[2021-03-16 00:53:33] Seeded: 5
[2021-03-16 00:53:33] Created experiment 0:
[2021-03-16 00:53:33]  - Model: mlp
[2021-03-16 00:53:33]  - Acquisition function: sp-coreset
[2021-03-16 00:53:33] Loading blobsthin test set...
[2021-03-16 00:53:33] Generated 5000 blobs:
[2021-03-16 00:53:33] Class 0: 200
[2021-03-16 00:53:33] Class 1: 300
[2021-03-16 00:53:33] Class 2: 1000
[2021-03-16 00:53:33] Class 3: 100
[2021-03-16 00:53:33] Class 4: 50
[2021-03-16 00:53:33] Class 5: 200
[2021-03-16 00:53:33] Class 6: 400
[2021-03-16 00:53:33] Class 7: 1800
[2021-03-16 00:53:33] Class 8: 50
[2021-03-16 00:53:33] Class 9: 900
[2021-03-16 00:53:33] Saved figure to: ../data/blobsthin_data.png
[2021-03-16 00:53:33] Generated 5000 test points.
[2021-03-16 00:53:33] Saved figure to: ../data/blobsthin_test.png
[2021-03-16 00:53:33] Experiment repeat 1/2
[2021-03-16 00:53:33] Randomly labelled 50/5000
[2021-03-16 00:53:33] Showing labelled: True (50/5000 visible)
[2021-03-16 00:53:33] Running: experiment 0
[2021-03-16 00:53:33] Showing labelled: True (50/5000 visible)
[2021-03-16 00:53:33] mlp: initialized 1482 parameters.
[2021-03-16 00:53:33] Creating trainer with model on device: cuda
[2021-03-16 00:53:39] Training mlp across 50 data points in blobsthin...
[2021-03-16 00:53:39] Training loss: 1.5627
[2021-03-16 00:53:39] Testing on 5000 data points...
[2021-03-16 00:53:39] Test accuracy: 0.5000
[2021-03-16 00:53:39] Score for 50 labels: 0.5000
[2021-03-16 00:53:39] Showing labelled: False (4950/5000 visible)
[2021-03-16 00:53:39] Found 4950 unlabelled features.
[2021-03-16 00:53:39] Training GMM (k=50) on labelled features...
[2021-03-16 00:53:39] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:53:40] Showing labelled: True (75/5000 visible)
[2021-03-16 00:53:40] Training mlp across 75 data points in blobsthin...
[2021-03-16 00:53:40] Training loss: 0.6830
[2021-03-16 00:53:40] Testing on 5000 data points...
[2021-03-16 00:53:40] Test accuracy: 0.7000
[2021-03-16 00:53:40] Score for 75 labels: 0.7000
[2021-03-16 00:53:40] Showing labelled: False (4925/5000 visible)
[2021-03-16 00:53:40] Found 4925 unlabelled features.
[2021-03-16 00:53:40] Training GMM (k=50) on labelled features...
[2021-03-16 00:53:40] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:53:41] Showing labelled: True (100/5000 visible)
[2021-03-16 00:53:41] Training mlp across 100 data points in blobsthin...
[2021-03-16 00:53:42] Training loss: 0.3053
[2021-03-16 00:53:42] Testing on 5000 data points...
[2021-03-16 00:53:42] Test accuracy: 0.7980
[2021-03-16 00:53:42] Score for 100 labels: 0.7980
[2021-03-16 00:53:42] Showing labelled: False (4900/5000 visible)
[2021-03-16 00:53:42] Found 4900 unlabelled features.
[2021-03-16 00:53:42] Training GMM (k=50) on labelled features...
[2021-03-16 00:53:42] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:53:43] Showing labelled: True (125/5000 visible)
[2021-03-16 00:53:43] Training mlp across 125 data points in blobsthin...
[2021-03-16 00:53:43] Training loss: 0.1787
[2021-03-16 00:53:43] Testing on 5000 data points...
[2021-03-16 00:53:43] Test accuracy: 0.7990
[2021-03-16 00:53:43] Score for 125 labels: 0.7990
[2021-03-16 00:53:43] Showing labelled: False (4875/5000 visible)
[2021-03-16 00:53:43] Found 4875 unlabelled features.
[2021-03-16 00:53:44] Training GMM (k=50) on labelled features...
[2021-03-16 00:53:44] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:53:44] Showing labelled: True (150/5000 visible)
[2021-03-16 00:53:44] Training mlp across 150 data points in blobsthin...
[2021-03-16 00:53:45] Training loss: 0.1238
[2021-03-16 00:53:45] Testing on 5000 data points...
[2021-03-16 00:53:45] Test accuracy: 0.8562
[2021-03-16 00:53:45] Score for 150 labels: 0.8562
[2021-03-16 00:53:45] Showing labelled: False (4850/5000 visible)
[2021-03-16 00:53:45] Found 4850 unlabelled features.
[2021-03-16 00:53:45] Training GMM (k=50) on labelled features...
[2021-03-16 00:53:45] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:53:46] Showing labelled: True (175/5000 visible)
[2021-03-16 00:53:46] Training mlp across 175 data points in blobsthin...
[2021-03-16 00:53:47] Training loss: 0.0712
[2021-03-16 00:53:47] Testing on 5000 data points...
[2021-03-16 00:53:47] Test accuracy: 0.8846
[2021-03-16 00:53:47] Score for 175 labels: 0.8846
[2021-03-16 00:53:47] Showing labelled: False (4825/5000 visible)
[2021-03-16 00:53:47] Found 4825 unlabelled features.
[2021-03-16 00:53:47] Training GMM (k=50) on labelled features...
[2021-03-16 00:53:48] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:53:48] Showing labelled: True (200/5000 visible)
[2021-03-16 00:53:48] Training mlp across 200 data points in blobsthin...
[2021-03-16 00:53:49] Training loss: 0.0396
[2021-03-16 00:53:49] Testing on 5000 data points...
[2021-03-16 00:53:49] Test accuracy: 0.8892
[2021-03-16 00:53:49] Score for 200 labels: 0.8892
[2021-03-16 00:53:49] Showing labelled: False (4800/5000 visible)
[2021-03-16 00:53:49] Found 4800 unlabelled features.
[2021-03-16 00:53:49] Training GMM (k=50) on labelled features...
[2021-03-16 00:53:50] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:53:50] Showing labelled: True (225/5000 visible)
[2021-03-16 00:53:50] Training mlp across 225 data points in blobsthin...
[2021-03-16 00:53:51] Training loss: 0.0255
[2021-03-16 00:53:51] Testing on 5000 data points...
[2021-03-16 00:53:52] Test accuracy: 0.8920
[2021-03-16 00:53:52] Score for 225 labels: 0.8920
[2021-03-16 00:53:52] Showing labelled: False (4775/5000 visible)
[2021-03-16 00:53:52] Found 4775 unlabelled features.
[2021-03-16 00:53:52] Training GMM (k=50) on labelled features...
[2021-03-16 00:53:52] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:53:52] Showing labelled: True (250/5000 visible)
[2021-03-16 00:53:52] Training mlp across 250 data points in blobsthin...
[2021-03-16 00:53:53] Training loss: 0.0168
[2021-03-16 00:53:53] Testing on 5000 data points...
[2021-03-16 00:53:54] Test accuracy: 0.8922
[2021-03-16 00:53:54] Score for 250 labels: 0.8922
[2021-03-16 00:53:54] Showing labelled: False (4750/5000 visible)
[2021-03-16 00:53:54] Found 4750 unlabelled features.
[2021-03-16 00:53:54] Training GMM (k=50) on labelled features...
[2021-03-16 00:53:54] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:53:54] Showing labelled: True (275/5000 visible)
[2021-03-16 00:53:54] Training mlp across 275 data points in blobsthin...
[2021-03-16 00:53:56] Training loss: 0.0124
[2021-03-16 00:53:56] Testing on 5000 data points...
[2021-03-16 00:53:56] Test accuracy: 0.8936
[2021-03-16 00:53:56] Score for 275 labels: 0.8936
[2021-03-16 00:53:56] Showing labelled: False (4725/5000 visible)
[2021-03-16 00:53:56] Found 4725 unlabelled features.
[2021-03-16 00:53:56] Training GMM (k=50) on labelled features...
[2021-03-16 00:53:56] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:53:57] Showing labelled: True (300/5000 visible)
[2021-03-16 00:53:57] Training mlp across 300 data points in blobsthin...
[2021-03-16 00:53:58] Training loss: 0.0199
[2021-03-16 00:53:58] Testing on 5000 data points...
[2021-03-16 00:53:58] Test accuracy: 0.8938
[2021-03-16 00:53:58] Score for 300 labels: 0.8938
[2021-03-16 00:53:58] Experiment repeat 2/2
[2021-03-16 00:53:58] Randomly labelled 50/5000
[2021-03-16 00:53:58] Showing labelled: True (50/5000 visible)
[2021-03-16 00:53:58] Running: experiment 0
[2021-03-16 00:53:58] Showing labelled: True (50/5000 visible)
[2021-03-16 00:53:58] mlp: initialized 1482 parameters.
[2021-03-16 00:53:58] Creating trainer with model on device: cuda
[2021-03-16 00:53:58] Training mlp across 50 data points in blobsthin...
[2021-03-16 00:53:58] Training loss: 1.4453
[2021-03-16 00:53:58] Testing on 5000 data points...
[2021-03-16 00:53:58] Test accuracy: 0.5000
[2021-03-16 00:53:58] Score for 50 labels: 0.5000
[2021-03-16 00:53:58] Showing labelled: False (4950/5000 visible)
[2021-03-16 00:53:58] Found 4950 unlabelled features.
[2021-03-16 00:53:58] Training GMM (k=50) on labelled features...
[2021-03-16 00:53:59] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:53:59] Showing labelled: True (75/5000 visible)
[2021-03-16 00:53:59] Training mlp across 75 data points in blobsthin...
[2021-03-16 00:54:00] Training loss: 0.4818
[2021-03-16 00:54:00] Testing on 5000 data points...
[2021-03-16 00:54:00] Test accuracy: 0.7000
[2021-03-16 00:54:00] Score for 75 labels: 0.7000
[2021-03-16 00:54:00] Showing labelled: False (4925/5000 visible)
[2021-03-16 00:54:00] Found 4925 unlabelled features.
[2021-03-16 00:54:00] Training GMM (k=50) on labelled features...
[2021-03-16 00:54:00] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:54:00] Showing labelled: True (100/5000 visible)
[2021-03-16 00:54:00] Training mlp across 100 data points in blobsthin...
[2021-03-16 00:54:01] Training loss: 0.2145
[2021-03-16 00:54:01] Testing on 5000 data points...
[2021-03-16 00:54:01] Test accuracy: 0.8504
[2021-03-16 00:54:01] Score for 100 labels: 0.8504
[2021-03-16 00:54:01] Showing labelled: False (4900/5000 visible)
[2021-03-16 00:54:01] Found 4900 unlabelled features.
[2021-03-16 00:54:01] Training GMM (k=50) on labelled features...
[2021-03-16 00:54:01] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:54:02] Showing labelled: True (125/5000 visible)
[2021-03-16 00:54:02] Training mlp across 125 data points in blobsthin...
[2021-03-16 00:54:02] Training loss: 0.1147
[2021-03-16 00:54:02] Testing on 5000 data points...
[2021-03-16 00:54:02] Test accuracy: 0.8976
[2021-03-16 00:54:02] Score for 125 labels: 0.8976
[2021-03-16 00:54:02] Showing labelled: False (4875/5000 visible)
[2021-03-16 00:54:02] Found 4875 unlabelled features.
[2021-03-16 00:54:02] Training GMM (k=50) on labelled features...
[2021-03-16 00:54:02] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:54:03] Showing labelled: True (150/5000 visible)
[2021-03-16 00:54:03] Training mlp across 150 data points in blobsthin...
[2021-03-16 00:54:04] Training loss: 0.0854
[2021-03-16 00:54:04] Testing on 5000 data points...
[2021-03-16 00:54:04] Test accuracy: 0.8988
[2021-03-16 00:54:04] Score for 150 labels: 0.8988
[2021-03-16 00:54:04] Showing labelled: False (4850/5000 visible)
[2021-03-16 00:54:04] Found 4850 unlabelled features.
[2021-03-16 00:54:04] Training GMM (k=50) on labelled features...
[2021-03-16 00:54:04] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:54:04] Showing labelled: True (175/5000 visible)
[2021-03-16 00:54:04] Training mlp across 175 data points in blobsthin...
[2021-03-16 00:54:05] Training loss: 0.0661
[2021-03-16 00:54:05] Testing on 5000 data points...
[2021-03-16 00:54:05] Test accuracy: 0.8994
[2021-03-16 00:54:05] Score for 175 labels: 0.8994
[2021-03-16 00:54:05] Showing labelled: False (4825/5000 visible)
[2021-03-16 00:54:05] Found 4825 unlabelled features.
[2021-03-16 00:54:05] Training GMM (k=50) on labelled features...
[2021-03-16 00:54:06] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:54:06] Showing labelled: True (200/5000 visible)
[2021-03-16 00:54:06] Training mlp across 200 data points in blobsthin...
[2021-03-16 00:54:07] Training loss: 0.0554
[2021-03-16 00:54:07] Testing on 5000 data points...
[2021-03-16 00:54:07] Test accuracy: 0.9516
[2021-03-16 00:54:07] Score for 200 labels: 0.9516
[2021-03-16 00:54:07] Showing labelled: False (4800/5000 visible)
[2021-03-16 00:54:07] Found 4800 unlabelled features.
[2021-03-16 00:54:07] Training GMM (k=50) on labelled features...
[2021-03-16 00:54:07] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:54:08] Showing labelled: True (225/5000 visible)
[2021-03-16 00:54:08] Training mlp across 225 data points in blobsthin...
[2021-03-16 00:54:09] Training loss: 0.0372
[2021-03-16 00:54:09] Testing on 5000 data points...
[2021-03-16 00:54:09] Test accuracy: 0.9854
[2021-03-16 00:54:09] Score for 225 labels: 0.9854
[2021-03-16 00:54:09] Showing labelled: False (4775/5000 visible)
[2021-03-16 00:54:09] Found 4775 unlabelled features.
[2021-03-16 00:54:09] Training GMM (k=50) on labelled features...
[2021-03-16 00:54:09] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:54:10] Showing labelled: True (250/5000 visible)
[2021-03-16 00:54:10] Training mlp across 250 data points in blobsthin...
[2021-03-16 00:54:11] Training loss: 0.0261
[2021-03-16 00:54:11] Testing on 5000 data points...
[2021-03-16 00:54:11] Test accuracy: 0.9910
[2021-03-16 00:54:11] Score for 250 labels: 0.9910
[2021-03-16 00:54:11] Showing labelled: False (4750/5000 visible)
[2021-03-16 00:54:11] Found 4750 unlabelled features.
[2021-03-16 00:54:11] Training GMM (k=50) on labelled features...
[2021-03-16 00:54:11] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:54:12] Showing labelled: True (275/5000 visible)
[2021-03-16 00:54:12] Training mlp across 275 data points in blobsthin...
[2021-03-16 00:54:13] Training loss: 0.0177
[2021-03-16 00:54:13] Testing on 5000 data points...
[2021-03-16 00:54:13] Test accuracy: 0.9938
[2021-03-16 00:54:13] Score for 275 labels: 0.9938
[2021-03-16 00:54:13] Showing labelled: False (4725/5000 visible)
[2021-03-16 00:54:13] Found 4725 unlabelled features.
[2021-03-16 00:54:13] Training GMM (k=50) on labelled features...
[2021-03-16 00:54:14] Training GMM (k=50) on unlabelled features...
[2021-03-16 00:54:14] Showing labelled: True (300/5000 visible)
[2021-03-16 00:54:14] Training mlp across 300 data points in blobsthin...
[2021-03-16 00:54:16] Training loss: 0.0133
[2021-03-16 00:54:16] Testing on 5000 data points...
[2021-03-16 00:54:16] Test accuracy: 0.9948
[2021-03-16 00:54:16] Score for 300 labels: 0.9948
[2021-03-16 00:54:16] Updated results: ../results/sp_coreset_focus/results.json
