============================================================ CIFAR-10 Conv-SNN Experiment Job ID: 14363510 | Node: gpub011 Start: Mon Dec 29 10:47:15 CST 2025 ============================================================ NVIDIA A40, 46068 MiB ============================================================ ====================================================================== CIFAR-10 Conv-SNN Experiment ====================================================================== Model: simple Timesteps: 25 Epochs: 50 Device: cuda ====================================================================== Loading CIFAR-10... Train: 50000, Test: 10000 ============================================================ Depth = 2 conv layers, channels = [64, 128] ============================================================ Training VANILLA... Model: simple, params: 157,770 Epoch 1: train=0.294 test=0.387 (68.7s) Epoch 2: train=0.363 test=0.461 (66.7s) Epoch 3: train=0.430 test=0.510 (66.7s) Epoch 4: train=0.508 test=0.502 (66.7s) Epoch 5: train=0.537 test=0.575 (66.7s) Epoch 6: train=0.560 test=0.551 (66.7s) Epoch 7: train=0.579 test=0.577 (66.7s) Epoch 8: train=0.592 test=0.616 (66.7s) Epoch 9: train=0.604 test=0.620 (66.7s) Epoch 10: train=0.615 test=0.620 (66.7s) Epoch 11: train=0.623 test=0.637 (66.7s) Epoch 12: train=0.633 test=0.664 (66.7s) Epoch 13: train=0.639 test=0.660 (66.7s) Epoch 14: train=0.647 test=0.635 (66.8s) Epoch 15: train=0.652 test=0.679 (66.7s) Epoch 16: train=0.657 test=0.677 (66.8s) Epoch 17: train=0.664 test=0.678 (66.8s) Epoch 18: train=0.669 test=0.693 (66.8s) Epoch 19: train=0.674 test=0.697 (66.8s) Epoch 20: train=0.677 test=0.688 (66.8s) Epoch 21: train=0.682 test=0.705 (66.8s) Epoch 22: train=0.681 test=0.700 (66.8s) Epoch 23: train=0.684 test=0.708 (66.8s) Epoch 24: train=0.691 test=0.695 (66.7s) Epoch 25: train=0.695 test=0.708 (66.8s) Epoch 26: train=0.696 test=0.702 (66.8s) Epoch 27: train=0.699 test=0.713 (66.8s) Epoch 28: train=0.703 test=0.720 (66.8s) Epoch 29: train=0.704 test=0.718 (66.8s) Epoch 30: train=0.706 test=0.723 (66.8s) Epoch 31: train=0.711 test=0.736 (66.8s) Epoch 32: train=0.711 test=0.732 (66.8s) Epoch 33: train=0.715 test=0.733 (66.8s) Epoch 34: train=0.717 test=0.728 (66.8s) Epoch 35: train=0.720 test=0.730 (66.8s) Epoch 36: train=0.723 test=0.735 (66.8s) Epoch 37: train=0.720 test=0.727 (66.8s) Epoch 38: train=0.726 test=0.731 (66.8s) Epoch 39: train=0.726 test=0.739 (66.8s) Epoch 40: train=0.728 test=0.738 (66.8s) Epoch 41: train=0.728 test=0.738 (66.7s) Epoch 42: train=0.730 test=0.735 (66.7s) Epoch 43: train=0.729 test=0.737 (66.8s) Epoch 44: train=0.731 test=0.742 (66.8s) Epoch 45: train=0.730 test=0.750 (66.8s) Epoch 46: train=0.733 test=0.739 (66.8s) Epoch 47: train=0.731 test=0.748 (66.8s) Epoch 48: train=0.731 test=0.746 (66.8s) Epoch 49: train=0.736 test=0.747 (66.8s) Epoch 50: train=0.735 test=0.737 (66.8s) Best test accuracy: 0.750 Training LYAPUNOV... Model: simple, params: 157,770 Epoch 1: train=0.290 test=0.412 λ=1.266 (177.1s) Epoch 2: train=0.361 test=0.485 λ=1.141 (177.1s) Epoch 3: train=0.405 test=0.508 λ=0.808 (177.0s) Epoch 4: train=0.490 test=0.580 λ=0.641 (177.1s) Epoch 5: train=0.532 test=0.588 λ=0.617 (177.1s) Epoch 6: train=0.552 test=0.611 λ=0.610 (177.2s) Epoch 7: train=0.568 test=0.632 λ=0.602 (177.1s) Epoch 8: train=0.578 test=0.628 λ=0.600 (177.1s) Epoch 9: train=0.590 test=0.639 λ=0.593 (177.1s) Epoch 10: train=0.594 test=0.643 λ=0.588 (177.1s) Epoch 11: train=0.605 test=0.662 λ=0.587 (177.1s) Epoch 12: train=0.612 test=0.657 λ=0.585 (177.1s) Epoch 13: train=0.615 test=0.663 λ=0.586 (177.1s) Epoch 14: train=0.624 test=0.642 λ=0.584 (177.1s) Epoch 15: train=0.627 test=0.673 λ=0.580 (177.1s) Epoch 16: train=0.629 test=0.662 λ=0.582 (177.1s) Epoch 17: train=0.636 test=0.671 λ=0.580 (177.1s) Epoch 18: train=0.639 test=0.683 λ=0.579 (177.2s) Epoch 19: train=0.644 test=0.673 λ=0.580 (177.2s) Epoch 20: train=0.649 test=0.686 λ=0.581 (177.2s) Epoch 21: train=0.650 test=0.689 λ=0.575 (177.2s) Epoch 22: train=0.653 test=0.696 λ=0.572 (177.2s) Epoch 23: train=0.657 test=0.695 λ=0.579 (177.2s) Epoch 24: train=0.663 test=0.683 λ=0.574 (177.2s) Epoch 25: train=0.662 test=0.706 λ=0.577 (177.2s) Epoch 26: train=0.667 test=0.702 λ=0.572 (177.2s) Epoch 27: train=0.669 test=0.701 λ=0.572 (177.2s) Epoch 28: train=0.675 test=0.702 λ=0.572 (177.2s) Epoch 29: train=0.675 test=0.701 λ=0.570 (177.2s) Epoch 30: train=0.674 test=0.701 λ=0.573 (177.2s) Epoch 31: train=0.679 test=0.718 λ=0.573 (177.2s) Epoch 32: train=0.680 test=0.708 λ=0.569 (177.2s) Epoch 33: train=0.684 test=0.716 λ=0.569 (177.2s) Epoch 34: train=0.684 test=0.710 λ=0.570 (177.2s) Epoch 35: train=0.687 test=0.719 λ=0.573 (177.2s) Epoch 36: train=0.691 test=0.724 λ=0.573 (177.2s) Epoch 37: train=0.692 test=0.701 λ=0.570 (177.2s) Epoch 38: train=0.694 test=0.716 λ=0.569 (177.2s) Epoch 39: train=0.694 test=0.726 λ=0.569 (177.2s) Epoch 40: train=0.695 test=0.720 λ=0.569 (177.2s) Epoch 41: train=0.697 test=0.728 λ=0.569 (177.2s) Epoch 42: train=0.698 test=0.714 λ=0.571 (177.3s) Epoch 43: train=0.697 test=0.719 λ=0.569 (177.2s) Epoch 44: train=0.700 test=0.719 λ=0.571 (177.2s) Epoch 45: train=0.699 test=0.721 λ=0.568 (177.2s) Epoch 46: train=0.702 test=0.722 λ=0.572 (177.2s) Epoch 47: train=0.702 test=0.727 λ=0.569 (177.2s) Epoch 48: train=0.702 test=0.726 λ=0.571 (177.2s) Epoch 49: train=0.701 test=0.728 λ=0.572 (177.2s) Epoch 50: train=0.700 test=0.714 λ=0.570 (177.2s) Best test accuracy: 0.728 ============================================================ Depth = 3 conv layers, channels = [64, 128, 256] ============================================================ Training VANILLA... Model: simple, params: 412,234 Epoch 1: train=0.293 test=0.431 (81.7s) Epoch 2: train=0.377 test=0.464 (81.6s) Epoch 3: train=0.449 test=0.478 (81.6s) Epoch 4: train=0.499 test=0.577 (81.6s) Epoch 5: train=0.563 test=0.615 (81.6s) Epoch 6: train=0.618 test=0.665 (81.6s) Epoch 7: train=0.642 test=0.670 (81.6s) Epoch 8: train=0.663 test=0.693 (81.6s) Epoch 9: train=0.679 test=0.687 (81.6s) Epoch 10: train=0.693 test=0.715 (81.6s) Epoch 11: train=0.704 test=0.704 (96.8s) Epoch 12: train=0.714 test=0.701 (81.6s) Epoch 13: train=0.718 test=0.723 (91.9s) Epoch 14: train=0.728 test=0.734 (81.6s) Epoch 15: train=0.734 test=0.720 (81.6s) Epoch 16: train=0.742 test=0.730 (91.9s) Epoch 17: train=0.748 test=0.726 (81.6s) Epoch 18: train=0.752 test=0.754 (81.6s) Epoch 19: train=0.758 test=0.746 (81.6s) Epoch 20: train=0.764 test=0.760 (81.6s) Epoch 21: train=0.770 test=0.757 (81.6s) Epoch 22: train=0.776 test=0.756 (81.6s) Epoch 23: train=0.781 test=0.730 (81.6s) Epoch 24: train=0.786 test=0.752 (81.6s) Epoch 25: train=0.790 test=0.771 (81.6s) Epoch 26: train=0.795 test=0.772 (81.6s) Epoch 27: train=0.796 test=0.766 (81.6s) Epoch 28: train=0.800 test=0.760 (81.7s) Epoch 29: train=0.804 test=0.755 (81.6s) Epoch 30: train=0.808 test=0.754 (81.6s) Epoch 31: train=0.810 test=0.760 (81.6s) Epoch 32: train=0.813 test=0.771 (81.6s) Epoch 33: train=0.817 test=0.753 (81.6s) Epoch 34: train=0.820 test=0.784 (81.6s) Epoch 35: train=0.819 test=0.779 (81.6s) Epoch 36: train=0.823 test=0.773 (81.6s) Epoch 37: train=0.823 test=0.773 (81.6s) Epoch 38: train=0.829 test=0.778 (81.6s) Epoch 39: train=0.831 test=0.776 (81.9s) Epoch 40: train=0.832 test=0.785 (81.7s) Epoch 41: train=0.831 test=0.787 (81.6s) Epoch 42: train=0.835 test=0.780 (81.6s) Epoch 43: train=0.836 test=0.775 (81.6s) Epoch 44: train=0.838 test=0.779 (81.6s) Epoch 45: train=0.837 test=0.774 (81.6s) Epoch 46: train=0.837 test=0.777 (81.6s) Epoch 47: train=0.839 test=0.796 (81.6s) Epoch 48: train=0.840 test=0.775 (81.7s) Epoch 49: train=0.839 test=0.776 (81.6s) Epoch 50: train=0.837 test=0.777 (81.6s) Best test accuracy: 0.796 Training LYAPUNOV... Model: simple, params: 412,234 Epoch 1: train=0.268 test=0.346 λ=1.850 (219.1s) Epoch 2: train=0.321 test=0.407 λ=1.465 (219.1s) Epoch 3: train=0.381 test=0.417 λ=1.248 (219.1s) Epoch 4: train=0.419 test=0.461 λ=1.133 (219.1s) Epoch 5: train=0.445 test=0.511 λ=1.062 (219.1s) Epoch 6: train=0.483 test=0.564 λ=0.977 (219.1s) Epoch 7: train=0.517 test=0.573 λ=0.942 (219.0s) Epoch 8: train=0.532 test=0.581 λ=0.921 (219.1s) Epoch 9: train=0.539 test=0.596 λ=0.909 (219.1s) Epoch 10: train=0.549 test=0.601 λ=0.907 (219.1s) Epoch 11: train=0.557 test=0.610 λ=0.902 (219.0s) Epoch 12: train=0.568 test=0.611 λ=0.897 (219.1s) Epoch 13: train=0.574 test=0.612 λ=0.900 (219.1s) Epoch 14: train=0.579 test=0.622 λ=0.894 (219.1s) Epoch 15: train=0.580 test=0.627 λ=0.895 (219.1s) Epoch 16: train=0.586 test=0.637 λ=0.890 (219.1s) Epoch 17: train=0.592 test=0.645 λ=0.891 (219.1s) Epoch 18: train=0.593 test=0.638 λ=0.889 (219.1s) Epoch 19: train=0.599 test=0.641 λ=0.887 (219.0s) Epoch 20: train=0.603 test=0.644 λ=0.883 (219.0s) Epoch 21: train=0.606 test=0.645 λ=0.879 (219.1s) Epoch 22: train=0.610 test=0.655 λ=0.882 (219.1s) Epoch 23: train=0.612 test=0.649 λ=0.888 (219.1s) Epoch 24: train=0.616 test=0.649 λ=0.878 (219.1s) Epoch 25: train=0.619 test=0.661 λ=0.885 (219.1s) Epoch 26: train=0.622 test=0.657 λ=0.882 (219.1s) Epoch 27: train=0.626 test=0.662 λ=0.881 (219.1s) Epoch 28: train=0.627 test=0.662 λ=0.875 (219.1s) Epoch 29: train=0.627 test=0.660 λ=0.882 (219.1s) Epoch 30: train=0.634 test=0.672 λ=0.879 (219.1s) Epoch 31: train=0.633 test=0.673 λ=0.879 (219.2s) Epoch 32: train=0.637 test=0.674 λ=0.877 (219.2s) Epoch 33: train=0.639 test=0.664 λ=0.880 (219.2s) Epoch 34: train=0.640 test=0.677 λ=0.883 (219.1s) Epoch 35: train=0.639 test=0.674 λ=0.877 (219.1s) Epoch 36: train=0.645 test=0.678 λ=0.877 (219.1s) Epoch 37: train=0.645 test=0.665 λ=0.878 (219.1s) Epoch 38: train=0.644 test=0.669 λ=0.875 (219.1s) Epoch 39: train=0.649 test=0.673 λ=0.877 (219.1s) Epoch 40: train=0.648 test=0.680 λ=0.876 (219.1s) Epoch 41: train=0.651 test=0.674 λ=0.878 (219.1s) Epoch 42: train=0.649 test=0.679 λ=0.881 (219.1s) Epoch 43: train=0.651 test=0.676 λ=0.873 (219.1s) Epoch 44: train=0.650 test=0.680 λ=0.878 (219.1s) Epoch 45: train=0.649 test=0.684 λ=0.875 (219.1s) Epoch 46: train=0.652 test=0.682 λ=0.882 (219.1s) Epoch 47: train=0.653 test=0.681 λ=0.876 (219.1s) Epoch 48: train=0.650 test=0.684 λ=0.876 (219.1s) Epoch 49: train=0.653 test=0.679 λ=0.872 (219.1s) Epoch 50: train=0.653 test=0.679 λ=0.877 (219.1s) Best test accuracy: 0.684 ============================================================ Depth = 4 conv layers, channels = [64, 128, 256, 512] ============================================================ Training VANILLA... Model: simple, params: 1,572,426 Epoch 1: train=0.293 test=0.421 (91.2s) Epoch 2: train=0.406 test=0.522 (91.1s) Epoch 3: train=0.490 test=0.491 (91.1s) Epoch 4: train=0.543 test=0.591 (91.2s) Epoch 5: train=0.598 test=0.635 (91.1s) Epoch 6: train=0.637 test=0.648 (91.2s) Epoch 7: train=0.671 test=0.657 (91.1s) Epoch 8: train=0.696 test=0.683 (91.2s) Epoch 9: train=0.714 test=0.688 (91.1s) Epoch 10: train=0.731 test=0.686 (91.1s) Epoch 11: train=0.742 test=0.730 (91.2s) Epoch 12: train=0.755 test=0.718 (91.1s) Epoch 13: train=0.763 test=0.707 (91.2s) Epoch 14: train=0.775 test=0.723 (91.1s) Epoch 15: train=0.781 test=0.748 (91.1s) Epoch 16: train=0.787 test=0.730 (91.2s) Epoch 17: train=0.795 test=0.694 (91.1s) Epoch 18: train=0.802 test=0.716 (91.1s) Epoch 19: train=0.808 test=0.759 (91.1s) Epoch 20: train=0.815 test=0.751 (91.2s) Epoch 21: train=0.822 test=0.756 (91.1s) Epoch 22: train=0.825 test=0.748 (91.1s) Epoch 23: train=0.835 test=0.743 (91.2s) Epoch 24: train=0.837 test=0.762 (91.1s) Epoch 25: train=0.843 test=0.751 (91.1s) Epoch 26: train=0.845 test=0.763 (91.1s) Epoch 27: train=0.852 test=0.782 (91.1s) Epoch 28: train=0.858 test=0.760 (91.1s) Epoch 29: train=0.862 test=0.774 (91.1s) Epoch 30: train=0.865 test=0.766 (91.1s) Epoch 31: train=0.868 test=0.748 (91.1s) Epoch 32: train=0.870 test=0.776 (91.1s) Epoch 33: train=0.876 test=0.763 (91.1s) Epoch 34: train=0.880 test=0.766 (91.1s) Epoch 35: train=0.883 test=0.768 (91.1s) Epoch 36: train=0.887 test=0.772 (91.1s) Epoch 37: train=0.891 test=0.767 (91.2s) Epoch 38: train=0.890 test=0.773 (91.1s) Epoch 39: train=0.894 test=0.774 (91.1s) Epoch 40: train=0.893 test=0.757 (91.1s) Epoch 41: train=0.899 test=0.773 (91.1s) Epoch 42: train=0.900 test=0.787 (91.1s) Epoch 43: train=0.902 test=0.764 (91.1s) Epoch 44: train=0.901 test=0.778 (91.1s) Epoch 45: train=0.904 test=0.754 (91.0s) Epoch 46: train=0.905 test=0.786 (91.1s) Epoch 47: train=0.906 test=0.775 (91.1s) Epoch 48: train=0.906 test=0.790 (91.1s) Epoch 49: train=0.906 test=0.778 (91.1s) Epoch 50: train=0.907 test=0.785 (91.0s) Best test accuracy: 0.790 Training LYAPUNOV... Model: simple, params: 1,572,426 Epoch 1: train=0.226 test=0.341 λ=2.642 (243.8s) Epoch 2: train=0.238 test=0.338 λ=1.931 (243.8s) Epoch 3: train=0.249 test=0.292 λ=1.659 (243.8s) Epoch 4: train=0.257 test=0.327 λ=1.510 (243.8s) Epoch 5: train=0.265 test=0.313 λ=1.386 (243.8s) Epoch 6: train=0.282 test=0.349 λ=1.291 (243.8s) Epoch 7: train=0.320 test=0.367 λ=1.211 (243.8s) Epoch 8: train=0.342 test=0.389 λ=1.171 (243.9s) Epoch 9: train=0.355 test=0.392 λ=1.157 (243.8s) Epoch 10: train=0.361 test=0.396 λ=1.140 (243.7s) Epoch 11: train=0.362 test=0.393 λ=1.139 (243.8s) Epoch 12: train=0.367 test=0.398 λ=1.131 (243.8s) Epoch 13: train=0.370 test=0.417 λ=1.123 (243.8s) Epoch 14: train=0.373 test=0.420 λ=1.124 (243.8s) Epoch 15: train=0.378 test=0.409 λ=1.120 (243.8s) Epoch 16: train=0.379 test=0.408 λ=1.119 (243.8s) Epoch 17: train=0.382 test=0.422 λ=1.117 (243.8s) Epoch 18: train=0.382 test=0.419 λ=1.120 (243.8s) Epoch 19: train=0.386 test=0.421 λ=1.119 (243.8s) Epoch 20: train=0.388 test=0.421 λ=1.121 (243.8s) Epoch 21: train=0.391 test=0.433 λ=1.127 (243.8s) Epoch 22: train=0.395 test=0.436 λ=1.130 (243.8s) Epoch 23: train=0.396 test=0.441 λ=1.125 (243.9s) Epoch 24: train=0.398 test=0.437 λ=1.129 (243.8s) Epoch 25: train=0.400 test=0.437 λ=1.136 (243.9s) Epoch 26: train=0.399 test=0.438 λ=1.141 (243.8s) Epoch 27: train=0.405 test=0.428 λ=1.137 (243.8s) Epoch 28: train=0.407 test=0.441 λ=1.143 (243.8s) Epoch 29: train=0.405 test=0.450 λ=1.139 (243.8s) Epoch 30: train=0.410 test=0.443 λ=1.139 (243.8s) Epoch 31: train=0.413 test=0.446 λ=1.139 (243.8s) Epoch 32: train=0.418 test=0.453 λ=1.138 (243.8s) Epoch 33: train=0.416 test=0.455 λ=1.138 (243.9s) Epoch 34: train=0.419 test=0.455 λ=1.139 (243.8s) Epoch 35: train=0.417 test=0.459 λ=1.138 (243.8s) Epoch 36: train=0.420 test=0.459 λ=1.141 (243.8s) Epoch 37: train=0.421 test=0.458 λ=1.139 (243.8s) Epoch 38: train=0.425 test=0.452 λ=1.140 (243.8s) Epoch 39: train=0.425 test=0.457 λ=1.135 (243.8s) Epoch 40: train=0.424 test=0.458 λ=1.141 (243.9s) Epoch 41: train=0.429 test=0.461 λ=1.138 (243.8s) Epoch 42: train=0.428 test=0.461 λ=1.134 (243.8s) Epoch 43: train=0.427 test=0.457 λ=1.136 (244.0s) Epoch 44: train=0.431 test=0.461 λ=1.136 (243.8s) Epoch 45: train=0.431 test=0.462 λ=1.138 (243.8s) Epoch 46: train=0.432 test=0.460 λ=1.139 (243.8s) Epoch 47: train=0.431 test=0.467 λ=1.142 (243.8s) Epoch 48: train=0.432 test=0.470 λ=1.136 (243.8s) Epoch 49: train=0.430 test=0.462 λ=1.136 (243.8s) Epoch 50: train=0.430 test=0.463 λ=1.142 (243.8s) Best test accuracy: 0.470 ====================================================================== SUMMARY: CIFAR-10 Conv-SNN Results ====================================================================== Depth Vanilla Lyapunov Improvement ---------------------------------------------------------------------- 2 0.737 0.714 -0.023 3 0.777 0.679 -0.098 4 0.785 0.463 -0.323 ====================================================================== Results saved to runs/cifar10_conv/20251229-230105 ============================================================ Finished: Mon Dec 29 23:01:08 CST 2025 ============================================================