[2021-03-23 01:02:01] Created experiment 0:
[2021-03-23 01:02:01]  - Model: vgg-pt
[2021-03-23 01:02:01]  - Acquisition function: random
[2021-03-23 01:02:01] Created experiment 1:
[2021-03-23 01:02:01]  - Model: vgg-pt
[2021-03-23 01:02:01]  - Acquisition function: greedy-coreset
[2021-03-23 01:02:01] Created experiment 2:
[2021-03-23 01:02:01]  - Model: vgg-pt
[2021-03-23 01:02:01]  - Acquisition function: lc-beam-pweighted-coreset (beams=10)
[2021-03-23 01:02:01] Loading svhn test set...
[2021-03-23 01:02:03] Experiment repeat 1/5
[2021-03-23 01:02:03] Seeded: 5
[2021-03-23 01:02:03] Using 6.83% labels of the dataset (5000/73257)
[2021-03-23 01:02:03] Randomly labelled 5000/73257
[2021-03-23 01:02:03] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 01:02:03] Seeded: 6
[2021-03-23 01:02:03] Running: experiment 0
[2021-03-23 01:02:03] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 01:02:03] Creating pretrained=True VGG16...
[2021-03-23 01:02:05] No parameter reset since we are using a pretrained model.
[2021-03-23 01:02:05] vgg-pretrained: initialized 15245130 parameters.
[2021-03-23 01:02:05] Creating trainer with model on device: cuda
[2021-03-23 01:02:10] Training with expected score >= 0.9900
[2021-03-23 01:02:10] Training vgg-pretrained across 5000 data points in svhn...
[2021-03-23 01:05:09] Training accuracy: 0.9932
[2021-03-23 01:05:09] Testing on 26032 data points...
[2021-03-23 01:05:14] Test score for 5000 training labels: 0.8935
[2021-03-23 01:05:14] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 01:05:14] Found 73257 unlabelled features.
[2021-03-23 01:05:20] Showing labelled: True (9631/73257 visible, 369 redundant)
[2021-03-23 01:05:20] Creating trainer with model on device: cuda
[2021-03-23 01:05:20] Training with expected score >= 0.9900
[2021-03-23 01:05:20] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 01:07:46] Training accuracy: 0.9920
[2021-03-23 01:07:46] Testing on 26032 data points...
[2021-03-23 01:07:51] Test score for 10000 training labels: 0.9136
[2021-03-23 01:07:51] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 01:07:51] Found 73257 unlabelled features.
[2021-03-23 01:07:57] Showing labelled: True (13969/73257 visible, 1031 redundant)
[2021-03-23 01:07:57] Creating trainer with model on device: cuda
[2021-03-23 01:07:57] Training with expected score >= 0.9900
[2021-03-23 01:07:57] Training vgg-pretrained across 15000 data points in svhn...
[2021-03-23 01:11:36] Training accuracy: 0.9978
[2021-03-23 01:11:36] Testing on 26032 data points...
[2021-03-23 01:11:40] Test score for 15000 training labels: 0.9226
[2021-03-23 01:11:40] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 01:11:40] Found 73257 unlabelled features.
[2021-03-23 01:11:47] Showing labelled: True (18015/73257 visible, 1985 redundant)
[2021-03-23 01:11:47] Creating trainer with model on device: cuda
[2021-03-23 01:11:47] Training with expected score >= 0.9900
[2021-03-23 01:11:47] Training vgg-pretrained across 20000 data points in svhn...
[2021-03-23 01:16:38] Training accuracy: 0.9982
[2021-03-23 01:16:38] Testing on 26032 data points...
[2021-03-23 01:16:42] Test score for 20000 training labels: 0.9293
[2021-03-23 01:16:42] Seeded: 6
[2021-03-23 01:16:42] Running: experiment 1
[2021-03-23 01:16:42] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 01:16:42] Creating pretrained=True VGG16...
[2021-03-23 01:16:43] No parameter reset since we are using a pretrained model.
[2021-03-23 01:16:43] vgg-pretrained: initialized 15245130 parameters.
[2021-03-23 01:16:43] Creating trainer with model on device: cuda
[2021-03-23 01:16:43] Training with expected score >= 0.9900
[2021-03-23 01:16:43] Training vgg-pretrained across 5000 data points in svhn...
[2021-03-23 01:19:46] Training accuracy: 0.9933
[2021-03-23 01:19:46] Testing on 26032 data points...
[2021-03-23 01:19:51] Test score for 5000 training labels: 0.8935
[2021-03-23 01:19:51] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 01:19:51] Found 73257 unlabelled features.
[2021-03-23 01:20:04] Computing distance between 5000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 01:20:27] Searching for coresets greedily...
[2021-03-23 01:20:42] Showing labelled: True (10000/73257 visible, 0 redundant)
[2021-03-23 01:20:42] Creating trainer with model on device: cuda
[2021-03-23 01:20:42] Training with expected score >= 0.9900
[2021-03-23 01:20:42] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 01:23:09] Training accuracy: 0.9906
[2021-03-23 01:23:09] Testing on 26032 data points...
[2021-03-23 01:23:13] Test score for 10000 training labels: 0.9266
[2021-03-23 01:23:13] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 01:23:13] Found 73257 unlabelled features.
[2021-03-23 01:23:27] Computing distance between 10000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 01:24:13] Searching for coresets greedily...
[2021-03-23 01:24:28] Showing labelled: True (15000/73257 visible, 0 redundant)
[2021-03-23 01:24:28] Creating trainer with model on device: cuda
[2021-03-23 01:24:28] Training with expected score >= 0.9900
[2021-03-23 01:24:28] Training vgg-pretrained across 15000 data points in svhn...
[2021-03-23 01:28:07] Training accuracy: 0.9915
[2021-03-23 01:28:07] Testing on 26032 data points...
[2021-03-23 01:28:11] Test score for 15000 training labels: 0.9410
[2021-03-23 01:28:11] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 01:28:11] Found 73257 unlabelled features.
[2021-03-23 01:28:26] Computing distance between 15000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 01:29:28] Searching for coresets greedily...
[2021-03-23 01:29:43] Showing labelled: True (20000/73257 visible, 0 redundant)
[2021-03-23 01:29:43] Creating trainer with model on device: cuda
[2021-03-23 01:29:43] Training with expected score >= 0.9900
[2021-03-23 01:29:43] Training vgg-pretrained across 20000 data points in svhn...
[2021-03-23 01:34:34] Training accuracy: 0.9963
[2021-03-23 01:34:34] Testing on 26032 data points...
[2021-03-23 01:34:39] Test score for 20000 training labels: 0.9455
[2021-03-23 01:34:39] Seeded: 6
[2021-03-23 01:34:39] Running: experiment 2
[2021-03-23 01:34:39] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 01:34:39] Creating pretrained=True VGG16...
[2021-03-23 01:34:40] No parameter reset since we are using a pretrained model.
[2021-03-23 01:34:40] vgg-pretrained: initialized 15245130 parameters.
[2021-03-23 01:34:40] Creating trainer with model on device: cuda
[2021-03-23 01:34:40] Training with expected score >= 0.9900
[2021-03-23 01:34:40] Training vgg-pretrained across 5000 data points in svhn...
[2021-03-23 01:37:42] Training accuracy: 0.9938
[2021-03-23 01:37:42] Testing on 26032 data points...
[2021-03-23 01:37:47] Test score for 5000 training labels: 0.8926
[2021-03-23 01:37:47] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 01:37:47] Found 73257 unlabelled features.
[2021-03-23 01:38:00] Computing distance between 5000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 01:38:21] Searching for coresets with 10 beams...
[2021-03-23 01:42:23] Showing labelled: True (10000/73257 visible, 0 redundant)
[2021-03-23 01:42:23] Creating trainer with model on device: cuda
[2021-03-23 01:42:23] Training with expected score >= 0.9900
[2021-03-23 01:42:23] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 01:44:49] Training accuracy: 0.9831
[2021-03-23 01:44:49] Training score (0.9831) was below expectations. Retraining...
[2021-03-23 01:44:49] Creating trainer with model on device: cuda
[2021-03-23 01:44:49] Training with expected score >= 0.9900
[2021-03-23 01:44:49] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 01:47:16] Training accuracy: 0.9960
[2021-03-23 01:47:16] Testing on 26032 data points...
[2021-03-23 01:47:20] Test score for 10000 training labels: 0.9368
[2021-03-23 01:47:20] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 01:47:20] Found 73257 unlabelled features.
[2021-03-23 01:47:34] Computing distance between 10000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 01:48:16] Searching for coresets with 10 beams...
[2021-03-23 01:52:18] Showing labelled: True (15000/73257 visible, 0 redundant)
[2021-03-23 01:52:18] Creating trainer with model on device: cuda
[2021-03-23 01:52:18] Training with expected score >= 0.9900
[2021-03-23 01:52:18] Training vgg-pretrained across 15000 data points in svhn...
[2021-03-23 01:55:57] Training accuracy: 0.9932
[2021-03-23 01:55:57] Testing on 26032 data points...
[2021-03-23 01:56:02] Test score for 15000 training labels: 0.9461
[2021-03-23 01:56:02] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 01:56:02] Found 73257 unlabelled features.
[2021-03-23 01:56:17] Computing distance between 15000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 01:57:20] Searching for coresets with 10 beams...
[2021-03-23 02:01:22] Showing labelled: True (20000/73257 visible, 0 redundant)
[2021-03-23 02:01:22] Creating trainer with model on device: cuda
[2021-03-23 02:01:22] Training with expected score >= 0.9900
[2021-03-23 02:01:22] Training vgg-pretrained across 20000 data points in svhn...
[2021-03-23 02:06:12] Training accuracy: 0.9971
[2021-03-23 02:06:12] Testing on 26032 data points...
[2021-03-23 02:06:17] Test score for 20000 training labels: 0.9499
[2021-03-23 02:06:17] Experiment repeat 2/5
[2021-03-23 02:06:17] Seeded: 6
[2021-03-23 02:06:17] Using 6.83% labels of the dataset (5000/73257)
[2021-03-23 02:06:17] Randomly labelled 5000/73257
[2021-03-23 02:06:17] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 02:06:17] Seeded: 7
[2021-03-23 02:06:17] Running: experiment 0
[2021-03-23 02:06:17] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 02:06:17] Creating pretrained=True VGG16...
[2021-03-23 02:06:24] No parameter reset since we are using a pretrained model.
[2021-03-23 02:06:24] vgg-pretrained: initialized 15245130 parameters.
[2021-03-23 02:06:24] Creating trainer with model on device: cuda
[2021-03-23 02:06:24] Training with expected score >= 0.9900
[2021-03-23 02:06:24] Training vgg-pretrained across 5000 data points in svhn...
[2021-03-23 02:09:27] Training accuracy: 0.9923
[2021-03-23 02:09:27] Testing on 26032 data points...
[2021-03-23 02:09:31] Test score for 5000 training labels: 0.8991
[2021-03-23 02:09:31] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 02:09:31] Found 73257 unlabelled features.
[2021-03-23 02:09:37] Showing labelled: True (9655/73257 visible, 345 redundant)
[2021-03-23 02:09:37] Creating trainer with model on device: cuda
[2021-03-23 02:09:37] Training with expected score >= 0.9900
[2021-03-23 02:09:37] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 02:12:04] Training accuracy: 0.9940
[2021-03-23 02:12:04] Testing on 26032 data points...
[2021-03-23 02:12:09] Test score for 10000 training labels: 0.9166
[2021-03-23 02:12:09] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 02:12:09] Found 73257 unlabelled features.
[2021-03-23 02:12:16] Showing labelled: True (14009/73257 visible, 991 redundant)
[2021-03-23 02:12:16] Creating trainer with model on device: cuda
[2021-03-23 02:12:16] Training with expected score >= 0.9900
[2021-03-23 02:12:16] Training vgg-pretrained across 15000 data points in svhn...
[2021-03-23 02:15:56] Training accuracy: 0.9975
[2021-03-23 02:15:56] Testing on 26032 data points...
[2021-03-23 02:16:00] Test score for 15000 training labels: 0.9254
[2021-03-23 02:16:00] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 02:16:00] Found 73257 unlabelled features.
[2021-03-23 02:16:07] Showing labelled: True (18051/73257 visible, 1949 redundant)
[2021-03-23 02:16:07] Creating trainer with model on device: cuda
[2021-03-23 02:16:07] Training with expected score >= 0.9900
[2021-03-23 02:16:07] Training vgg-pretrained across 20000 data points in svhn...
[2021-03-23 02:20:59] Training accuracy: 0.9998
[2021-03-23 02:20:59] Testing on 26032 data points...
[2021-03-23 02:21:04] Test score for 20000 training labels: 0.9284
[2021-03-23 02:21:04] Seeded: 7
[2021-03-23 02:21:04] Running: experiment 1
[2021-03-23 02:21:04] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 02:21:04] Creating pretrained=True VGG16...
[2021-03-23 02:21:06] No parameter reset since we are using a pretrained model.
[2021-03-23 02:21:06] vgg-pretrained: initialized 15245130 parameters.
[2021-03-23 02:21:06] Creating trainer with model on device: cuda
[2021-03-23 02:21:06] Training with expected score >= 0.9900
[2021-03-23 02:21:06] Training vgg-pretrained across 5000 data points in svhn...
[2021-03-23 02:24:08] Training accuracy: 0.9925
[2021-03-23 02:24:08] Testing on 26032 data points...
[2021-03-23 02:24:12] Test score for 5000 training labels: 0.8975
[2021-03-23 02:24:12] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 02:24:12] Found 73257 unlabelled features.
[2021-03-23 02:24:26] Computing distance between 5000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 02:24:46] Searching for coresets greedily...
[2021-03-23 02:25:00] Showing labelled: True (10000/73257 visible, 0 redundant)
[2021-03-23 02:25:00] Creating trainer with model on device: cuda
[2021-03-23 02:25:00] Training with expected score >= 0.9900
[2021-03-23 02:25:00] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 02:27:27] Training accuracy: 0.9915
[2021-03-23 02:27:27] Testing on 26032 data points...
[2021-03-23 02:27:32] Test score for 10000 training labels: 0.9263
[2021-03-23 02:27:32] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 02:27:32] Found 73257 unlabelled features.
[2021-03-23 02:27:47] Computing distance between 10000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 02:28:30] Searching for coresets greedily...
[2021-03-23 02:28:44] Showing labelled: True (15000/73257 visible, 0 redundant)
[2021-03-23 02:28:44] Creating trainer with model on device: cuda
[2021-03-23 02:28:44] Training with expected score >= 0.9900
[2021-03-23 02:28:44] Training vgg-pretrained across 15000 data points in svhn...
[2021-03-23 02:32:25] Training accuracy: 0.9954
[2021-03-23 02:32:25] Testing on 26032 data points...
[2021-03-23 02:32:30] Test score for 15000 training labels: 0.9422
[2021-03-23 02:32:30] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 02:32:30] Found 73257 unlabelled features.
[2021-03-23 02:32:45] Computing distance between 15000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 02:33:47] Searching for coresets greedily...
[2021-03-23 02:34:01] Showing labelled: True (20000/73257 visible, 0 redundant)
[2021-03-23 02:34:01] Creating trainer with model on device: cuda
[2021-03-23 02:34:01] Training with expected score >= 0.9900
[2021-03-23 02:34:01] Training vgg-pretrained across 20000 data points in svhn...
[2021-03-23 02:38:53] Training accuracy: 0.9974
[2021-03-23 02:38:53] Testing on 26032 data points...
[2021-03-23 02:38:58] Test score for 20000 training labels: 0.9472
[2021-03-23 02:38:58] Seeded: 7
[2021-03-23 02:38:58] Running: experiment 2
[2021-03-23 02:38:58] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 02:38:58] Creating pretrained=True VGG16...
[2021-03-23 02:38:59] No parameter reset since we are using a pretrained model.
[2021-03-23 02:38:59] vgg-pretrained: initialized 15245130 parameters.
[2021-03-23 02:38:59] Creating trainer with model on device: cuda
[2021-03-23 02:38:59] Training with expected score >= 0.9900
[2021-03-23 02:38:59] Training vgg-pretrained across 5000 data points in svhn...
[2021-03-23 02:42:02] Training accuracy: 0.9915
[2021-03-23 02:42:02] Testing on 26032 data points...
[2021-03-23 02:42:06] Test score for 5000 training labels: 0.8994
[2021-03-23 02:42:06] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 02:42:06] Found 73257 unlabelled features.
[2021-03-23 02:42:20] Computing distance between 5000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 02:42:40] Searching for coresets with 10 beams...
[2021-03-23 02:46:42] Showing labelled: True (10000/73257 visible, 0 redundant)
[2021-03-23 02:46:42] Creating trainer with model on device: cuda
[2021-03-23 02:46:42] Training with expected score >= 0.9900
[2021-03-23 02:46:42] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 02:49:08] Training accuracy: 0.9818
[2021-03-23 02:49:08] Training score (0.9818) was below expectations. Retraining...
[2021-03-23 02:49:08] Creating trainer with model on device: cuda
[2021-03-23 02:49:08] Training with expected score >= 0.9900
[2021-03-23 02:49:08] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 02:51:34] Training accuracy: 0.9980
[2021-03-23 02:51:34] Testing on 26032 data points...
[2021-03-23 02:51:39] Test score for 10000 training labels: 0.9387
[2021-03-23 02:51:39] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 02:51:39] Found 73257 unlabelled features.
[2021-03-23 02:51:53] Computing distance between 10000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 02:52:37] Searching for coresets with 10 beams...
[2021-03-23 02:56:39] Showing labelled: True (15000/73257 visible, 0 redundant)
[2021-03-23 02:56:39] Creating trainer with model on device: cuda
[2021-03-23 02:56:39] Training with expected score >= 0.9900
[2021-03-23 02:56:39] Training vgg-pretrained across 15000 data points in svhn...
[2021-03-23 03:00:18] Training accuracy: 0.9927
[2021-03-23 03:00:18] Testing on 26032 data points...
[2021-03-23 03:00:23] Test score for 15000 training labels: 0.9482
[2021-03-23 03:00:23] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 03:00:23] Found 73257 unlabelled features.
[2021-03-23 03:00:38] Computing distance between 15000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 03:01:40] Searching for coresets with 10 beams...
[2021-03-23 03:05:43] Showing labelled: True (20000/73257 visible, 0 redundant)
[2021-03-23 03:05:43] Creating trainer with model on device: cuda
[2021-03-23 03:05:43] Training with expected score >= 0.9900
[2021-03-23 03:05:43] Training vgg-pretrained across 20000 data points in svhn...
[2021-03-23 03:10:34] Training accuracy: 0.9987
[2021-03-23 03:10:34] Testing on 26032 data points...
[2021-03-23 03:10:39] Test score for 20000 training labels: 0.9512
[2021-03-23 03:10:39] Experiment repeat 3/5
[2021-03-23 03:10:39] Seeded: 7
[2021-03-23 03:10:39] Using 6.83% labels of the dataset (5000/73257)
[2021-03-23 03:10:39] Randomly labelled 5000/73257
[2021-03-23 03:10:39] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 03:10:39] Seeded: 8
[2021-03-23 03:10:39] Running: experiment 0
[2021-03-23 03:10:39] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 03:10:39] Creating pretrained=True VGG16...
[2021-03-23 03:10:40] No parameter reset since we are using a pretrained model.
[2021-03-23 03:10:40] vgg-pretrained: initialized 15245130 parameters.
[2021-03-23 03:10:40] Creating trainer with model on device: cuda
[2021-03-23 03:10:40] Training with expected score >= 0.9900
[2021-03-23 03:10:40] Training vgg-pretrained across 5000 data points in svhn...
[2021-03-23 03:13:43] Training accuracy: 0.9945
[2021-03-23 03:13:43] Testing on 26032 data points...
[2021-03-23 03:13:47] Test score for 5000 training labels: 0.8930
[2021-03-23 03:13:47] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 03:13:47] Found 73257 unlabelled features.
[2021-03-23 03:13:54] Showing labelled: True (9690/73257 visible, 310 redundant)
[2021-03-23 03:13:54] Creating trainer with model on device: cuda
[2021-03-23 03:13:54] Training with expected score >= 0.9900
[2021-03-23 03:13:54] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 03:16:21] Training accuracy: 0.9938
[2021-03-23 03:16:21] Testing on 26032 data points...
[2021-03-23 03:16:25] Test score for 10000 training labels: 0.9136
[2021-03-23 03:16:25] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 03:16:25] Found 73257 unlabelled features.
[2021-03-23 03:16:32] Showing labelled: True (14024/73257 visible, 976 redundant)
[2021-03-23 03:16:32] Creating trainer with model on device: cuda
[2021-03-23 03:16:32] Training with expected score >= 0.9900
[2021-03-23 03:16:32] Training vgg-pretrained across 15000 data points in svhn...
[2021-03-23 03:20:11] Training accuracy: 0.9979
[2021-03-23 03:20:11] Testing on 26032 data points...
[2021-03-23 03:20:16] Test score for 15000 training labels: 0.9218
[2021-03-23 03:20:16] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 03:20:16] Found 73257 unlabelled features.
[2021-03-23 03:20:22] Showing labelled: True (18118/73257 visible, 1882 redundant)
[2021-03-23 03:20:22] Creating trainer with model on device: cuda
[2021-03-23 03:20:22] Training with expected score >= 0.9900
[2021-03-23 03:20:22] Training vgg-pretrained across 20000 data points in svhn...
[2021-03-23 03:25:14] Training accuracy: 0.9998
[2021-03-23 03:25:14] Testing on 26032 data points...
[2021-03-23 03:25:19] Test score for 20000 training labels: 0.9292
[2021-03-23 03:25:19] Seeded: 8
[2021-03-23 03:25:19] Running: experiment 1
[2021-03-23 03:25:19] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 03:25:19] Creating pretrained=True VGG16...
[2021-03-23 03:25:21] No parameter reset since we are using a pretrained model.
[2021-03-23 03:25:21] vgg-pretrained: initialized 15245130 parameters.
[2021-03-23 03:25:21] Creating trainer with model on device: cuda
[2021-03-23 03:25:21] Training with expected score >= 0.9900
[2021-03-23 03:25:21] Training vgg-pretrained across 5000 data points in svhn...
[2021-03-23 03:28:23] Training accuracy: 0.9947
[2021-03-23 03:28:23] Testing on 26032 data points...
[2021-03-23 03:28:28] Test score for 5000 training labels: 0.8940
[2021-03-23 03:28:28] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 03:28:28] Found 73257 unlabelled features.
[2021-03-23 03:28:41] Computing distance between 5000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 03:29:02] Searching for coresets greedily...
[2021-03-23 03:29:16] Showing labelled: True (10000/73257 visible, 0 redundant)
[2021-03-23 03:29:16] Creating trainer with model on device: cuda
[2021-03-23 03:29:16] Training with expected score >= 0.9900
[2021-03-23 03:29:16] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 03:31:43] Training accuracy: 0.9884
[2021-03-23 03:31:43] Training score (0.9884) was below expectations. Retraining...
[2021-03-23 03:31:43] Creating trainer with model on device: cuda
[2021-03-23 03:31:43] Training with expected score >= 0.9900
[2021-03-23 03:31:43] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 03:34:10] Training accuracy: 0.9972
[2021-03-23 03:34:10] Testing on 26032 data points...
[2021-03-23 03:34:15] Test score for 10000 training labels: 0.9293
[2021-03-23 03:34:15] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 03:34:15] Found 73257 unlabelled features.
[2021-03-23 03:34:29] Computing distance between 10000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 03:35:10] Searching for coresets greedily...
[2021-03-23 03:35:25] Showing labelled: True (15000/73257 visible, 0 redundant)
[2021-03-23 03:35:25] Creating trainer with model on device: cuda
[2021-03-23 03:35:25] Training with expected score >= 0.9900
[2021-03-23 03:35:25] Training vgg-pretrained across 15000 data points in svhn...
[2021-03-23 03:39:04] Training accuracy: 0.9951
[2021-03-23 03:39:04] Testing on 26032 data points...
[2021-03-23 03:39:09] Test score for 15000 training labels: 0.9412
[2021-03-23 03:39:09] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 03:39:09] Found 73257 unlabelled features.
[2021-03-23 03:39:26] Computing distance between 15000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 03:40:30] Searching for coresets greedily...
[2021-03-23 03:40:45] Showing labelled: True (20000/73257 visible, 0 redundant)
[2021-03-23 03:40:45] Creating trainer with model on device: cuda
[2021-03-23 03:40:45] Training with expected score >= 0.9900
[2021-03-23 03:40:45] Training vgg-pretrained across 20000 data points in svhn...
[2021-03-23 03:45:37] Training accuracy: 0.9986
[2021-03-23 03:45:37] Testing on 26032 data points...
[2021-03-23 03:45:41] Test score for 20000 training labels: 0.9456
[2021-03-23 03:45:41] Seeded: 8
[2021-03-23 03:45:41] Running: experiment 2
[2021-03-23 03:45:41] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 03:45:41] Creating pretrained=True VGG16...
[2021-03-23 03:45:43] No parameter reset since we are using a pretrained model.
[2021-03-23 03:45:43] vgg-pretrained: initialized 15245130 parameters.
[2021-03-23 03:45:43] Creating trainer with model on device: cuda
[2021-03-23 03:45:43] Training with expected score >= 0.9900
[2021-03-23 03:45:43] Training vgg-pretrained across 5000 data points in svhn...
[2021-03-23 03:48:47] Training accuracy: 0.9938
[2021-03-23 03:48:47] Testing on 26032 data points...
[2021-03-23 03:48:52] Test score for 5000 training labels: 0.8939
[2021-03-23 03:48:52] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 03:48:52] Found 73257 unlabelled features.
[2021-03-23 03:49:05] Computing distance between 5000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 03:49:26] Searching for coresets with 10 beams...
[2021-03-23 03:53:28] Showing labelled: True (10000/73257 visible, 0 redundant)
[2021-03-23 03:53:28] Creating trainer with model on device: cuda
[2021-03-23 03:53:28] Training with expected score >= 0.9900
[2021-03-23 03:53:28] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 03:55:54] Training accuracy: 0.9796
[2021-03-23 03:55:54] Training score (0.9796) was below expectations. Retraining...
[2021-03-23 03:55:54] Creating trainer with model on device: cuda
[2021-03-23 03:55:54] Training with expected score >= 0.9900
[2021-03-23 03:55:54] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 03:58:21] Training accuracy: 0.9992
[2021-03-23 03:58:21] Testing on 26032 data points...
[2021-03-23 03:58:25] Test score for 10000 training labels: 0.9358
[2021-03-23 03:58:25] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 03:58:25] Found 73257 unlabelled features.
[2021-03-23 03:58:39] Computing distance between 10000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 03:59:21] Searching for coresets with 10 beams...
[2021-03-23 04:03:23] Showing labelled: True (15000/73257 visible, 0 redundant)
[2021-03-23 04:03:23] Creating trainer with model on device: cuda
[2021-03-23 04:03:23] Training with expected score >= 0.9900
[2021-03-23 04:03:23] Training vgg-pretrained across 15000 data points in svhn...
[2021-03-23 04:07:02] Training accuracy: 0.9958
[2021-03-23 04:07:02] Testing on 26032 data points...
[2021-03-23 04:07:07] Test score for 15000 training labels: 0.9473
[2021-03-23 04:07:07] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 04:07:07] Found 73257 unlabelled features.
[2021-03-23 04:07:22] Computing distance between 15000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 04:08:27] Searching for coresets with 10 beams...
[2021-03-23 04:12:30] Showing labelled: True (20000/73257 visible, 0 redundant)
[2021-03-23 04:12:30] Creating trainer with model on device: cuda
[2021-03-23 04:12:30] Training with expected score >= 0.9900
[2021-03-23 04:12:30] Training vgg-pretrained across 20000 data points in svhn...
[2021-03-23 04:17:22] Training accuracy: 0.9955
[2021-03-23 04:17:22] Testing on 26032 data points...
[2021-03-23 04:17:26] Test score for 20000 training labels: 0.9492
[2021-03-23 04:17:26] Experiment repeat 4/5
[2021-03-23 04:17:26] Seeded: 8
[2021-03-23 04:17:26] Using 6.83% labels of the dataset (5000/73257)
[2021-03-23 04:17:26] Randomly labelled 5000/73257
[2021-03-23 04:17:26] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 04:17:26] Seeded: 9
[2021-03-23 04:17:26] Running: experiment 0
[2021-03-23 04:17:26] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 04:17:26] Creating pretrained=True VGG16...
[2021-03-23 04:17:28] No parameter reset since we are using a pretrained model.
[2021-03-23 04:17:28] vgg-pretrained: initialized 15245130 parameters.
[2021-03-23 04:17:28] Creating trainer with model on device: cuda
[2021-03-23 04:17:28] Training with expected score >= 0.9900
[2021-03-23 04:17:28] Training vgg-pretrained across 5000 data points in svhn...
[2021-03-23 04:20:31] Training accuracy: 0.9920
[2021-03-23 04:20:31] Testing on 26032 data points...
[2021-03-23 04:20:35] Test score for 5000 training labels: 0.8974
[2021-03-23 04:20:35] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 04:20:35] Found 73257 unlabelled features.
[2021-03-23 04:20:42] Showing labelled: True (9656/73257 visible, 344 redundant)
[2021-03-23 04:20:42] Creating trainer with model on device: cuda
[2021-03-23 04:20:42] Training with expected score >= 0.9900
[2021-03-23 04:20:42] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 04:23:09] Training accuracy: 0.9929
[2021-03-23 04:23:09] Testing on 26032 data points...
[2021-03-23 04:23:13] Test score for 10000 training labels: 0.9146
[2021-03-23 04:23:13] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 04:23:13] Found 73257 unlabelled features.
[2021-03-23 04:23:20] Showing labelled: True (14027/73257 visible, 973 redundant)
[2021-03-23 04:23:20] Creating trainer with model on device: cuda
[2021-03-23 04:23:20] Training with expected score >= 0.9900
[2021-03-23 04:23:20] Training vgg-pretrained across 15000 data points in svhn...
[2021-03-23 04:27:00] Training accuracy: 0.9961
[2021-03-23 04:27:00] Testing on 26032 data points...
[2021-03-23 04:27:05] Test score for 15000 training labels: 0.9232
[2021-03-23 04:27:05] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 04:27:05] Found 73257 unlabelled features.
[2021-03-23 04:27:11] Showing labelled: True (18064/73257 visible, 1936 redundant)
[2021-03-23 04:27:11] Creating trainer with model on device: cuda
[2021-03-23 04:27:11] Training with expected score >= 0.9900
[2021-03-23 04:27:11] Training vgg-pretrained across 20000 data points in svhn...
[2021-03-23 04:32:04] Training accuracy: 0.9974
[2021-03-23 04:32:04] Testing on 26032 data points...
[2021-03-23 04:32:09] Test score for 20000 training labels: 0.9292
[2021-03-23 04:32:09] Seeded: 9
[2021-03-23 04:32:09] Running: experiment 1
[2021-03-23 04:32:09] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 04:32:09] Creating pretrained=True VGG16...
[2021-03-23 04:32:10] No parameter reset since we are using a pretrained model.
[2021-03-23 04:32:10] vgg-pretrained: initialized 15245130 parameters.
[2021-03-23 04:32:10] Creating trainer with model on device: cuda
[2021-03-23 04:32:10] Training with expected score >= 0.9900
[2021-03-23 04:32:10] Training vgg-pretrained across 5000 data points in svhn...
[2021-03-23 04:35:13] Training accuracy: 0.9925
[2021-03-23 04:35:13] Testing on 26032 data points...
[2021-03-23 04:35:17] Test score for 5000 training labels: 0.8975
[2021-03-23 04:35:17] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 04:35:17] Found 73257 unlabelled features.
[2021-03-23 04:35:31] Computing distance between 5000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 04:35:52] Searching for coresets greedily...
[2021-03-23 04:36:06] Showing labelled: True (10000/73257 visible, 0 redundant)
[2021-03-23 04:36:06] Creating trainer with model on device: cuda
[2021-03-23 04:36:06] Training with expected score >= 0.9900
[2021-03-23 04:36:06] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 04:38:33] Training accuracy: 0.9907
[2021-03-23 04:38:33] Testing on 26032 data points...
[2021-03-23 04:38:37] Test score for 10000 training labels: 0.9288
[2021-03-23 04:38:37] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 04:38:37] Found 73257 unlabelled features.
[2021-03-23 04:38:51] Computing distance between 10000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 04:39:33] Searching for coresets greedily...
[2021-03-23 04:39:47] Showing labelled: True (15000/73257 visible, 0 redundant)
[2021-03-23 04:39:47] Creating trainer with model on device: cuda
[2021-03-23 04:39:47] Training with expected score >= 0.9900
[2021-03-23 04:39:47] Training vgg-pretrained across 15000 data points in svhn...
[2021-03-23 04:43:27] Training accuracy: 0.9946
[2021-03-23 04:43:27] Testing on 26032 data points...
[2021-03-23 04:43:31] Test score for 15000 training labels: 0.9417
[2021-03-23 04:43:32] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 04:43:32] Found 73257 unlabelled features.
[2021-03-23 04:43:47] Computing distance between 15000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 04:44:49] Searching for coresets greedily...
[2021-03-23 04:45:03] Showing labelled: True (20000/73257 visible, 0 redundant)
[2021-03-23 04:45:03] Creating trainer with model on device: cuda
[2021-03-23 04:45:03] Training with expected score >= 0.9900
[2021-03-23 04:45:03] Training vgg-pretrained across 20000 data points in svhn...
[2021-03-23 04:49:56] Training accuracy: 0.9952
[2021-03-23 04:49:56] Testing on 26032 data points...
[2021-03-23 04:50:00] Test score for 20000 training labels: 0.9469
[2021-03-23 04:50:00] Seeded: 9
[2021-03-23 04:50:00] Running: experiment 2
[2021-03-23 04:50:00] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 04:50:00] Creating pretrained=True VGG16...
[2021-03-23 04:50:02] No parameter reset since we are using a pretrained model.
[2021-03-23 04:50:02] vgg-pretrained: initialized 15245130 parameters.
[2021-03-23 04:50:02] Creating trainer with model on device: cuda
[2021-03-23 04:50:02] Training with expected score >= 0.9900
[2021-03-23 04:50:02] Training vgg-pretrained across 5000 data points in svhn...
[2021-03-23 04:53:05] Training accuracy: 0.9919
[2021-03-23 04:53:05] Testing on 26032 data points...
[2021-03-23 04:53:09] Test score for 5000 training labels: 0.8974
[2021-03-23 04:53:09] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 04:53:09] Found 73257 unlabelled features.
[2021-03-23 04:53:23] Computing distance between 5000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 04:53:43] Searching for coresets with 10 beams...
[2021-03-23 04:57:46] Showing labelled: True (10000/73257 visible, 0 redundant)
[2021-03-23 04:57:46] Creating trainer with model on device: cuda
[2021-03-23 04:57:46] Training with expected score >= 0.9900
[2021-03-23 04:57:46] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 05:00:12] Training accuracy: 0.9725
[2021-03-23 05:00:12] Training score (0.9725) was below expectations. Retraining...
[2021-03-23 05:00:12] Creating trainer with model on device: cuda
[2021-03-23 05:00:12] Training with expected score >= 0.9900
[2021-03-23 05:00:12] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 05:02:39] Training accuracy: 0.9959
[2021-03-23 05:02:39] Testing on 26032 data points...
[2021-03-23 05:02:43] Test score for 10000 training labels: 0.9367
[2021-03-23 05:02:43] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 05:02:43] Found 73257 unlabelled features.
[2021-03-23 05:02:57] Computing distance between 10000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 05:03:39] Searching for coresets with 10 beams...
[2021-03-23 05:07:41] Showing labelled: True (15000/73257 visible, 0 redundant)
[2021-03-23 05:07:41] Creating trainer with model on device: cuda
[2021-03-23 05:07:41] Training with expected score >= 0.9900
[2021-03-23 05:07:41] Training vgg-pretrained across 15000 data points in svhn...
[2021-03-23 05:11:20] Training accuracy: 0.9950
[2021-03-23 05:11:20] Testing on 26032 data points...
[2021-03-23 05:11:24] Test score for 15000 training labels: 0.9462
[2021-03-23 05:11:24] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 05:11:24] Found 73257 unlabelled features.
[2021-03-23 05:11:39] Computing distance between 15000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 05:12:42] Searching for coresets with 10 beams...
[2021-03-23 05:16:44] Showing labelled: True (20000/73257 visible, 0 redundant)
[2021-03-23 05:16:44] Creating trainer with model on device: cuda
[2021-03-23 05:16:44] Training with expected score >= 0.9900
[2021-03-23 05:16:44] Training vgg-pretrained across 20000 data points in svhn...
[2021-03-23 05:21:36] Training accuracy: 0.9991
[2021-03-23 05:21:36] Testing on 26032 data points...
[2021-03-23 05:21:40] Test score for 20000 training labels: 0.9489
[2021-03-23 05:21:40] Experiment repeat 5/5
[2021-03-23 05:21:40] Seeded: 9
[2021-03-23 05:21:40] Using 6.83% labels of the dataset (5000/73257)
[2021-03-23 05:21:40] Randomly labelled 5000/73257
[2021-03-23 05:21:40] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 05:21:40] Seeded: 10
[2021-03-23 05:21:40] Running: experiment 0
[2021-03-23 05:21:40] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 05:21:40] Creating pretrained=True VGG16...
[2021-03-23 05:21:42] No parameter reset since we are using a pretrained model.
[2021-03-23 05:21:42] vgg-pretrained: initialized 15245130 parameters.
[2021-03-23 05:21:42] Creating trainer with model on device: cuda
[2021-03-23 05:21:42] Training with expected score >= 0.9900
[2021-03-23 05:21:42] Training vgg-pretrained across 5000 data points in svhn...
[2021-03-23 05:24:44] Training accuracy: 0.9911
[2021-03-23 05:24:44] Testing on 26032 data points...
[2021-03-23 05:24:49] Test score for 5000 training labels: 0.8960
[2021-03-23 05:24:49] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 05:24:49] Found 73257 unlabelled features.
[2021-03-23 05:24:55] Showing labelled: True (9651/73257 visible, 349 redundant)
[2021-03-23 05:24:55] Creating trainer with model on device: cuda
[2021-03-23 05:24:55] Training with expected score >= 0.9900
[2021-03-23 05:24:55] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 05:27:23] Training accuracy: 0.9963
[2021-03-23 05:27:23] Testing on 26032 data points...
[2021-03-23 05:27:27] Test score for 10000 training labels: 0.9132
[2021-03-23 05:27:27] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 05:27:27] Found 73257 unlabelled features.
[2021-03-23 05:27:35] Showing labelled: True (13990/73257 visible, 1010 redundant)
[2021-03-23 05:27:35] Creating trainer with model on device: cuda
[2021-03-23 05:27:35] Training with expected score >= 0.9900
[2021-03-23 05:27:35] Training vgg-pretrained across 15000 data points in svhn...
[2021-03-23 05:31:17] Training accuracy: 0.9973
[2021-03-23 05:31:17] Testing on 26032 data points...
[2021-03-23 05:31:22] Test score for 15000 training labels: 0.9200
[2021-03-23 05:31:22] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 05:31:22] Found 73257 unlabelled features.
[2021-03-23 05:31:28] Showing labelled: True (18060/73257 visible, 1940 redundant)
[2021-03-23 05:31:28] Creating trainer with model on device: cuda
[2021-03-23 05:31:28] Training with expected score >= 0.9900
[2021-03-23 05:31:28] Training vgg-pretrained across 20000 data points in svhn...
[2021-03-23 05:36:21] Training accuracy: 0.9975
[2021-03-23 05:36:21] Testing on 26032 data points...
[2021-03-23 05:36:26] Test score for 20000 training labels: 0.9252
[2021-03-23 05:36:26] Seeded: 10
[2021-03-23 05:36:26] Running: experiment 1
[2021-03-23 05:36:26] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 05:36:26] Creating pretrained=True VGG16...
[2021-03-23 05:36:27] No parameter reset since we are using a pretrained model.
[2021-03-23 05:36:27] vgg-pretrained: initialized 15245130 parameters.
[2021-03-23 05:36:27] Creating trainer with model on device: cuda
[2021-03-23 05:36:27] Training with expected score >= 0.9900
[2021-03-23 05:36:27] Training vgg-pretrained across 5000 data points in svhn...
[2021-03-23 05:39:30] Training accuracy: 0.9907
[2021-03-23 05:39:30] Testing on 26032 data points...
[2021-03-23 05:39:34] Test score for 5000 training labels: 0.8971
[2021-03-23 05:39:34] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 05:39:34] Found 73257 unlabelled features.
[2021-03-23 05:39:48] Computing distance between 5000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 05:40:08] Searching for coresets greedily...
[2021-03-23 05:40:23] Showing labelled: True (10000/73257 visible, 0 redundant)
[2021-03-23 05:40:23] Creating trainer with model on device: cuda
[2021-03-23 05:40:23] Training with expected score >= 0.9900
[2021-03-23 05:40:23] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 05:42:49] Training accuracy: 0.9910
[2021-03-23 05:42:49] Testing on 26032 data points...
[2021-03-23 05:42:54] Test score for 10000 training labels: 0.9245
[2021-03-23 05:42:54] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 05:42:54] Found 73257 unlabelled features.
[2021-03-23 05:43:08] Computing distance between 10000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 05:43:50] Searching for coresets greedily...
[2021-03-23 05:44:04] Showing labelled: True (15000/73257 visible, 0 redundant)
[2021-03-23 05:44:04] Creating trainer with model on device: cuda
[2021-03-23 05:44:04] Training with expected score >= 0.9900
[2021-03-23 05:44:04] Training vgg-pretrained across 15000 data points in svhn...
[2021-03-23 05:47:45] Training accuracy: 0.9921
[2021-03-23 05:47:45] Testing on 26032 data points...
[2021-03-23 05:47:49] Test score for 15000 training labels: 0.9411
[2021-03-23 05:47:49] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 05:47:49] Found 73257 unlabelled features.
[2021-03-23 05:48:05] Computing distance between 15000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 05:49:07] Searching for coresets greedily...
[2021-03-23 05:49:21] Showing labelled: True (20000/73257 visible, 0 redundant)
[2021-03-23 05:49:21] Creating trainer with model on device: cuda
[2021-03-23 05:49:21] Training with expected score >= 0.9900
[2021-03-23 05:49:21] Training vgg-pretrained across 20000 data points in svhn...
[2021-03-23 05:54:14] Training accuracy: 0.9965
[2021-03-23 05:54:14] Testing on 26032 data points...
[2021-03-23 05:54:18] Test score for 20000 training labels: 0.9476
[2021-03-23 05:54:18] Seeded: 10
[2021-03-23 05:54:18] Running: experiment 2
[2021-03-23 05:54:18] Showing labelled: True (5000/73257 visible, 0 redundant)
[2021-03-23 05:54:18] Creating pretrained=True VGG16...
[2021-03-23 05:54:20] No parameter reset since we are using a pretrained model.
[2021-03-23 05:54:20] vgg-pretrained: initialized 15245130 parameters.
[2021-03-23 05:54:20] Creating trainer with model on device: cuda
[2021-03-23 05:54:20] Training with expected score >= 0.9900
[2021-03-23 05:54:20] Training vgg-pretrained across 5000 data points in svhn...
[2021-03-23 05:57:22] Training accuracy: 0.9921
[2021-03-23 05:57:22] Testing on 26032 data points...
[2021-03-23 05:57:27] Test score for 5000 training labels: 0.8959
[2021-03-23 05:57:27] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 05:57:27] Found 73257 unlabelled features.
[2021-03-23 05:57:41] Computing distance between 5000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 05:58:04] Searching for coresets with 10 beams...
[2021-03-23 06:02:06] Showing labelled: True (10000/73257 visible, 0 redundant)
[2021-03-23 06:02:06] Creating trainer with model on device: cuda
[2021-03-23 06:02:06] Training with expected score >= 0.9900
[2021-03-23 06:02:06] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 06:04:33] Training accuracy: 0.9755
[2021-03-23 06:04:33] Training score (0.9755) was below expectations. Retraining...
[2021-03-23 06:04:33] Creating trainer with model on device: cuda
[2021-03-23 06:04:33] Training with expected score >= 0.9900
[2021-03-23 06:04:33] Training vgg-pretrained across 10000 data points in svhn...
[2021-03-23 06:06:59] Training accuracy: 0.9970
[2021-03-23 06:06:59] Testing on 26032 data points...
[2021-03-23 06:07:04] Test score for 10000 training labels: 0.9355
[2021-03-23 06:07:04] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 06:07:04] Found 73257 unlabelled features.
[2021-03-23 06:07:18] Computing distance between 10000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 06:08:00] Searching for coresets with 10 beams...
[2021-03-23 06:12:01] Showing labelled: True (15000/73257 visible, 0 redundant)
[2021-03-23 06:12:01] Creating trainer with model on device: cuda
[2021-03-23 06:12:01] Training with expected score >= 0.9900
[2021-03-23 06:12:01] Training vgg-pretrained across 15000 data points in svhn...
[2021-03-23 06:15:41] Training accuracy: 0.9945
[2021-03-23 06:15:41] Testing on 26032 data points...
[2021-03-23 06:15:45] Test score for 15000 training labels: 0.9478
[2021-03-23 06:15:45] Showing labelled: False (73257/73257 visible, 0 redundant)
[2021-03-23 06:15:45] Found 73257 unlabelled features.
[2021-03-23 06:16:02] Computing distance between 15000 labelled and 73257 unlabelled vectors of length 512...
[2021-03-23 06:17:11] Searching for coresets with 10 beams...
[2021-03-23 06:21:13] Showing labelled: True (20000/73257 visible, 0 redundant)
[2021-03-23 06:21:13] Creating trainer with model on device: cuda
[2021-03-23 06:21:13] Training with expected score >= 0.9900
[2021-03-23 06:21:13] Training vgg-pretrained across 20000 data points in svhn...
[2021-03-23 06:26:06] Training accuracy: 0.9977
[2021-03-23 06:26:06] Testing on 26032 data points...
[2021-03-23 06:26:10] Test score for 20000 training labels: 0.9511
[2021-03-23 06:26:10] Updated results: ../results/svhn/main/results.json
