Classification on CIFAR10 using ResNet18 with hyperparameter tuning
1) Image Augmentation used with CutOut being one of the transformation
2) Found the best Learning Rate to train the model
Best Learning Rate: 0.01584893192461112
3) Used SGD with momentum with momentum = 0.9
4) Trained the ResNet-18 model for 50 epochs
EPOCH: 1 Loss=2.15325665473938 Batch_id=390 Accuracy=43.85: 100%|██████████| 391/391 [02:34<00:00, 2.52it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0097, Accuracy: 5419/10000 (54.19%)
EPOCH: 2 Loss=2.012441873550415 Batch_id=390 Accuracy=60.02: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0077, Accuracy: 6524/10000 (65.24%)
EPOCH: 3 Loss=1.6384177207946777 Batch_id=390 Accuracy=67.32: 100%|██████████| 391/391 [02:34<00:00, 2.52it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0082, Accuracy: 6519/10000 (65.19%)
EPOCH: 4 Loss=1.7387962341308594 Batch_id=390 Accuracy=71.24: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0064, Accuracy: 7264/10000 (72.64%)
EPOCH: 5 Loss=1.4806256294250488 Batch_id=390 Accuracy=73.87: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0069, Accuracy: 7144/10000 (71.44%)
EPOCH: 6 Loss=1.2457722425460815 Batch_id=390 Accuracy=75.62: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0052, Accuracy: 7764/10000 (77.64%)
EPOCH: 7 Loss=1.3323636054992676 Batch_id=390 Accuracy=77.13: 100%|██████████| 391/391 [02:34<00:00, 2.52it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0047, Accuracy: 7911/10000 (79.11%)
EPOCH: 8 Loss=1.1756545305252075 Batch_id=390 Accuracy=78.16: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0043, Accuracy: 8188/10000 (81.88%)
EPOCH: 9 Loss=1.0852481126785278 Batch_id=390 Accuracy=79.02: 100%|██████████| 391/391 [02:35<00:00, 2.52it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0041, Accuracy: 8255/10000 (82.55%)
EPOCH: 10 Loss=1.4269263744354248 Batch_id=390 Accuracy=79.67: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0037, Accuracy: 8453/10000 (84.53%)
EPOCH: 11 Loss=0.9477803707122803 Batch_id=390 Accuracy=80.62: 100%|██████████| 391/391 [02:34<00:00, 2.52it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0043, Accuracy: 8126/10000 (81.26%)
EPOCH: 12 Loss=0.9541343450546265 Batch_id=390 Accuracy=81.57: 100%|██████████| 391/391 [02:33<00:00, 2.54it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0042, Accuracy: 8166/10000 (81.66%)
EPOCH: 13 Loss=0.9508116841316223 Batch_id=390 Accuracy=81.75: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0042, Accuracy: 8194/10000 (81.94%)
EPOCH: 14 Loss=0.907364010810852 Batch_id=390 Accuracy=82.21: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0048, Accuracy: 8018/10000 (80.18%)
EPOCH: 15 Loss=0.833751916885376 Batch_id=390 Accuracy=82.52: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0037, Accuracy: 8390/10000 (83.90%)
EPOCH: 16 Loss=0.6455802917480469 Batch_id=390 Accuracy=82.90: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0036, Accuracy: 8447/10000 (84.47%)
EPOCH: 17 Loss=0.9995672702789307 Batch_id=390 Accuracy=83.17: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0034, Accuracy: 8517/10000 (85.17%)
EPOCH: 18 Loss=0.872093141078949 Batch_id=390 Accuracy=83.89: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0035, Accuracy: 8450/10000 (84.50%)
EPOCH: 19 Loss=0.8379157781600952 Batch_id=390 Accuracy=83.82: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0036, Accuracy: 8513/10000 (85.13%)
EPOCH: 20 Loss=0.9011598825454712 Batch_id=390 Accuracy=84.30: 100%|██████████| 391/391 [02:35<00:00, 2.51it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0035, Accuracy: 8483/10000 (84.83%)
EPOCH: 21 Loss=0.7680745124816895 Batch_id=390 Accuracy=84.55: 100%|██████████| 391/391 [02:34<00:00, 2.52it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0043, Accuracy: 8184/10000 (81.84%)
EPOCH: 22 Loss=0.6993612051010132 Batch_id=390 Accuracy=84.64: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0033, Accuracy: 8578/10000 (85.78%)
EPOCH: 23 Loss=0.7411786317825317 Batch_id=390 Accuracy=84.69: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0034, Accuracy: 8553/10000 (85.53%)
EPOCH: 24 Loss=0.7328760027885437 Batch_id=390 Accuracy=85.22: 100%|██████████| 391/391 [02:34<00:00, 2.54it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0034, Accuracy: 8571/10000 (85.71%)
EPOCH: 25 Loss=0.7580845355987549 Batch_id=390 Accuracy=85.41: 100%|██████████| 391/391 [02:34<00:00, 2.54it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0030, Accuracy: 8702/10000 (87.02%)
EPOCH: 26 Loss=0.7757035493850708 Batch_id=390 Accuracy=85.39: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0030, Accuracy: 8689/10000 (86.89%)
EPOCH: 27 Loss=0.5850611925125122 Batch_id=390 Accuracy=85.91: 100%|██████████| 391/391 [02:35<00:00, 2.52it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0038, Accuracy: 8449/10000 (84.49%)
EPOCH: 28 Loss=0.6602646112442017 Batch_id=390 Accuracy=85.84: 100%|██████████| 391/391 [02:35<00:00, 2.52it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0030, Accuracy: 8726/10000 (87.26%)
EPOCH: 29 Loss=0.708653450012207 Batch_id=390 Accuracy=86.08: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0035, Accuracy: 8514/10000 (85.14%)
EPOCH: 30 Loss=0.8008662462234497 Batch_id=390 Accuracy=86.35: 100%|██████████| 391/391 [02:34<00:00, 2.52it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0028, Accuracy: 8833/10000 (88.33%)
EPOCH: 31 Loss=0.636769711971283 Batch_id=390 Accuracy=86.49: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0028, Accuracy: 8800/10000 (88.00%)
EPOCH: 32 Loss=0.7092130184173584 Batch_id=390 Accuracy=86.39: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0029, Accuracy: 8792/10000 (87.92%)
EPOCH: 33 Loss=0.752608060836792 Batch_id=390 Accuracy=86.69: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0036, Accuracy: 8503/10000 (85.03%)
EPOCH: 34 Loss=1.0032578706741333 Batch_id=390 Accuracy=86.82: 100%|██████████| 391/391 [02:34<00:00, 2.54it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0031, Accuracy: 8683/10000 (86.83%)
EPOCH: 35 Loss=0.7567118406295776 Batch_id=390 Accuracy=86.98: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0025, Accuracy: 8964/10000 (89.64%)
EPOCH: 36 Loss=0.7628611326217651 Batch_id=390 Accuracy=87.04: 100%|██████████| 391/391 [02:34<00:00, 2.52it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0025, Accuracy: 8957/10000 (89.57%)
EPOCH: 37 Loss=0.7143936157226562 Batch_id=390 Accuracy=87.22: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0034, Accuracy: 8568/10000 (85.68%)
EPOCH: 38 Loss=0.6786922812461853 Batch_id=390 Accuracy=87.30: 100%|██████████| 391/391 [02:34<00:00, 2.54it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0027, Accuracy: 8850/10000 (88.50%)
EPOCH: 39 Loss=0.797775387763977 Batch_id=390 Accuracy=87.53: 100%|██████████| 391/391 [02:33<00:00, 2.54it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0027, Accuracy: 8826/10000 (88.26%)
EPOCH: 40 Loss=0.7024500966072083 Batch_id=390 Accuracy=87.49: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0036, Accuracy: 8526/10000 (85.26%)
EPOCH: 41 Loss=0.6100055575370789 Batch_id=390 Accuracy=87.59: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0025, Accuracy: 8939/10000 (89.39%)
EPOCH: 42 Loss=0.5923382043838501 Batch_id=390 Accuracy=87.90: 100%|██████████| 391/391 [02:35<00:00, 2.51it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0031, Accuracy: 8674/10000 (86.74%)
EPOCH: 43 Loss=0.6672704815864563 Batch_id=390 Accuracy=87.78: 100%|██████████| 391/391 [02:34<00:00, 2.54it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0028, Accuracy: 8787/10000 (87.87%)
EPOCH: 44 Loss=0.5858694911003113 Batch_id=390 Accuracy=87.72: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0028, Accuracy: 8819/10000 (88.19%)
EPOCH: 45 Loss=0.5993590950965881 Batch_id=390 Accuracy=88.05: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0027, Accuracy: 8878/10000 (88.78%)
EPOCH: 46 Loss=0.480996310710907 Batch_id=390 Accuracy=87.98: 100%|██████████| 391/391 [02:34<00:00, 2.52it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0026, Accuracy: 8868/10000 (88.68%)
EPOCH: 47 Loss=0.4863784909248352 Batch_id=390 Accuracy=91.49: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0017, Accuracy: 9288/10000 (92.88%)
EPOCH: 48 Loss=0.5819729566574097 Batch_id=390 Accuracy=92.66: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0016, Accuracy: 9330/10000 (93.30%)
EPOCH: 49 Loss=0.4391074776649475 Batch_id=390 Accuracy=93.26: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] 0%| | 0/391 [00:00<?, ?it/s] Test set: Average loss: 0.0016, Accuracy: 9347/10000 (93.47%)
EPOCH: 50 Loss=0.48527491092681885 Batch_id=390 Accuracy=93.74: 100%|██████████| 391/391 [02:34<00:00, 2.53it/s] Test set: Average loss: 0.0015, Accuracy: 9367/10000 (93.67%)
5) Training Vs Test Accuracy
6) Graphs of Loss and Accuracy for training and testing
7) Test Accuracy reached: 93.67%
8) GradCAM on 25 misclassified images
