[2021-03-20 22:24:27] Created experiment 0:
[2021-03-20 22:24:27]  - Model: vgg-pretrained
[2021-03-20 22:24:27]  - Acquisition function: lc-beam-pweighted-relconf-coreset
[2021-03-20 22:24:27] Created experiment 1:
[2021-03-20 22:24:27]  - Model: vgg-pretrained
[2021-03-20 22:24:27]  - Acquisition function: random
[2021-03-20 22:24:27] Created experiment 2:
[2021-03-20 22:24:27]  - Model: vgg-pretrained
[2021-03-20 22:24:27]  - Acquisition function: greedy-coreset
[2021-03-20 22:24:27] Created experiment 3:
[2021-03-20 22:24:27]  - Model: vgg-pretrained
[2021-03-20 22:24:27]  - Acquisition function: lc-beam-pweighted-relconf-coreset
[2021-03-20 22:24:27] Created experiment 4:
[2021-03-20 22:24:27]  - Model: vgg-pretrained
[2021-03-20 22:24:27]  - Acquisition function: lc-beam-pweighted-coreset
[2021-03-20 22:24:27] Loading cifar10 test set...
[2021-03-20 22:24:28] Experiment repeat 1/1
[2021-03-20 22:24:28] Seeded: 5
[2021-03-20 22:24:28] Using 10.00% labels of the dataset (5000/50000)
[2021-03-20 22:24:28] Randomly labelled 5000/50000
[2021-03-20 22:24:28] Showing labelled: True (5000/50000 visible, 0 redundant)
[2021-03-20 22:24:28] Seeded: 6
[2021-03-20 22:24:28] Running: experiment 0
[2021-03-20 22:24:28] Showing labelled: True (5000/50000 visible, 0 redundant)
[2021-03-20 22:24:28] Creating pretrained=True VGG16...
[2021-03-20 22:24:29] No parameter reset since we are using a pretrained model.
[2021-03-20 22:24:29] vgg-pretrained: initialized 15245130 parameters.
[2021-03-20 22:24:29] Creating trainer with model on device: cuda
[2021-03-20 22:24:34] Training vgg-pretrained across 5000 data points in cifar10...
[2021-03-20 22:25:50] Training accuracy: 0.9949
[2021-03-20 22:25:50] Testing on 10000 data points...
[2021-03-20 22:25:52] Test score for 5000 training labels: 0.8054
[2021-03-20 22:25:52] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-20 22:25:52] Found 50000 unlabelled features.
[2021-03-20 22:26:04] Computing distance between 5000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-20 22:26:19] Searching for coresets with 10 beams...
[2021-03-20 22:29:03] Showing labelled: True (10000/50000 visible, 0 redundant)
[2021-03-20 22:29:03] Creating trainer with model on device: cuda
[2021-03-20 22:29:03] Training vgg-pretrained across 10000 data points in cifar10...
[2021-03-20 22:30:19] Training accuracy: 0.9917
[2021-03-20 22:30:19] Testing on 10000 data points...
[2021-03-20 22:30:21] Test score for 10000 training labels: 0.8547
[2021-03-20 22:30:21] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-20 22:30:21] Found 50000 unlabelled features.
[2021-03-20 22:30:34] Computing distance between 10000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-20 22:31:06] Searching for coresets with 10 beams...
[2021-03-20 22:33:50] Showing labelled: True (15000/50000 visible, 0 redundant)
[2021-03-20 22:33:50] Creating trainer with model on device: cuda
[2021-03-20 22:33:50] Training vgg-pretrained across 15000 data points in cifar10...
[2021-03-20 22:35:43] Training accuracy: 0.9970
[2021-03-20 22:35:43] Testing on 10000 data points...
[2021-03-20 22:35:45] Test score for 15000 training labels: 0.8745
[2021-03-20 22:35:45] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-20 22:35:45] Found 50000 unlabelled features.
[2021-03-20 22:35:58] Computing distance between 15000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-20 22:36:45] Searching for coresets with 10 beams...
[2021-03-20 22:39:30] Showing labelled: True (20000/50000 visible, 0 redundant)
[2021-03-20 22:39:30] Creating trainer with model on device: cuda
[2021-03-20 22:39:30] Training vgg-pretrained across 20000 data points in cifar10...
[2021-03-20 22:42:00] Training accuracy: 0.9975
[2021-03-20 22:42:00] Testing on 10000 data points...
[2021-03-20 22:42:02] Test score for 20000 training labels: 0.8867
[2021-03-20 22:42:02] Seeded: 6
[2021-03-20 22:42:02] Running: experiment 1
[2021-03-20 22:42:02] Showing labelled: True (5000/50000 visible, 0 redundant)
[2021-03-20 22:42:02] Creating pretrained=True VGG16...
[2021-03-20 22:42:04] No parameter reset since we are using a pretrained model.
[2021-03-20 22:42:04] vgg-pretrained: initialized 15245130 parameters.
[2021-03-20 22:42:04] Creating trainer with model on device: cuda
[2021-03-20 22:42:04] Training vgg-pretrained across 5000 data points in cifar10...
[2021-03-20 22:43:20] Training accuracy: 0.9938
[2021-03-20 22:43:20] Testing on 10000 data points...
[2021-03-20 22:43:22] Test score for 5000 training labels: 0.8059
[2021-03-20 22:43:22] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-20 22:43:22] Found 50000 unlabelled features.
[2021-03-20 22:43:28] Showing labelled: True (9512/50000 visible, 488 redundant)
[2021-03-20 22:43:28] Creating trainer with model on device: cuda
[2021-03-20 22:43:28] Training vgg-pretrained across 10000 data points in cifar10...
[2021-03-20 22:44:44] Training accuracy: 0.9921
[2021-03-20 22:44:44] Testing on 10000 data points...
[2021-03-20 22:44:46] Test score for 10000 training labels: 0.8298
[2021-03-20 22:44:46] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-20 22:44:46] Found 50000 unlabelled features.
[2021-03-20 22:44:51] Showing labelled: True (13552/50000 visible, 1448 redundant)
[2021-03-20 22:44:51] Creating trainer with model on device: cuda
[2021-03-20 22:44:51] Training vgg-pretrained across 15000 data points in cifar10...
[2021-03-20 22:46:45] Training accuracy: 0.9998
[2021-03-20 22:46:45] Testing on 10000 data points...
[2021-03-20 22:46:47] Test score for 15000 training labels: 0.8448
[2021-03-20 22:46:47] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-20 22:46:47] Found 50000 unlabelled features.
[2021-03-20 22:46:52] Showing labelled: True (17219/50000 visible, 2781 redundant)
[2021-03-20 22:46:52] Creating trainer with model on device: cuda
[2021-03-20 22:46:52] Training vgg-pretrained across 20000 data points in cifar10...
[2021-03-20 22:49:24] Training accuracy: 0.9995
[2021-03-20 22:49:24] Testing on 10000 data points...
[2021-03-20 22:49:26] Test score for 20000 training labels: 0.8533
[2021-03-20 22:49:26] Seeded: 6
[2021-03-20 22:49:26] Running: experiment 2
[2021-03-20 22:49:26] Showing labelled: True (5000/50000 visible, 0 redundant)
[2021-03-20 22:49:26] Creating pretrained=True VGG16...
[2021-03-20 22:49:28] No parameter reset since we are using a pretrained model.
[2021-03-20 22:49:28] vgg-pretrained: initialized 15245130 parameters.
[2021-03-20 22:49:28] Creating trainer with model on device: cuda
[2021-03-20 22:49:28] Training vgg-pretrained across 5000 data points in cifar10...
[2021-03-20 22:50:44] Training accuracy: 0.9942
[2021-03-20 22:50:44] Testing on 10000 data points...
[2021-03-20 22:50:46] Test score for 5000 training labels: 0.8036
[2021-03-20 22:50:46] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-20 22:50:46] Found 50000 unlabelled features.
[2021-03-20 22:50:58] Computing distance between 5000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-20 22:51:12] Searching for coresets greedily...
[2021-03-20 22:51:22] Showing labelled: True (10000/50000 visible, 0 redundant)
[2021-03-20 22:51:22] Creating trainer with model on device: cuda
[2021-03-20 22:51:22] Training vgg-pretrained across 10000 data points in cifar10...
[2021-03-20 22:52:37] Training accuracy: 0.9944
[2021-03-20 22:52:37] Testing on 10000 data points...
[2021-03-20 22:52:39] Test score for 10000 training labels: 0.8395
[2021-03-20 22:52:39] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-20 22:52:39] Found 50000 unlabelled features.
[2021-03-20 22:52:52] Computing distance between 10000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-20 22:53:21] Searching for coresets greedily...
[2021-03-20 22:53:30] Showing labelled: True (15000/50000 visible, 0 redundant)
[2021-03-20 22:53:30] Creating trainer with model on device: cuda
[2021-03-20 22:53:30] Training vgg-pretrained across 15000 data points in cifar10...
[2021-03-20 22:55:26] Training accuracy: 0.9949
[2021-03-20 22:55:26] Testing on 10000 data points...
[2021-03-20 22:55:28] Test score for 15000 training labels: 0.8625
[2021-03-20 22:55:28] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-20 22:55:28] Found 50000 unlabelled features.
[2021-03-20 22:55:42] Computing distance between 15000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-20 22:56:28] Searching for coresets greedily...
[2021-03-20 22:56:38] Showing labelled: True (20000/50000 visible, 0 redundant)
[2021-03-20 22:56:38] Creating trainer with model on device: cuda
[2021-03-20 22:56:38] Training vgg-pretrained across 20000 data points in cifar10...
[2021-03-20 22:59:08] Training accuracy: 0.9974
[2021-03-20 22:59:08] Testing on 10000 data points...
[2021-03-20 22:59:11] Test score for 20000 training labels: 0.8784
[2021-03-20 22:59:11] Seeded: 6
[2021-03-20 22:59:11] Running: experiment 3
[2021-03-20 22:59:11] Showing labelled: True (5000/50000 visible, 0 redundant)
[2021-03-20 22:59:11] Creating pretrained=True VGG16...
[2021-03-20 22:59:12] No parameter reset since we are using a pretrained model.
[2021-03-20 22:59:12] vgg-pretrained: initialized 15245130 parameters.
[2021-03-20 22:59:12] Creating trainer with model on device: cuda
[2021-03-20 22:59:12] Training vgg-pretrained across 5000 data points in cifar10...
[2021-03-20 23:00:28] Training accuracy: 0.9919
[2021-03-20 23:00:28] Testing on 10000 data points...
[2021-03-20 23:00:30] Test score for 5000 training labels: 0.8051
[2021-03-20 23:00:30] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-20 23:00:30] Found 50000 unlabelled features.
[2021-03-20 23:00:42] Computing distance between 5000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-20 23:00:56] Searching for coresets with 1 beams...
[2021-03-20 23:01:12] Showing labelled: True (10000/50000 visible, 0 redundant)
[2021-03-20 23:01:12] Creating trainer with model on device: cuda
[2021-03-20 23:01:12] Training vgg-pretrained across 10000 data points in cifar10...
[2021-03-20 23:02:29] Training accuracy: 0.9924
[2021-03-20 23:02:29] Testing on 10000 data points...
[2021-03-20 23:02:31] Test score for 10000 training labels: 0.8532
[2021-03-20 23:02:31] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-20 23:02:31] Found 50000 unlabelled features.
[2021-03-20 23:02:44] Computing distance between 10000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-20 23:03:15] Searching for coresets with 1 beams...
[2021-03-20 23:03:31] Showing labelled: True (15000/50000 visible, 0 redundant)
[2021-03-20 23:03:31] Creating trainer with model on device: cuda
[2021-03-20 23:03:31] Training vgg-pretrained across 15000 data points in cifar10...
[2021-03-20 23:07:30] Training accuracy: 0.9960
[2021-03-20 23:07:30] Testing on 10000 data points...
[2021-03-20 23:07:32] Test score for 15000 training labels: 0.8746
[2021-03-20 23:07:32] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-20 23:07:32] Found 50000 unlabelled features.
[2021-03-20 23:07:46] Computing distance between 15000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-20 23:08:32] Searching for coresets with 1 beams...
[2021-03-20 23:08:47] Showing labelled: True (20000/50000 visible, 0 redundant)
[2021-03-20 23:08:47] Creating trainer with model on device: cuda
[2021-03-20 23:08:47] Training vgg-pretrained across 20000 data points in cifar10...
[2021-03-20 23:11:16] Training accuracy: 0.9992
[2021-03-20 23:11:16] Testing on 10000 data points...
[2021-03-20 23:11:18] Test score for 20000 training labels: 0.8845
[2021-03-20 23:11:18] Seeded: 6
[2021-03-20 23:11:18] Running: experiment 4
[2021-03-20 23:11:18] Showing labelled: True (5000/50000 visible, 0 redundant)
[2021-03-20 23:11:18] Creating pretrained=True VGG16...
[2021-03-20 23:11:20] No parameter reset since we are using a pretrained model.
[2021-03-20 23:11:20] vgg-pretrained: initialized 15245130 parameters.
[2021-03-20 23:11:20] Creating trainer with model on device: cuda
[2021-03-20 23:11:20] Training vgg-pretrained across 5000 data points in cifar10...
[2021-03-20 23:12:34] Training accuracy: 0.9931
[2021-03-20 23:12:34] Testing on 10000 data points...
[2021-03-20 23:12:36] Test score for 5000 training labels: 0.8073
[2021-03-20 23:12:36] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-20 23:12:36] Found 50000 unlabelled features.
[2021-03-20 23:12:47] Computing distance between 5000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-20 23:13:01] Searching for coresets with 10 beams...
[2021-03-20 23:15:39] Showing labelled: True (10000/50000 visible, 0 redundant)
[2021-03-20 23:15:39] Creating trainer with model on device: cuda
[2021-03-20 23:15:39] Training vgg-pretrained across 10000 data points in cifar10...
[2021-03-20 23:16:55] Training accuracy: 0.9897
[2021-03-20 23:16:55] Testing on 10000 data points...
[2021-03-20 23:16:57] Test score for 10000 training labels: 0.8576
[2021-03-20 23:16:57] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-20 23:16:57] Found 50000 unlabelled features.
[2021-03-20 23:17:10] Computing distance between 10000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-20 23:17:38] Searching for coresets with 10 beams...
[2021-03-20 23:20:16] Showing labelled: True (15000/50000 visible, 0 redundant)
[2021-03-20 23:20:16] Creating trainer with model on device: cuda
[2021-03-20 23:20:16] Training vgg-pretrained across 15000 data points in cifar10...
[2021-03-20 23:22:09] Training accuracy: 0.9955
[2021-03-20 23:22:09] Testing on 10000 data points...
[2021-03-20 23:22:11] Test score for 15000 training labels: 0.8759
[2021-03-20 23:22:11] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-20 23:22:11] Found 50000 unlabelled features.
[2021-03-20 23:22:24] Computing distance between 15000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-20 23:23:10] Searching for coresets with 10 beams...
[2021-03-20 23:25:48] Showing labelled: True (20000/50000 visible, 0 redundant)
[2021-03-20 23:25:48] Creating trainer with model on device: cuda
[2021-03-20 23:25:48] Training vgg-pretrained across 20000 data points in cifar10...
[2021-03-20 23:28:19] Training accuracy: 0.9987
[2021-03-20 23:28:19] Testing on 10000 data points...
[2021-03-20 23:28:21] Test score for 20000 training labels: 0.8882
[2021-03-20 23:28:21] Updated results: ../results/cifar10/5k-steps/results.json
