[2021-03-18 13:29:51] Created experiment 0:
[2021-03-18 13:29:51]  - Model: vgg-pretrained
[2021-03-18 13:29:51]  - Acquisition function: random
[2021-03-18 13:29:51] Created experiment 1:
[2021-03-18 13:29:51]  - Model: vgg-pretrained
[2021-03-18 13:29:51]  - Acquisition function: greedy-coreset
[2021-03-18 13:29:51] Created experiment 2:
[2021-03-18 13:29:51]  - Model: vgg-pretrained
[2021-03-18 13:29:51]  - Acquisition function: lc-beam-coreset
[2021-03-18 13:29:51] Created experiment 3:
[2021-03-18 13:29:51]  - Model: vgg-pretrained
[2021-03-18 13:29:51]  - Acquisition function: least-confidence
[2021-03-18 13:29:51] Loading cifar10 test set...
[2021-03-18 13:29:52] Experiment repeat 1/1
[2021-03-18 13:29:52] Seeded: 5
[2021-03-18 13:29:52] Using 10.00% labels of the dataset (5000/50000)
[2021-03-18 13:29:52] Randomly labelled 5000/50000
[2021-03-18 13:29:52] Showing labelled: True (5000/50000 visible)
[2021-03-18 13:29:52] Seeded: 5
[2021-03-18 13:29:52] Running: experiment 0
[2021-03-18 13:29:52] Showing labelled: True (5000/50000 visible)
[2021-03-18 13:29:52] Creating pretrained=True VGG16...
[2021-03-18 13:29:54] No parameter reset since we are using a pretrained model.
[2021-03-18 13:29:54] vgg-pretrained: initialized 15250250 parameters.
[2021-03-18 13:29:54] Creating trainer with model on device: cuda
[2021-03-18 13:29:59] Training vgg-pretrained across 5000 data points in cifar10...
[2021-03-18 13:30:43] Training accuracy: 0.9975
[2021-03-18 13:30:43] Testing on 10000 data points...
[2021-03-18 13:30:45] Test score for 5000 training labels: 0.7909
[2021-03-18 13:30:45] Showing labelled: False (45000/50000 visible)
[2021-03-18 13:30:45] Found 45000 unlabelled features.
[2021-03-18 13:30:51] Showing labelled: True (10000/50000 visible)
[2021-03-18 13:30:51] Creating pretrained=True VGG16...
[2021-03-18 13:30:52] No parameter reset since we are using a pretrained model.
[2021-03-18 13:30:52] vgg-pretrained: initialized 15250250 parameters.
[2021-03-18 13:30:52] Creating trainer with model on device: cuda
[2021-03-18 13:30:52] Training vgg-pretrained across 10000 data points in cifar10...
[2021-03-18 13:32:22] Training accuracy: 0.9999
[2021-03-18 13:32:22] Testing on 10000 data points...
[2021-03-18 13:32:24] Test score for 10000 training labels: 0.8171
[2021-03-18 13:32:24] Showing labelled: False (40000/50000 visible)
[2021-03-18 13:32:24] Found 40000 unlabelled features.
[2021-03-18 13:32:30] Showing labelled: True (15000/50000 visible)
[2021-03-18 13:32:30] Creating pretrained=True VGG16...
[2021-03-18 13:32:31] No parameter reset since we are using a pretrained model.
[2021-03-18 13:32:31] vgg-pretrained: initialized 15250250 parameters.
[2021-03-18 13:32:31] Creating trainer with model on device: cuda
[2021-03-18 13:32:31] Training vgg-pretrained across 15000 data points in cifar10...
[2021-03-18 13:34:46] Training accuracy: 0.9993
[2021-03-18 13:34:46] Testing on 10000 data points...
[2021-03-18 13:34:48] Test score for 15000 training labels: 0.8421
[2021-03-18 13:34:48] Showing labelled: False (35000/50000 visible)
[2021-03-18 13:34:48] Found 35000 unlabelled features.
[2021-03-18 13:34:52] Showing labelled: True (20000/50000 visible)
[2021-03-18 13:34:52] Creating pretrained=True VGG16...
[2021-03-18 13:34:54] No parameter reset since we are using a pretrained model.
[2021-03-18 13:34:54] vgg-pretrained: initialized 15250250 parameters.
[2021-03-18 13:34:54] Creating trainer with model on device: cuda
[2021-03-18 13:34:54] Training vgg-pretrained across 20000 data points in cifar10...
[2021-03-18 13:37:56] Training accuracy: 0.9992
[2021-03-18 13:37:56] Testing on 10000 data points...
[2021-03-18 13:37:58] Test score for 20000 training labels: 0.8512
[2021-03-18 13:37:58] Seeded: 5
[2021-03-18 13:37:58] Running: experiment 1
[2021-03-18 13:37:58] Showing labelled: True (5000/50000 visible)
[2021-03-18 13:37:58] Creating pretrained=True VGG16...
[2021-03-18 13:38:00] No parameter reset since we are using a pretrained model.
[2021-03-18 13:38:00] vgg-pretrained: initialized 15250250 parameters.
[2021-03-18 13:38:00] Creating trainer with model on device: cuda
[2021-03-18 13:38:00] Training vgg-pretrained across 5000 data points in cifar10...
[2021-03-18 13:38:46] Training accuracy: 0.9974
[2021-03-18 13:38:46] Testing on 10000 data points...
[2021-03-18 13:38:48] Test score for 5000 training labels: 0.7880
[2021-03-18 13:38:48] Showing labelled: False (45000/50000 visible)
[2021-03-18 13:38:48] Found 45000 unlabelled features.
[2021-03-18 13:38:58] Computing distance between 5000 labelled and 45000 unlabelled vectors of length 512...
[2021-03-18 13:39:11] Searching for coresets greedily...
[2021-03-18 13:39:20] Showing labelled: True (10000/50000 visible)
[2021-03-18 13:39:20] Creating pretrained=True VGG16...
[2021-03-18 13:39:21] No parameter reset since we are using a pretrained model.
[2021-03-18 13:39:21] vgg-pretrained: initialized 15250250 parameters.
[2021-03-18 13:39:21] Creating trainer with model on device: cuda
[2021-03-18 13:39:21] Training vgg-pretrained across 10000 data points in cifar10...
[2021-03-18 13:40:53] Training accuracy: 0.9993
[2021-03-18 13:40:53] Testing on 10000 data points...
[2021-03-18 13:40:55] Test score for 10000 training labels: 0.8216
[2021-03-18 13:40:55] Showing labelled: False (40000/50000 visible)
[2021-03-18 13:40:55] Found 40000 unlabelled features.
[2021-03-18 13:41:06] Computing distance between 10000 labelled and 40000 unlabelled vectors of length 512...
[2021-03-18 13:41:30] Searching for coresets greedily...
[2021-03-18 13:41:38] Showing labelled: True (15000/50000 visible)
[2021-03-18 13:41:38] Creating pretrained=True VGG16...
[2021-03-18 13:41:39] No parameter reset since we are using a pretrained model.
[2021-03-18 13:41:39] vgg-pretrained: initialized 15250250 parameters.
[2021-03-18 13:41:39] Creating trainer with model on device: cuda
[2021-03-18 13:41:39] Training vgg-pretrained across 15000 data points in cifar10...
[2021-03-18 13:43:55] Training accuracy: 0.9989
[2021-03-18 13:43:55] Testing on 10000 data points...
[2021-03-18 13:43:57] Test score for 15000 training labels: 0.8457
[2021-03-18 13:43:57] Showing labelled: False (35000/50000 visible)
[2021-03-18 13:43:57] Found 35000 unlabelled features.
[2021-03-18 13:44:08] Computing distance between 15000 labelled and 35000 unlabelled vectors of length 512...
[2021-03-18 13:44:39] Searching for coresets greedily...
[2021-03-18 13:44:46] Showing labelled: True (20000/50000 visible)
[2021-03-18 13:44:46] Creating pretrained=True VGG16...
[2021-03-18 13:44:48] No parameter reset since we are using a pretrained model.
[2021-03-18 13:44:48] vgg-pretrained: initialized 15250250 parameters.
[2021-03-18 13:44:48] Creating trainer with model on device: cuda
[2021-03-18 13:44:48] Training vgg-pretrained across 20000 data points in cifar10...
[2021-03-18 13:47:50] Training accuracy: 0.9876
[2021-03-18 13:47:50] Testing on 10000 data points...
[2021-03-18 13:47:52] Test score for 20000 training labels: 0.8446
[2021-03-18 13:47:52] Seeded: 5
[2021-03-18 13:47:52] Running: experiment 2
[2021-03-18 13:47:52] Showing labelled: True (5000/50000 visible)
[2021-03-18 13:47:52] Creating pretrained=True VGG16...
[2021-03-18 13:47:53] No parameter reset since we are using a pretrained model.
[2021-03-18 13:47:53] vgg-pretrained: initialized 15250250 parameters.
[2021-03-18 13:47:53] Creating trainer with model on device: cuda
[2021-03-18 13:47:53] Training vgg-pretrained across 5000 data points in cifar10...
[2021-03-18 13:48:40] Training accuracy: 0.9458
[2021-03-18 13:48:40] Testing on 10000 data points...
[2021-03-18 13:48:42] Test score for 5000 training labels: 0.7580
[2021-03-18 13:48:42] Showing labelled: False (45000/50000 visible)
[2021-03-18 13:48:42] Found 45000 unlabelled features.
[2021-03-18 13:48:52] Computing distance between 5000 labelled and 45000 unlabelled vectors of length 512...
[2021-03-18 13:49:04] Searching for coresets with 10 beams...
[2021-03-18 13:51:31] Showing labelled: True (10000/50000 visible)
[2021-03-18 13:51:31] Creating pretrained=True VGG16...
[2021-03-18 13:51:33] No parameter reset since we are using a pretrained model.
[2021-03-18 13:51:33] vgg-pretrained: initialized 15250250 parameters.
[2021-03-18 13:51:33] Creating trainer with model on device: cuda
[2021-03-18 13:51:33] Training vgg-pretrained across 10000 data points in cifar10...
[2021-03-18 13:53:04] Training accuracy: 0.9997
[2021-03-18 13:53:04] Testing on 10000 data points...
[2021-03-18 13:53:06] Test score for 10000 training labels: 0.8170
[2021-03-18 13:53:06] Showing labelled: False (40000/50000 visible)
[2021-03-18 13:53:06] Found 40000 unlabelled features.
[2021-03-18 13:53:16] Computing distance between 10000 labelled and 40000 unlabelled vectors of length 512...
[2021-03-18 13:53:40] Searching for coresets with 10 beams...
[2021-03-18 13:55:49] Showing labelled: True (15000/50000 visible)
[2021-03-18 13:55:49] Creating pretrained=True VGG16...
[2021-03-18 13:55:50] No parameter reset since we are using a pretrained model.
[2021-03-18 13:55:50] vgg-pretrained: initialized 15250250 parameters.
[2021-03-18 13:55:50] Creating trainer with model on device: cuda
[2021-03-18 13:55:50] Training vgg-pretrained across 15000 data points in cifar10...
[2021-03-18 13:58:09] Training accuracy: 0.9990
[2021-03-18 13:58:09] Testing on 10000 data points...
[2021-03-18 13:58:11] Test score for 15000 training labels: 0.8404
[2021-03-18 13:58:11] Showing labelled: False (35000/50000 visible)
[2021-03-18 13:58:11] Found 35000 unlabelled features.
[2021-03-18 13:58:22] Computing distance between 15000 labelled and 35000 unlabelled vectors of length 512...
[2021-03-18 13:58:53] Searching for coresets with 10 beams...
[2021-03-18 14:00:45] Showing labelled: True (20000/50000 visible)
[2021-03-18 14:00:45] Creating pretrained=True VGG16...
[2021-03-18 14:00:47] No parameter reset since we are using a pretrained model.
[2021-03-18 14:00:47] vgg-pretrained: initialized 15250250 parameters.
[2021-03-18 14:00:47] Creating trainer with model on device: cuda
[2021-03-18 14:00:47] Training vgg-pretrained across 20000 data points in cifar10...
[2021-03-18 14:03:51] Training accuracy: 0.9985
[2021-03-18 14:03:51] Testing on 10000 data points...
[2021-03-18 14:03:53] Test score for 20000 training labels: 0.8580
[2021-03-18 14:03:53] Seeded: 5
[2021-03-18 14:03:53] Running: experiment 3
[2021-03-18 14:03:53] Showing labelled: True (5000/50000 visible)
[2021-03-18 14:03:53] Creating pretrained=True VGG16...
[2021-03-18 14:03:54] No parameter reset since we are using a pretrained model.
[2021-03-18 14:03:54] vgg-pretrained: initialized 15250250 parameters.
[2021-03-18 14:03:54] Creating trainer with model on device: cuda
[2021-03-18 14:03:54] Training vgg-pretrained across 5000 data points in cifar10...
[2021-03-18 14:04:41] Training accuracy: 0.9961
[2021-03-18 14:04:41] Testing on 10000 data points...
[2021-03-18 14:04:43] Test score for 5000 training labels: 0.7892
[2021-03-18 14:04:43] Showing labelled: False (45000/50000 visible)
[2021-03-18 14:04:43] Found 45000 unlabelled features.
[2021-03-18 14:04:53] Showing labelled: True (10000/50000 visible)
[2021-03-18 14:04:53] Creating pretrained=True VGG16...
[2021-03-18 14:04:54] No parameter reset since we are using a pretrained model.
[2021-03-18 14:04:54] vgg-pretrained: initialized 15250250 parameters.
[2021-03-18 14:04:54] Creating trainer with model on device: cuda
[2021-03-18 14:04:54] Training vgg-pretrained across 10000 data points in cifar10...
[2021-03-18 14:06:27] Training accuracy: 0.9965
[2021-03-18 14:06:27] Testing on 10000 data points...
[2021-03-18 14:06:29] Test score for 10000 training labels: 0.8448
[2021-03-18 14:06:29] Showing labelled: False (40000/50000 visible)
[2021-03-18 14:06:29] Found 40000 unlabelled features.
[2021-03-18 14:06:38] Showing labelled: True (15000/50000 visible)
[2021-03-18 14:06:38] Creating pretrained=True VGG16...
[2021-03-18 14:06:39] No parameter reset since we are using a pretrained model.
[2021-03-18 14:06:39] vgg-pretrained: initialized 15250250 parameters.
[2021-03-18 14:06:39] Creating trainer with model on device: cuda
[2021-03-18 14:06:39] Training vgg-pretrained across 15000 data points in cifar10...
[2021-03-18 14:08:56] Training accuracy: 0.9977
[2021-03-18 14:08:56] Testing on 10000 data points...
[2021-03-18 14:08:58] Test score for 15000 training labels: 0.8684
[2021-03-18 14:08:58] Showing labelled: False (35000/50000 visible)
[2021-03-18 14:08:58] Found 35000 unlabelled features.
[2021-03-18 14:09:06] Showing labelled: True (20000/50000 visible)
[2021-03-18 14:09:06] Creating pretrained=True VGG16...
[2021-03-18 14:09:08] No parameter reset since we are using a pretrained model.
[2021-03-18 14:09:08] vgg-pretrained: initialized 15250250 parameters.
[2021-03-18 14:09:08] Creating trainer with model on device: cuda
[2021-03-18 14:09:08] Training vgg-pretrained across 20000 data points in cifar10...
[2021-03-18 14:12:10] Training accuracy: 0.9966
[2021-03-18 14:12:10] Testing on 10000 data points...
[2021-03-18 14:12:11] Test score for 20000 training labels: 0.8808
[2021-03-18 14:12:11] Updated results: ../results/cifar10/vgg_pretrained/results.json
