| 07:59:34 args: {'variant': 'wrong_text', 'epochs': 30, 'bs': 512, 'lr': 0.001, 'lamb': 0.02, 'proj_dim': 128, 'num_workers': 12, 'grad_clip': 1.0, 'ckpt_every': 5, 'data_dir': '/vast/adi/tclejepa/v1', 'wandb': 1, 'push_hf': 1} |
| 07:59:36 train=50000 val=10000 |
| 08:00:23 ep 0 | loss=5.4125 inv=0.1978 sig=29.6598 lejepa=0.7870 probe=4.6256 probe_acc=0.1108 gnorm=2.39 lr=5.05e-04 dt=43.0s |
| 08:00:59 ep 1 | loss=4.9444 inv=0.1023 sig=16.6005 lejepa=0.4322 probe=4.5121 probe_acc=0.1196 gnorm=1.64 lr=1.00e-03 dt=36.0s |
| 08:01:33 ep 2 | loss=4.9069 inv=0.0779 sig=17.2545 lejepa=0.4214 probe=4.4855 probe_acc=0.1316 gnorm=1.12 lr=9.97e-04 dt=34.1s |
| 08:02:08 ep 3 | loss=4.8383 inv=0.0484 sig=14.5019 lejepa=0.3375 probe=4.5008 probe_acc=0.1236 gnorm=0.72 lr=9.88e-04 dt=35.0s |
| 08:02:43 ep 4 | loss=4.7880 inv=0.0362 sig=13.3460 lejepa=0.3024 probe=4.4855 probe_acc=0.1196 gnorm=0.71 lr=9.72e-04 dt=34.8s |
| 08:02:43 saved /vast/adi/tclejepa/v1/checkpoints/wrong_text/epoch_004.pt |
| 08:03:24 ep 5 | loss=4.6976 inv=0.0269 sig=11.8305 lejepa=0.2629 probe=4.4347 probe_acc=0.1200 gnorm=0.60 lr=9.51e-04 dt=34.7s |
| 08:03:59 ep 6 | loss=4.6444 inv=0.0233 sig=11.1553 lejepa=0.2459 probe=4.3986 probe_acc=0.1340 gnorm=0.66 lr=9.24e-04 dt=35.2s |
| 08:04:35 ep 7 | loss=4.6318 inv=0.0216 sig=10.9162 lejepa=0.2395 probe=4.3923 probe_acc=0.1328 gnorm=0.62 lr=8.92e-04 dt=35.8s |
| 08:05:10 ep 8 | loss=4.5947 inv=0.0221 sig=10.2030 lejepa=0.2257 probe=4.3690 probe_acc=0.1340 gnorm=0.58 lr=8.55e-04 dt=34.8s |
| 08:05:44 ep 9 | loss=4.5846 inv=0.0210 sig=10.1070 lejepa=0.2227 probe=4.3619 probe_acc=0.1276 gnorm=0.55 lr=8.14e-04 dt=34.3s |
| 08:05:44 saved /vast/adi/tclejepa/v1/checkpoints/wrong_text/epoch_009.pt |
| 08:06:25 ep 10 | loss=4.5552 inv=0.0189 sig=9.6517 lejepa=0.2116 probe=4.3436 probe_acc=0.1316 gnorm=0.45 lr=7.68e-04 dt=35.7s |
| 08:07:00 ep 11 | loss=4.5690 inv=0.0157 sig=8.9997 lejepa=0.1954 probe=4.3736 probe_acc=0.1296 gnorm=0.46 lr=7.20e-04 dt=34.8s |
| 08:07:34 ep 12 | loss=4.5968 inv=0.0147 sig=9.1279 lejepa=0.1969 probe=4.3999 probe_acc=0.1348 gnorm=0.50 lr=6.68e-04 dt=33.5s |
| 08:08:07 ep 13 | loss=4.5613 inv=0.0139 sig=8.7310 lejepa=0.1883 probe=4.3730 probe_acc=0.1368 gnorm=0.48 lr=6.15e-04 dt=33.7s |
| 08:08:42 ep 14 | loss=4.5303 inv=0.0134 sig=8.5209 lejepa=0.1836 probe=4.3468 probe_acc=0.1360 gnorm=0.49 lr=5.60e-04 dt=34.7s |
| 08:08:42 saved /vast/adi/tclejepa/v1/checkpoints/wrong_text/epoch_014.pt |
| 08:09:24 ep 15 | loss=4.5311 inv=0.0124 sig=8.1395 lejepa=0.1750 probe=4.3561 probe_acc=0.1428 gnorm=0.40 lr=5.05e-04 dt=36.6s |
| 08:10:00 ep 16 | loss=4.5071 inv=0.0113 sig=7.9526 lejepa=0.1701 probe=4.3370 probe_acc=0.1348 gnorm=0.47 lr=4.50e-04 dt=36.0s |
| 08:10:34 ep 17 | loss=4.5072 inv=0.0105 sig=7.7581 lejepa=0.1655 probe=4.3417 probe_acc=0.1328 gnorm=0.40 lr=3.95e-04 dt=34.1s |
| 08:11:09 ep 18 | loss=4.4853 inv=0.0105 sig=7.4932 lejepa=0.1601 probe=4.3251 probe_acc=0.1368 gnorm=0.49 lr=3.42e-04 dt=34.7s |
| 08:11:45 ep 19 | loss=4.4828 inv=0.0103 sig=7.3843 lejepa=0.1577 probe=4.3250 probe_acc=0.1436 gnorm=0.42 lr=2.90e-04 dt=36.2s |
| 08:11:46 saved /vast/adi/tclejepa/v1/checkpoints/wrong_text/epoch_019.pt |
| 08:12:26 ep 20 | loss=4.4611 inv=0.0096 sig=7.1566 lejepa=0.1525 probe=4.3086 probe_acc=0.1424 gnorm=0.37 lr=2.42e-04 dt=35.7s |
| 08:13:01 ep 21 | loss=4.4512 inv=0.0095 sig=7.0477 lejepa=0.1502 probe=4.3010 probe_acc=0.1380 gnorm=0.40 lr=1.96e-04 dt=34.9s |
| 08:13:37 ep 22 | loss=4.4386 inv=0.0096 sig=6.8138 lejepa=0.1457 probe=4.2929 probe_acc=0.1364 gnorm=0.40 lr=1.55e-04 dt=36.2s |
| 08:14:12 ep 23 | loss=4.4403 inv=0.0094 sig=6.7716 lejepa=0.1446 probe=4.2957 probe_acc=0.1444 gnorm=0.43 lr=1.18e-04 dt=35.0s |
| 08:14:46 ep 24 | loss=4.4298 inv=0.0092 sig=6.6408 lejepa=0.1418 probe=4.2880 probe_acc=0.1420 gnorm=0.36 lr=8.59e-05 dt=33.6s |
| 08:14:46 saved /vast/adi/tclejepa/v1/checkpoints/wrong_text/epoch_024.pt |
| 08:15:26 ep 25 | loss=4.4158 inv=0.0093 sig=6.5544 lejepa=0.1402 probe=4.2756 probe_acc=0.1468 gnorm=0.34 lr=5.90e-05 dt=35.0s |
| 08:16:01 ep 26 | loss=4.4149 inv=0.0090 sig=6.4381 lejepa=0.1376 probe=4.2773 probe_acc=0.1412 gnorm=0.31 lr=3.78e-05 dt=34.6s |
| 08:16:37 ep 27 | loss=4.4083 inv=0.0091 sig=6.4133 lejepa=0.1372 probe=4.2711 probe_acc=0.1420 gnorm=0.34 lr=2.24e-05 dt=35.8s |
| 08:17:11 ep 28 | loss=4.4077 inv=0.0091 sig=6.4079 lejepa=0.1371 probe=4.2706 probe_acc=0.1424 gnorm=0.31 lr=1.31e-05 dt=34.5s |
| 08:17:46 ep 29 | loss=4.4015 inv=0.0091 sig=6.3412 lejepa=0.1357 probe=4.2657 probe_acc=0.1456 gnorm=0.31 lr=1.00e-05 dt=34.4s |
| 08:17:46 saved /vast/adi/tclejepa/v1/checkpoints/wrong_text/epoch_029.pt |
| 08:17:51 DONE in 1091s |
|
|
