Create constellation_a_test_trainer_cifar10_output.txt
Browse files
constellation_a_test_trainer_cifar10_output.txt
ADDED
|
@@ -0,0 +1,223 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
============================================================
|
| 2 |
+
GeoLIP Core β Conv + ConstellationCore
|
| 3 |
+
Encoder: 6-layer conv β 256-d sphere
|
| 4 |
+
Constellation: 64 anchors, 8Γ64 patchwork
|
| 5 |
+
Activation: squared_relu
|
| 6 |
+
Loss: CE + InfoNCE + attract + CV(0.22) + spread
|
| 7 |
+
Batch: 256, LR: 0.0003, Epochs: 100
|
| 8 |
+
Push: every 100 batches, lr=0.1
|
| 9 |
+
Device: cuda
|
| 10 |
+
============================================================
|
| 11 |
+
Train: 50,000 Val: 10,000
|
| 12 |
+
Parameters: 1,706,058 (encoder: 1,213,504, core: 492,554)
|
| 13 |
+
|
| 14 |
+
============================================================
|
| 15 |
+
TRAINING β 100 epochs
|
| 16 |
+
============================================================
|
| 17 |
+
E 1/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.35b/s, acc=23%, cos=0.340, loss=4.4095, nce=0.63, ordered=1, push=1]
|
| 18 |
+
E 1: train=23.2% val=42.8% loss=4.3573 nce=0.63 cos=0.271 cv=0.2052(β) anch=36/64 push=1 (11s) β
|
| 19 |
+
E 2/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.53b/s, acc=45%, cos=0.393, loss=2.0887, nce=0.97, ordered=1, push=3]
|
| 20 |
+
E 2: train=45.5% val=49.6% loss=2.0830 nce=0.97 cos=0.388 cv=0.2057(β) anch=41/64 push=3 (11s) β
|
| 21 |
+
E 3/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.62b/s, acc=55%, cos=0.486, loss=1.6869, nce=0.99, ordered=1, push=5]
|
| 22 |
+
E 3: train=55.2% val=57.5% loss=1.6840 nce=0.99 cos=0.449 cv=0.2135(β) anch=34/64 push=5 (10s) β
|
| 23 |
+
E 4/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.58b/s, acc=62%, cos=0.520, loss=1.4310, nce=1.00, ordered=1, push=7]
|
| 24 |
+
E 4: train=62.0% val=65.3% loss=1.4285 nce=1.00 cos=0.503 cv=0.1861(β) anch=45/64 push=7 (10s) β
|
| 25 |
+
E 5/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.70b/s, acc=66%, cos=0.543, loss=1.2767, nce=1.00, ordered=1, push=9]
|
| 26 |
+
E 5: train=66.4% val=68.6% loss=1.2743 nce=1.00 cos=0.533 cv=0.1827(β) anch=49/64 push=9 (10s) β
|
| 27 |
+
E 6/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.48b/s, acc=70%, cos=0.539, loss=1.1631, nce=1.00, ordered=1, push=11]
|
| 28 |
+
E 6: train=70.1% val=69.0% loss=1.1617 nce=1.00 cos=0.550 cv=0.2212(β) anch=56/64 push=11 (11s) β
|
| 29 |
+
E 7/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.58b/s, acc=73%, cos=0.562, loss=1.0645, nce=1.00, ordered=1, push=13]
|
| 30 |
+
E 7: train=73.4% val=72.4% loss=1.0644 nce=1.00 cos=0.563 cv=0.1889(β) anch=57/64 push=13 (10s) β
|
| 31 |
+
E 8/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.39b/s, acc=75%, cos=0.582, loss=0.9999, nce=1.00, ordered=1, push=15]
|
| 32 |
+
E 8: train=75.2% val=74.1% loss=0.9993 nce=1.00 cos=0.573 cv=0.1984(β) anch=61/64 push=15 (11s) β
|
| 33 |
+
E 9/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.68b/s, acc=78%, cos=0.602, loss=0.9310, nce=1.00, ordered=1, push=17]
|
| 34 |
+
E 9: train=77.6% val=76.4% loss=0.9296 nce=1.00 cos=0.582 cv=0.1825(β) anch=61/64 push=17 (10s) β
|
| 35 |
+
E 10/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.05b/s, acc=79%, cos=0.590, loss=0.8909, nce=1.00, ordered=1, push=19]
|
| 36 |
+
E 10: train=78.8% val=75.8% loss=0.8898 nce=1.00 cos=0.589 cv=0.1858(β) anch=61/64 push=19 (11s)
|
| 37 |
+
E 11/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.77b/s, acc=80%, cos=0.586, loss=0.8450, nce=1.00, ordered=1, push=21]
|
| 38 |
+
E 11: train=80.1% val=79.3% loss=0.8431 nce=1.00 cos=0.595 cv=0.1865(β) anch=63/64 push=21 (10s) β
|
| 39 |
+
E 12/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.75b/s, acc=81%, cos=0.605, loss=0.8123, nce=1.00, ordered=1, push=23]
|
| 40 |
+
E 12: train=81.1% val=76.7% loss=0.8127 nce=1.00 cos=0.603 cv=0.2004(β) anch=64/64 push=23 (10s)
|
| 41 |
+
E 13/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.48b/s, acc=82%, cos=0.613, loss=0.7768, nce=1.00, ordered=1, push=25]
|
| 42 |
+
E 13: train=82.5% val=78.8% loss=0.7767 nce=1.00 cos=0.604 cv=0.1858(β) anch=63/64 push=25 (11s)
|
| 43 |
+
E 14/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.61b/s, acc=83%, cos=0.613, loss=0.7525, nce=1.00, ordered=1, push=27]
|
| 44 |
+
E 14: train=83.0% val=81.9% loss=0.7522 nce=1.00 cos=0.609 cv=0.1733(β) anch=64/64 push=27 (10s) β
|
| 45 |
+
E 15/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.61b/s, acc=84%, cos=0.621, loss=0.7223, nce=1.00, ordered=1, push=29]
|
| 46 |
+
E 15: train=84.0% val=80.3% loss=0.7226 nce=1.00 cos=0.612 cv=0.1799(β) anch=64/64 push=29 (10s)
|
| 47 |
+
E 16/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.66b/s, acc=85%, cos=0.613, loss=0.6988, nce=1.00, ordered=1, push=31]
|
| 48 |
+
E 16: train=84.9% val=81.8% loss=0.6975 nce=1.00 cos=0.612 cv=0.1844(β) anch=64/64 push=31 (10s)
|
| 49 |
+
E 17/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.53b/s, acc=85%, cos=0.617, loss=0.6724, nce=1.00, ordered=1, push=33]
|
| 50 |
+
E 17: train=85.5% val=83.1% loss=0.6723 nce=1.00 cos=0.615 cv=0.1803(β) anch=64/64 push=33 (11s) β
|
| 51 |
+
E 18/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.71b/s, acc=86%, cos=0.609, loss=0.6584, nce=1.00, ordered=1, push=35]
|
| 52 |
+
E 18: train=85.9% val=80.7% loss=0.6588 nce=1.00 cos=0.618 cv=0.1910(β) anch=64/64 push=35 (10s)
|
| 53 |
+
E 19/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.51b/s, acc=87%, cos=0.617, loss=0.6362, nce=1.00, ordered=1, push=37]
|
| 54 |
+
E 19: train=86.6% val=84.1% loss=0.6367 nce=1.00 cos=0.619 cv=0.1784(β) anch=64/64 push=37 (11s) β
|
| 55 |
+
E 20/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.24b/s, acc=87%, cos=0.621, loss=0.6196, nce=1.00, ordered=1, push=38]
|
| 56 |
+
E 20: train=87.2% val=83.6% loss=0.6198 nce=1.00 cos=0.620 cv=0.1750(β) anch=64/64 push=39 (11s)
|
| 57 |
+
E 21/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.64b/s, acc=88%, cos=0.637, loss=0.6056, nce=1.00, ordered=1, push=40]
|
| 58 |
+
E 21: train=87.8% val=85.2% loss=0.6048 nce=1.00 cos=0.621 cv=0.1749(β) anch=64/64 push=40 (10s) β
|
| 59 |
+
E 22/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.65b/s, acc=88%, cos=0.617, loss=0.5867, nce=1.00, ordered=1, push=42]
|
| 60 |
+
E 22: train=88.2% val=84.3% loss=0.5862 nce=1.00 cos=0.622 cv=0.1712(β) anch=64/64 push=42 (10s)
|
| 61 |
+
E 23/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.63b/s, acc=89%, cos=0.629, loss=0.5721, nce=1.00, ordered=1, push=44]
|
| 62 |
+
E 23: train=88.8% val=84.2% loss=0.5725 nce=1.00 cos=0.622 cv=0.1765(β) anch=64/64 push=44 (10s)
|
| 63 |
+
E 24/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.36b/s, acc=89%, cos=0.633, loss=0.5566, nce=1.00, ordered=1, push=46]
|
| 64 |
+
E 24: train=89.2% val=83.2% loss=0.5570 nce=1.00 cos=0.623 cv=0.1863(β) anch=64/64 push=46 (11s)
|
| 65 |
+
E 25/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.62b/s, acc=89%, cos=0.648, loss=0.5531, nce=1.00, ordered=1, push=48]
|
| 66 |
+
E 25: train=89.2% val=86.3% loss=0.5515 nce=1.00 cos=0.624 cv=0.1615(β) anch=64/64 push=48 (10s) β
|
| 67 |
+
E 26/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.60b/s, acc=90%, cos=0.633, loss=0.5364, nce=1.00, ordered=1, push=50]
|
| 68 |
+
E 26: train=90.0% val=85.6% loss=0.5358 nce=1.00 cos=0.623 cv=0.1764(β) anch=64/64 push=50 (10s)
|
| 69 |
+
E 27/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.56b/s, acc=90%, cos=0.617, loss=0.5244, nce=1.00, ordered=1, push=52]
|
| 70 |
+
E 27: train=90.2% val=86.3% loss=0.5247 nce=1.00 cos=0.623 cv=0.1668(β) anch=64/64 push=52 (11s)
|
| 71 |
+
E 28/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.56b/s, acc=91%, cos=0.648, loss=0.5100, nce=1.00, ordered=1, push=54]
|
| 72 |
+
E 28: train=90.8% val=85.6% loss=0.5098 nce=1.00 cos=0.626 cv=0.1699(β) anch=64/64 push=54 (11s)
|
| 73 |
+
E 29/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.28b/s, acc=91%, cos=0.625, loss=0.5029, nce=1.00, ordered=1, push=56]
|
| 74 |
+
E 29: train=91.0% val=86.1% loss=0.5025 nce=1.00 cos=0.626 cv=0.1664(β) anch=64/64 push=56 (11s)
|
| 75 |
+
E 30/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.76b/s, acc=92%, cos=0.621, loss=0.4870, nce=1.00, ordered=1, push=58]
|
| 76 |
+
E 30: train=91.6% val=85.5% loss=0.4879 nce=1.00 cos=0.625 cv=0.1603(β) anch=64/64 push=58 (10s)
|
| 77 |
+
E 31/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.65b/s, acc=92%, cos=0.629, loss=0.4800, nce=1.00, ordered=1, push=60]
|
| 78 |
+
E 31: train=91.8% val=86.7% loss=0.4803 nce=1.00 cos=0.625 cv=0.1659(β) anch=64/64 push=60 (10s) β
|
| 79 |
+
E 32/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.32b/s, acc=92%, cos=0.617, loss=0.4728, nce=1.00, ordered=1, push=62]
|
| 80 |
+
E 32: train=92.0% val=86.0% loss=0.4727 nce=1.00 cos=0.626 cv=0.1686(β) anch=64/64 push=62 (11s)
|
| 81 |
+
E 33/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.77b/s, acc=92%, cos=0.637, loss=0.4637, nce=1.00, ordered=1, push=64]
|
| 82 |
+
E 33: train=92.2% val=85.5% loss=0.4636 nce=1.00 cos=0.626 cv=0.1861(β) anch=64/64 push=64 (10s)
|
| 83 |
+
E 34/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.28b/s, acc=92%, cos=0.645, loss=0.4606, nce=1.00, ordered=1, push=66]
|
| 84 |
+
E 34: train=92.3% val=87.1% loss=0.4609 nce=1.00 cos=0.628 cv=0.1733(β) anch=64/64 push=66 (11s) β
|
| 85 |
+
E 35/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.67b/s, acc=93%, cos=0.625, loss=0.4478, nce=1.00, ordered=1, push=68]
|
| 86 |
+
E 35: train=92.8% val=86.9% loss=0.4485 nce=1.00 cos=0.627 cv=0.1702(β) anch=64/64 push=68 (10s)
|
| 87 |
+
E 36/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.50b/s, acc=93%, cos=0.629, loss=0.4374, nce=1.00, ordered=1, push=70]
|
| 88 |
+
E 36: train=93.2% val=87.3% loss=0.4370 nce=1.00 cos=0.629 cv=0.1721(β) anch=64/64 push=70 (11s) β
|
| 89 |
+
E 37/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.65b/s, acc=93%, cos=0.621, loss=0.4285, nce=1.00, ordered=1, push=72]
|
| 90 |
+
E 37: train=93.3% val=86.5% loss=0.4290 nce=1.00 cos=0.629 cv=0.1605(β) anch=63/64 push=72 (10s)
|
| 91 |
+
E 38/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.67b/s, acc=94%, cos=0.613, loss=0.4240, nce=1.00, ordered=1, push=74]
|
| 92 |
+
E 38: train=93.6% val=87.4% loss=0.4232 nce=1.00 cos=0.627 cv=0.1755(β) anch=63/64 push=74 (10s) β
|
| 93 |
+
E 39/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.70b/s, acc=94%, cos=0.621, loss=0.4182, nce=1.00, ordered=1, push=76]
|
| 94 |
+
E 39: train=93.7% val=86.8% loss=0.4185 nce=1.00 cos=0.628 cv=0.1719(β) anch=63/64 push=76 (10s)
|
| 95 |
+
E 40/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.65b/s, acc=94%, cos=0.633, loss=0.4094, nce=1.00, ordered=1, push=77]
|
| 96 |
+
E 40: train=94.1% val=87.6% loss=0.4097 nce=1.00 cos=0.628 cv=0.1657(β) anch=64/64 push=78 (10s) β
|
| 97 |
+
E 41/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.19b/s, acc=94%, cos=0.625, loss=0.4040, nce=1.00, ordered=1, push=79]
|
| 98 |
+
E 41: train=94.3% val=87.6% loss=0.4040 nce=1.00 cos=0.628 cv=0.1570(β) anch=64/64 push=79 (11s) β
|
| 99 |
+
E 42/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.65b/s, acc=94%, cos=0.648, loss=0.3969, nce=1.00, ordered=1, push=81]
|
| 100 |
+
E 42: train=94.4% val=86.2% loss=0.3968 nce=1.00 cos=0.628 cv=0.1591(β) anch=62/64 push=81 (10s)
|
| 101 |
+
E 43/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.57b/s, acc=95%, cos=0.621, loss=0.3887, nce=1.00, ordered=1, push=83]
|
| 102 |
+
E 43: train=94.8% val=88.0% loss=0.3889 nce=1.00 cos=0.629 cv=0.1589(β) anch=64/64 push=83 (11s) β
|
| 103 |
+
E 44/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.61b/s, acc=95%, cos=0.625, loss=0.3825, nce=1.00, ordered=1, push=85]
|
| 104 |
+
E 44: train=94.9% val=87.9% loss=0.3823 nce=1.00 cos=0.628 cv=0.1651(β) anch=62/64 push=85 (10s)
|
| 105 |
+
E 45/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.68b/s, acc=95%, cos=0.621, loss=0.3788, nce=1.00, ordered=1, push=87]
|
| 106 |
+
E 45: train=94.9% val=87.4% loss=0.3794 nce=1.00 cos=0.628 cv=0.1500(β) anch=62/64 push=87 (10s)
|
| 107 |
+
E 46/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.76b/s, acc=95%, cos=0.633, loss=0.3709, nce=1.00, ordered=1, push=89]
|
| 108 |
+
E 46: train=95.2% val=86.9% loss=0.3707 nce=1.00 cos=0.629 cv=0.1664(β) anch=63/64 push=89 (10s)
|
| 109 |
+
E 47/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.59b/s, acc=95%, cos=0.621, loss=0.3676, nce=1.00, ordered=1, push=91]
|
| 110 |
+
E 47: train=95.4% val=87.3% loss=0.3680 nce=1.00 cos=0.629 cv=0.1675(β) anch=62/64 push=91 (10s)
|
| 111 |
+
E 48/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.63b/s, acc=96%, cos=0.633, loss=0.3614, nce=1.00, ordered=1, push=93]
|
| 112 |
+
E 48: train=95.7% val=88.1% loss=0.3612 nce=1.00 cos=0.630 cv=0.1787(β) anch=61/64 push=93 (10s) β
|
| 113 |
+
E 49/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.57b/s, acc=96%, cos=0.629, loss=0.3580, nce=1.00, ordered=1, push=95]
|
| 114 |
+
E 49: train=95.7% val=87.3% loss=0.3587 nce=1.00 cos=0.629 cv=0.1675(β) anch=61/64 push=95 (11s)
|
| 115 |
+
E 50/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.45b/s, acc=96%, cos=0.629, loss=0.3546, nce=1.00, ordered=1, push=97]
|
| 116 |
+
E 50: train=95.8% val=87.9% loss=0.3543 nce=1.00 cos=0.630 cv=0.1666(β) anch=62/64 push=97 (11s)
|
| 117 |
+
E 51/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.57b/s, acc=96%, cos=0.641, loss=0.3463, nce=1.00, ordered=1, push=99]
|
| 118 |
+
E 51: train=96.2% val=87.6% loss=0.3463 nce=1.00 cos=0.629 cv=0.1779(β) anch=62/64 push=99 (10s)
|
| 119 |
+
E 52/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.54b/s, acc=96%, cos=0.637, loss=0.3408, nce=1.00, ordered=1, push=101]
|
| 120 |
+
E 52: train=96.3% val=88.5% loss=0.3415 nce=1.00 cos=0.629 cv=0.1613(β) anch=62/64 push=101 (11s) β
|
| 121 |
+
E 53/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.78b/s, acc=97%, cos=0.625, loss=0.3344, nce=1.00, ordered=1, push=103]
|
| 122 |
+
E 53: train=96.5% val=88.4% loss=0.3354 nce=1.00 cos=0.628 cv=0.1543(β) anch=61/64 push=103 (10s)
|
| 123 |
+
E 54/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.61b/s, acc=97%, cos=0.621, loss=0.3326, nce=1.00, ordered=1, push=105]
|
| 124 |
+
E 54: train=96.6% val=86.7% loss=0.3323 nce=1.00 cos=0.629 cv=0.1658(β) anch=61/64 push=105 (10s)
|
| 125 |
+
E 55/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.69b/s, acc=97%, cos=0.633, loss=0.3310, nce=1.00, ordered=1, push=107]
|
| 126 |
+
E 55: train=96.7% val=88.5% loss=0.3311 nce=1.00 cos=0.629 cv=0.1676(β) anch=62/64 push=107 (10s)
|
| 127 |
+
E 56/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 17.96b/s, acc=97%, cos=0.629, loss=0.3264, nce=1.00, ordered=1, push=109]
|
| 128 |
+
E 56: train=96.8% val=88.5% loss=0.3263 nce=1.00 cos=0.628 cv=0.1533(β) anch=60/64 push=109 (11s)
|
| 129 |
+
E 57/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.60b/s, acc=97%, cos=0.629, loss=0.3218, nce=1.00, ordered=1, push=111]
|
| 130 |
+
E 57: train=97.0% val=88.5% loss=0.3221 nce=1.00 cos=0.628 cv=0.1593(β) anch=64/64 push=111 (10s)
|
| 131 |
+
E 58/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.57b/s, acc=97%, cos=0.637, loss=0.3172, nce=1.00, ordered=1, push=113]
|
| 132 |
+
E 58: train=97.2% val=89.0% loss=0.3174 nce=1.00 cos=0.628 cv=0.1582(β) anch=64/64 push=113 (11s) β
|
| 133 |
+
E 59/100: 100%|ββββββββββ| 195/195 [00:11<00:00, 16.91b/s, acc=97%, cos=0.633, loss=0.3149, nce=1.00, ordered=1, push=115]
|
| 134 |
+
E 59: train=97.1% val=88.4% loss=0.3151 nce=1.00 cos=0.628 cv=0.1496(β) anch=62/64 push=115 (12s)
|
| 135 |
+
E 60/100: 100%|ββββββββββ| 195/195 [00:11<00:00, 17.72b/s, acc=98%, cos=0.629, loss=0.3079, nce=1.00, ordered=1, push=116]
|
| 136 |
+
E 60: train=97.6% val=88.5% loss=0.3079 nce=1.00 cos=0.627 cv=0.1644(β) anch=60/64 push=117 (11s)
|
| 137 |
+
E 61/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 17.73b/s, acc=97%, cos=0.609, loss=0.3066, nce=1.00, ordered=1, push=118]
|
| 138 |
+
E 61: train=97.5% val=89.4% loss=0.3066 nce=1.00 cos=0.628 cv=0.1601(β) anch=62/64 push=118 (11s) β
|
| 139 |
+
E 62/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.64b/s, acc=98%, cos=0.637, loss=0.3010, nce=1.00, ordered=1, push=120]
|
| 140 |
+
E 62: train=97.7% val=89.2% loss=0.3014 nce=1.00 cos=0.628 cv=0.1527(β) anch=60/64 push=120 (10s)
|
| 141 |
+
E 63/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.49b/s, acc=98%, cos=0.625, loss=0.3013, nce=1.00, ordered=1, push=122]
|
| 142 |
+
E 63: train=97.8% val=88.9% loss=0.3012 nce=1.00 cos=0.628 cv=0.1679(β) anch=62/64 push=122 (11s)
|
| 143 |
+
E 64/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.46b/s, acc=98%, cos=0.625, loss=0.3013, nce=1.00, ordered=1, push=124]
|
| 144 |
+
E 64: train=97.7% val=88.9% loss=0.3014 nce=1.00 cos=0.627 cv=0.1576(β) anch=62/64 push=124 (11s)
|
| 145 |
+
E 65/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.58b/s, acc=98%, cos=0.633, loss=0.2960, nce=1.00, ordered=1, push=126]
|
| 146 |
+
E 65: train=97.9% val=88.9% loss=0.2964 nce=1.00 cos=0.627 cv=0.1604(β) anch=61/64 push=126 (10s)
|
| 147 |
+
E 66/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.74b/s, acc=98%, cos=0.633, loss=0.2937, nce=1.00, ordered=1, push=128]
|
| 148 |
+
E 66: train=98.0% val=88.9% loss=0.2941 nce=1.00 cos=0.625 cv=0.1633(β) anch=61/64 push=128 (10s)
|
| 149 |
+
E 67/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.64b/s, acc=98%, cos=0.625, loss=0.2903, nce=1.00, ordered=1, push=130]
|
| 150 |
+
E 67: train=98.1% val=89.2% loss=0.2900 nce=1.00 cos=0.627 cv=0.1642(β) anch=62/64 push=130 (10s)
|
| 151 |
+
E 68/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.57b/s, acc=98%, cos=0.621, loss=0.2857, nce=1.00, ordered=1, push=132]
|
| 152 |
+
E 68: train=98.2% val=89.0% loss=0.2857 nce=1.00 cos=0.626 cv=0.1784(β) anch=64/64 push=132 (11s)
|
| 153 |
+
E 69/100: 100%|ββββββββββ| 195/195 [00:11<00:00, 17.71b/s, acc=98%, cos=0.625, loss=0.2835, nce=1.00, ordered=1, push=134]
|
| 154 |
+
E 69: train=98.3% val=89.6% loss=0.2836 nce=1.00 cos=0.626 cv=0.1552(β) anch=62/64 push=134 (11s) β
|
| 155 |
+
E 70/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.66b/s, acc=98%, cos=0.641, loss=0.2845, nce=1.00, ordered=1, push=136]
|
| 156 |
+
E 70: train=98.3% val=88.9% loss=0.2846 nce=1.00 cos=0.626 cv=0.1506(β) anch=64/64 push=136 (10s)
|
| 157 |
+
E 71/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.67b/s, acc=98%, cos=0.613, loss=0.2826, nce=1.00, ordered=1, push=138]
|
| 158 |
+
E 71: train=98.4% val=88.9% loss=0.2824 nce=1.00 cos=0.625 cv=0.1662(β) anch=63/64 push=138 (10s)
|
| 159 |
+
E 72/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.76b/s, acc=99%, cos=0.629, loss=0.2785, nce=1.00, ordered=1, push=140]
|
| 160 |
+
E 72: train=98.6% val=89.2% loss=0.2783 nce=1.00 cos=0.625 cv=0.1694(β) anch=60/64 push=140 (10s)
|
| 161 |
+
E 73/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.52b/s, acc=98%, cos=0.621, loss=0.2790, nce=1.00, ordered=1, push=142]
|
| 162 |
+
E 73: train=98.5% val=89.2% loss=0.2790 nce=1.00 cos=0.624 cv=0.1657(β) anch=61/64 push=142 (11s)
|
| 163 |
+
E 74/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.68b/s, acc=99%, cos=0.613, loss=0.2763, nce=1.00, ordered=1, push=144]
|
| 164 |
+
E 74: train=98.6% val=89.1% loss=0.2763 nce=1.00 cos=0.624 cv=0.1499(β) anch=62/64 push=144 (10s)
|
| 165 |
+
E 75/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.74b/s, acc=99%, cos=0.617, loss=0.2734, nce=1.00, ordered=1, push=146]
|
| 166 |
+
E 75: train=98.7% val=89.2% loss=0.2735 nce=1.00 cos=0.623 cv=0.1513(β) anch=59/64 push=146 (10s)
|
| 167 |
+
E 76/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.74b/s, acc=99%, cos=0.621, loss=0.2726, nce=1.00, ordered=1, push=148]
|
| 168 |
+
E 76: train=98.7% val=89.4% loss=0.2727 nce=1.00 cos=0.624 cv=0.1747(β) anch=61/64 push=148 (10s)
|
| 169 |
+
E 77/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.66b/s, acc=99%, cos=0.629, loss=0.2716, nce=1.00, ordered=1, push=150]
|
| 170 |
+
E 77: train=98.8% val=89.5% loss=0.2715 nce=1.00 cos=0.623 cv=0.1729(β) anch=61/64 push=150 (10s)
|
| 171 |
+
E 78/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.60b/s, acc=99%, cos=0.621, loss=0.2705, nce=1.00, ordered=1, push=152]
|
| 172 |
+
E 78: train=98.9% val=89.5% loss=0.2706 nce=1.00 cos=0.622 cv=0.1662(β) anch=63/64 push=152 (10s)
|
| 173 |
+
E 79/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.52b/s, acc=99%, cos=0.617, loss=0.2695, nce=1.00, ordered=1, push=154]
|
| 174 |
+
E 79: train=98.8% val=89.6% loss=0.2697 nce=1.00 cos=0.620 cv=0.1543(β) anch=63/64 push=154 (11s) β
|
| 175 |
+
E 80/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.51b/s, acc=99%, cos=0.633, loss=0.2671, nce=1.00, ordered=1, push=155]
|
| 176 |
+
E 80: train=99.0% val=89.6% loss=0.2670 nce=1.00 cos=0.621 cv=0.1497(β) anch=64/64 push=156 (11s)
|
| 177 |
+
E 81/100: 100%|ββββββοΏ½οΏ½βββ| 195/195 [00:10<00:00, 18.49b/s, acc=99%, cos=0.621, loss=0.2659, nce=1.00, ordered=1, push=157]
|
| 178 |
+
E 81: train=99.0% val=89.7% loss=0.2660 nce=1.00 cos=0.620 cv=0.1483(β) anch=64/64 push=157 (11s) β
|
| 179 |
+
E 82/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.60b/s, acc=99%, cos=0.617, loss=0.2648, nce=1.00, ordered=1, push=159]
|
| 180 |
+
E 82: train=99.1% val=89.5% loss=0.2651 nce=1.00 cos=0.619 cv=0.1481(β) anch=63/64 push=159 (10s)
|
| 181 |
+
E 83/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.57b/s, acc=99%, cos=0.617, loss=0.2662, nce=1.00, ordered=1, push=161]
|
| 182 |
+
E 83: train=99.0% val=89.7% loss=0.2663 nce=1.00 cos=0.619 cv=0.1515(β) anch=63/64 push=161 (11s)
|
| 183 |
+
E 84/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.70b/s, acc=99%, cos=0.625, loss=0.2635, nce=1.00, ordered=1, push=163]
|
| 184 |
+
E 84: train=99.2% val=89.8% loss=0.2636 nce=1.00 cos=0.618 cv=0.1581(β) anch=64/64 push=163 (10s) β
|
| 185 |
+
E 85/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.60b/s, acc=99%, cos=0.621, loss=0.2636, nce=1.00, ordered=1, push=165]
|
| 186 |
+
E 85: train=99.2% val=89.5% loss=0.2635 nce=1.00 cos=0.617 cv=0.1610(β) anch=63/64 push=165 (10s)
|
| 187 |
+
E 86/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.70b/s, acc=99%, cos=0.613, loss=0.2635, nce=1.00, ordered=1, push=167]
|
| 188 |
+
E 86: train=99.2% val=89.9% loss=0.2635 nce=1.00 cos=0.616 cv=0.1661(β) anch=64/64 push=167 (10s) β
|
| 189 |
+
E 87/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.27b/s, acc=99%, cos=0.613, loss=0.2626, nce=1.00, ordered=1, push=169]
|
| 190 |
+
E 87: train=99.2% val=89.7% loss=0.2628 nce=1.00 cos=0.615 cv=0.1491(β) anch=64/64 push=169 (11s)
|
| 191 |
+
E 88/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.61b/s, acc=99%, cos=0.617, loss=0.2625, nce=1.00, ordered=1, push=171]
|
| 192 |
+
E 88: train=99.2% val=89.8% loss=0.2626 nce=1.00 cos=0.614 cv=0.1519(β) anch=64/64 push=171 (10s)
|
| 193 |
+
E 89/100: 100%|ββββββββββ| 195/195 [00:11<00:00, 17.43b/s, acc=99%, cos=0.605, loss=0.2627, nce=1.00, ordered=1, push=173]
|
| 194 |
+
E 89: train=99.2% val=89.8% loss=0.2629 nce=1.00 cos=0.614 cv=0.1712(β) anch=63/64 push=173 (11s)
|
| 195 |
+
E 90/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.31b/s, acc=99%, cos=0.617, loss=0.2620, nce=1.00, ordered=1, push=175]
|
| 196 |
+
E 90: train=99.3% val=89.6% loss=0.2620 nce=1.00 cos=0.613 cv=0.1522(β) anch=63/64 push=175 (11s)
|
| 197 |
+
E 91/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.65b/s, acc=99%, cos=0.621, loss=0.2614, nce=1.00, ordered=1, push=177]
|
| 198 |
+
E 91: train=99.3% val=89.9% loss=0.2613 nce=1.00 cos=0.613 cv=0.1674(β) anch=64/64 push=177 (10s) β
|
| 199 |
+
E 92/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.55b/s, acc=99%, cos=0.609, loss=0.2616, nce=1.00, ordered=1, push=179]
|
| 200 |
+
E 92: train=99.3% val=90.0% loss=0.2616 nce=1.00 cos=0.611 cv=0.1468(β) anch=64/64 push=179 (11s) β
|
| 201 |
+
E 93/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.75b/s, acc=99%, cos=0.609, loss=0.2612, nce=1.00, ordered=1, push=181]
|
| 202 |
+
E 93: train=99.4% val=89.9% loss=0.2611 nce=1.00 cos=0.611 cv=0.1644(β) anch=64/64 push=181 (10s)
|
| 203 |
+
E 94/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.48b/s, acc=99%, cos=0.613, loss=0.2626, nce=1.00, ordered=1, push=183]
|
| 204 |
+
E 94: train=99.3% val=89.8% loss=0.2624 nce=1.00 cos=0.610 cv=0.1587(β) anch=64/64 push=183 (11s)
|
| 205 |
+
E 95/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.75b/s, acc=99%, cos=0.617, loss=0.2614, nce=1.00, ordered=1, push=185]
|
| 206 |
+
E 95: train=99.4% val=89.7% loss=0.2612 nce=1.00 cos=0.609 cv=0.1335(β) anch=64/64 push=185 (10s)
|
| 207 |
+
E 96/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.59b/s, acc=99%, cos=0.594, loss=0.2617, nce=1.00, ordered=1, push=187]
|
| 208 |
+
E 96: train=99.4% val=89.8% loss=0.2616 nce=1.00 cos=0.609 cv=0.1648(β) anch=64/64 push=187 (10s)
|
| 209 |
+
E 97/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.54b/s, acc=99%, cos=0.605, loss=0.2628, nce=1.00, ordered=1, push=189]
|
| 210 |
+
E 97: train=99.4% val=89.9% loss=0.2627 nce=1.00 cos=0.608 cv=0.1569(β) anch=64/64 push=189 (11s)
|
| 211 |
+
E 98/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.79b/s, acc=99%, cos=0.605, loss=0.2622, nce=1.00, ordered=1, push=191]
|
| 212 |
+
E 98: train=99.4% val=89.8% loss=0.2620 nce=1.00 cos=0.608 cv=0.1534(β) anch=64/64 push=191 (10s)
|
| 213 |
+
E 99/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.61b/s, acc=99%, cos=0.605, loss=0.2624, nce=1.00, ordered=1, push=193]
|
| 214 |
+
E 99: train=99.4% val=89.8% loss=0.2626 nce=1.00 cos=0.607 cv=0.1433(β) anch=64/64 push=193 (10s)
|
| 215 |
+
E100/100: 100%|ββββββββββ| 195/195 [00:10<00:00, 18.53b/s, acc=99%, cos=0.602, loss=0.2639, nce=1.00, ordered=1, push=194]
|
| 216 |
+
E100: train=99.3% val=89.9% loss=0.2641 nce=1.00 cos=0.607 cv=0.1606(β) anch=64/64 push=195 (11s)
|
| 217 |
+
|
| 218 |
+
Best val accuracy: 90.0%
|
| 219 |
+
Parameters: 1,706,058
|
| 220 |
+
|
| 221 |
+
============================================================
|
| 222 |
+
DONE
|
| 223 |
+
============================================================
|