[2021-03-21 12:30:11] Created experiment 0:
[2021-03-21 12:30:11]  - Model: vgg-pretrained
[2021-03-21 12:30:11]  - Acquisition function: lc-beam-pweighted-relconf-coreset
[2021-03-21 12:30:11] Loading cifar10 test set...
[2021-03-21 12:30:12] Experiment repeat 1/5
[2021-03-21 12:30:12] Seeded: 5
[2021-03-21 12:30:12] Using 2.00% labels of the dataset (1000/50000)
[2021-03-21 12:30:12] Randomly labelled 1000/50000
[2021-03-21 12:30:12] Showing labelled: True (1000/50000 visible, 0 redundant)
[2021-03-21 12:30:12] Seeded: 6
[2021-03-21 12:30:12] Running: experiment 0
[2021-03-21 12:30:12] Showing labelled: True (1000/50000 visible, 0 redundant)
[2021-03-21 12:30:12] Creating pretrained=True VGG16...
[2021-03-21 12:30:14] No parameter reset since we are using a pretrained model.
[2021-03-21 12:30:14] vgg-pretrained: initialized 15245130 parameters.
[2021-03-21 12:30:14] Creating trainer with model on device: cuda
[2021-03-21 12:30:19] Training vgg-pretrained across 1000 data points in cifar10...
[2021-03-21 12:31:18] Training accuracy: 0.9995
[2021-03-21 12:31:18] Testing on 10000 data points...
[2021-03-21 12:31:20] Test score for 1000 training labels: 0.7222
[2021-03-21 12:31:20] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 12:31:20] Found 50000 unlabelled features.
[2021-03-21 12:31:31] Computing distance between 1000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 12:31:34] Searching for coresets with 20 beams...
[2021-03-21 12:32:03] Showing labelled: True (1400/50000 visible, 0 redundant)
[2021-03-21 12:32:03] Creating trainer with model on device: cuda
[2021-03-21 12:32:03] Training vgg-pretrained across 1400 data points in cifar10...
[2021-03-21 12:32:24] Training accuracy: 0.9915
[2021-03-21 12:32:24] Testing on 10000 data points...
[2021-03-21 12:32:26] Test score for 1400 training labels: 0.7325
[2021-03-21 12:32:26] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 12:32:26] Found 50000 unlabelled features.
[2021-03-21 12:32:36] Computing distance between 1400 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 12:32:40] Searching for coresets with 20 beams...
[2021-03-21 12:33:09] Showing labelled: True (1800/50000 visible, 0 redundant)
[2021-03-21 12:33:09] Creating trainer with model on device: cuda
[2021-03-21 12:33:09] Training vgg-pretrained across 1800 data points in cifar10...
[2021-03-21 12:33:36] Training accuracy: 0.9979
[2021-03-21 12:33:36] Testing on 10000 data points...
[2021-03-21 12:33:38] Test score for 1800 training labels: 0.7668
[2021-03-21 12:33:38] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 12:33:38] Found 50000 unlabelled features.
[2021-03-21 12:33:49] Computing distance between 1800 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 12:33:54] Searching for coresets with 20 beams...
[2021-03-21 12:34:23] Showing labelled: True (2200/50000 visible, 0 redundant)
[2021-03-21 12:34:23] Creating trainer with model on device: cuda
[2021-03-21 12:34:23] Training vgg-pretrained across 2200 data points in cifar10...
[2021-03-21 12:34:57] Training accuracy: 0.9992
[2021-03-21 12:34:57] Testing on 10000 data points...
[2021-03-21 12:34:59] Test score for 2200 training labels: 0.7771
[2021-03-21 12:34:59] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 12:34:59] Found 50000 unlabelled features.
[2021-03-21 12:35:11] Computing distance between 2200 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 12:35:17] Searching for coresets with 20 beams...
[2021-03-21 12:35:46] Showing labelled: True (2600/50000 visible, 0 redundant)
[2021-03-21 12:35:46] Creating trainer with model on device: cuda
[2021-03-21 12:35:46] Training vgg-pretrained across 2600 data points in cifar10...
[2021-03-21 12:36:26] Training accuracy: 1.0000
[2021-03-21 12:36:26] Testing on 10000 data points...
[2021-03-21 12:36:28] Test score for 2600 training labels: 0.7955
[2021-03-21 12:36:28] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 12:36:28] Found 50000 unlabelled features.
[2021-03-21 12:36:39] Computing distance between 2600 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 12:36:47] Searching for coresets with 20 beams...
[2021-03-21 12:37:16] Showing labelled: True (3000/50000 visible, 0 redundant)
[2021-03-21 12:37:16] Creating trainer with model on device: cuda
[2021-03-21 12:37:16] Training vgg-pretrained across 3000 data points in cifar10...
[2021-03-21 12:38:01] Training accuracy: 1.0000
[2021-03-21 12:38:01] Testing on 10000 data points...
[2021-03-21 12:38:03] Test score for 3000 training labels: 0.8058
[2021-03-21 12:38:03] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 12:38:03] Found 50000 unlabelled features.
[2021-03-21 12:38:14] Computing distance between 3000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 12:38:23] Searching for coresets with 20 beams...
[2021-03-21 12:38:52] Showing labelled: True (3400/50000 visible, 0 redundant)
[2021-03-21 12:38:52] Creating trainer with model on device: cuda
[2021-03-21 12:38:52] Training vgg-pretrained across 3400 data points in cifar10...
[2021-03-21 12:39:44] Training accuracy: 1.0000
[2021-03-21 12:39:44] Testing on 10000 data points...
[2021-03-21 12:39:46] Test score for 3400 training labels: 0.8174
[2021-03-21 12:39:46] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 12:39:46] Found 50000 unlabelled features.
[2021-03-21 12:39:58] Computing distance between 3400 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 12:40:09] Searching for coresets with 20 beams...
[2021-03-21 12:40:38] Showing labelled: True (3800/50000 visible, 0 redundant)
[2021-03-21 12:40:38] Creating trainer with model on device: cuda
[2021-03-21 12:40:38] Training vgg-pretrained across 3800 data points in cifar10...
[2021-03-21 12:41:36] Training accuracy: 1.0000
[2021-03-21 12:41:36] Testing on 10000 data points...
[2021-03-21 12:41:38] Test score for 3800 training labels: 0.8209
[2021-03-21 12:41:38] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 12:41:38] Found 50000 unlabelled features.
[2021-03-21 12:41:49] Computing distance between 3800 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 12:42:01] Searching for coresets with 20 beams...
[2021-03-21 12:42:30] Showing labelled: True (4200/50000 visible, 0 redundant)
[2021-03-21 12:42:30] Creating trainer with model on device: cuda
[2021-03-21 12:42:30] Training vgg-pretrained across 4200 data points in cifar10...
[2021-03-21 12:43:33] Training accuracy: 1.0000
[2021-03-21 12:43:33] Testing on 10000 data points...
[2021-03-21 12:43:35] Test score for 4200 training labels: 0.8275
[2021-03-21 12:43:35] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 12:43:35] Found 50000 unlabelled features.
[2021-03-21 12:43:47] Computing distance between 4200 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 12:43:59] Searching for coresets with 20 beams...
[2021-03-21 12:44:28] Showing labelled: True (4600/50000 visible, 0 redundant)
[2021-03-21 12:44:28] Creating trainer with model on device: cuda
[2021-03-21 12:44:28] Training vgg-pretrained across 4600 data points in cifar10...
[2021-03-21 12:45:44] Training accuracy: 0.9992
[2021-03-21 12:45:44] Testing on 10000 data points...
[2021-03-21 12:45:46] Test score for 4600 training labels: 0.8286
[2021-03-21 12:45:46] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 12:45:46] Found 50000 unlabelled features.
[2021-03-21 12:45:58] Computing distance between 4600 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 12:46:12] Searching for coresets with 20 beams...
[2021-03-21 12:46:41] Showing labelled: True (5000/50000 visible, 0 redundant)
[2021-03-21 12:46:41] Creating trainer with model on device: cuda
[2021-03-21 12:46:41] Training vgg-pretrained across 5000 data points in cifar10...
[2021-03-21 12:47:56] Training accuracy: 1.0000
[2021-03-21 12:47:56] Testing on 10000 data points...
[2021-03-21 12:47:58] Test score for 5000 training labels: 0.8351
[2021-03-21 12:47:58] Experiment repeat 2/5
[2021-03-21 12:47:58] Seeded: 6
[2021-03-21 12:47:58] Using 2.00% labels of the dataset (1000/50000)
[2021-03-21 12:47:58] Randomly labelled 1000/50000
[2021-03-21 12:47:58] Showing labelled: True (1000/50000 visible, 0 redundant)
[2021-03-21 12:47:58] Seeded: 7
[2021-03-21 12:47:58] Running: experiment 0
[2021-03-21 12:47:58] Showing labelled: True (1000/50000 visible, 0 redundant)
[2021-03-21 12:47:58] Creating pretrained=True VGG16...
[2021-03-21 12:48:00] No parameter reset since we are using a pretrained model.
[2021-03-21 12:48:00] vgg-pretrained: initialized 15245130 parameters.
[2021-03-21 12:48:00] Creating trainer with model on device: cuda
[2021-03-21 12:48:00] Training vgg-pretrained across 1000 data points in cifar10...
[2021-03-21 12:49:00] Training accuracy: 0.9993
[2021-03-21 12:49:00] Testing on 10000 data points...
[2021-03-21 12:49:02] Test score for 1000 training labels: 0.7370
[2021-03-21 12:49:02] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 12:49:02] Found 50000 unlabelled features.
[2021-03-21 12:49:12] Computing distance between 1000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 12:49:16] Searching for coresets with 20 beams...
[2021-03-21 12:49:45] Showing labelled: True (1400/50000 visible, 0 redundant)
[2021-03-21 12:49:45] Creating trainer with model on device: cuda
[2021-03-21 12:49:45] Training vgg-pretrained across 1400 data points in cifar10...
[2021-03-21 12:50:06] Training accuracy: 0.9967
[2021-03-21 12:50:06] Testing on 10000 data points...
[2021-03-21 12:50:08] Test score for 1400 training labels: 0.7635
[2021-03-21 12:50:08] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 12:50:08] Found 50000 unlabelled features.
[2021-03-21 12:50:19] Computing distance between 1400 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 12:50:23] Searching for coresets with 20 beams...
[2021-03-21 12:50:52] Showing labelled: True (1800/50000 visible, 0 redundant)
[2021-03-21 12:50:52] Creating trainer with model on device: cuda
[2021-03-21 12:50:52] Training vgg-pretrained across 1800 data points in cifar10...
[2021-03-21 12:51:20] Training accuracy: 0.9990
[2021-03-21 12:51:20] Testing on 10000 data points...
[2021-03-21 12:51:22] Test score for 1800 training labels: 0.7763
[2021-03-21 12:51:22] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 12:51:22] Found 50000 unlabelled features.
[2021-03-21 12:51:33] Computing distance between 1800 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 12:51:38] Searching for coresets with 20 beams...
[2021-03-21 12:52:07] Showing labelled: True (2200/50000 visible, 0 redundant)
[2021-03-21 12:52:07] Creating trainer with model on device: cuda
[2021-03-21 12:52:07] Training vgg-pretrained across 2200 data points in cifar10...
[2021-03-21 12:52:41] Training accuracy: 0.9989
[2021-03-21 12:52:41] Testing on 10000 data points...
[2021-03-21 12:52:43] Test score for 2200 training labels: 0.7929
[2021-03-21 12:52:43] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 12:52:43] Found 50000 unlabelled features.
[2021-03-21 12:52:54] Computing distance between 2200 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 12:53:01] Searching for coresets with 20 beams...
[2021-03-21 12:53:30] Showing labelled: True (2600/50000 visible, 0 redundant)
[2021-03-21 12:53:30] Creating trainer with model on device: cuda
[2021-03-21 12:53:30] Training vgg-pretrained across 2600 data points in cifar10...
[2021-03-21 12:54:09] Training accuracy: 0.9996
[2021-03-21 12:54:09] Testing on 10000 data points...
[2021-03-21 12:54:12] Test score for 2600 training labels: 0.8040
[2021-03-21 12:54:12] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 12:54:12] Found 50000 unlabelled features.
[2021-03-21 12:54:22] Computing distance between 2600 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 12:54:30] Searching for coresets with 20 beams...
[2021-03-21 12:54:59] Showing labelled: True (3000/50000 visible, 0 redundant)
[2021-03-21 12:54:59] Creating trainer with model on device: cuda
[2021-03-21 12:54:59] Training vgg-pretrained across 3000 data points in cifar10...
[2021-03-21 12:55:44] Training accuracy: 0.9986
[2021-03-21 12:55:44] Testing on 10000 data points...
[2021-03-21 12:55:46] Test score for 3000 training labels: 0.8034
[2021-03-21 12:55:47] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 12:55:47] Found 50000 unlabelled features.
[2021-03-21 12:55:58] Computing distance between 3000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 12:56:06] Searching for coresets with 20 beams...
[2021-03-21 12:56:35] Showing labelled: True (3400/50000 visible, 0 redundant)
[2021-03-21 12:56:35] Creating trainer with model on device: cuda
[2021-03-21 12:56:35] Training vgg-pretrained across 3400 data points in cifar10...
[2021-03-21 12:57:28] Training accuracy: 1.0000
[2021-03-21 12:57:28] Testing on 10000 data points...
[2021-03-21 12:57:30] Test score for 3400 training labels: 0.8134
[2021-03-21 12:57:30] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 12:57:30] Found 50000 unlabelled features.
[2021-03-21 12:57:41] Computing distance between 3400 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 12:57:52] Searching for coresets with 20 beams...
[2021-03-21 12:58:21] Showing labelled: True (3800/50000 visible, 0 redundant)
[2021-03-21 12:58:21] Creating trainer with model on device: cuda
[2021-03-21 12:58:21] Training vgg-pretrained across 3800 data points in cifar10...
[2021-03-21 12:59:42] Training accuracy: 1.0000
[2021-03-21 12:59:42] Testing on 10000 data points...
[2021-03-21 12:59:44] Test score for 3800 training labels: 0.8196
[2021-03-21 12:59:44] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 12:59:44] Found 50000 unlabelled features.
[2021-03-21 12:59:56] Computing distance between 3800 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:00:06] Searching for coresets with 20 beams...
[2021-03-21 13:00:34] Showing labelled: True (4200/50000 visible, 0 redundant)
[2021-03-21 13:00:34] Creating trainer with model on device: cuda
[2021-03-21 13:00:34] Training vgg-pretrained across 4200 data points in cifar10...
[2021-03-21 13:01:37] Training accuracy: 1.0000
[2021-03-21 13:01:37] Testing on 10000 data points...
[2021-03-21 13:01:39] Test score for 4200 training labels: 0.8230
[2021-03-21 13:01:39] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:01:39] Found 50000 unlabelled features.
[2021-03-21 13:01:51] Computing distance between 4200 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:02:03] Searching for coresets with 20 beams...
[2021-03-21 13:02:31] Showing labelled: True (4600/50000 visible, 0 redundant)
[2021-03-21 13:02:31] Creating trainer with model on device: cuda
[2021-03-21 13:02:31] Training vgg-pretrained across 4600 data points in cifar10...
[2021-03-21 13:03:40] Training accuracy: 1.0000
[2021-03-21 13:03:40] Testing on 10000 data points...
[2021-03-21 13:03:42] Test score for 4600 training labels: 0.8313
[2021-03-21 13:03:42] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:03:42] Found 50000 unlabelled features.
[2021-03-21 13:03:54] Computing distance between 4600 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:04:07] Searching for coresets with 20 beams...
[2021-03-21 13:04:35] Showing labelled: True (5000/50000 visible, 0 redundant)
[2021-03-21 13:04:35] Creating trainer with model on device: cuda
[2021-03-21 13:04:35] Training vgg-pretrained across 5000 data points in cifar10...
[2021-03-21 13:05:50] Training accuracy: 1.0000
[2021-03-21 13:05:50] Testing on 10000 data points...
[2021-03-21 13:05:52] Test score for 5000 training labels: 0.8371
[2021-03-21 13:05:52] Experiment repeat 3/5
[2021-03-21 13:05:52] Seeded: 7
[2021-03-21 13:05:52] Using 2.00% labels of the dataset (1000/50000)
[2021-03-21 13:05:52] Randomly labelled 1000/50000
[2021-03-21 13:05:52] Showing labelled: True (1000/50000 visible, 0 redundant)
[2021-03-21 13:05:52] Seeded: 8
[2021-03-21 13:05:52] Running: experiment 0
[2021-03-21 13:05:52] Showing labelled: True (1000/50000 visible, 0 redundant)
[2021-03-21 13:05:52] Creating pretrained=True VGG16...
[2021-03-21 13:05:54] No parameter reset since we are using a pretrained model.
[2021-03-21 13:05:54] vgg-pretrained: initialized 15245130 parameters.
[2021-03-21 13:05:54] Creating trainer with model on device: cuda
[2021-03-21 13:05:54] Training vgg-pretrained across 1000 data points in cifar10...
[2021-03-21 13:06:53] Training accuracy: 0.9999
[2021-03-21 13:06:53] Testing on 10000 data points...
[2021-03-21 13:06:55] Test score for 1000 training labels: 0.7369
[2021-03-21 13:06:55] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:06:55] Found 50000 unlabelled features.
[2021-03-21 13:07:06] Computing distance between 1000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:07:09] Searching for coresets with 20 beams...
[2021-03-21 13:07:37] Showing labelled: True (1400/50000 visible, 0 redundant)
[2021-03-21 13:07:37] Creating trainer with model on device: cuda
[2021-03-21 13:07:37] Training vgg-pretrained across 1400 data points in cifar10...
[2021-03-21 13:07:58] Training accuracy: 0.9955
[2021-03-21 13:07:58] Testing on 10000 data points...
[2021-03-21 13:08:00] Test score for 1400 training labels: 0.7604
[2021-03-21 13:08:00] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:08:00] Found 50000 unlabelled features.
[2021-03-21 13:08:11] Computing distance between 1400 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:08:15] Searching for coresets with 20 beams...
[2021-03-21 13:08:43] Showing labelled: True (1800/50000 visible, 0 redundant)
[2021-03-21 13:08:43] Creating trainer with model on device: cuda
[2021-03-21 13:08:43] Training vgg-pretrained across 1800 data points in cifar10...
[2021-03-21 13:09:10] Training accuracy: 0.9988
[2021-03-21 13:09:10] Testing on 10000 data points...
[2021-03-21 13:09:12] Test score for 1800 training labels: 0.7788
[2021-03-21 13:09:12] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:09:12] Found 50000 unlabelled features.
[2021-03-21 13:09:23] Computing distance between 1800 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:09:29] Searching for coresets with 20 beams...
[2021-03-21 13:09:57] Showing labelled: True (2200/50000 visible, 0 redundant)
[2021-03-21 13:09:57] Creating trainer with model on device: cuda
[2021-03-21 13:09:57] Training vgg-pretrained across 2200 data points in cifar10...
[2021-03-21 13:10:31] Training accuracy: 0.8739
[2021-03-21 13:10:31] Testing on 10000 data points...
[2021-03-21 13:10:33] Test score for 2200 training labels: 0.7264
[2021-03-21 13:10:33] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:10:33] Found 50000 unlabelled features.
[2021-03-21 13:10:45] Computing distance between 2200 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:10:51] Searching for coresets with 20 beams...
[2021-03-21 13:11:19] Showing labelled: True (2600/50000 visible, 0 redundant)
[2021-03-21 13:11:19] Creating trainer with model on device: cuda
[2021-03-21 13:11:19] Training vgg-pretrained across 2600 data points in cifar10...
[2021-03-21 13:11:58] Training accuracy: 0.9997
[2021-03-21 13:11:58] Testing on 10000 data points...
[2021-03-21 13:12:00] Test score for 2600 training labels: 0.8006
[2021-03-21 13:12:00] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:12:00] Found 50000 unlabelled features.
[2021-03-21 13:12:11] Computing distance between 2600 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:12:19] Searching for coresets with 20 beams...
[2021-03-21 13:12:47] Showing labelled: True (3000/50000 visible, 0 redundant)
[2021-03-21 13:12:47] Creating trainer with model on device: cuda
[2021-03-21 13:12:47] Training vgg-pretrained across 3000 data points in cifar10...
[2021-03-21 13:13:32] Training accuracy: 0.9997
[2021-03-21 13:13:32] Testing on 10000 data points...
[2021-03-21 13:13:34] Test score for 3000 training labels: 0.8047
[2021-03-21 13:13:34] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:13:34] Found 50000 unlabelled features.
[2021-03-21 13:13:45] Computing distance between 3000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:13:54] Searching for coresets with 20 beams...
[2021-03-21 13:14:22] Showing labelled: True (3400/50000 visible, 0 redundant)
[2021-03-21 13:14:22] Creating trainer with model on device: cuda
[2021-03-21 13:14:22] Training vgg-pretrained across 3400 data points in cifar10...
[2021-03-21 13:15:14] Training accuracy: 0.9990
[2021-03-21 13:15:14] Testing on 10000 data points...
[2021-03-21 13:15:16] Test score for 3400 training labels: 0.8135
[2021-03-21 13:15:16] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:15:16] Found 50000 unlabelled features.
[2021-03-21 13:15:27] Computing distance between 3400 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:15:36] Searching for coresets with 20 beams...
[2021-03-21 13:16:05] Showing labelled: True (3800/50000 visible, 0 redundant)
[2021-03-21 13:16:05] Creating trainer with model on device: cuda
[2021-03-21 13:16:05] Training vgg-pretrained across 3800 data points in cifar10...
[2021-03-21 13:17:03] Training accuracy: 0.9999
[2021-03-21 13:17:03] Testing on 10000 data points...
[2021-03-21 13:17:05] Test score for 3800 training labels: 0.8166
[2021-03-21 13:17:05] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:17:05] Found 50000 unlabelled features.
[2021-03-21 13:17:16] Computing distance between 3800 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:17:28] Searching for coresets with 20 beams...
[2021-03-21 13:17:56] Showing labelled: True (4200/50000 visible, 0 redundant)
[2021-03-21 13:17:56] Creating trainer with model on device: cuda
[2021-03-21 13:17:56] Training vgg-pretrained across 4200 data points in cifar10...
[2021-03-21 13:18:59] Training accuracy: 1.0000
[2021-03-21 13:18:59] Testing on 10000 data points...
[2021-03-21 13:19:01] Test score for 4200 training labels: 0.8273
[2021-03-21 13:19:01] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:19:01] Found 50000 unlabelled features.
[2021-03-21 13:19:13] Computing distance between 4200 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:19:25] Searching for coresets with 20 beams...
[2021-03-21 13:19:53] Showing labelled: True (4600/50000 visible, 0 redundant)
[2021-03-21 13:19:53] Creating trainer with model on device: cuda
[2021-03-21 13:19:53] Training vgg-pretrained across 4600 data points in cifar10...
[2021-03-21 13:21:02] Training accuracy: 1.0000
[2021-03-21 13:21:02] Testing on 10000 data points...
[2021-03-21 13:21:04] Test score for 4600 training labels: 0.8309
[2021-03-21 13:21:04] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:21:04] Found 50000 unlabelled features.
[2021-03-21 13:21:15] Computing distance between 4600 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:21:29] Searching for coresets with 20 beams...
[2021-03-21 13:21:57] Showing labelled: True (5000/50000 visible, 0 redundant)
[2021-03-21 13:21:57] Creating trainer with model on device: cuda
[2021-03-21 13:21:57] Training vgg-pretrained across 5000 data points in cifar10...
[2021-03-21 13:23:14] Training accuracy: 1.0000
[2021-03-21 13:23:14] Testing on 10000 data points...
[2021-03-21 13:23:16] Test score for 5000 training labels: 0.8344
[2021-03-21 13:23:16] Experiment repeat 4/5
[2021-03-21 13:23:16] Seeded: 8
[2021-03-21 13:23:16] Using 2.00% labels of the dataset (1000/50000)
[2021-03-21 13:23:16] Randomly labelled 1000/50000
[2021-03-21 13:23:16] Showing labelled: True (1000/50000 visible, 0 redundant)
[2021-03-21 13:23:16] Seeded: 9
[2021-03-21 13:23:16] Running: experiment 0
[2021-03-21 13:23:16] Showing labelled: True (1000/50000 visible, 0 redundant)
[2021-03-21 13:23:16] Creating pretrained=True VGG16...
[2021-03-21 13:23:17] No parameter reset since we are using a pretrained model.
[2021-03-21 13:23:17] vgg-pretrained: initialized 15245130 parameters.
[2021-03-21 13:23:17] Creating trainer with model on device: cuda
[2021-03-21 13:23:17] Training vgg-pretrained across 1000 data points in cifar10...
[2021-03-21 13:24:17] Training accuracy: 0.9991
[2021-03-21 13:24:17] Testing on 10000 data points...
[2021-03-21 13:24:19] Test score for 1000 training labels: 0.7288
[2021-03-21 13:24:19] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:24:19] Found 50000 unlabelled features.
[2021-03-21 13:24:30] Computing distance between 1000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:24:33] Searching for coresets with 20 beams...
[2021-03-21 13:25:01] Showing labelled: True (1400/50000 visible, 0 redundant)
[2021-03-21 13:25:01] Creating trainer with model on device: cuda
[2021-03-21 13:25:01] Training vgg-pretrained across 1400 data points in cifar10...
[2021-03-21 13:25:22] Training accuracy: 0.9961
[2021-03-21 13:25:22] Testing on 10000 data points...
[2021-03-21 13:25:24] Test score for 1400 training labels: 0.7343
[2021-03-21 13:25:24] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:25:24] Found 50000 unlabelled features.
[2021-03-21 13:25:35] Computing distance between 1400 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:25:39] Searching for coresets with 20 beams...
[2021-03-21 13:26:07] Showing labelled: True (1800/50000 visible, 0 redundant)
[2021-03-21 13:26:07] Creating trainer with model on device: cuda
[2021-03-21 13:26:07] Training vgg-pretrained across 1800 data points in cifar10...
[2021-03-21 13:26:35] Training accuracy: 0.8949
[2021-03-21 13:26:35] Testing on 10000 data points...
[2021-03-21 13:26:37] Test score for 1800 training labels: 0.6863
[2021-03-21 13:26:37] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:26:37] Found 50000 unlabelled features.
[2021-03-21 13:26:48] Computing distance between 1800 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:26:53] Searching for coresets with 20 beams...
[2021-03-21 13:27:21] Showing labelled: True (2200/50000 visible, 0 redundant)
[2021-03-21 13:27:21] Creating trainer with model on device: cuda
[2021-03-21 13:27:21] Training vgg-pretrained across 2200 data points in cifar10...
[2021-03-21 13:27:55] Training accuracy: 0.9996
[2021-03-21 13:27:55] Testing on 10000 data points...
[2021-03-21 13:27:57] Test score for 2200 training labels: 0.7767
[2021-03-21 13:27:57] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:27:57] Found 50000 unlabelled features.
[2021-03-21 13:28:09] Computing distance between 2200 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:28:15] Searching for coresets with 20 beams...
[2021-03-21 13:28:43] Showing labelled: True (2600/50000 visible, 0 redundant)
[2021-03-21 13:28:43] Creating trainer with model on device: cuda
[2021-03-21 13:28:43] Training vgg-pretrained across 2600 data points in cifar10...
[2021-03-21 13:29:22] Training accuracy: 0.9991
[2021-03-21 13:29:22] Testing on 10000 data points...
[2021-03-21 13:29:24] Test score for 2600 training labels: 0.7842
[2021-03-21 13:29:24] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:29:24] Found 50000 unlabelled features.
[2021-03-21 13:29:36] Computing distance between 2600 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:29:43] Searching for coresets with 20 beams...
[2021-03-21 13:30:11] Showing labelled: True (3000/50000 visible, 0 redundant)
[2021-03-21 13:30:11] Creating trainer with model on device: cuda
[2021-03-21 13:30:11] Training vgg-pretrained across 3000 data points in cifar10...
[2021-03-21 13:30:57] Training accuracy: 0.9994
[2021-03-21 13:30:57] Testing on 10000 data points...
[2021-03-21 13:30:59] Test score for 3000 training labels: 0.7788
[2021-03-21 13:30:59] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:30:59] Found 50000 unlabelled features.
[2021-03-21 13:31:09] Computing distance between 3000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:31:18] Searching for coresets with 20 beams...
[2021-03-21 13:31:46] Showing labelled: True (3400/50000 visible, 0 redundant)
[2021-03-21 13:31:46] Creating trainer with model on device: cuda
[2021-03-21 13:31:46] Training vgg-pretrained across 3400 data points in cifar10...
[2021-03-21 13:32:38] Training accuracy: 0.9996
[2021-03-21 13:32:38] Testing on 10000 data points...
[2021-03-21 13:32:40] Test score for 3400 training labels: 0.7995
[2021-03-21 13:32:40] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:32:40] Found 50000 unlabelled features.
[2021-03-21 13:32:51] Computing distance between 3400 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:33:02] Searching for coresets with 20 beams...
[2021-03-21 13:33:30] Showing labelled: True (3800/50000 visible, 0 redundant)
[2021-03-21 13:33:30] Creating trainer with model on device: cuda
[2021-03-21 13:33:30] Training vgg-pretrained across 3800 data points in cifar10...
[2021-03-21 13:34:29] Training accuracy: 0.9996
[2021-03-21 13:34:29] Testing on 10000 data points...
[2021-03-21 13:34:31] Test score for 3800 training labels: 0.8107
[2021-03-21 13:34:31] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:34:31] Found 50000 unlabelled features.
[2021-03-21 13:34:42] Computing distance between 3800 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:34:53] Searching for coresets with 20 beams...
[2021-03-21 13:35:22] Showing labelled: True (4200/50000 visible, 0 redundant)
[2021-03-21 13:35:22] Creating trainer with model on device: cuda
[2021-03-21 13:35:22] Training vgg-pretrained across 4200 data points in cifar10...
[2021-03-21 13:36:25] Training accuracy: 1.0000
[2021-03-21 13:36:25] Testing on 10000 data points...
[2021-03-21 13:36:27] Test score for 4200 training labels: 0.8161
[2021-03-21 13:36:27] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:36:27] Found 50000 unlabelled features.
[2021-03-21 13:36:38] Computing distance between 4200 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:36:50] Searching for coresets with 20 beams...
[2021-03-21 13:37:18] Showing labelled: True (4600/50000 visible, 0 redundant)
[2021-03-21 13:37:18] Creating trainer with model on device: cuda
[2021-03-21 13:37:18] Training vgg-pretrained across 4600 data points in cifar10...
[2021-03-21 13:38:28] Training accuracy: 0.9996
[2021-03-21 13:38:28] Testing on 10000 data points...
[2021-03-21 13:38:30] Test score for 4600 training labels: 0.8207
[2021-03-21 13:38:30] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:38:30] Found 50000 unlabelled features.
[2021-03-21 13:38:41] Computing distance between 4600 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:38:55] Searching for coresets with 20 beams...
[2021-03-21 13:39:23] Showing labelled: True (5000/50000 visible, 0 redundant)
[2021-03-21 13:39:23] Creating trainer with model on device: cuda
[2021-03-21 13:39:23] Training vgg-pretrained across 5000 data points in cifar10...
[2021-03-21 13:40:39] Training accuracy: 1.0000
[2021-03-21 13:40:39] Testing on 10000 data points...
[2021-03-21 13:40:41] Test score for 5000 training labels: 0.8255
[2021-03-21 13:40:41] Experiment repeat 5/5
[2021-03-21 13:40:41] Seeded: 9
[2021-03-21 13:40:41] Using 2.00% labels of the dataset (1000/50000)
[2021-03-21 13:40:41] Randomly labelled 1000/50000
[2021-03-21 13:40:41] Showing labelled: True (1000/50000 visible, 0 redundant)
[2021-03-21 13:40:41] Seeded: 10
[2021-03-21 13:40:41] Running: experiment 0
[2021-03-21 13:40:41] Showing labelled: True (1000/50000 visible, 0 redundant)
[2021-03-21 13:40:41] Creating pretrained=True VGG16...
[2021-03-21 13:40:43] No parameter reset since we are using a pretrained model.
[2021-03-21 13:40:43] vgg-pretrained: initialized 15245130 parameters.
[2021-03-21 13:40:43] Creating trainer with model on device: cuda
[2021-03-21 13:40:43] Training vgg-pretrained across 1000 data points in cifar10...
[2021-03-21 13:41:43] Training accuracy: 1.0000
[2021-03-21 13:41:43] Testing on 10000 data points...
[2021-03-21 13:41:45] Test score for 1000 training labels: 0.7344
[2021-03-21 13:41:46] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:41:46] Found 50000 unlabelled features.
[2021-03-21 13:41:56] Computing distance between 1000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:41:59] Searching for coresets with 20 beams...
[2021-03-21 13:42:27] Showing labelled: True (1400/50000 visible, 0 redundant)
[2021-03-21 13:42:27] Creating trainer with model on device: cuda
[2021-03-21 13:42:27] Training vgg-pretrained across 1400 data points in cifar10...
[2021-03-21 13:42:48] Training accuracy: 0.9953
[2021-03-21 13:42:48] Testing on 10000 data points...
[2021-03-21 13:42:50] Test score for 1400 training labels: 0.7550
[2021-03-21 13:42:50] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:42:50] Found 50000 unlabelled features.
[2021-03-21 13:43:01] Computing distance between 1400 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:43:05] Searching for coresets with 20 beams...
[2021-03-21 13:43:33] Showing labelled: True (1800/50000 visible, 0 redundant)
[2021-03-21 13:43:33] Creating trainer with model on device: cuda
[2021-03-21 13:43:33] Training vgg-pretrained across 1800 data points in cifar10...
[2021-03-21 13:44:00] Training accuracy: 0.9999
[2021-03-21 13:44:00] Testing on 10000 data points...
[2021-03-21 13:44:03] Test score for 1800 training labels: 0.7767
[2021-03-21 13:44:03] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:44:03] Found 50000 unlabelled features.
[2021-03-21 13:44:14] Computing distance between 1800 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:44:19] Searching for coresets with 20 beams...
[2021-03-21 13:44:47] Showing labelled: True (2200/50000 visible, 0 redundant)
[2021-03-21 13:44:47] Creating trainer with model on device: cuda
[2021-03-21 13:44:47] Training vgg-pretrained across 2200 data points in cifar10...
[2021-03-21 13:45:21] Training accuracy: 0.9988
[2021-03-21 13:45:21] Testing on 10000 data points...
[2021-03-21 13:45:23] Test score for 2200 training labels: 0.7913
[2021-03-21 13:45:23] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:45:23] Found 50000 unlabelled features.
[2021-03-21 13:45:34] Computing distance between 2200 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:45:40] Searching for coresets with 20 beams...
[2021-03-21 13:46:08] Showing labelled: True (2600/50000 visible, 0 redundant)
[2021-03-21 13:46:08] Creating trainer with model on device: cuda
[2021-03-21 13:46:08] Training vgg-pretrained across 2600 data points in cifar10...
[2021-03-21 13:46:47] Training accuracy: 0.9996
[2021-03-21 13:46:47] Testing on 10000 data points...
[2021-03-21 13:46:50] Test score for 2600 training labels: 0.7949
[2021-03-21 13:46:50] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:46:50] Found 50000 unlabelled features.
[2021-03-21 13:47:01] Computing distance between 2600 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:47:08] Searching for coresets with 20 beams...
[2021-03-21 13:47:36] Showing labelled: True (3000/50000 visible, 0 redundant)
[2021-03-21 13:47:36] Creating trainer with model on device: cuda
[2021-03-21 13:47:36] Training vgg-pretrained across 3000 data points in cifar10...
[2021-03-21 13:48:22] Training accuracy: 0.9995
[2021-03-21 13:48:22] Testing on 10000 data points...
[2021-03-21 13:48:24] Test score for 3000 training labels: 0.8047
[2021-03-21 13:48:24] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:48:24] Found 50000 unlabelled features.
[2021-03-21 13:48:35] Computing distance between 3000 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:48:43] Searching for coresets with 20 beams...
[2021-03-21 13:49:11] Showing labelled: True (3400/50000 visible, 0 redundant)
[2021-03-21 13:49:11] Creating trainer with model on device: cuda
[2021-03-21 13:49:11] Training vgg-pretrained across 3400 data points in cifar10...
[2021-03-21 13:50:02] Training accuracy: 0.9998
[2021-03-21 13:50:02] Testing on 10000 data points...
[2021-03-21 13:50:04] Test score for 3400 training labels: 0.8072
[2021-03-21 13:50:04] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:50:04] Found 50000 unlabelled features.
[2021-03-21 13:50:16] Computing distance between 3400 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:50:25] Searching for coresets with 20 beams...
[2021-03-21 13:50:53] Showing labelled: True (3800/50000 visible, 0 redundant)
[2021-03-21 13:50:53] Creating trainer with model on device: cuda
[2021-03-21 13:50:53] Training vgg-pretrained across 3800 data points in cifar10...
[2021-03-21 13:51:51] Training accuracy: 1.0000
[2021-03-21 13:51:51] Testing on 10000 data points...
[2021-03-21 13:51:53] Test score for 3800 training labels: 0.8163
[2021-03-21 13:51:53] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:51:53] Found 50000 unlabelled features.
[2021-03-21 13:52:04] Computing distance between 3800 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:52:16] Searching for coresets with 20 beams...
[2021-03-21 13:52:44] Showing labelled: True (4200/50000 visible, 0 redundant)
[2021-03-21 13:52:44] Creating trainer with model on device: cuda
[2021-03-21 13:52:44] Training vgg-pretrained across 4200 data points in cifar10...
[2021-03-21 13:53:47] Training accuracy: 0.9999
[2021-03-21 13:53:47] Testing on 10000 data points...
[2021-03-21 13:53:49] Test score for 4200 training labels: 0.8260
[2021-03-21 13:53:49] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:53:49] Found 50000 unlabelled features.
[2021-03-21 13:53:59] Computing distance between 4200 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:54:12] Searching for coresets with 20 beams...
[2021-03-21 13:54:40] Showing labelled: True (4600/50000 visible, 0 redundant)
[2021-03-21 13:54:40] Creating trainer with model on device: cuda
[2021-03-21 13:54:40] Training vgg-pretrained across 4600 data points in cifar10...
[2021-03-21 13:55:49] Training accuracy: 1.0000
[2021-03-21 13:55:49] Testing on 10000 data points...
[2021-03-21 13:55:51] Test score for 4600 training labels: 0.8306
[2021-03-21 13:55:51] Showing labelled: False (50000/50000 visible, 0 redundant)
[2021-03-21 13:55:51] Found 50000 unlabelled features.
[2021-03-21 13:56:03] Computing distance between 4600 labelled and 50000 unlabelled vectors of length 512...
[2021-03-21 13:56:17] Searching for coresets with 20 beams...
[2021-03-21 13:56:45] Showing labelled: True (5000/50000 visible, 0 redundant)
[2021-03-21 13:56:45] Creating trainer with model on device: cuda
[2021-03-21 13:56:45] Training vgg-pretrained across 5000 data points in cifar10...
[2021-03-21 13:58:00] Training accuracy: 1.0000
[2021-03-21 13:58:00] Testing on 10000 data points...
[2021-03-21 13:58:02] Test score for 5000 training labels: 0.8341
[2021-03-21 13:58:02] Updated results: ../results/cifar10/main-relconf/results.json
