + [ + 111, + 388, + 499 + ], + [ + 134, + 382, + 511 + ], + [ + 130, + 375, + 591 + ], + [ + 125, + 340, + 510 + ], + [ + 132, + 379, + 555 + ], + [ + 124, + 380, + 526 + ], + [ + 113, + 380, + 545 + ], + [ + 127, + 379, + 537 + ], + [ + 127, + 387, + 543 + ], + [ + 116, + 390, + 597 + ], + [ + 85, + 395, + 435 + ], + [ + 116, + 379, + 567 + ], + [ + 130, + 381, + 589 + ], + [ + 95, + 322, + 242 + ], + [ + 98, + 321, + 242 + ], + [ + 96, + 318, + 243 + ], + [ + 96, + 321, + 246 + ], + [ + 96, + 316, + 241 + ], + [ + 85, + 316, + 242 + ], + [ + 96, + 318, + 242 + ], + [ + 88, + 319, + 240 + ], + [ + 94, + 320, + 246 + ], + [ + 97, + 317, + 251 + ], + [ + 96, + 323, + 243 + ], + [ + 95, + 319, + 242 + ], + [ + 95, + 317, + 241 + ], + [ + 89, + 317, + 241 + ], + [ + 89, + 317, + 256 + ], + [ + 105, + 347, + 253 + ], + [ + 96, + 318, + 250 + ], + [ + 95, + 319, + 245 + ], + [ + 82, + 318, + 243 + ], + [ + 94, + 321, + 243 + ], + [ + 96, + 318, + 248 + ], + [ + 94, + 320, + 244 + ], + [ + 96, + 318, + 242 + ], + [ + 94, + 315, + 238 + ], + [ + 89, + 317, + 244 + ], + [ + 96, + 320, + 249 + ], + [ + 96, + 316, + 246 + ], + [ + 96, + 320, + 245 + ], + [ + 95, + 321, + 246 + ], + [ + 88, + 320, + 243 + ], + [ + 96, + 318, + 241 + ], + [ + 92, + 321, + 243 + ], + [ + 54, + 258, + 250 + ], + [ + 48, + 260, + 258 + ], + [ + 65, + 295, + 281 + ], + [ + 57, + 262, + 256 + ], + [ + 58, + 250, + 245 + ], + [ + 65, + 286, + 287 + ], + [ + 66, + 261, + 249 + ], + [ + 50, + 242, + 243 + ], + [ + 66, + 297, + 300 + ], + [ + 63, + 265, + 254 + ], + [ + 70, + 272, + 256 + ], + [ + 55, + 253, + 254 + ], + [ + 54, + 246, + 245 + ], + [ + 61, + 298, + 290 + ], + [ + 58, + 300, + 297 + ], + [ + 69, + 280, + 262 + ], + [ + 66, + 259, + 249 + ], + [ + 64, + 299, + 288 + ], + [ + 52, + 254, + 255 + ], + [ + 83, + 310, + 288 + ], + [ + 71, + 282, + 297 + ], + [ + 60, + 287, + 288 + ], + [ + 72, + 307, + 288 + ], + [ + 65, + 292, + 280 + ], + [ + 55, + 256, + 253 + ], + [ + 56, + 263, + 257 + ], + [ + 49, + 242, + 239 + ], + [ + 69, + 296, + 296 + ], + [ + 75, + 328, + 311 + ], + [ + 45, + 260, + 260 + ], + [ + 63, + 295, + 290 + ], + [ + 63, + 285, + 279 + ], + [ + 78, + 318, + 293 + ], + [ + 55, + 280, + 279 + ], + [ + 56, + 253, + 244 + ], + [ + 60, + 262, + 257 + ], + [ + 58, + 266, + 261 + ], + [ + 69, + 297, + 284 + ], + [ + 79, + 304, + 288 + ], + [ + 58, + 253, + 247 + ], + [ + 58, + 283, + 280 + ], + [ + 56, + 252, + 256 + ], + [ + 76, + 297, + 298 + ], + [ + 58, + 291, + 288 + ], + [ + 53, + 284, + 285 + ], + [ + 66, + 272, + 253 + ], + [ + 49, + 315, + 315 + ], + [ + 68, + 260, + 251 + ], + [ + 66, + 311, + 297 + ], + [ + 68, + 300, + 297 + ], + [ + 64, + 285, + 286 + ], + [ + 64, + 269, + 254 + ], + [ + 48, + 265, + 265 + ], + [ + 41, + 241, + 242 + ], + [ + 61, + 273, + 262 + ], + [ + 67, + 305, + 296 + ], + [ + 41, + 302, + 303 + ], + [ + 55, + 258, + 251 + ], + [ + 52, + 256, + 257 + ], + [ + 59, + 252, + 244 + ], + [ + 56, + 249, + 249 + ], + [ + 62, + 265, + 260 + ], + [ + 58, + 295, + 295 + ], + [ + 63, + 251, + 250 + ], + [ + 68, + 306, + 295 + ], + [ + 88, + 280, + 273 + ], + [ + 89, + 273, + 286 + ], + [ + 90, + 282, + 275 + ], + [ + 88, + 277, + 270 + ], + [ + 89, + 278, + 272 + ], + [ + 89, + 276, + 273 + ], + [ + 88, + 271, + 277 + ], + [ + 92, + 269, + 287 + ], + [ + 91, + 268, + 284 + ], + [ + 90, + 276, + 286 + ], + [ + 89, + 278, + 283 + ], + [ + 91, + 279, + 277 + ], + [ + 89, + 270, + 279 + ], + [ + 89, + 272, + 280 + ], + [ + 88, + 270, + 279 + ], + [ + 89, + 279, + 271 + ], + [ + 89, + 281, + 270 + ], + [ + 88, + 268, + 276 + ], + [ + 88, + 270, + 277 + ], + [ + 90, + 267, + 280 + ], + [ + 88, + 270, + 277 + ], + [ + 88, + 268, + 276 + ], + [ + 88, + 271, + 277 + ], + [ + 89, + 274, + 281 + ], + [ + 89, + 271, + 280 + ], + [ + 89, + 271, + 279 + ], + [ + 90, + 273, + 277 + ], + [ + 90, + 279, + 268 + ], + [ + 89, + 276, + 275 + ], + [ + 89, + 270, + 279 + ], + [ + 89, + 280, + 272 + ], + [ + 90, + 279, + 271 + ], + [ + 88, + 267, + 277 + ], + [ + 88, + 271, + 278 + ], + [ + 89, + 280, + 271 + ], + [ + 89, + 279, + 274 + ], + [ + 90, + 272, + 279 + ], + [ + 88, + 270, + 277 + ], + [ + 90, + 275, + 287 + ], + [ + 88, + 271, + 280 + ], + [ + 89, + 270, + 278 + ], + [ + 88, + 272, + 277 + ], + [ + 90, + 279, + 281 + ], + [ + 97, + 272, + 281 + ], + [ + 89, + 271, + 279 + ], + [ + 89, + 277, + 283 + ], + [ + 89, + 269, + 276 + ], + [ + 89, + 272, + 278 + ], + [ + 91, + 278, + 277 + ], + [ + 97, + 302, + 276 + ], + [ + 89, + 268, + 279 + ], + [ + 88, + 268, + 277 + ], + [ + 98, + 271, + 283 + ], + [ + 88, + 285, + 275 + ], + [ + 100, + 319, + 280 + ], + [ + 88, + 268, + 278 + ], + [ + 89, + 273, + 279 + ], + [ + 87, + 272, + 279 + ], + [ + 89, + 270, + 280 + ], + [ + 91, + 274, + 286 + ], + [ + 90, + 270, + 284 + ], + [ + 90, + 271, + 280 + ], + [ + 90, + 277, + 289 + ], + [ + 88, + 269, + 277 + ], + [ + 88, + 269, + 277 + ] + ], + "spacings": [ + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ] + ] +} \ No newline at end of file diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/checkpoint_final.pth b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/checkpoint_final.pth new file mode 100644 index 0000000000000000000000000000000000000000..f242922667cdf23bac00bb4a3a80977c2d128bb7 --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/checkpoint_final.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5f4523753a388b22f13f591954caa280be393921ee3ec53ef17dc2dd027fed6f +size 815340333 diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/debug.json b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/debug.json new file mode 100644 index 0000000000000000000000000000000000000000..77ef71885da227c0984d490983a8ba0b7f012d08 --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/debug.json @@ -0,0 +1,58 @@ +{ + "_best_ema": "0.39465758", + "aim_run": ">", + "batch_size": "2", + "configuration_manager": "{'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [56, 192, 160], 'median_image_size_in_voxels': [89.0, 296.0, 279.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}, 'deep_supervision': False}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False}", + "configuration_name": "3d_fullres", + "cudnn_version": 90501, + "current_epoch": "1000", + "dataloader_train": "", + "dataloader_train.generator": "", + "dataloader_train.num_processes": "12", + "dataloader_train.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataloader_val": "", + "dataloader_val.generator": "", + "dataloader_val.num_processes": "6", + "dataloader_val.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataset_json": "{'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 221, 'file_ending': '.mha'}", + "debug_custom": "{'run_hash': '1f094a83d32a4f4bbe106404'}", + "decoder_type": "standard", + "device": "cuda:0", + "disable_checkpointing": "False", + "enable_deep_supervision": "False", + "fold": "0", + "folder_with_segs_from_previous_stage": "None", + "gpu_name": "NVIDIA H100", + "grad_scaler": "", + "hostname": "g043", + "image_loss_weight": "0.5", + "inference_allowed_mirroring_axes": "None", + "initial_lr": "0.01", + "is_cascaded": "False", + "is_ddp": "False", + "label_manager": "", + "local_rank": "0", + "log_file": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/training_log_2025_8_13_01_35_46.txt", + "logger": "", + "loss": "SynPerceptionLoss(\n (seg_model): ResidualEncoderUNet(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (decoder): UNetDecoder(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (stages): ModuleList(\n (0): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (1): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (2): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (3): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (4): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (transpconvs): ModuleList(\n (0): ConvTranspose3d(320, 320, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n (1): ConvTranspose3d(320, 256, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (2): ConvTranspose3d(256, 128, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (3): ConvTranspose3d(128, 64, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (4): ConvTranspose3d(64, 32, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n )\n (seg_layers): ModuleList(\n (0): Conv3d(320, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (1): Conv3d(256, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (2): Conv3d(128, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (3): Conv3d(64, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (4): Conv3d(32, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n )\n )\n )\n (L1): L1Loss()\n (image_loss): myMaskedMSE(\n (mse): myMSE()\n )\n)", + "lr_scheduler": "", + "my_init_kwargs": "{'plans': {'dataset_name': 'Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [89, 296, 279], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 30, 'patch_size': [320, 320], 'median_image_size_in_voxels': [296.0, 279.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [56, 192, 160], 'median_image_size_in_voxels': [89.0, 296.0, 279.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 75.25260162353516, 'median': 0.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1222.0, 'std': 197.03660583496094}}}, 'configuration': '3d_fullres', 'fold': 0, 'dataset_json': {'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 221, 'file_ending': '.mha'}, 'unpack_dataset': True, 'device': device(type='cuda')}", + "network": "ResidualEncoderUNet", + "num_epochs": "1000", + "num_input_channels": "1", + "num_iterations_per_epoch": "250", + "num_val_iterations_per_epoch": "50", + "optimizer": "SGD (\nParameter Group 0\n dampening: 0\n differentiable: False\n foreach: None\n fused: None\n initial_lr: 0.01\n lr: 1.995262314968881e-05\n maximize: False\n momentum: 0.99\n nesterov: True\n weight_decay: 3e-05\n)", + "output_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0", + "output_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres", + "oversample_foreground_percent": "0.33", + "perception_masked": "True", + "plans_manager": "{'dataset_name': 'Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [89, 296, 279], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 30, 'patch_size': [320, 320], 'median_image_size_in_voxels': [296.0, 279.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [56, 192, 160], 'median_image_size_in_voxels': [89.0, 296.0, 279.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 75.25260162353516, 'median': 0.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1222.0, 'std': 197.03660583496094}}}", + "preprocessed_dataset_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetPlans_3d_fullres", + "preprocessed_dataset_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked", + "save_every": "50", + "torch_version": "2.7.1+cu126", + "unpack_dataset": "True", + "was_initialized": "True", + "weight_decay": "3e-05" +} \ No newline at end of file diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/progress.png b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/progress.png new file mode 100644 index 0000000000000000000000000000000000000000..1bf8d3281ea1585c498a6df168e127dbf36cd9c4 --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/progress.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:50befff8c975cab64e39909e1854a1d858218b37e0c66666638f2a040c00590f +size 464577 diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/validation/summary.json b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/validation/summary.json new file mode 100644 index 0000000000000000000000000000000000000000..280a4469ac09f74fe9a608b9718ce2832d1a2051 --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/validation/summary.json @@ -0,0 +1,52 @@ +{ + "mae": { + "mean": 77.26506009860128, + "std": 29.047413697702925, + "max": 178.38790131889363, + "min": 47.132550452794426, + "25pc": 57.471762288010716, + "50pc": 63.945487494455776, + "75pc": 90.56736272468387, + "count": 45 + }, + "psnr": { + "mean": 28.235542021901257, + "std": 2.690380844350698, + "max": 32.000148173802174, + "min": 21.6541829845217, + "25pc": 26.41537212009891, + "50pc": 29.23759860873585, + "75pc": 30.31196145523424, + "count": 45 + }, + "ms_ssim": { + "mean": 0.948246476628697, + "std": 0.0469000801835997, + "max": 0.9842433749327716, + "min": 0.7856202409498639, + "25pc": 0.9423137105579374, + "50pc": 0.9720659616183045, + "75pc": 0.9758630404130089, + "count": 45 + }, + "DICE": { + "mean": 0.8557620300187005, + "std": 0.0942056960163186, + "max": 0.9376087188720703, + "min": 0.4086264967918396, + "25pc": 0.8338607549667358, + "50pc": 0.8933591246604919, + "75pc": 0.9052481651306152, + "count": 45 + }, + "HD95": { + "mean": 2.011283122168647, + "std": 1.3636167809435726, + "max": 8.224165916442871, + "min": 1.1031426191329956, + "25pc": 1.3593908548355103, + "50pc": 1.4717508554458618, + "75pc": 1.7875702381134033, + "count": 45 + } +} \ No newline at end of file diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/checkpoint_final.pth b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/checkpoint_final.pth new file mode 100644 index 0000000000000000000000000000000000000000..d8f4783e4685ee41dcb335ca8999b6e2fb0eaf96 --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/checkpoint_final.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7ec4047be4932626d5c405635279e5cff3144a6ff8ff09642e67d1fb767b23eb +size 815576941 diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/debug.json b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/debug.json new file mode 100644 index 0000000000000000000000000000000000000000..8b786c9acff7ac85667c4cd7c7af420a2803597c --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/debug.json @@ -0,0 +1,58 @@ +{ + "_best_ema": "None", + "aim_run": ">", + "batch_size": "2", + "configuration_manager": "{'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [56, 192, 160], 'median_image_size_in_voxels': [89.0, 296.0, 279.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}, 'deep_supervision': False}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False}", + "configuration_name": "3d_fullres", + "cudnn_version": 90501, + "current_epoch": "0", + "dataloader_train": "", + "dataloader_train.generator": "", + "dataloader_train.num_processes": "12", + "dataloader_train.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataloader_val": "", + "dataloader_val.generator": "", + "dataloader_val.num_processes": "6", + "dataloader_val.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataset_json": "{'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 221, 'file_ending': '.mha'}", + "debug_custom": "{'run_hash': 'a0da8d2e97f041d498d4b019'}", + "decoder_type": "standard", + "device": "cuda:0", + "disable_checkpointing": "False", + "enable_deep_supervision": "False", + "fold": "1", + "folder_with_segs_from_previous_stage": "None", + "gpu_name": "NVIDIA H100", + "grad_scaler": "", + "hostname": "g043", + "image_loss_weight": "0.5", + "inference_allowed_mirroring_axes": "None", + "initial_lr": "0.01", + "is_cascaded": "False", + "is_ddp": "False", + "label_manager": "", + "local_rank": "0", + "log_file": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/training_log_2025_8_13_01_36_42.txt", + "logger": "", + "loss": "SynPerceptionLoss(\n (seg_model): ResidualEncoderUNet(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (decoder): UNetDecoder(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (stages): ModuleList(\n (0): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (1): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (2): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (3): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (4): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (transpconvs): ModuleList(\n (0): ConvTranspose3d(320, 320, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n (1): ConvTranspose3d(320, 256, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (2): ConvTranspose3d(256, 128, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (3): ConvTranspose3d(128, 64, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (4): ConvTranspose3d(64, 32, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n )\n (seg_layers): ModuleList(\n (0): Conv3d(320, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (1): Conv3d(256, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (2): Conv3d(128, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (3): Conv3d(64, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (4): Conv3d(32, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n )\n )\n )\n (L1): L1Loss()\n (image_loss): myMaskedMSE(\n (mse): myMSE()\n )\n)", + "lr_scheduler": "", + "my_init_kwargs": "{'plans': {'dataset_name': 'Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [89, 296, 279], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 30, 'patch_size': [320, 320], 'median_image_size_in_voxels': [296.0, 279.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [56, 192, 160], 'median_image_size_in_voxels': [89.0, 296.0, 279.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 75.25260162353516, 'median': 0.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1222.0, 'std': 197.03660583496094}}}, 'configuration': '3d_fullres', 'fold': 1, 'dataset_json': {'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 221, 'file_ending': '.mha'}, 'unpack_dataset': True, 'device': device(type='cuda')}", + "network": "ResidualEncoderUNet", + "num_epochs": "1000", + "num_input_channels": "1", + "num_iterations_per_epoch": "250", + "num_val_iterations_per_epoch": "50", + "optimizer": "SGD (\nParameter Group 0\n dampening: 0\n differentiable: False\n foreach: None\n fused: None\n initial_lr: 0.01\n lr: 0.01\n maximize: False\n momentum: 0.99\n nesterov: True\n weight_decay: 3e-05\n)", + "output_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1", + "output_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres", + "oversample_foreground_percent": "0.33", + "perception_masked": "True", + "plans_manager": "{'dataset_name': 'Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [89, 296, 279], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 30, 'patch_size': [320, 320], 'median_image_size_in_voxels': [296.0, 279.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [56, 192, 160], 'median_image_size_in_voxels': [89.0, 296.0, 279.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 75.25260162353516, 'median': 0.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1222.0, 'std': 197.03660583496094}}}", + "preprocessed_dataset_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetPlans_3d_fullres", + "preprocessed_dataset_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked", + "save_every": "50", + "torch_version": "2.7.1+cu126", + "unpack_dataset": "True", + "was_initialized": "True", + "weight_decay": "3e-05" +} \ No newline at end of file diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/progress.png b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/progress.png new file mode 100644 index 0000000000000000000000000000000000000000..9859e011f91d9ab31d3a21d1c77530c57dfc0654 --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/progress.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b67429a508f85faef76d96e5a4b5ec357eb99c1c4ef725e25287a4058b5c3ac7 +size 476576 diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/validation/summary.json b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/validation/summary.json new file mode 100644 index 0000000000000000000000000000000000000000..2d4a9ee631ffce5a2addaca06ff24c4a590f59ae --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/validation/summary.json @@ -0,0 +1,52 @@ +{ + "mae": { + "mean": 77.08031060795992, + "std": 37.7342786118798, + "max": 226.5298721192156, + "min": 46.869953341959786, + "25pc": 55.65348859505815, + "50pc": 62.47728006085707, + "75pc": 88.62038012916034, + "count": 44 + }, + "psnr": { + "mean": 28.683509931304478, + "std": 2.958996696745497, + "max": 32.156225455274416, + "min": 20.389197370147905, + "25pc": 26.594943919752982, + "50pc": 29.73280855814989, + "75pc": 30.704535229599596, + "count": 44 + }, + "ms_ssim": { + "mean": 0.9434731013352579, + "std": 0.06479873540042186, + "max": 0.9861069725352045, + "min": 0.6694157708855971, + "25pc": 0.9370740939667938, + "50pc": 0.9685755157528508, + "75pc": 0.9761700734939637, + "count": 44 + }, + "DICE": { + "mean": 0.8469922440973195, + "std": 0.1222063219809408, + "max": 0.9264131188392639, + "min": 0.2666594386100769, + "25pc": 0.8274439722299576, + "50pc": 0.8889445066452026, + "75pc": 0.9080726206302643, + "count": 44 + }, + "HD95": { + "mean": 2.2272169671275397, + "std": 1.7176528510543398, + "max": 9.024123191833496, + "min": 1.125, + "25pc": 1.2868992388248444, + "50pc": 1.5151956677436829, + "75pc": 2.1825499534606934, + "count": 44 + } +} \ No newline at end of file diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/checkpoint_final.pth b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/checkpoint_final.pth new file mode 100644 index 0000000000000000000000000000000000000000..9beda6c64965d20e7721e49785b8e377a66844ca --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/checkpoint_final.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6d49278421ea20efb027574e1fb74adf1a3310e6854326ba3a43981d823a1116 +size 815576941 diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/debug.json b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/debug.json new file mode 100644 index 0000000000000000000000000000000000000000..1c09e830ee235dcd68b906438daef5b084ca06ec --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/debug.json @@ -0,0 +1,58 @@ +{ + "_best_ema": "None", + "aim_run": ">", + "batch_size": "2", + "configuration_manager": "{'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [56, 192, 160], 'median_image_size_in_voxels': [89.0, 296.0, 279.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}, 'deep_supervision': False}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False}", + "configuration_name": "3d_fullres", + "cudnn_version": 90501, + "current_epoch": "0", + "dataloader_train": "", + "dataloader_train.generator": "", + "dataloader_train.num_processes": "12", + "dataloader_train.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataloader_val": "", + "dataloader_val.generator": "", + "dataloader_val.num_processes": "6", + "dataloader_val.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataset_json": "{'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 221, 'file_ending': '.mha'}", + "debug_custom": "{'run_hash': '59407d6c5463471fa96d8244'}", + "decoder_type": "standard", + "device": "cuda:0", + "disable_checkpointing": "False", + "enable_deep_supervision": "False", + "fold": "2", + "folder_with_segs_from_previous_stage": "None", + "gpu_name": "NVIDIA H100", + "grad_scaler": "", + "hostname": "g031", + "image_loss_weight": "0.5", + "inference_allowed_mirroring_axes": "None", + "initial_lr": "0.01", + "is_cascaded": "False", + "is_ddp": "False", + "label_manager": "", + "local_rank": "0", + "log_file": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/training_log_2025_8_13_01_36_48.txt", + "logger": "", + "loss": "SynPerceptionLoss(\n (seg_model): ResidualEncoderUNet(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (decoder): UNetDecoder(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (stages): ModuleList(\n (0): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (1): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (2): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (3): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (4): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (transpconvs): ModuleList(\n (0): ConvTranspose3d(320, 320, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n (1): ConvTranspose3d(320, 256, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (2): ConvTranspose3d(256, 128, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (3): ConvTranspose3d(128, 64, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (4): ConvTranspose3d(64, 32, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n )\n (seg_layers): ModuleList(\n (0): Conv3d(320, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (1): Conv3d(256, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (2): Conv3d(128, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (3): Conv3d(64, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (4): Conv3d(32, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n )\n )\n )\n (L1): L1Loss()\n (image_loss): myMaskedMSE(\n (mse): myMSE()\n )\n)", + "lr_scheduler": "", + "my_init_kwargs": "{'plans': {'dataset_name': 'Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [89, 296, 279], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 30, 'patch_size': [320, 320], 'median_image_size_in_voxels': [296.0, 279.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [56, 192, 160], 'median_image_size_in_voxels': [89.0, 296.0, 279.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 75.25260162353516, 'median': 0.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1222.0, 'std': 197.03660583496094}}}, 'configuration': '3d_fullres', 'fold': 2, 'dataset_json': {'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 221, 'file_ending': '.mha'}, 'unpack_dataset': True, 'device': device(type='cuda')}", + "network": "ResidualEncoderUNet", + "num_epochs": "1000", + "num_input_channels": "1", + "num_iterations_per_epoch": "250", + "num_val_iterations_per_epoch": "50", + "optimizer": "SGD (\nParameter Group 0\n dampening: 0\n differentiable: False\n foreach: None\n fused: None\n initial_lr: 0.01\n lr: 0.01\n maximize: False\n momentum: 0.99\n nesterov: True\n weight_decay: 3e-05\n)", + "output_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2", + "output_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres", + "oversample_foreground_percent": "0.33", + "perception_masked": "True", + "plans_manager": "{'dataset_name': 'Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [89, 296, 279], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 30, 'patch_size': [320, 320], 'median_image_size_in_voxels': [296.0, 279.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [56, 192, 160], 'median_image_size_in_voxels': [89.0, 296.0, 279.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 75.25260162353516, 'median': 0.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1222.0, 'std': 197.03660583496094}}}", + "preprocessed_dataset_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetPlans_3d_fullres", + "preprocessed_dataset_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked", + "save_every": "50", + "torch_version": "2.7.1+cu126", + "unpack_dataset": "True", + "was_initialized": "True", + "weight_decay": "3e-05" +} \ No newline at end of file diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/progress.png b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/progress.png new file mode 100644 index 0000000000000000000000000000000000000000..ca962c7ae998768e8435c03bec7863c21cfcedbb --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/progress.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:79f08816255a2608176049556b703474ea6d2b1c8f0618d53a5a8892624aada2 +size 466092 diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/validation/summary.json b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/validation/summary.json new file mode 100644 index 0000000000000000000000000000000000000000..73f3bc0143ad2bfd9fa8f32c241f6b2ec9b8b6cd --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/validation/summary.json @@ -0,0 +1,52 @@ +{ + "mae": { + "mean": 74.65292392337206, + "std": 27.21601270921252, + "max": 180.22424320443434, + "min": 50.03639991050344, + "25pc": 55.20103966286182, + "50pc": 60.57595211267504, + "75pc": 95.06778034175494, + "count": 44 + }, + "psnr": { + "mean": 28.46884222939042, + "std": 2.740618880237753, + "max": 31.788027766490927, + "min": 21.581445444266258, + "25pc": 25.956340082853398, + "50pc": 29.589134728975914, + "75pc": 30.78935038430233, + "count": 44 + }, + "ms_ssim": { + "mean": 0.9470716773141884, + "std": 0.04705583669583442, + "max": 0.9837417913911773, + "min": 0.7408830509741683, + "25pc": 0.9323297519024558, + "50pc": 0.9680664744199681, + "75pc": 0.9773946177781223, + "count": 44 + }, + "DICE": { + "mean": 0.8484438739039681, + "std": 0.11799668647372488, + "max": 0.9286674857139587, + "min": 0.18524354696273804, + "25pc": 0.821153312921524, + "50pc": 0.8878902792930603, + "75pc": 0.911092683672905, + "count": 44 + }, + "HD95": { + "mean": 2.199756370349364, + "std": 1.6979196164112982, + "max": 11.68980884552002, + "min": 1.0883883237838745, + "25pc": 1.2982812225818634, + "50pc": 1.5779862999916077, + "75pc": 2.510884642601013, + "count": 44 + } +} \ No newline at end of file diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/checkpoint_final.pth b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/checkpoint_final.pth new file mode 100644 index 0000000000000000000000000000000000000000..bb4ada78cd64539e4682ed3049f35b8ba62b38d0 --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/checkpoint_final.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:04ff8055f73f71f37b01529006091cc697ab989e8591f585a122975342d8bee9 +size 815577069 diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/debug.json b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/debug.json new file mode 100644 index 0000000000000000000000000000000000000000..f68d8137f1f11c2a0dff4885363e80dcdd3b1355 --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/debug.json @@ -0,0 +1,58 @@ +{ + "_best_ema": "None", + "aim_run": ">", + "batch_size": "2", + "configuration_manager": "{'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [56, 192, 160], 'median_image_size_in_voxels': [89.0, 296.0, 279.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}, 'deep_supervision': False}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False}", + "configuration_name": "3d_fullres", + "cudnn_version": 90501, + "current_epoch": "0", + "dataloader_train": "", + "dataloader_train.generator": "", + "dataloader_train.num_processes": "12", + "dataloader_train.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataloader_val": "", + "dataloader_val.generator": "", + "dataloader_val.num_processes": "6", + "dataloader_val.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataset_json": "{'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 221, 'file_ending': '.mha'}", + "debug_custom": "{'run_hash': 'c2d61ffb0a944f139d1af9da'}", + "decoder_type": "standard", + "device": "cuda:0", + "disable_checkpointing": "False", + "enable_deep_supervision": "False", + "fold": "3", + "folder_with_segs_from_previous_stage": "None", + "gpu_name": "NVIDIA H100", + "grad_scaler": "", + "hostname": "g031", + "image_loss_weight": "0.5", + "inference_allowed_mirroring_axes": "None", + "initial_lr": "0.01", + "is_cascaded": "False", + "is_ddp": "False", + "label_manager": "", + "local_rank": "0", + "log_file": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/training_log_2025_8_13_01_36_51.txt", + "logger": "", + "loss": "SynPerceptionLoss(\n (seg_model): ResidualEncoderUNet(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (decoder): UNetDecoder(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (stages): ModuleList(\n (0): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (1): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (2): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (3): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (4): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (transpconvs): ModuleList(\n (0): ConvTranspose3d(320, 320, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n (1): ConvTranspose3d(320, 256, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (2): ConvTranspose3d(256, 128, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (3): ConvTranspose3d(128, 64, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (4): ConvTranspose3d(64, 32, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n )\n (seg_layers): ModuleList(\n (0): Conv3d(320, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (1): Conv3d(256, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (2): Conv3d(128, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (3): Conv3d(64, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (4): Conv3d(32, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n )\n )\n )\n (L1): L1Loss()\n (image_loss): myMaskedMSE(\n (mse): myMSE()\n )\n)", + "lr_scheduler": "", + "my_init_kwargs": "{'plans': {'dataset_name': 'Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [89, 296, 279], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 30, 'patch_size': [320, 320], 'median_image_size_in_voxels': [296.0, 279.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [56, 192, 160], 'median_image_size_in_voxels': [89.0, 296.0, 279.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 75.25260162353516, 'median': 0.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1222.0, 'std': 197.03660583496094}}}, 'configuration': '3d_fullres', 'fold': 3, 'dataset_json': {'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 221, 'file_ending': '.mha'}, 'unpack_dataset': True, 'device': device(type='cuda')}", + "network": "ResidualEncoderUNet", + "num_epochs": "1000", + "num_input_channels": "1", + "num_iterations_per_epoch": "250", + "num_val_iterations_per_epoch": "50", + "optimizer": "SGD (\nParameter Group 0\n dampening: 0\n differentiable: False\n foreach: None\n fused: None\n initial_lr: 0.01\n lr: 0.01\n maximize: False\n momentum: 0.99\n nesterov: True\n weight_decay: 3e-05\n)", + "output_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3", + "output_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres", + "oversample_foreground_percent": "0.33", + "perception_masked": "True", + "plans_manager": "{'dataset_name': 'Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [89, 296, 279], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 30, 'patch_size': [320, 320], 'median_image_size_in_voxels': [296.0, 279.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [56, 192, 160], 'median_image_size_in_voxels': [89.0, 296.0, 279.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 75.25260162353516, 'median': 0.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1222.0, 'std': 197.03660583496094}}}", + "preprocessed_dataset_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetPlans_3d_fullres", + "preprocessed_dataset_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked", + "save_every": "50", + "torch_version": "2.7.1+cu126", + "unpack_dataset": "True", + "was_initialized": "True", + "weight_decay": "3e-05" +} \ No newline at end of file diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/progress.png b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/progress.png new file mode 100644 index 0000000000000000000000000000000000000000..bfd5bee7793e599bacd4eb87455a59f864de4f5c --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/progress.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e8ad5f4bb1c655d6f4886294a6f7db19c4a6be416190c1f5a83592eed2635c9 +size 465407 diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/validation/summary.json b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/validation/summary.json new file mode 100644 index 0000000000000000000000000000000000000000..3a535424d2b72784807f9e4fdb5557f00dea8cc6 --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/validation/summary.json @@ -0,0 +1,52 @@ +{ + "mae": { + "mean": 65.43940683318942, + "std": 19.30033067412743, + "max": 133.17917855274558, + "min": 49.35271362166326, + "25pc": 54.440088390029416, + "50pc": 57.81114012880869, + "75pc": 66.98212477099185, + "count": 44 + }, + "psnr": { + "mean": 29.552400000618967, + "std": 2.1467199540790727, + "max": 31.802603406072585, + "min": 22.912650964081408, + "25pc": 29.284818134323984, + "50pc": 30.2391087792035, + "75pc": 30.816039260804658, + "count": 44 + }, + "ms_ssim": { + "mean": 0.9629728119067432, + "std": 0.03201789619518063, + "max": 0.9824830027178535, + "min": 0.8321075191950946, + "25pc": 0.9701554374794714, + "50pc": 0.9754116594674357, + "75pc": 0.9778994457327609, + "count": 44 + }, + "DICE": { + "mean": 0.8776735744693063, + "std": 0.060516889606336846, + "max": 0.9218840003013611, + "min": 0.59428471326828, + "25pc": 0.872382253408432, + "50pc": 0.8999759256839752, + "75pc": 0.9089413583278656, + "count": 44 + }, + "HD95": { + "mean": 1.9350836900147526, + "std": 1.2071419560361742, + "max": 5.903610706329346, + "min": 1.0942808389663696, + "25pc": 1.2886926531791687, + "50pc": 1.3760043382644653, + "75pc": 1.8396624028682709, + "count": 44 + } +} \ No newline at end of file diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/checkpoint_final.pth b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/checkpoint_final.pth new file mode 100644 index 0000000000000000000000000000000000000000..1f3042188f4caae1eb75ae2b398b8badfc5673d9 --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/checkpoint_final.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a53eb93ead6be5736e97b05407ffcc92829de1605140e0383b2c20a52916cd14 +size 815577005 diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/debug.json b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/debug.json new file mode 100644 index 0000000000000000000000000000000000000000..fd291cb0f3481431b377f2f5fee431553aa10b95 --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/debug.json @@ -0,0 +1,58 @@ +{ + "_best_ema": "None", + "aim_run": ">", + "batch_size": "2", + "configuration_manager": "{'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [56, 192, 160], 'median_image_size_in_voxels': [89.0, 296.0, 279.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}, 'deep_supervision': False}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False}", + "configuration_name": "3d_fullres", + "cudnn_version": 90501, + "current_epoch": "0", + "dataloader_train": "", + "dataloader_train.generator": "", + "dataloader_train.num_processes": "12", + "dataloader_train.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataloader_val": "", + "dataloader_val.generator": "", + "dataloader_val.num_processes": "6", + "dataloader_val.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataset_json": "{'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 221, 'file_ending': '.mha'}", + "debug_custom": "{'run_hash': '5d726a9fb5f2480ba80f2655'}", + "decoder_type": "standard", + "device": "cuda:0", + "disable_checkpointing": "False", + "enable_deep_supervision": "False", + "fold": "4", + "folder_with_segs_from_previous_stage": "None", + "gpu_name": "NVIDIA H100", + "grad_scaler": "", + "hostname": "g032", + "image_loss_weight": "0.5", + "inference_allowed_mirroring_axes": "None", + "initial_lr": "0.01", + "is_cascaded": "False", + "is_ddp": "False", + "label_manager": "", + "local_rank": "0", + "log_file": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/training_log_2025_8_13_01_37_31.txt", + "logger": "", + "loss": "SynPerceptionLoss(\n (seg_model): ResidualEncoderUNet(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (decoder): UNetDecoder(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (stages): ModuleList(\n (0): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (1): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (2): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (3): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (4): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (transpconvs): ModuleList(\n (0): ConvTranspose3d(320, 320, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n (1): ConvTranspose3d(320, 256, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (2): ConvTranspose3d(256, 128, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (3): ConvTranspose3d(128, 64, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (4): ConvTranspose3d(64, 32, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n )\n (seg_layers): ModuleList(\n (0): Conv3d(320, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (1): Conv3d(256, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (2): Conv3d(128, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (3): Conv3d(64, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (4): Conv3d(32, 32, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n )\n )\n )\n (L1): L1Loss()\n (image_loss): myMaskedMSE(\n (mse): myMSE()\n )\n)", + "lr_scheduler": "", + "my_init_kwargs": "{'plans': {'dataset_name': 'Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [89, 296, 279], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 30, 'patch_size': [320, 320], 'median_image_size_in_voxels': [296.0, 279.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [56, 192, 160], 'median_image_size_in_voxels': [89.0, 296.0, 279.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 75.25260162353516, 'median': 0.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1222.0, 'std': 197.03660583496094}}}, 'configuration': '3d_fullres', 'fold': 4, 'dataset_json': {'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 221, 'file_ending': '.mha'}, 'unpack_dataset': True, 'device': device(type='cuda')}", + "network": "ResidualEncoderUNet", + "num_epochs": "1000", + "num_input_channels": "1", + "num_iterations_per_epoch": "250", + "num_val_iterations_per_epoch": "50", + "optimizer": "SGD (\nParameter Group 0\n dampening: 0\n differentiable: False\n foreach: None\n fused: None\n initial_lr: 0.01\n lr: 0.01\n maximize: False\n momentum: 0.99\n nesterov: True\n weight_decay: 3e-05\n)", + "output_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4", + "output_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres", + "oversample_foreground_percent": "0.33", + "perception_masked": "True", + "plans_manager": "{'dataset_name': 'Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [89, 296, 279], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 30, 'patch_size': [320, 320], 'median_image_size_in_voxels': [296.0, 279.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [56, 192, 160], 'median_image_size_in_voxels': [89.0, 296.0, 279.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 4142.0, 'mean': 75.25260162353516, 'median': 0.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1222.0, 'std': 197.03660583496094}}}", + "preprocessed_dataset_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked/nnUNetPlans_3d_fullres", + "preprocessed_dataset_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked", + "save_every": "50", + "torch_version": "2.7.1+cu126", + "unpack_dataset": "True", + "was_initialized": "True", + "weight_decay": "3e-05" +} \ No newline at end of file diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/progress.png b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/progress.png new file mode 100644 index 0000000000000000000000000000000000000000..2d9a230d28c33aea2f56fa2a8c0820c75dd2bc91 --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/progress.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bbbf2a4d4c0e219c6dec76aa7e7f009cc6028effa9a1490189b98f384f82f4cd +size 474172 diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/validation/summary.json b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/validation/summary.json new file mode 100644 index 0000000000000000000000000000000000000000..e8e91f4aad673b88c20b87692ab8eb04b4ebae52 --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/validation/summary.json @@ -0,0 +1,52 @@ +{ + "mae": { + "mean": 79.01048282643166, + "std": 25.814640283626616, + "max": 136.18736974287893, + "min": 48.097312855892966, + "25pc": 56.43015459462086, + "50pc": 70.26204491277278, + "75pc": 100.98830817898656, + "count": 44 + }, + "psnr": { + "mean": 28.16481870353614, + "std": 2.7080036499500273, + "max": 31.785287333787497, + "min": 23.128256951315468, + "25pc": 25.650981331586912, + "50pc": 28.757230793948047, + "75pc": 30.502033048682343, + "count": 44 + }, + "ms_ssim": { + "mean": 0.9456753494688663, + "std": 0.03748245085540397, + "max": 0.9847741314776761, + "min": 0.8537015800968059, + "25pc": 0.9215156936718175, + "50pc": 0.9640495073026166, + "75pc": 0.9752980815606082, + "count": 44 + }, + "DICE": { + "mean": 0.8262107798999007, + "std": 0.11717953033145945, + "max": 0.9291353821754456, + "min": 0.43098184466362, + "25pc": 0.7794158011674881, + "50pc": 0.8882485926151276, + "75pc": 0.906611979007721, + "count": 44 + }, + "HD95": { + "mean": 2.321003418077122, + "std": 1.4156960699276717, + "max": 7.2080817222595215, + "min": 1.0942808389663696, + "25pc": 1.3163809180259705, + "50pc": 1.8190175890922546, + "75pc": 2.7259116172790527, + "count": 44 + } +} \ No newline at end of file diff --git a/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/plans.json b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/plans.json new file mode 100644 index 0000000000000000000000000000000000000000..5a5a48b6d288b12870c9cd4f4cbce479cc06e601 --- /dev/null +++ b/Dataset262/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/plans.json @@ -0,0 +1,356 @@ +{ + "dataset_name": "Dataset262_synthrad2025_task1_MR_HN_pre_v2r_stitched_masked", + "plans_name": "nnUNetResEncUNetLPlans", + "original_median_spacing_after_transp": [ + 3.0, + 1.0, + 1.0 + ], + "original_median_shape_after_transp": [ + 89, + 296, + 279 + ], + "image_reader_writer": "SimpleITKIO", + "transpose_forward": [ + 0, + 1, + 2 + ], + "transpose_backward": [ + 0, + 1, + 2 + ], + "configurations": { + "2d": { + "data_identifier": "nnUNetPlans_2d", + "preprocessor_name": "DefaultPreprocessor", + "batch_size": 30, + "patch_size": [ + 320, + 320 + ], + "median_image_size_in_voxels": [ + 296.0, + 279.0 + ], + "spacing": [ + 1.0, + 1.0 + ], + "normalization_schemes": [ + "ZScoreNormalization" + ], + "use_mask_for_norm": [ + false + ], + "resampling_fn_data": "resample_data_or_seg_to_shape", + "resampling_fn_seg": "resample_data_or_seg_to_shape", + "resampling_fn_data_kwargs": { + "is_seg": false, + "order": 3, + "order_z": 0, + "force_separate_z": null + }, + "resampling_fn_seg_kwargs": { + "is_seg": true, + "order": 1, + "order_z": 0, + "force_separate_z": null + }, + "resampling_fn_probabilities": "resample_data_or_seg_to_shape", + "resampling_fn_probabilities_kwargs": { + "is_seg": false, + "order": 1, + "order_z": 0, + "force_separate_z": null + }, + "architecture": { + "network_class_name": "dynamic_network_architectures.architectures.unet.ResidualEncoderUNet", + "arch_kwargs": { + "n_stages": 7, + "features_per_stage": [ + 32, + 64, + 128, + 256, + 512, + 512, + 512 + ], + "conv_op": "torch.nn.modules.conv.Conv2d", + "kernel_sizes": [ + [ + 3, + 3 + ], + [ + 3, + 3 + ], + [ + 3, + 3 + ], + [ + 3, + 3 + ], + [ + 3, + 3 + ], + [ + 3, + 3 + ], + [ + 3, + 3 + ] + ], + "strides": [ + [ + 1, + 1 + ], + [ + 2, + 2 + ], + [ + 2, + 2 + ], + [ + 2, + 2 + ], + [ + 2, + 2 + ], + [ + 2, + 2 + ], + [ + 2, + 2 + ] + ], + "n_blocks_per_stage": [ + 1, + 3, + 4, + 6, + 6, + 6, + 6 + ], + "n_conv_per_stage_decoder": [ + 1, + 1, + 1, + 1, + 1, + 1 + ], + "conv_bias": true, + "norm_op": "torch.nn.modules.instancenorm.InstanceNorm2d", + "norm_op_kwargs": { + "eps": 1e-05, + "affine": true + }, + "dropout_op": null, + "dropout_op_kwargs": null, + "nonlin": "torch.nn.LeakyReLU", + "nonlin_kwargs": { + "inplace": true + } + }, + "_kw_requires_import": [ + "conv_op", + "norm_op", + "dropout_op", + "nonlin" + ] + }, + "batch_dice": true + }, + "3d_fullres": { + "data_identifier": "nnUNetPlans_3d_fullres", + "preprocessor_name": "DefaultPreprocessor", + "batch_size": 2, + "patch_size": [ + 56, + 192, + 160 + ], + "median_image_size_in_voxels": [ + 89.0, + 296.0, + 279.0 + ], + "spacing": [ + 3.0, + 1.0, + 1.0 + ], + "normalization_schemes": [ + "ZScoreNormalization" + ], + "use_mask_for_norm": [ + false + ], + "resampling_fn_data": "resample_data_or_seg_to_shape", + "resampling_fn_seg": "resample_data_or_seg_to_shape", + "resampling_fn_data_kwargs": { + "is_seg": false, + "order": 3, + "order_z": 0, + "force_separate_z": null + }, + "resampling_fn_seg_kwargs": { + "is_seg": true, + "order": 1, + "order_z": 0, + "force_separate_z": null + }, + "resampling_fn_probabilities": "resample_data_or_seg_to_shape", + "resampling_fn_probabilities_kwargs": { + "is_seg": false, + "order": 1, + "order_z": 0, + "force_separate_z": null + }, + "architecture": { + "network_class_name": "dynamic_network_architectures.architectures.unet.ResidualEncoderUNet", + "arch_kwargs": { + "n_stages": 6, + "features_per_stage": [ + 32, + 64, + 128, + 256, + 320, + 320 + ], + "conv_op": "torch.nn.modules.conv.Conv3d", + "kernel_sizes": [ + [ + 1, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ] + ], + "strides": [ + [ + 1, + 1, + 1 + ], + [ + 1, + 2, + 2 + ], + [ + 2, + 2, + 2 + ], + [ + 2, + 2, + 2 + ], + [ + 2, + 2, + 2 + ], + [ + 1, + 2, + 2 + ] + ], + "n_blocks_per_stage": [ + 1, + 3, + 4, + 6, + 6, + 6 + ], + "n_conv_per_stage_decoder": [ + 1, + 1, + 1, + 1, + 1 + ], + "conv_bias": true, + "norm_op": "torch.nn.modules.instancenorm.InstanceNorm3d", + "norm_op_kwargs": { + "eps": 1e-05, + "affine": true + }, + "dropout_op": null, + "dropout_op_kwargs": null, + "nonlin": "torch.nn.LeakyReLU", + "nonlin_kwargs": { + "inplace": true + } + }, + "_kw_requires_import": [ + "conv_op", + "norm_op", + "dropout_op", + "nonlin" + ] + }, + "batch_dice": false + } + }, + "experiment_planner_used": "nnUNetPlannerResEncL", + "label_manager": "LabelManager", + "foreground_intensity_properties_per_channel": { + "0": { + "max": 4142.0, + "mean": 75.25260162353516, + "median": 0.0, + "min": 0.0, + "percentile_00_5": 0.0, + "percentile_99_5": 1222.0, + "std": 197.03660583496094 + } + } +} \ No newline at end of file diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/dataset.json b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/dataset.json new file mode 100644 index 0000000000000000000000000000000000000000..773bc7f5c8b44c6aad0585afea545abe10e1c5a8 --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/dataset.json @@ -0,0 +1,14 @@ +{ + "labels": { + "label_001": "1", + "background": 0 + }, + "channel_names": { + "0": "MR" + }, + "numTraining": 182, + "file_ending": ".nii.gz", + "test_set": [ + "case_0000" + ] +} \ No newline at end of file diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/dataset_fingerprint.json b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/dataset_fingerprint.json new file mode 100644 index 0000000000000000000000000000000000000000..a1ec9b955f82edbcdaee62c0d96e62044fcf286a --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/dataset_fingerprint.json @@ -0,0 +1,1838 @@ +{ + "foreground_intensity_properties_per_channel": { + "0": { + "max": 5250.0, + "mean": 123.740478515625, + "median": 1.0, + "min": 0.0, + "percentile_00_5": 0.0, + "percentile_99_5": 1584.0, + "std": 316.9871520996094 + } + }, + "median_relative_size_after_cropping": 1.0, + "shapes_after_crop": [ + [ + 103, + 313, + 499 + ], + [ + 97, + 360, + 492 + ], + [ + 97, + 366, + 473 + ], + [ + 100, + 376, + 495 + ], + [ + 93, + 321, + 423 + ], + [ + 96, + 355, + 467 + ], + [ + 100, + 394, + 524 + ], + [ + 103, + 356, + 509 + ], + [ + 73, + 321, + 489 + ], + [ + 99, + 336, + 451 + ], + [ + 99, + 375, + 528 + ], + [ + 105, + 310, + 467 + ], + [ + 90, + 303, + 495 + ], + [ + 99, + 333, + 505 + ], + [ + 104, + 348, + 523 + ], + [ + 100, + 412, + 455 + ], + [ + 106, + 410, + 553 + ], + [ + 102, + 404, + 556 + ], + [ + 108, + 430, + 552 + ], + [ + 96, + 409, + 554 + ], + [ + 101, + 403, + 546 + ], + [ + 105, + 417, + 559 + ], + [ + 101, + 408, + 544 + ], + [ + 104, + 406, + 555 + ], + [ + 112, + 417, + 559 + ], + [ + 104, + 405, + 548 + ], + [ + 107, + 403, + 550 + ], + [ + 107, + 401, + 553 + ], + [ + 100, + 406, + 545 + ], + [ + 100, + 404, + 556 + ], + [ + 96, + 390, + 544 + ], + [ + 103, + 401, + 548 + ], + [ + 104, + 407, + 546 + ], + [ + 100, + 404, + 544 + ], + [ + 101, + 398, + 551 + ], + [ + 92, + 479, + 486 + ], + [ + 88, + 479, + 499 + ], + [ + 93, + 512, + 535 + ], + [ + 90, + 447, + 455 + ], + [ + 93, + 492, + 494 + ], + [ + 91, + 493, + 503 + ], + [ + 99, + 489, + 483 + ], + [ + 97, + 501, + 497 + ], + [ + 106, + 481, + 477 + ], + [ + 88, + 463, + 467 + ], + [ + 80, + 553, + 547 + ], + [ + 91, + 393, + 443 + ], + [ + 95, + 474, + 471 + ], + [ + 101, + 532, + 556 + ], + [ + 71, + 461, + 461 + ], + [ + 106, + 465, + 472 + ], + [ + 106, + 480, + 487 + ], + [ + 90, + 471, + 471 + ], + [ + 83, + 475, + 469 + ], + [ + 101, + 461, + 478 + ], + [ + 93, + 458, + 458 + ], + [ + 104, + 481, + 468 + ], + [ + 106, + 467, + 465 + ], + [ + 85, + 459, + 460 + ], + [ + 92, + 473, + 465 + ], + [ + 110, + 471, + 472 + ], + [ + 102, + 444, + 467 + ], + [ + 89, + 465, + 456 + ], + [ + 109, + 461, + 483 + ], + [ + 111, + 458, + 475 + ], + [ + 89, + 479, + 467 + ], + [ + 96, + 456, + 459 + ], + [ + 103, + 468, + 465 + ], + [ + 105, + 474, + 466 + ], + [ + 92, + 462, + 462 + ], + [ + 91, + 467, + 465 + ], + [ + 102, + 473, + 481 + ], + [ + 109, + 469, + 477 + ], + [ + 90, + 471, + 485 + ], + [ + 95, + 481, + 467 + ], + [ + 93, + 477, + 477 + ], + [ + 91, + 451, + 461 + ], + [ + 105, + 459, + 464 + ], + [ + 115, + 448, + 458 + ], + [ + 104, + 470, + 477 + ], + [ + 103, + 458, + 462 + ], + [ + 78, + 441, + 458 + ], + [ + 107, + 456, + 468 + ], + [ + 104, + 467, + 471 + ], + [ + 103, + 460, + 463 + ], + [ + 93, + 461, + 460 + ], + [ + 108, + 468, + 470 + ], + [ + 96, + 447, + 470 + ], + [ + 107, + 448, + 466 + ], + [ + 104, + 479, + 461 + ], + [ + 80, + 474, + 465 + ], + [ + 152, + 520, + 547 + ], + [ + 150, + 526, + 588 + ], + [ + 118, + 518, + 587 + ], + [ + 116, + 449, + 439 + ], + [ + 106, + 516, + 585 + ], + [ + 114, + 511, + 579 + ], + [ + 127, + 506, + 571 + ], + [ + 118, + 511, + 546 + ], + [ + 126, + 517, + 541 + ], + [ + 120, + 492, + 540 + ], + [ + 142, + 507, + 549 + ], + [ + 91, + 511, + 570 + ], + [ + 150, + 515, + 577 + ], + [ + 145, + 505, + 579 + ], + [ + 128, + 533, + 549 + ], + [ + 122, + 508, + 583 + ], + [ + 134, + 378, + 378 + ], + [ + 123, + 514, + 546 + ], + [ + 121, + 491, + 546 + ], + [ + 133, + 506, + 549 + ], + [ + 123, + 537, + 592 + ], + [ + 118, + 527, + 590 + ], + [ + 115, + 489, + 527 + ], + [ + 116, + 489, + 538 + ], + [ + 113, + 526, + 547 + ], + [ + 145, + 500, + 538 + ], + [ + 91, + 485, + 496 + ], + [ + 120, + 496, + 544 + ], + [ + 136, + 489, + 545 + ], + [ + 119, + 490, + 542 + ], + [ + 151, + 520, + 587 + ], + [ + 142, + 527, + 592 + ], + [ + 128, + 515, + 569 + ], + [ + 115, + 508, + 549 + ], + [ + 145, + 520, + 547 + ], + [ + 127, + 491, + 537 + ], + [ + 131, + 463, + 454 + ], + [ + 116, + 507, + 556 + ], + [ + 138, + 534, + 549 + ], + [ + 153, + 546, + 591 + ], + [ + 143, + 526, + 546 + ], + [ + 115, + 500, + 545 + ], + [ + 132, + 486, + 499 + ], + [ + 133, + 521, + 531 + ], + [ + 121, + 507, + 543 + ], + [ + 133, + 440, + 443 + ], + [ + 121, + 444, + 441 + ], + [ + 116, + 512, + 584 + ], + [ + 120, + 447, + 449 + ], + [ + 128, + 448, + 448 + ], + [ + 121, + 510, + 543 + ], + [ + 115, + 515, + 605 + ], + [ + 142, + 506, + 533 + ], + [ + 102, + 500, + 540 + ], + [ + 154, + 503, + 539 + ], + [ + 126, + 512, + 586 + ], + [ + 123, + 501, + 550 + ], + [ + 140, + 506, + 547 + ], + [ + 95, + 501, + 542 + ], + [ + 136, + 512, + 541 + ], + [ + 97, + 524, + 545 + ], + [ + 146, + 518, + 590 + ], + [ + 108, + 442, + 440 + ], + [ + 123, + 400, + 390 + ], + [ + 71, + 500, + 499 + ], + [ + 159, + 402, + 396 + ], + [ + 134, + 527, + 587 + ], + [ + 144, + 493, + 561 + ], + [ + 120, + 496, + 544 + ], + [ + 150, + 506, + 541 + ], + [ + 145, + 520, + 594 + ], + [ + 118, + 459, + 458 + ], + [ + 108, + 399, + 400 + ], + [ + 119, + 421, + 421 + ], + [ + 131, + 497, + 546 + ], + [ + 121, + 496, + 552 + ], + [ + 108, + 509, + 579 + ], + [ + 141, + 374, + 572 + ], + [ + 115, + 508, + 580 + ], + [ + 103, + 518, + 596 + ], + [ + 131, + 512, + 583 + ], + [ + 118, + 490, + 542 + ], + [ + 117, + 502, + 545 + ], + [ + 125, + 498, + 547 + ], + [ + 126, + 519, + 576 + ], + [ + 128, + 445, + 446 + ], + [ + 126, + 527, + 591 + ], + [ + 115, + 525, + 594 + ], + [ + 150, + 507, + 590 + ], + [ + 109, + 508, + 585 + ], + [ + 125, + 444, + 450 + ] + ], + "spacings": [ + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ], + [ + 3.0, + 1.0, + 1.0 + ] + ] +} \ No newline at end of file diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/checkpoint_final.pth b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/checkpoint_final.pth new file mode 100644 index 0000000000000000000000000000000000000000..ada97968d274ba1844b42cffec5f3c00bb977dcf --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/checkpoint_final.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f0c0045dc35f9b1a552f45255d20da86a1cb481a3b175dd8ef6d72b73529cf66 +size 815341933 diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/debug.json b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/debug.json new file mode 100644 index 0000000000000000000000000000000000000000..fb149f4bde8090eaa7e2691c9e66003d9215ae73 --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/debug.json @@ -0,0 +1,57 @@ +{ + "_best_ema": "None", + "aim_run": ">", + "batch_size": "2", + "configuration_manager": "{'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [107.5, 476.0, 536.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}, 'deep_supervision': False}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}", + "configuration_name": "3d_fullres", + "cudnn_version": 90501, + "current_epoch": "0", + "dataloader_train": "", + "dataloader_train.generator": "", + "dataloader_train.num_processes": "12", + "dataloader_train.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataloader_val": "", + "dataloader_val.generator": "", + "dataloader_val.num_processes": "6", + "dataloader_val.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataset_json": "{'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 182, 'file_ending': '.mha'}", + "debug_custom": "{'run_hash': '8416c427d4594a37bd4e01bc'}", + "decoder_type": "standard", + "device": "cuda:0", + "disable_checkpointing": "False", + "enable_deep_supervision": "False", + "fold": "0", + "folder_with_segs_from_previous_stage": "None", + "gpu_name": "NVIDIA H100", + "grad_scaler": "", + "hostname": "g039", + "inference_allowed_mirroring_axes": "None", + "initial_lr": "0.01", + "is_cascaded": "False", + "is_ddp": "False", + "label_manager": "", + "local_rank": "0", + "log_file": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/training_log_2025_8_11_11_13_37.txt", + "logger": "", + "loss": "SynPerceptionLoss(\n (seg_model): ResidualEncoderUNet(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (decoder): UNetDecoder(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (stages): ModuleList(\n (0): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (1): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (2): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (3): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (4): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (transpconvs): ModuleList(\n (0): ConvTranspose3d(320, 320, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n (1): ConvTranspose3d(320, 256, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (2): ConvTranspose3d(256, 128, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (3): ConvTranspose3d(128, 64, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (4): ConvTranspose3d(64, 32, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n )\n (seg_layers): ModuleList(\n (0): Conv3d(320, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (1): Conv3d(256, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (2): Conv3d(128, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (3): Conv3d(64, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (4): Conv3d(32, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n )\n )\n )\n (L1): L1Loss()\n (image_loss): myMaskedMSE(\n (mse): myMSE()\n )\n)", + "lr_scheduler": "", + "my_init_kwargs": "{'plans': {'dataset_name': 'Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [108, 476, 536], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [476.0, 536.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [98, 288, 324], 'spacing': [3.278181, 1.652847632271752, 1.652847632271752], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [107.5, 476.0, 536.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 5250.0, 'mean': 123.740478515625, 'median': 1.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1584.0, 'std': 316.9871520996094}}}, 'configuration': '3d_fullres', 'fold': 0, 'dataset_json': {'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 182, 'file_ending': '.mha'}, 'unpack_dataset': True, 'device': device(type='cuda')}", + "network": "ResidualEncoderUNet", + "num_epochs": "1000", + "num_input_channels": "1", + "num_iterations_per_epoch": "250", + "num_val_iterations_per_epoch": "50", + "optimizer": "SGD (\nParameter Group 0\n dampening: 0\n differentiable: False\n foreach: None\n fused: None\n initial_lr: 0.01\n lr: 0.01\n maximize: False\n momentum: 0.99\n nesterov: True\n weight_decay: 3e-05\n)", + "output_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0", + "output_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres", + "oversample_foreground_percent": "0.33", + "perception_masked": "True", + "plans_manager": "{'dataset_name': 'Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [108, 476, 536], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [476.0, 536.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [98, 288, 324], 'spacing': [3.278181, 1.652847632271752, 1.652847632271752], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [107.5, 476.0, 536.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 5250.0, 'mean': 123.740478515625, 'median': 1.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1584.0, 'std': 316.9871520996094}}}", + "preprocessed_dataset_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetPlans_3d_fullres", + "preprocessed_dataset_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked", + "save_every": "50", + "torch_version": "2.7.1+cu126", + "unpack_dataset": "True", + "was_initialized": "True", + "weight_decay": "3e-05" +} \ No newline at end of file diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/progress.png b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/progress.png new file mode 100644 index 0000000000000000000000000000000000000000..7924d216fedf095fd0cdb3f886fab4f8717673d4 --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/progress.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0790c3c5a5de2d4cc5588f2ec072146e839e16b8def36d0bc251a1bd33b19c28 +size 582616 diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/validation/summary.json b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/validation/summary.json new file mode 100644 index 0000000000000000000000000000000000000000..53ef20eef69b8e731c28bb1b3f1ecef3a5ba3f0a --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_0/validation/summary.json @@ -0,0 +1,52 @@ +{ + "mae": { + "mean": 65.54925162984547, + "std": 22.711604553199436, + "max": 149.49946940572337, + "min": 38.819947088801925, + "25pc": 50.36406383246631, + "50pc": 60.48019173397385, + "75pc": 71.47043812610828, + "count": 37 + }, + "psnr": { + "mean": 29.35849474712559, + "std": 2.786161453159081, + "max": 33.9541516841314, + "min": 21.939775479876317, + "25pc": 27.72671622823851, + "50pc": 29.678033989984232, + "75pc": 31.58941879973888, + "count": 37 + }, + "ms_ssim": { + "mean": 0.919854566652931, + "std": 0.06330094671092198, + "max": 0.9771491893960651, + "min": 0.672010500472794, + "25pc": 0.9086998715256039, + "50pc": 0.9384153070912361, + "75pc": 0.966504391562742, + "count": 37 + }, + "DICE": { + "mean": 0.7462090616290634, + "std": 0.11164662261476627, + "max": 0.8656862378120422, + "min": 0.2728056311607361, + "25pc": 0.709774911403656, + "50pc": 0.7784290909767151, + "75pc": 0.8168733716011047, + "count": 37 + }, + "HD95": { + "mean": 4.027943765794909, + "std": 2.293875912076277, + "max": 10.72010612487793, + "min": 1.8405364751815796, + "25pc": 2.515124559402466, + "50pc": 3.436922550201416, + "75pc": 4.268289566040039, + "count": 37 + } +} \ No newline at end of file diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/checkpoint_final.pth b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/checkpoint_final.pth new file mode 100644 index 0000000000000000000000000000000000000000..7d01b34b7f8870f7334783438a25b7398c583bb5 --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/checkpoint_final.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a72cf7c1f030b10eb0f2eeed2251e976f559f81b663f412aafc073be9bcc30b5 +size 815580013 diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/debug.json b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/debug.json new file mode 100644 index 0000000000000000000000000000000000000000..474bc397567de08e2ccfcc1096b90035dfe76943 --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/debug.json @@ -0,0 +1,58 @@ +{ + "_best_ema": "None", + "aim_run": ">", + "batch_size": "2", + "configuration_manager": "{'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [107.5, 476.0, 536.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}, 'deep_supervision': False}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}", + "configuration_name": "3d_fullres", + "cudnn_version": 90501, + "current_epoch": "0", + "dataloader_train": "", + "dataloader_train.generator": "", + "dataloader_train.num_processes": "12", + "dataloader_train.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataloader_val": "", + "dataloader_val.generator": "", + "dataloader_val.num_processes": "6", + "dataloader_val.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataset_json": "{'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 182, 'file_ending': '.mha'}", + "debug_custom": "{'run_hash': '604ebfdd8ed74c9c88c7839a'}", + "decoder_type": "standard", + "device": "cuda:0", + "disable_checkpointing": "False", + "enable_deep_supervision": "False", + "fold": "1", + "folder_with_segs_from_previous_stage": "None", + "gpu_name": "NVIDIA H100", + "grad_scaler": "", + "hostname": "g032", + "image_loss_weight": "0.5", + "inference_allowed_mirroring_axes": "None", + "initial_lr": "0.01", + "is_cascaded": "False", + "is_ddp": "False", + "label_manager": "", + "local_rank": "0", + "log_file": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/training_log_2025_8_13_01_37_53.txt", + "logger": "", + "loss": "SynPerceptionLoss(\n (seg_model): ResidualEncoderUNet(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (decoder): UNetDecoder(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (stages): ModuleList(\n (0): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (1): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (2): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (3): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (4): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (transpconvs): ModuleList(\n (0): ConvTranspose3d(320, 320, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n (1): ConvTranspose3d(320, 256, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (2): ConvTranspose3d(256, 128, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (3): ConvTranspose3d(128, 64, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (4): ConvTranspose3d(64, 32, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n )\n (seg_layers): ModuleList(\n (0): Conv3d(320, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (1): Conv3d(256, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (2): Conv3d(128, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (3): Conv3d(64, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (4): Conv3d(32, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n )\n )\n )\n (L1): L1Loss()\n (image_loss): myMaskedMSE(\n (mse): myMSE()\n )\n)", + "lr_scheduler": "", + "my_init_kwargs": "{'plans': {'dataset_name': 'Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [108, 476, 536], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [476.0, 536.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [98, 288, 324], 'spacing': [3.278181, 1.652847632271752, 1.652847632271752], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [107.5, 476.0, 536.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 5250.0, 'mean': 123.740478515625, 'median': 1.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1584.0, 'std': 316.9871520996094}}}, 'configuration': '3d_fullres', 'fold': 1, 'dataset_json': {'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 182, 'file_ending': '.mha'}, 'unpack_dataset': True, 'device': device(type='cuda')}", + "network": "ResidualEncoderUNet", + "num_epochs": "1000", + "num_input_channels": "1", + "num_iterations_per_epoch": "250", + "num_val_iterations_per_epoch": "50", + "optimizer": "SGD (\nParameter Group 0\n dampening: 0\n differentiable: False\n foreach: None\n fused: None\n initial_lr: 0.01\n lr: 0.01\n maximize: False\n momentum: 0.99\n nesterov: True\n weight_decay: 3e-05\n)", + "output_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1", + "output_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres", + "oversample_foreground_percent": "0.33", + "perception_masked": "True", + "plans_manager": "{'dataset_name': 'Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [108, 476, 536], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [476.0, 536.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [98, 288, 324], 'spacing': [3.278181, 1.652847632271752, 1.652847632271752], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [107.5, 476.0, 536.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 5250.0, 'mean': 123.740478515625, 'median': 1.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1584.0, 'std': 316.9871520996094}}}", + "preprocessed_dataset_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetPlans_3d_fullres", + "preprocessed_dataset_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked", + "save_every": "50", + "torch_version": "2.7.1+cu126", + "unpack_dataset": "True", + "was_initialized": "True", + "weight_decay": "3e-05" +} \ No newline at end of file diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/progress.png b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/progress.png new file mode 100644 index 0000000000000000000000000000000000000000..d4d984b34eef0ea64306274e21ffffce5ae0adf9 --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/progress.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5a8489f95a78f8f8d0afea6bcc33dcdcb83bfdbbc7bbc0edc62de4e0abf5e013 +size 586663 diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/validation/summary.json b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/validation/summary.json new file mode 100644 index 0000000000000000000000000000000000000000..c38e186f80bf3eb7caca425eb6e327fecf15cd3f --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_1/validation/summary.json @@ -0,0 +1,52 @@ +{ + "mae": { + "mean": 68.42680998438118, + "std": 29.296221265690853, + "max": 159.9648282446066, + "min": 39.51275675792069, + "25pc": 49.07730820733404, + "50pc": 58.01609527537543, + "75pc": 72.2934481798172, + "count": 37 + }, + "psnr": { + "mean": 29.2831915868661, + "std": 3.1145729083930394, + "max": 34.03873729293684, + "min": 22.60916865803456, + "25pc": 27.5709519599854, + "50pc": 30.123124065186477, + "75pc": 31.815726939773683, + "count": 37 + }, + "ms_ssim": { + "mean": 0.9183165510203545, + "std": 0.06999156156634556, + "max": 0.981221500586083, + "min": 0.717303776111277, + "25pc": 0.8976345151432202, + "50pc": 0.9396216703209193, + "75pc": 0.9634983997081166, + "count": 37 + }, + "DICE": { + "mean": 0.7525587774611808, + "std": 0.10446370560565381, + "max": 0.864150881767273, + "min": 0.3753151297569275, + "25pc": 0.7059541344642639, + "50pc": 0.7906670570373535, + "75pc": 0.8190597295761108, + "count": 37 + }, + "HD95": { + "mean": 3.7715794685724617, + "std": 1.8167927346559494, + "max": 9.583288192749023, + "min": 1.7149158716201782, + "25pc": 2.5365164279937744, + "50pc": 3.0820226669311523, + "75pc": 4.5893988609313965, + "count": 37 + } +} \ No newline at end of file diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/checkpoint_final.pth b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/checkpoint_final.pth new file mode 100644 index 0000000000000000000000000000000000000000..b2e08ec7a73c22cc56875f2ef6bff1cbd00c7539 --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/checkpoint_final.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fb55a05d0457d6d2215f8a958917fa299ec567d756564fb2f538ef61673a767c +size 815579501 diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/debug.json b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/debug.json new file mode 100644 index 0000000000000000000000000000000000000000..6077eea194cf09e0641fa40d1e3e6d02af45f5e6 --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/debug.json @@ -0,0 +1,58 @@ +{ + "_best_ema": "None", + "aim_run": ">", + "batch_size": "2", + "configuration_manager": "{'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [107.5, 476.0, 536.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}, 'deep_supervision': False}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}", + "configuration_name": "3d_fullres", + "cudnn_version": 90501, + "current_epoch": "0", + "dataloader_train": "", + "dataloader_train.generator": "", + "dataloader_train.num_processes": "12", + "dataloader_train.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataloader_val": "", + "dataloader_val.generator": "", + "dataloader_val.num_processes": "6", + "dataloader_val.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataset_json": "{'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 182, 'file_ending': '.mha'}", + "debug_custom": "{'run_hash': '4803481847cb4ca9bd0d8acb'}", + "decoder_type": "standard", + "device": "cuda:0", + "disable_checkpointing": "False", + "enable_deep_supervision": "False", + "fold": "2", + "folder_with_segs_from_previous_stage": "None", + "gpu_name": "NVIDIA H100", + "grad_scaler": "", + "hostname": "g036", + "image_loss_weight": "0.5", + "inference_allowed_mirroring_axes": "None", + "initial_lr": "0.01", + "is_cascaded": "False", + "is_ddp": "False", + "label_manager": "", + "local_rank": "0", + "log_file": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/training_log_2025_8_13_01_38_33.txt", + "logger": "", + "loss": "SynPerceptionLoss(\n (seg_model): ResidualEncoderUNet(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (decoder): UNetDecoder(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (stages): ModuleList(\n (0): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (1): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (2): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (3): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (4): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (transpconvs): ModuleList(\n (0): ConvTranspose3d(320, 320, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n (1): ConvTranspose3d(320, 256, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (2): ConvTranspose3d(256, 128, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (3): ConvTranspose3d(128, 64, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (4): ConvTranspose3d(64, 32, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n )\n (seg_layers): ModuleList(\n (0): Conv3d(320, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (1): Conv3d(256, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (2): Conv3d(128, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (3): Conv3d(64, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (4): Conv3d(32, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n )\n )\n )\n (L1): L1Loss()\n (image_loss): myMaskedMSE(\n (mse): myMSE()\n )\n)", + "lr_scheduler": "", + "my_init_kwargs": "{'plans': {'dataset_name': 'Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [108, 476, 536], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [476.0, 536.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [98, 288, 324], 'spacing': [3.278181, 1.652847632271752, 1.652847632271752], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [107.5, 476.0, 536.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 5250.0, 'mean': 123.740478515625, 'median': 1.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1584.0, 'std': 316.9871520996094}}}, 'configuration': '3d_fullres', 'fold': 2, 'dataset_json': {'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 182, 'file_ending': '.mha'}, 'unpack_dataset': True, 'device': device(type='cuda')}", + "network": "ResidualEncoderUNet", + "num_epochs": "1000", + "num_input_channels": "1", + "num_iterations_per_epoch": "250", + "num_val_iterations_per_epoch": "50", + "optimizer": "SGD (\nParameter Group 0\n dampening: 0\n differentiable: False\n foreach: None\n fused: None\n initial_lr: 0.01\n lr: 0.01\n maximize: False\n momentum: 0.99\n nesterov: True\n weight_decay: 3e-05\n)", + "output_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2", + "output_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres", + "oversample_foreground_percent": "0.33", + "perception_masked": "True", + "plans_manager": "{'dataset_name': 'Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [108, 476, 536], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [476.0, 536.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [98, 288, 324], 'spacing': [3.278181, 1.652847632271752, 1.652847632271752], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [107.5, 476.0, 536.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 5250.0, 'mean': 123.740478515625, 'median': 1.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1584.0, 'std': 316.9871520996094}}}", + "preprocessed_dataset_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetPlans_3d_fullres", + "preprocessed_dataset_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked", + "save_every": "50", + "torch_version": "2.7.1+cu126", + "unpack_dataset": "True", + "was_initialized": "True", + "weight_decay": "3e-05" +} \ No newline at end of file diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/progress.png b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/progress.png new file mode 100644 index 0000000000000000000000000000000000000000..cf67480ce1bdc97d826e826db1b9c849f84ec44f --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/progress.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c2e553ac72c399e1c9d23e64d08ca40816945b54c0030b2098445ca1231669a0 +size 549268 diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/validation/summary.json b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/validation/summary.json new file mode 100644 index 0000000000000000000000000000000000000000..09fd842af992bfffacc4e3800fd24254376b3d97 --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_2/validation/summary.json @@ -0,0 +1,52 @@ +{ + "mae": { + "mean": 63.8707099520243, + "std": 29.344427486480047, + "max": 198.07948562127422, + "min": 33.32851431796214, + "25pc": 46.581016657823035, + "50pc": 57.87843714114535, + "75pc": 70.05121648091296, + "count": 36 + }, + "psnr": { + "mean": 29.68457783171023, + "std": 2.971449374801366, + "max": 35.45539921689221, + "min": 20.4766768270116, + "25pc": 28.23257943723521, + "50pc": 29.8280142189061, + "75pc": 31.727413984889985, + "count": 36 + }, + "ms_ssim": { + "mean": 0.9198675013587881, + "std": 0.07759512869660178, + "max": 0.9826847628438435, + "min": 0.5727009206661299, + "25pc": 0.9114057024453344, + "50pc": 0.9407359314147499, + "75pc": 0.9630996771625975, + "count": 36 + }, + "DICE": { + "mean": 0.7543800191746818, + "std": 0.09705718003692082, + "max": 0.8731817603111267, + "min": 0.43767493963241577, + "25pc": 0.73861064016819, + "50pc": 0.7844387590885162, + "75pc": 0.8136635422706604, + "count": 36 + }, + "HD95": { + "mean": 3.4939137597878775, + "std": 1.1462497679842092, + "max": 6.230932235717773, + "min": 1.5738335847854614, + "25pc": 2.6403597593307495, + "50pc": 3.133875846862793, + "75pc": 4.266846299171448, + "count": 36 + } +} \ No newline at end of file diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/checkpoint_final.pth b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/checkpoint_final.pth new file mode 100644 index 0000000000000000000000000000000000000000..0f0c113a075df189d162b379c1f5432d17cd90ed --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/checkpoint_final.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e45131331b0060dce7601493ffd44e4a167a31194ba57fb06ce5a73385fba42b +size 815579565 diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/debug.json b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/debug.json new file mode 100644 index 0000000000000000000000000000000000000000..c34df15af83a4730b38fd7499320d29644cbc10d --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/debug.json @@ -0,0 +1,58 @@ +{ + "_best_ema": "None", + "aim_run": ">", + "batch_size": "2", + "configuration_manager": "{'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [107.5, 476.0, 536.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}, 'deep_supervision': False}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}", + "configuration_name": "3d_fullres", + "cudnn_version": 90501, + "current_epoch": "0", + "dataloader_train": "", + "dataloader_train.generator": "", + "dataloader_train.num_processes": "12", + "dataloader_train.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataloader_val": "", + "dataloader_val.generator": "", + "dataloader_val.num_processes": "6", + "dataloader_val.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataset_json": "{'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 182, 'file_ending': '.mha'}", + "debug_custom": "{'run_hash': 'cd9c881b35db4318bd8569ce'}", + "decoder_type": "standard", + "device": "cuda:0", + "disable_checkpointing": "False", + "enable_deep_supervision": "False", + "fold": "3", + "folder_with_segs_from_previous_stage": "None", + "gpu_name": "NVIDIA H100", + "grad_scaler": "", + "hostname": "g037", + "image_loss_weight": "0.5", + "inference_allowed_mirroring_axes": "None", + "initial_lr": "0.01", + "is_cascaded": "False", + "is_ddp": "False", + "label_manager": "", + "local_rank": "0", + "log_file": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/training_log_2025_8_13_01_38_33.txt", + "logger": "", + "loss": "SynPerceptionLoss(\n (seg_model): ResidualEncoderUNet(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (decoder): UNetDecoder(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (stages): ModuleList(\n (0): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (1): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (2): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (3): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (4): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (transpconvs): ModuleList(\n (0): ConvTranspose3d(320, 320, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n (1): ConvTranspose3d(320, 256, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (2): ConvTranspose3d(256, 128, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (3): ConvTranspose3d(128, 64, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (4): ConvTranspose3d(64, 32, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n )\n (seg_layers): ModuleList(\n (0): Conv3d(320, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (1): Conv3d(256, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (2): Conv3d(128, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (3): Conv3d(64, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (4): Conv3d(32, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n )\n )\n )\n (L1): L1Loss()\n (image_loss): myMaskedMSE(\n (mse): myMSE()\n )\n)", + "lr_scheduler": "", + "my_init_kwargs": "{'plans': {'dataset_name': 'Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [108, 476, 536], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [476.0, 536.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [98, 288, 324], 'spacing': [3.278181, 1.652847632271752, 1.652847632271752], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [107.5, 476.0, 536.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 5250.0, 'mean': 123.740478515625, 'median': 1.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1584.0, 'std': 316.9871520996094}}}, 'configuration': '3d_fullres', 'fold': 3, 'dataset_json': {'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 182, 'file_ending': '.mha'}, 'unpack_dataset': True, 'device': device(type='cuda')}", + "network": "ResidualEncoderUNet", + "num_epochs": "1000", + "num_input_channels": "1", + "num_iterations_per_epoch": "250", + "num_val_iterations_per_epoch": "50", + "optimizer": "SGD (\nParameter Group 0\n dampening: 0\n differentiable: False\n foreach: None\n fused: None\n initial_lr: 0.01\n lr: 0.01\n maximize: False\n momentum: 0.99\n nesterov: True\n weight_decay: 3e-05\n)", + "output_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3", + "output_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres", + "oversample_foreground_percent": "0.33", + "perception_masked": "True", + "plans_manager": "{'dataset_name': 'Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [108, 476, 536], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [476.0, 536.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [98, 288, 324], 'spacing': [3.278181, 1.652847632271752, 1.652847632271752], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [107.5, 476.0, 536.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 5250.0, 'mean': 123.740478515625, 'median': 1.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1584.0, 'std': 316.9871520996094}}}", + "preprocessed_dataset_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetPlans_3d_fullres", + "preprocessed_dataset_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked", + "save_every": "50", + "torch_version": "2.7.1+cu126", + "unpack_dataset": "True", + "was_initialized": "True", + "weight_decay": "3e-05" +} \ No newline at end of file diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/progress.png b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/progress.png new file mode 100644 index 0000000000000000000000000000000000000000..cbb8093cc4e8b1d8b3ed7d812a6c017a557b959d --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/progress.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:da85269076f3d4e5b51b7e9322f2e4d6d12afe6769cf40f6d54ef614204d9b77 +size 531875 diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/validation/summary.json b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/validation/summary.json new file mode 100644 index 0000000000000000000000000000000000000000..62c982bd62f66d07be6dc94a474a4d95166a4c64 --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_3/validation/summary.json @@ -0,0 +1,52 @@ +{ + "mae": { + "mean": 63.31832781524968, + "std": 24.15074190662186, + "max": 133.6028552469902, + "min": 34.86351526159813, + "25pc": 47.173206561319105, + "50pc": 55.72438918455102, + "75pc": 66.03616182068274, + "count": 36 + }, + "psnr": { + "mean": 29.6443801409689, + "std": 3.0305464517352876, + "max": 35.61330928880079, + "min": 23.072966783681615, + "25pc": 28.062260880883578, + "50pc": 29.88319523698734, + "75pc": 31.671291149802208, + "count": 36 + }, + "ms_ssim": { + "mean": 0.9239241422498191, + "std": 0.058300796420574195, + "max": 0.9828489260707507, + "min": 0.7412400545439081, + "25pc": 0.9162496387426491, + "50pc": 0.9407468017112399, + "75pc": 0.9615914052039394, + "count": 36 + }, + "DICE": { + "mean": 0.7821378691328896, + "std": 0.0704044486988359, + "max": 0.8884629607200623, + "min": 0.5152435898780823, + "25pc": 0.7445724755525589, + "50pc": 0.7949909865856171, + "75pc": 0.8273429274559021, + "count": 36 + }, + "HD95": { + "mean": 3.9563661648167505, + "std": 3.494689735389124, + "max": 22.816848754882812, + "min": 1.5855196714401245, + "25pc": 2.407499074935913, + "50pc": 2.9286502599716187, + "75pc": 4.124189734458923, + "count": 36 + } +} \ No newline at end of file diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/checkpoint_final.pth b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/checkpoint_final.pth new file mode 100644 index 0000000000000000000000000000000000000000..d1251b07c9e2822424126cf70b28d7f8812c289a --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/checkpoint_final.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c83391668633b7eae001a720d7005dfce7125d25e9b696e62f8b1a46bd4ecbbb +size 815579309 diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/debug.json b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/debug.json new file mode 100644 index 0000000000000000000000000000000000000000..88528e0534b32885b2bec712926bc7af60848af9 --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/debug.json @@ -0,0 +1,58 @@ +{ + "_best_ema": "None", + "aim_run": ">", + "batch_size": "2", + "configuration_manager": "{'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [107.5, 476.0, 536.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}, 'deep_supervision': False}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}", + "configuration_name": "3d_fullres", + "cudnn_version": 90501, + "current_epoch": "0", + "dataloader_train": "", + "dataloader_train.generator": "", + "dataloader_train.num_processes": "12", + "dataloader_train.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataloader_val": "", + "dataloader_val.generator": "", + "dataloader_val.num_processes": "6", + "dataloader_val.transform": "Compose ( [RemoveLabelTransform( output_key = 'seg', input_key = 'seg', replace_with = 0, remove_label = -1 ), RenameTransform( delete_old = True, out_key = 'target', in_key = 'seg' ), NumpyToTensor( keys = ['data', 'target', 'mask'], cast_to = 'float' )] )", + "dataset_json": "{'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 182, 'file_ending': '.mha'}", + "debug_custom": "{'run_hash': 'd4328565aa884546af60571b'}", + "decoder_type": "standard", + "device": "cuda:0", + "disable_checkpointing": "False", + "enable_deep_supervision": "False", + "fold": "4", + "folder_with_segs_from_previous_stage": "None", + "gpu_name": "NVIDIA H100", + "grad_scaler": "", + "hostname": "g037", + "image_loss_weight": "0.5", + "inference_allowed_mirroring_axes": "None", + "initial_lr": "0.01", + "is_cascaded": "False", + "is_ddp": "False", + "label_manager": "", + "local_rank": "0", + "log_file": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/training_log_2025_8_13_01_38_33.txt", + "logger": "", + "loss": "SynPerceptionLoss(\n (seg_model): ResidualEncoderUNet(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (decoder): UNetDecoder(\n (encoder): ResidualEncoder(\n (stem): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(1, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (stages): Sequential(\n (0): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (1): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(32, 64, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (2): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(64, 128, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (3): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(128, 256, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (4): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(3, 3, 3), stride=(2, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[2, 2, 2], stride=[2, 2, 2], padding=0)\n (1): ConvDropoutNormReLU(\n (conv): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(256, 320, kernel_size=(1, 1, 1), stride=(1, 1, 1), bias=False)\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n (5): StackedResidualBlocks(\n (blocks): Sequential(\n (0): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 2, 2), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n (skip): Sequential(\n (0): AvgPool3d(kernel_size=[1, 2, 2], stride=[1, 2, 2], padding=0)\n )\n )\n (1): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (2): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (3): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (4): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n (5): BasicBlockD(\n (conv1): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n (conv2): ConvDropoutNormReLU(\n (conv): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (all_modules): Sequential(\n (0): Conv3d(320, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n )\n )\n (nonlin2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (stages): ModuleList(\n (0): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(640, 320, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (1): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(512, 256, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (2): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(256, 128, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (3): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (norm): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(128, 64, kernel_size=(3, 3, 3), stride=(1, 1, 1), padding=(1, 1, 1))\n (1): InstanceNorm3d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n (4): StackedConvBlocks(\n (convs): Sequential(\n (0): ConvDropoutNormReLU(\n (conv): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (norm): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (nonlin): LeakyReLU(negative_slope=0.01, inplace=True)\n (all_modules): Sequential(\n (0): Conv3d(64, 32, kernel_size=(1, 3, 3), stride=(1, 1, 1), padding=(0, 1, 1))\n (1): InstanceNorm3d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=False)\n (2): LeakyReLU(negative_slope=0.01, inplace=True)\n )\n )\n )\n )\n )\n (transpconvs): ModuleList(\n (0): ConvTranspose3d(320, 320, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n (1): ConvTranspose3d(320, 256, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (2): ConvTranspose3d(256, 128, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (3): ConvTranspose3d(128, 64, kernel_size=(2, 2, 2), stride=(2, 2, 2))\n (4): ConvTranspose3d(64, 32, kernel_size=(1, 2, 2), stride=(1, 2, 2))\n )\n (seg_layers): ModuleList(\n (0): Conv3d(320, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (1): Conv3d(256, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (2): Conv3d(128, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (3): Conv3d(64, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n (4): Conv3d(32, 62, kernel_size=(1, 1, 1), stride=(1, 1, 1))\n )\n )\n )\n (L1): L1Loss()\n (image_loss): myMaskedMSE(\n (mse): myMSE()\n )\n)", + "lr_scheduler": "", + "my_init_kwargs": "{'plans': {'dataset_name': 'Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [108, 476, 536], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [476.0, 536.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [98, 288, 324], 'spacing': [3.278181, 1.652847632271752, 1.652847632271752], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [107.5, 476.0, 536.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 5250.0, 'mean': 123.740478515625, 'median': 1.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1584.0, 'std': 316.9871520996094}}}, 'configuration': '3d_fullres', 'fold': 4, 'dataset_json': {'labels': {'label_001': '1', 'background': 0}, 'channel_names': {'0': 'MR'}, 'numTraining': 182, 'file_ending': '.mha'}, 'unpack_dataset': True, 'device': device(type='cuda')}", + "network": "ResidualEncoderUNet", + "num_epochs": "1000", + "num_input_channels": "1", + "num_iterations_per_epoch": "250", + "num_val_iterations_per_epoch": "50", + "optimizer": "SGD (\nParameter Group 0\n dampening: 0\n differentiable: False\n foreach: None\n fused: None\n initial_lr: 0.01\n lr: 0.01\n maximize: False\n momentum: 0.99\n nesterov: True\n weight_decay: 3e-05\n)", + "output_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4", + "output_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/results/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres", + "oversample_foreground_percent": "0.33", + "perception_masked": "True", + "plans_manager": "{'dataset_name': 'Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked', 'plans_name': 'nnUNetResEncUNetLPlans', 'original_median_spacing_after_transp': [3.0, 1.0, 1.0], 'original_median_shape_after_transp': [108, 476, 536], 'image_reader_writer': 'SimpleITKIO', 'transpose_forward': [0, 1, 2], 'transpose_backward': [0, 1, 2], 'configurations': {'2d': {'data_identifier': 'nnUNetPlans_2d', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 13, 'patch_size': [448, 512], 'median_image_size_in_voxels': [476.0, 536.0], 'spacing': [1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 7, 'features_per_stage': [32, 64, 128, 256, 512, 512, 512], 'conv_op': 'torch.nn.modules.conv.Conv2d', 'kernel_sizes': [[3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3], [3, 3]], 'strides': [[1, 1], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2], [2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm2d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_lowres': {'data_identifier': 'nnUNetResEncUNetLPlans_3d_lowres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [64, 192, 192], 'median_image_size_in_voxels': [98, 288, 324], 'spacing': [3.278181, 1.652847632271752, 1.652847632271752], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [2, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': False, 'next_stage': '3d_cascade_fullres'}, '3d_fullres': {'data_identifier': 'nnUNetPlans_3d_fullres', 'preprocessor_name': 'DefaultPreprocessor', 'batch_size': 2, 'patch_size': [40, 192, 224], 'median_image_size_in_voxels': [107.5, 476.0, 536.0], 'spacing': [3.0, 1.0, 1.0], 'normalization_schemes': ['ZScoreNormalization'], 'use_mask_for_norm': [False], 'resampling_fn_data': 'resample_data_or_seg_to_shape', 'resampling_fn_seg': 'resample_data_or_seg_to_shape', 'resampling_fn_data_kwargs': {'is_seg': False, 'order': 3, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_seg_kwargs': {'is_seg': True, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'resampling_fn_probabilities': 'resample_data_or_seg_to_shape', 'resampling_fn_probabilities_kwargs': {'is_seg': False, 'order': 1, 'order_z': 0, 'force_separate_z': None}, 'architecture': {'network_class_name': 'dynamic_network_architectures.architectures.unet.ResidualEncoderUNet', 'arch_kwargs': {'n_stages': 6, 'features_per_stage': [32, 64, 128, 256, 320, 320], 'conv_op': 'torch.nn.modules.conv.Conv3d', 'kernel_sizes': [[1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]], 'strides': [[1, 1, 1], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2], [1, 2, 2]], 'n_blocks_per_stage': [1, 3, 4, 6, 6, 6], 'n_conv_per_stage_decoder': [1, 1, 1, 1, 1], 'conv_bias': True, 'norm_op': 'torch.nn.modules.instancenorm.InstanceNorm3d', 'norm_op_kwargs': {'eps': 1e-05, 'affine': True}, 'dropout_op': None, 'dropout_op_kwargs': None, 'nonlin': 'torch.nn.LeakyReLU', 'nonlin_kwargs': {'inplace': True}}, '_kw_requires_import': ['conv_op', 'norm_op', 'dropout_op', 'nonlin']}, 'batch_dice': True}, '3d_cascade_fullres': {'inherits_from': '3d_fullres', 'previous_stage': '3d_lowres'}}, 'experiment_planner_used': 'nnUNetPlannerResEncL', 'label_manager': 'LabelManager', 'foreground_intensity_properties_per_channel': {'0': {'max': 5250.0, 'mean': 123.740478515625, 'median': 1.0, 'min': 0.0, 'percentile_00_5': 0.0, 'percentile_99_5': 1584.0, 'std': 316.9871520996094}}}", + "preprocessed_dataset_folder": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked/nnUNetPlans_3d_fullres", + "preprocessed_dataset_folder_base": "/datasets/work/hb-synthrad2023/work/synthrad2025/bw_workplace/data/nnunet_struct/preprocessed/Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked", + "save_every": "50", + "torch_version": "2.7.1+cu126", + "unpack_dataset": "True", + "was_initialized": "True", + "weight_decay": "3e-05" +} \ No newline at end of file diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/progress.png b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/progress.png new file mode 100644 index 0000000000000000000000000000000000000000..80f0e3d63ddd89ae13a72ed1c812eaffcb2763f6 --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/progress.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41a0af95a3e16833c339770c354ad73825c06b2d4215d4bb991bb9699b0c69f2 +size 521540 diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/validation/summary.json b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/validation/summary.json new file mode 100644 index 0000000000000000000000000000000000000000..2c430f90b13d56a60830560e879c7b84698aedea --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/fold_4/validation/summary.json @@ -0,0 +1,52 @@ +{ + "mae": { + "mean": 59.25234469230895, + "std": 17.071746210612847, + "max": 122.95905040097718, + "min": 37.58547386164849, + "25pc": 48.967471535994456, + "50pc": 53.033552020167036, + "75pc": 63.69078459860516, + "count": 36 + }, + "psnr": { + "mean": 30.15354383012489, + "std": 2.3524182073030557, + "max": 33.72105749188257, + "min": 23.59388887062555, + "25pc": 28.88756547983715, + "50pc": 30.39382479576161, + "75pc": 32.14567937370919, + "count": 36 + }, + "ms_ssim": { + "mean": 0.9334084994741705, + "std": 0.06478386403129642, + "max": 0.9803306785614478, + "min": 0.6084901323650287, + "25pc": 0.9228978499881949, + "50pc": 0.9545662546029308, + "75pc": 0.9669458677738854, + "count": 36 + }, + "DICE": { + "mean": 0.7706747170951631, + "std": 0.0912005046226504, + "max": 0.857702910900116, + "min": 0.4930849075317383, + "25pc": 0.7322952449321747, + "50pc": 0.806801050901413, + "75pc": 0.8347460627555847, + "count": 36 + }, + "HD95": { + "mean": 3.5273261037137775, + "std": 1.4436999983361203, + "max": 7.052619934082031, + "min": 1.9079017639160156, + "25pc": 2.3011980056762695, + "50pc": 3.0675822496414185, + "75pc": 4.525658965110779, + "count": 36 + } +} \ No newline at end of file diff --git a/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/plans.json b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/plans.json new file mode 100644 index 0000000000000000000000000000000000000000..5647721544689e7635095b83e016b31d5ac0ca6b --- /dev/null +++ b/Dataset264/nnUNetTrainerMRCT_loss_masked_perception_masked__nnUNetResEncUNetLPlans__3d_fullres/plans.json @@ -0,0 +1,521 @@ +{ + "dataset_name": "Dataset264_synthrad2025_task1_MR_TH_pre_v2r_stitched_masked", + "plans_name": "nnUNetResEncUNetLPlans", + "original_median_spacing_after_transp": [ + 3.0, + 1.0, + 1.0 + ], + "original_median_shape_after_transp": [ + 108, + 476, + 536 + ], + "image_reader_writer": "SimpleITKIO", + "transpose_forward": [ + 0, + 1, + 2 + ], + "transpose_backward": [ + 0, + 1, + 2 + ], + "configurations": { + "2d": { + "data_identifier": "nnUNetPlans_2d", + "preprocessor_name": "DefaultPreprocessor", + "batch_size": 13, + "patch_size": [ + 448, + 512 + ], + "median_image_size_in_voxels": [ + 476.0, + 536.0 + ], + "spacing": [ + 1.0, + 1.0 + ], + "normalization_schemes": [ + "ZScoreNormalization" + ], + "use_mask_for_norm": [ + false + ], + "resampling_fn_data": "resample_data_or_seg_to_shape", + "resampling_fn_seg": "resample_data_or_seg_to_shape", + "resampling_fn_data_kwargs": { + "is_seg": false, + "order": 3, + "order_z": 0, + "force_separate_z": null + }, + "resampling_fn_seg_kwargs": { + "is_seg": true, + "order": 1, + "order_z": 0, + "force_separate_z": null + }, + "resampling_fn_probabilities": "resample_data_or_seg_to_shape", + "resampling_fn_probabilities_kwargs": { + "is_seg": false, + "order": 1, + "order_z": 0, + "force_separate_z": null + }, + "architecture": { + "network_class_name": "dynamic_network_architectures.architectures.unet.ResidualEncoderUNet", + "arch_kwargs": { + "n_stages": 7, + "features_per_stage": [ + 32, + 64, + 128, + 256, + 512, + 512, + 512 + ], + "conv_op": "torch.nn.modules.conv.Conv2d", + "kernel_sizes": [ + [ + 3, + 3 + ], + [ + 3, + 3 + ], + [ + 3, + 3 + ], + [ + 3, + 3 + ], + [ + 3, + 3 + ], + [ + 3, + 3 + ], + [ + 3, + 3 + ] + ], + "strides": [ + [ + 1, + 1 + ], + [ + 2, + 2 + ], + [ + 2, + 2 + ], + [ + 2, + 2 + ], + [ + 2, + 2 + ], + [ + 2, + 2 + ], + [ + 2, + 2 + ] + ], + "n_blocks_per_stage": [ + 1, + 3, + 4, + 6, + 6, + 6, + 6 + ], + "n_conv_per_stage_decoder": [ + 1, + 1, + 1, + 1, + 1, + 1 + ], + "conv_bias": true, + "norm_op": "torch.nn.modules.instancenorm.InstanceNorm2d", + "norm_op_kwargs": { + "eps": 1e-05, + "affine": true + }, + "dropout_op": null, + "dropout_op_kwargs": null, + "nonlin": "torch.nn.LeakyReLU", + "nonlin_kwargs": { + "inplace": true + } + }, + "_kw_requires_import": [ + "conv_op", + "norm_op", + "dropout_op", + "nonlin" + ] + }, + "batch_dice": true + }, + "3d_lowres": { + "data_identifier": "nnUNetResEncUNetLPlans_3d_lowres", + "preprocessor_name": "DefaultPreprocessor", + "batch_size": 2, + "patch_size": [ + 64, + 192, + 192 + ], + "median_image_size_in_voxels": [ + 98, + 288, + 324 + ], + "spacing": [ + 3.278181, + 1.652847632271752, + 1.652847632271752 + ], + "normalization_schemes": [ + "ZScoreNormalization" + ], + "use_mask_for_norm": [ + false + ], + "resampling_fn_data": "resample_data_or_seg_to_shape", + "resampling_fn_seg": "resample_data_or_seg_to_shape", + "resampling_fn_data_kwargs": { + "is_seg": false, + "order": 3, + "order_z": 0, + "force_separate_z": null + }, + "resampling_fn_seg_kwargs": { + "is_seg": true, + "order": 1, + "order_z": 0, + "force_separate_z": null + }, + "resampling_fn_probabilities": "resample_data_or_seg_to_shape", + "resampling_fn_probabilities_kwargs": { + "is_seg": false, + "order": 1, + "order_z": 0, + "force_separate_z": null + }, + "architecture": { + "network_class_name": "dynamic_network_architectures.architectures.unet.ResidualEncoderUNet", + "arch_kwargs": { + "n_stages": 6, + "features_per_stage": [ + 32, + 64, + 128, + 256, + 320, + 320 + ], + "conv_op": "torch.nn.modules.conv.Conv3d", + "kernel_sizes": [ + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ] + ], + "strides": [ + [ + 1, + 1, + 1 + ], + [ + 2, + 2, + 2 + ], + [ + 2, + 2, + 2 + ], + [ + 2, + 2, + 2 + ], + [ + 2, + 2, + 2 + ], + [ + 1, + 2, + 2 + ] + ], + "n_blocks_per_stage": [ + 1, + 3, + 4, + 6, + 6, + 6 + ], + "n_conv_per_stage_decoder": [ + 1, + 1, + 1, + 1, + 1 + ], + "conv_bias": true, + "norm_op": "torch.nn.modules.instancenorm.InstanceNorm3d", + "norm_op_kwargs": { + "eps": 1e-05, + "affine": true + }, + "dropout_op": null, + "dropout_op_kwargs": null, + "nonlin": "torch.nn.LeakyReLU", + "nonlin_kwargs": { + "inplace": true + } + }, + "_kw_requires_import": [ + "conv_op", + "norm_op", + "dropout_op", + "nonlin" + ] + }, + "batch_dice": false, + "next_stage": "3d_cascade_fullres" + }, + "3d_fullres": { + "data_identifier": "nnUNetPlans_3d_fullres", + "preprocessor_name": "DefaultPreprocessor", + "batch_size": 2, + "patch_size": [ + 40, + 192, + 224 + ], + "median_image_size_in_voxels": [ + 107.5, + 476.0, + 536.0 + ], + "spacing": [ + 3.0, + 1.0, + 1.0 + ], + "normalization_schemes": [ + "ZScoreNormalization" + ], + "use_mask_for_norm": [ + false + ], + "resampling_fn_data": "resample_data_or_seg_to_shape", + "resampling_fn_seg": "resample_data_or_seg_to_shape", + "resampling_fn_data_kwargs": { + "is_seg": false, + "order": 3, + "order_z": 0, + "force_separate_z": null + }, + "resampling_fn_seg_kwargs": { + "is_seg": true, + "order": 1, + "order_z": 0, + "force_separate_z": null + }, + "resampling_fn_probabilities": "resample_data_or_seg_to_shape", + "resampling_fn_probabilities_kwargs": { + "is_seg": false, + "order": 1, + "order_z": 0, + "force_separate_z": null + }, + "architecture": { + "network_class_name": "dynamic_network_architectures.architectures.unet.ResidualEncoderUNet", + "arch_kwargs": { + "n_stages": 6, + "features_per_stage": [ + 32, + 64, + 128, + 256, + 320, + 320 + ], + "conv_op": "torch.nn.modules.conv.Conv3d", + "kernel_sizes": [ + [ + 1, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ], + [ + 3, + 3, + 3 + ] + ], + "strides": [ + [ + 1, + 1, + 1 + ], + [ + 1, + 2, + 2 + ], + [ + 2, + 2, + 2 + ], + [ + 2, + 2, + 2 + ], + [ + 2, + 2, + 2 + ], + [ + 1, + 2, + 2 + ] + ], + "n_blocks_per_stage": [ + 1, + 3, + 4, + 6, + 6, + 6 + ], + "n_conv_per_stage_decoder": [ + 1, + 1, + 1, + 1, + 1 + ], + "conv_bias": true, + "norm_op": "torch.nn.modules.instancenorm.InstanceNorm3d", + "norm_op_kwargs": { + "eps": 1e-05, + "affine": true + }, + "dropout_op": null, + "dropout_op_kwargs": null, + "nonlin": "torch.nn.LeakyReLU", + "nonlin_kwargs": { + "inplace": true + } + }, + "_kw_requires_import": [ + "conv_op", + "norm_op", + "dropout_op", + "nonlin" + ] + }, + "batch_dice": true + }, + "3d_cascade_fullres": { + "inherits_from": "3d_fullres", + "previous_stage": "3d_lowres" + } + }, + "experiment_planner_used": "nnUNetPlannerResEncL", + "label_manager": "LabelManager", + "foreground_intensity_properties_per_channel": { + "0": { + "max": 5250.0, + "mean": 123.740478515625, + "median": 1.0, + "min": 0.0, + "percentile_00_5": 0.0, + "percentile_99_5": 1584.0, + "std": 316.9871520996094 + } + } +} \ No newline at end of file