[env_hub] repo=/storage/home/jiaruiliu/local/git-repos/Open-Moe-Embedding venv=/home/jiaruiliu/local/envs/openmoeembedding data=/lustre-storage/fsx_it_0/users/jiaruiliu/open-moe-embedding-data need_gpu=1 host=h100-062-204 [flagship] nodes=4 (world 32) master=h100-062-204 config=configs/pretrain/flagship_h100_4x8.json [env_hub] repo=/storage/home/jiaruiliu/local/git-repos/Open-Moe-Embedding venv=/home/jiaruiliu/local/envs/openmoeembedding data=/lustre-storage/fsx_it_0/users/jiaruiliu/open-moe-embedding-data need_gpu=1 host=h100-064-159 [env_hub] repo=/storage/home/jiaruiliu/local/git-repos/Open-Moe-Embedding venv=/home/jiaruiliu/local/envs/openmoeembedding data=/lustre-storage/fsx_it_0/users/jiaruiliu/open-moe-embedding-data need_gpu=1 host=h100-062-204 [env_hub] repo=/storage/home/jiaruiliu/local/git-repos/Open-Moe-Embedding venv=/home/jiaruiliu/local/envs/openmoeembedding data=/lustre-storage/fsx_it_0/users/jiaruiliu/open-moe-embedding-data need_gpu=1 host=h100-064-152 [env_hub] repo=/storage/home/jiaruiliu/local/git-repos/Open-Moe-Embedding venv=/home/jiaruiliu/local/envs/openmoeembedding data=/lustre-storage/fsx_it_0/users/jiaruiliu/open-moe-embedding-data need_gpu=1 host=h100-063-109 W0802 14:17:24.451000 182029 /storage/home/jiaruiliu/local/envs/openmoeembedding/lib/python3.12/site-packages/torch/distributed/run.py:766] W0802 14:17:24.451000 182029 /storage/home/jiaruiliu/local/envs/openmoeembedding/lib/python3.12/site-packages/torch/distributed/run.py:766] ***************************************** W0802 14:17:24.451000 182029 /storage/home/jiaruiliu/local/envs/openmoeembedding/lib/python3.12/site-packages/torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. W0802 14:17:24.451000 182029 /storage/home/jiaruiliu/local/envs/openmoeembedding/lib/python3.12/site-packages/torch/distributed/run.py:766] ***************************************** W0802 14:17:24.471000 181518 /storage/home/jiaruiliu/local/envs/openmoeembedding/lib/python3.12/site-packages/torch/distributed/run.py:766] W0802 14:17:24.471000 181518 /storage/home/jiaruiliu/local/envs/openmoeembedding/lib/python3.12/site-packages/torch/distributed/run.py:766] ***************************************** W0802 14:17:24.471000 181518 /storage/home/jiaruiliu/local/envs/openmoeembedding/lib/python3.12/site-packages/torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. W0802 14:17:24.471000 181518 /storage/home/jiaruiliu/local/envs/openmoeembedding/lib/python3.12/site-packages/torch/distributed/run.py:766] ***************************************** W0802 14:17:24.497000 181486 /storage/home/jiaruiliu/local/envs/openmoeembedding/lib/python3.12/site-packages/torch/distributed/run.py:766] W0802 14:17:24.497000 181486 /storage/home/jiaruiliu/local/envs/openmoeembedding/lib/python3.12/site-packages/torch/distributed/run.py:766] ***************************************** W0802 14:17:24.497000 181486 /storage/home/jiaruiliu/local/envs/openmoeembedding/lib/python3.12/site-packages/torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. W0802 14:17:24.497000 181486 /storage/home/jiaruiliu/local/envs/openmoeembedding/lib/python3.12/site-packages/torch/distributed/run.py:766] ***************************************** W0802 14:17:24.530000 181506 /storage/home/jiaruiliu/local/envs/openmoeembedding/lib/python3.12/site-packages/torch/distributed/run.py:766] W0802 14:17:24.530000 181506 /storage/home/jiaruiliu/local/envs/openmoeembedding/lib/python3.12/site-packages/torch/distributed/run.py:766] ***************************************** W0802 14:17:24.530000 181506 /storage/home/jiaruiliu/local/envs/openmoeembedding/lib/python3.12/site-packages/torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. W0802 14:17:24.530000 181506 /storage/home/jiaruiliu/local/envs/openmoeembedding/lib/python3.12/site-packages/torch/distributed/run.py:766] ***************************************** [moe] lb_coef=0.01 z_loss_coef=0.001 [moe] lb_coef=0.01 z_loss_coef=0.001 [moe] lb_coef=0.01 z_loss_coef=0.001 [moe] lb_coef=0.01 z_loss_coef=0.001 NCCL version 2.26.2+cuda12.2 [pretrain] total_steps=40589 global_batch=16384 B/rank=512 chunk=128 world=32 [pretrain] total_steps=40589 global_batch=16384 B/rank=512 chunk=128 world=32 [pretrain] total_steps=40589 global_batch=16384 B/rank=512 chunk=128 world=32 [pretrain] total_steps=40589 global_batch=16384 B/rank=512 chunk=128 world=32 step 10/40589 loss 10.2263 acc 0.002 lb 2.155 z 12.131 lr 1.64e-06 | 1012 pairs/s | mem 59.8/79 GiB | 7.6d left step 20/40589 loss 9.9267 acc 0.003 lb 2.157 z 12.133 lr 3.29e-06 | 1179 pairs/s | mem 59.8/79 GiB | 6.5d left [routing] mean entropy_norm 0.803 max_load(range) 0.095-0.351 step 30/40589 loss 9.6212 acc 0.003 lb 2.164 z 12.133 lr 4.93e-06 | 1254 pairs/s | mem 61.1/79 GiB (min 59.4) | 6.1d left step 40/40589 loss 9.3044 acc 0.006 lb 2.162 z 12.133 lr 6.57e-06 | 1291 pairs/s | mem 61.7/79 GiB (min 59.4) | 6.0d left step 50/40589 loss 8.7640 acc 0.014 lb 2.156 z 12.139 lr 8.22e-06 | 1321 pairs/s | mem 61.7/79 GiB (min 59.4) | 5.8d left [routing] mean entropy_norm 0.801 max_load(range) 0.104-0.309 step 60/40589 loss 7.8590 acc 0.039 lb 2.146 z 12.145 lr 9.86e-06 | 1344 pairs/s | mem 61.7/79 GiB (min 59.4) | 5.7d left step 70/40589 loss 6.7272 acc 0.101 lb 2.152 z 12.147 lr 1.15e-05 | 1359 pairs/s | mem 61.9/79 GiB (min 59.5) | 5.7d left [routing] mean entropy_norm 0.811 max_load(range) 0.111-0.329 step 80/40589 loss 5.5873 acc 0.213 lb 2.154 z 12.140 lr 1.31e-05 | 1374 pairs/s | mem 61.9/79 GiB (min 59.6) | 5.6d left step 90/40589 loss 4.5187 acc 0.364 lb 2.156 z 12.139 lr 1.48e-05 | 1387 pairs/s | mem 61.9/79 GiB (min 59.6) | 5.5d left step 100/40589 loss 3.6640 acc 0.492 lb 2.161 z 12.139 lr 1.64e-05 | 1398 pairs/s | mem 61.9/79 GiB (min 59.6) | 5.5d left [routing] mean entropy_norm 0.776 max_load(range) 0.081-0.378 step 110/40589 loss 3.0155 acc 0.584 lb 2.170 z 12.135 lr 1.81e-05 | 1409 pairs/s | mem 61.9/79 GiB (min 60.1) | 5.4d left step 120/40589 loss 2.6492 acc 0.630 lb 2.181 z 12.128 lr 1.97e-05 | 1419 pairs/s | mem 61.9/79 GiB (min 60.1) | 5.4d left [routing] mean entropy_norm 0.756 max_load(range) 0.081-0.426 step 130/40589 loss 2.3510 acc 0.672 lb 2.189 z 12.132 lr 2.14e-05 | 1429 pairs/s | mem 61.9/79 GiB (min 60.2) | 5.4d left step 140/40589 loss 2.2557 acc 0.682 lb 2.195 z 12.137 lr 2.30e-05 | 1437 pairs/s | mem 61.9/79 GiB (min 60.2) | 5.3d left step 150/40589 loss 2.0533 acc 0.710 lb 2.201 z 12.139 lr 2.47e-05 | 1444 pairs/s | mem 61.9/79 GiB (min 61.0) | 5.3d left [routing] mean entropy_norm 0.729 max_load(range) 0.096-0.467 step 160/40589 loss 1.9237 acc 0.727 lb 2.207 z 12.140 lr 2.63e-05 | 1450 pairs/s | mem 61.9/79 GiB (min 61.0) | 5.3d left step 170/40589 loss 1.8082 acc 0.743 lb 2.205 z 12.145 lr 2.79e-05 | 1457 pairs/s | mem 61.9/79 GiB (min 61.0) | 5.3d left [routing] mean entropy_norm 0.612 max_load(range) 0.145-0.461 step 180/40589 loss 1.8105 acc 0.741 lb 2.205 z 12.152 lr 2.96e-05 | 1462 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.2d left step 190/40589 loss 1.6965 acc 0.755 lb 2.207 z 12.157 lr 3.12e-05 | 1468 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.2d left step 200/40589 loss 1.6200 acc 0.766 lb 2.212 z 12.159 lr 3.29e-05 | 1473 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.2d left [routing] mean entropy_norm 0.601 max_load(range) 0.179-0.459 step 210/40589 loss 1.5422 acc 0.777 lb 2.203 z 12.163 lr 3.45e-05 | 1477 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.2d left step 220/40589 loss 1.5712 acc 0.772 lb 2.204 z 12.170 lr 3.62e-05 | 1481 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.2d left [routing] mean entropy_norm 0.734 max_load(range) 0.159-0.355 step 230/40589 loss 1.4834 acc 0.782 lb 2.204 z 12.174 lr 3.78e-05 | 1485 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.2d left step 240/40589 loss 1.4383 acc 0.790 lb 2.203 z 12.173 lr 3.94e-05 | 1488 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left step 250/40589 loss 1.3676 acc 0.799 lb 2.194 z 12.178 lr 4.11e-05 | 1491 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left [routing] mean entropy_norm 0.735 max_load(range) 0.152-0.318 step 260/40589 loss 1.3926 acc 0.792 lb 2.193 z 12.185 lr 4.27e-05 | 1493 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left step 270/40589 loss 1.3181 acc 0.804 lb 2.195 z 12.187 lr 4.44e-05 | 1495 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left [routing] mean entropy_norm 0.704 max_load(range) 0.092-0.474 step 280/40589 loss 1.2867 acc 0.809 lb 2.192 z 12.186 lr 4.60e-05 | 1497 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left step 290/40589 loss 1.2277 acc 0.816 lb 2.180 z 12.187 lr 4.77e-05 | 1499 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left step 300/40589 loss 1.2626 acc 0.810 lb 2.179 z 12.189 lr 4.93e-05 | 1501 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left [routing] mean entropy_norm 0.762 max_load(range) 0.106-0.380 step 310/40589 loss 1.1986 acc 0.818 lb 2.181 z 12.189 lr 5.09e-05 | 1502 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left step 320/40589 loss 1.1629 acc 0.824 lb 2.174 z 12.185 lr 5.26e-05 | 1504 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left [routing] mean entropy_norm 0.692 max_load(range) 0.107-0.483 step 330/40589 loss 1.1089 acc 0.832 lb 2.161 z 12.180 lr 5.42e-05 | 1505 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left step 340/40589 loss 1.1469 acc 0.824 lb 2.161 z 12.181 lr 5.59e-05 | 1506 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left step 350/40589 loss 1.0944 acc 0.831 lb 2.164 z 12.177 lr 5.75e-05 | 1506 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left [routing] mean entropy_norm 0.713 max_load(range) 0.090-0.482 step 360/40589 loss 1.0516 acc 0.838 lb 2.158 z 12.173 lr 5.92e-05 | 1507 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left step 370/40589 loss 1.0143 acc 0.844 lb 2.143 z 12.166 lr 6.08e-05 | 1508 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left [routing] mean entropy_norm 0.612 max_load(range) 0.144-0.499 step 380/40589 loss 1.0573 acc 0.836 lb 2.144 z 12.162 lr 6.24e-05 | 1508 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left step 390/40589 loss 1.0040 acc 0.843 lb 2.146 z 12.157 lr 6.41e-05 | 1508 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left step 400/40589 loss 0.9773 acc 0.848 lb 2.147 z 12.147 lr 6.57e-05 | 1508 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left [routing] mean entropy_norm 0.615 max_load(range) 0.166-0.500 step 410/40589 loss 0.9322 acc 0.856 lb 2.127 z 12.136 lr 6.74e-05 | 1509 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left step 420/40589 loss 0.9798 acc 0.846 lb 2.129 z 12.129 lr 6.90e-05 | 1509 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.0d left [routing] mean entropy_norm 0.817 max_load(range) 0.122-0.362 step 430/40589 loss 0.9298 acc 0.854 lb 2.128 z 12.120 lr 7.07e-05 | 1509 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.0d left step 440/40589 loss 0.9103 acc 0.857 lb 2.125 z 12.109 lr 7.23e-05 | 1508 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.0d left step 450/40589 loss 0.8752 acc 0.862 lb 2.111 z 12.096 lr 7.40e-05 | 1508 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.0d left [routing] mean entropy_norm 0.822 max_load(range) 0.113-0.398 step 460/40589 loss 0.9061 acc 0.857 lb 2.113 z 12.088 lr 7.56e-05 | 1508 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.0d left step 470/40589 loss 0.8803 acc 0.860 lb 2.113 z 12.075 lr 7.72e-05 | 1507 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.0d left [routing] mean entropy_norm 0.734 max_load(range) 0.091-0.478 step 480/40589 loss 0.8510 acc 0.864 lb 2.110 z 12.061 lr 7.89e-05 | 1507 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.0d left step 490/40589 loss 0.8176 acc 0.870 lb 2.098 z 12.044 lr 8.05e-05 | 1506 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.0d left step 500/40589 loss 0.8608 acc 0.861 lb 2.103 z 12.035 lr 8.22e-05 | 1505 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.0d left [routing] mean entropy_norm 0.848 max_load(range) 0.071-0.425 step 510/40589 loss 0.8329 acc 0.866 lb 2.108 z 12.017 lr 8.38e-05 | 1505 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left step 520/40589 loss 0.8117 acc 0.870 lb 2.102 z 12.002 lr 8.55e-05 | 1504 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left [routing] mean entropy_norm 0.745 max_load(range) 0.090-0.478 step 530/40589 loss 0.7888 acc 0.873 lb 2.086 z 11.984 lr 8.71e-05 | 1504 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left step 540/40589 loss 0.8184 acc 0.868 lb 2.089 z 11.973 lr 8.87e-05 | 1503 pairs/s | mem 62.1/79 GiB (min 61.2) | 5.1d left step 550/40589 loss 0.7916 acc 0.872 lb 2.079 z 11.953 lr 9.04e-05 | 1502 pairs/s | mem 62.1/79 GiB (min 61.3) | 5.1d left [routing] mean entropy_norm 0.744 max_load(range) 0.095-0.482 step 560/40589 loss 0.7733 acc 0.875 lb 2.077 z 11.933 lr 9.20e-05 | 1501 pairs/s | mem 62.1/79 GiB (min 61.3) | 5.1d left step 570/40589 loss 0.7472 acc 0.880 lb 2.058 z 11.906 lr 9.37e-05 | 1501 pairs/s | mem 62.1/79 GiB (min 61.3) | 5.1d left [routing] mean entropy_norm 0.722 max_load(range) 0.107-0.499 step 580/40589 loss 0.7778 acc 0.874 lb 2.073 z 11.900 lr 9.53e-05 | 1500 pairs/s | mem 62.1/79 GiB (min 61.3) | 5.1d left step 590/40589 loss 0.7549 acc 0.877 lb 2.074 z 11.878 lr 9.70e-05 | 1499 pairs/s | mem 62.1/79 GiB (min 61.4) | 5.1d left step 600/40589 loss 0.7309 acc 0.881 lb 2.074 z 11.859 lr 9.86e-05 | 1498 pairs/s | mem 62.1/79 GiB (min 61.4) | 5.1d left [routing] mean entropy_norm 0.721 max_load(range) 0.101-0.487 step 610/40589 loss 0.7151 acc 0.883 lb 2.054 z 11.830 lr 1.00e-04 | 1498 pairs/s | mem 62.1/79 GiB (min 61.4) | 5.1d left step 620/40589 loss 0.7504 acc 0.877 lb 2.059 z 11.819 lr 1.02e-04 | 1497 pairs/s | mem 62.1/79 GiB (min 61.4) | 5.1d left [routing] mean entropy_norm 0.866 max_load(range) 0.065-0.332 step 630/40589 loss 0.7173 acc 0.882 lb 2.070 z 11.803 lr 1.04e-04 | 1496 pairs/s | mem 62.1/79 GiB (min 61.4) | 5.1d left step 640/40589 loss 0.7037 acc 0.885 lb 2.057 z 11.782 lr 1.05e-04 | 1495 pairs/s | mem 62.1/79 GiB (min 61.4) | 5.1d left step 650/40589 loss 0.6860 acc 0.888 lb 2.036 z 11.741 lr 1.07e-04 | 1495 pairs/s | mem 62.1/79 GiB (min 61.4) | 5.1d left [routing] mean entropy_norm 0.871 max_load(range) 0.073-0.340 step 660/40589 loss 0.7213 acc 0.879 lb 2.052 z 11.737 lr 1.08e-04 | 1494 pairs/s | mem 62.1/79 GiB (min 61.4) | 5.1d left step 670/40589 loss 0.6951 acc 0.884 lb 2.060 z 11.722 lr 1.10e-04 | 1493 pairs/s | mem 62.1/79 GiB (min 61.4) | 5.1d left [routing] mean entropy_norm 0.757 max_load(range) 0.102-0.483 step 680/40589 loss 0.6784 acc 0.888 lb 2.042 z 11.693 lr 1.12e-04 | 1493 pairs/s | mem 62.1/79 GiB (min 61.4) | 5.1d left step 690/40589 loss 0.6639 acc 0.890 lb 2.029 z 11.649 lr 1.13e-04 | 1492 pairs/s | mem 62.1/79 GiB (min 61.4) | 5.1d left step 700/40589 loss 0.6977 acc 0.884 lb 2.041 z 11.639 lr 1.15e-04 | 1491 pairs/s | mem 62.1/79 GiB (min 61.4) | 5.1d left [routing] mean entropy_norm 0.899 max_load(range) 0.067-0.327 step 710/40589 loss 0.6744 acc 0.887 lb 2.042 z 11.616 lr 1.17e-04 | 1491 pairs/s | mem 62.1/79 GiB (min 61.4) | 5.1d left step 720/40589 loss 0.6618 acc 0.890 lb 2.028 z 11.586 lr 1.18e-04 | 1490 pairs/s | mem 62.1/79 GiB (min 61.5) | 5.1d left [routing] mean entropy_norm 0.853 max_load(range) 0.083-0.398 step 730/40589 loss 0.6352 acc 0.895 lb 1.998 z 11.523 lr 1.20e-04 | 1489 pairs/s | mem 62.2/79 GiB (min 61.5) | 5.1d left step 740/40589 loss 0.6767 acc 0.887 lb 2.010 z 11.510 lr 1.22e-04 | 1489 pairs/s | mem 62.2/79 GiB (min 61.5) | 5.1d left step 750/40589 loss 0.6611 acc 0.891 lb 2.024 z 11.483 lr 1.23e-04 | 1488 pairs/s | mem 62.3/79 GiB (min 61.6) | 5.1d left [routing] mean entropy_norm 0.774 max_load(range) 0.114-0.479 step 760/40589 loss 0.6380 acc 0.893 lb 1.996 z 11.425 lr 1.25e-04 | 1487 pairs/s | mem 62.4/79 GiB (min 61.6) | 5.1d left step 770/40589 loss 0.6177 acc 0.898 lb 1.988 z 11.341 lr 1.27e-04 | 1487 pairs/s | mem 62.4/79 GiB (min 61.7) | 5.1d left [routing] mean entropy_norm 0.842 max_load(range) 0.091-0.435 step 780/40589 loss 0.6531 acc 0.891 lb 1.998 z 11.319 lr 1.28e-04 | 1486 pairs/s | mem 62.5/79 GiB (min 61.7) | 5.1d left step 790/40589 loss 0.6335 acc 0.894 lb 2.008 z 11.285 lr 1.30e-04 | 1485 pairs/s | mem 62.5/79 GiB (min 61.7) | 5.1d left step 800/40589 loss 0.6181 acc 0.896 lb 1.970 z 11.176 lr 1.31e-04 | 1485 pairs/s | mem 62.5/79 GiB (min 61.7) | 5.1d left [routing] mean entropy_norm 0.854 max_load(range) 0.087-0.332 step 810/40589 loss 0.6053 acc 0.899 lb 1.957 z 11.085 lr 1.33e-04 | 1484 pairs/s | mem 62.5/79 GiB (min 61.8) | 5.1d left step 820/40589 loss 0.6393 acc 0.893 lb 1.983 z 11.075 lr 1.35e-04 | 1484 pairs/s | mem 62.5/79 GiB (min 61.8) | 5.1d left [routing] mean entropy_norm 0.902 max_load(range) 0.077-0.285 step 830/40589 loss 0.6156 acc 0.897 lb 1.983 z 11.043 lr 1.36e-04 | 1483 pairs/s | mem 62.5/79 GiB (min 61.8) | 5.1d left step 840/40589 loss 0.6005 acc 0.898 lb 1.950 z 10.905 lr 1.38e-04 | 1482 pairs/s | mem 62.5/79 GiB (min 61.8) | 5.1d left step 850/40589 loss 0.5823 acc 0.902 lb 1.953 z 10.854 lr 1.40e-04 | 1482 pairs/s | mem 62.7/79 GiB (min 61.8) | 5.1d left [routing] mean entropy_norm 0.914 max_load(range) 0.061-0.282 step 860/40589 loss 0.6204 acc 0.895 lb 1.968 z 10.855 lr 1.41e-04 | 1481 pairs/s | mem 62.7/79 GiB (min 61.8) | 5.1d left step 870/40589 loss 0.5976 acc 0.899 lb 1.979 z 10.855 lr 1.43e-04 | 1481 pairs/s | mem 62.7/79 GiB (min 61.8) | 5.1d left [routing] mean entropy_norm 0.774 max_load(range) 0.109-0.478 step 880/40589 loss 0.5894 acc 0.899 lb 1.962 z 10.752 lr 1.45e-04 | 1480 pairs/s | mem 62.7/79 GiB (min 61.8) | 5.1d left step 890/40589 loss 0.5749 acc 0.903 lb 1.961 z 10.717 lr 1.46e-04 | 1480 pairs/s | mem 62.7/79 GiB (min 61.8) | 5.1d left step 900/40589 loss 0.6050 acc 0.897 lb 1.974 z 10.741 lr 1.48e-04 | 1479 pairs/s | mem 62.7/79 GiB (min 61.8) | 5.1d left [routing] mean entropy_norm 0.924 max_load(range) 0.064-0.262 step 910/40589 loss 0.5910 acc 0.900 lb 1.991 z 10.777 lr 1.50e-04 | 1479 pairs/s | mem 62.7/79 GiB (min 61.8) | 5.1d left step 920/40589 loss 0.5722 acc 0.902 lb 1.955 z 10.618 lr 1.51e-04 | 1478 pairs/s | mem 62.7/79 GiB (min 61.8) | 5.1d left [routing] mean entropy_norm 0.885 max_load(range) 0.077-0.323 step 930/40589 loss 0.5729 acc 0.905 lb 1.952 z 10.518 lr 1.53e-04 | 1478 pairs/s | mem 62.7/79 GiB (min 61.9) | 5.1d left step 940/40589 loss 0.6004 acc 0.899 lb 1.969 z 10.545 lr 1.54e-04 | 1477 pairs/s | mem 62.7/79 GiB (min 61.9) | 5.1d left step 950/40589 loss 0.5817 acc 0.901 lb 1.981 z 10.543 lr 1.56e-04 | 1477 pairs/s | mem 62.9/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.790 max_load(range) 0.108-0.469 step 960/40589 loss 0.5637 acc 0.905 lb 1.944 z 10.276 lr 1.58e-04 | 1476 pairs/s | mem 62.9/79 GiB (min 61.9) | 5.1d left step 970/40589 loss 0.5582 acc 0.906 lb 1.952 z 10.257 lr 1.59e-04 | 1476 pairs/s | mem 62.9/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.887 max_load(range) 0.074-0.203 step 980/40589 loss 0.5880 acc 0.901 lb 1.963 z 10.255 lr 1.61e-04 | 1476 pairs/s | mem 62.9/79 GiB (min 61.9) | 5.1d left step 990/40589 loss 0.5671 acc 0.904 lb 1.963 z 10.196 lr 1.63e-04 | 1475 pairs/s | mem 63.0/79 GiB (min 61.9) | 5.1d left step 1000/40589 loss 0.5548 acc 0.905 lb 1.945 z 9.997 lr 1.64e-04 | 1475 pairs/s | mem 63.0/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.876 max_load(range) 0.081-0.248 [save] runs/pretrain_flagship_h100_4x8/step1000 step 1010/40589 loss 0.5354 acc 0.909 lb 1.940 z 9.941 lr 1.66e-04 | 1471 pairs/s | mem 63.0/79 GiB (min 61.9) | 5.1d left step 1020/40589 loss 0.5717 acc 0.903 lb 1.948 z 9.937 lr 1.68e-04 | 1471 pairs/s | mem 63.0/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.928 max_load(range) 0.067-0.235 step 1030/40589 loss 0.5491 acc 0.907 lb 1.964 z 10.003 lr 1.69e-04 | 1470 pairs/s | mem 63.0/79 GiB (min 61.9) | 5.1d left step 1040/40589 loss 0.5370 acc 0.907 lb 1.937 z 9.687 lr 1.71e-04 | 1470 pairs/s | mem 63.0/79 GiB (min 61.9) | 5.1d left step 1050/40589 loss 0.5225 acc 0.911 lb 1.938 z 9.672 lr 1.73e-04 | 1470 pairs/s | mem 63.0/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.921 max_load(range) 0.073-0.274 step 1060/40589 loss 0.5609 acc 0.904 lb 1.944 z 9.661 lr 1.74e-04 | 1469 pairs/s | mem 63.0/79 GiB (min 61.9) | 5.1d left step 1070/40589 loss 0.5481 acc 0.906 lb 1.952 z 9.671 lr 1.76e-04 | 1469 pairs/s | mem 63.0/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.798 max_load(range) 0.105-0.472 step 1080/40589 loss 0.5305 acc 0.909 lb 1.933 z 9.431 lr 1.77e-04 | 1469 pairs/s | mem 63.0/79 GiB (min 61.9) | 5.1d left step 1090/40589 loss 0.5145 acc 0.912 lb 1.937 z 9.444 lr 1.79e-04 | 1468 pairs/s | mem 63.0/79 GiB (min 61.9) | 5.1d left step 1100/40589 loss 0.5503 acc 0.903 lb 1.947 z 9.463 lr 1.81e-04 | 1468 pairs/s | mem 63.0/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.935 max_load(range) 0.058-0.231 step 1110/40589 loss 0.5334 acc 0.909 lb 1.951 z 9.468 lr 1.82e-04 | 1468 pairs/s | mem 63.0/79 GiB (min 61.9) | 5.1d left step 1120/40589 loss 0.5166 acc 0.911 lb 1.932 z 9.189 lr 1.84e-04 | 1467 pairs/s | mem 63.0/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.892 max_load(range) 0.082-0.321 step 1130/40589 loss 0.5011 acc 0.913 lb 1.932 z 9.163 lr 1.86e-04 | 1467 pairs/s | mem 63.0/79 GiB (min 61.9) | 5.1d left step 1140/40589 loss 0.5379 acc 0.908 lb 1.942 z 9.209 lr 1.87e-04 | 1467 pairs/s | mem 63.0/79 GiB (min 61.9) | 5.1d left step 1150/40589 loss 0.5245 acc 0.909 lb 1.952 z 9.270 lr 1.89e-04 | 1466 pairs/s | mem 63.0/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.802 max_load(range) 0.116-0.466 step 1160/40589 loss 0.5159 acc 0.912 lb 1.932 z 8.948 lr 1.91e-04 | 1466 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1170/40589 loss 0.4982 acc 0.915 lb 1.925 z 8.857 lr 1.92e-04 | 1466 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.892 max_load(range) 0.084-0.213 step 1180/40589 loss 0.5362 acc 0.907 lb 1.937 z 8.949 lr 1.94e-04 | 1466 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1190/40589 loss 0.5182 acc 0.911 lb 1.951 z 9.046 lr 1.96e-04 | 1465 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1200/40589 loss 0.5066 acc 0.913 lb 1.932 z 8.726 lr 1.97e-04 | 1465 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.897 max_load(range) 0.079-0.223 step 1210/40589 loss 0.4912 acc 0.915 lb 1.933 z 8.725 lr 1.99e-04 | 1465 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1220/40589 loss 0.5297 acc 0.908 lb 1.934 z 8.707 lr 2.00e-04 | 1464 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.930 max_load(range) 0.067-0.234 step 1230/40589 loss 0.5126 acc 0.912 lb 1.949 z 8.844 lr 2.00e-04 | 1464 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1240/40589 loss 0.4972 acc 0.914 lb 1.916 z 8.353 lr 2.00e-04 | 1464 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1250/40589 loss 0.4876 acc 0.916 lb 1.933 z 8.523 lr 2.00e-04 | 1464 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.914 max_load(range) 0.055-0.308 step 1260/40589 loss 0.5169 acc 0.910 lb 1.929 z 8.445 lr 2.00e-04 | 1463 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1270/40589 loss 0.5008 acc 0.913 lb 1.946 z 8.663 lr 2.00e-04 | 1463 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.804 max_load(range) 0.105-0.462 step 1280/40589 loss 0.4868 acc 0.916 lb 1.926 z 8.298 lr 2.00e-04 | 1463 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1290/40589 loss 0.4765 acc 0.918 lb 1.930 z 8.323 lr 2.00e-04 | 1463 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1300/40589 loss 0.5115 acc 0.911 lb 1.930 z 8.297 lr 2.00e-04 | 1462 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.945 max_load(range) 0.058-0.209 step 1310/40589 loss 0.4879 acc 0.915 lb 1.934 z 8.328 lr 2.00e-04 | 1462 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1320/40589 loss 0.4817 acc 0.916 lb 1.919 z 8.010 lr 2.00e-04 | 1462 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.916 max_load(range) 0.072-0.278 step 1330/40589 loss 0.4715 acc 0.918 lb 1.916 z 7.990 lr 2.00e-04 | 1462 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1340/40589 loss 0.5025 acc 0.913 lb 1.932 z 8.159 lr 2.00e-04 | 1462 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1350/40589 loss 0.4874 acc 0.915 lb 1.941 z 8.276 lr 2.00e-04 | 1461 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.818 max_load(range) 0.105-0.457 step 1360/40589 loss 0.4745 acc 0.917 lb 1.917 z 7.806 lr 2.00e-04 | 1461 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1370/40589 loss 0.4631 acc 0.920 lb 1.919 z 7.854 lr 2.00e-04 | 1461 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.889 max_load(range) 0.080-0.210 step 1380/40589 loss 0.4910 acc 0.914 lb 1.929 z 7.977 lr 2.00e-04 | 1461 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1390/40589 loss 0.4798 acc 0.917 lb 1.937 z 8.069 lr 2.00e-04 | 1460 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1400/40589 loss 0.4692 acc 0.918 lb 1.919 z 7.670 lr 2.00e-04 | 1460 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.889 max_load(range) 0.084-0.249 step 1410/40589 loss 0.4622 acc 0.919 lb 1.915 z 7.654 lr 2.00e-04 | 1460 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1420/40589 loss 0.4927 acc 0.914 lb 1.932 z 7.841 lr 2.00e-04 | 1460 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.928 max_load(range) 0.065-0.247 step 1430/40589 loss 0.4752 acc 0.918 lb 1.938 z 7.888 lr 2.00e-04 | 1460 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1440/40589 loss 0.4635 acc 0.919 lb 1.923 z 7.577 lr 2.00e-04 | 1459 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1450/40589 loss 0.4536 acc 0.921 lb 1.920 z 7.574 lr 2.00e-04 | 1459 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.916 max_load(range) 0.052-0.312 step 1460/40589 loss 0.4814 acc 0.915 lb 1.930 z 7.728 lr 2.00e-04 | 1459 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1470/40589 loss 0.4662 acc 0.918 lb 1.934 z 7.734 lr 2.00e-04 | 1459 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.819 max_load(range) 0.099-0.471 step 1480/40589 loss 0.4520 acc 0.921 lb 1.910 z 7.264 lr 2.00e-04 | 1458 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1490/40589 loss 0.4485 acc 0.922 lb 1.918 z 7.364 lr 2.00e-04 | 1458 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1500/40589 loss 0.4779 acc 0.916 lb 1.927 z 7.490 lr 2.00e-04 | 1458 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.943 max_load(range) 0.056-0.201 step 1510/40589 loss 0.4581 acc 0.920 lb 1.928 z 7.507 lr 2.00e-04 | 1458 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1520/40589 loss 0.4479 acc 0.922 lb 1.904 z 7.050 lr 2.00e-04 | 1458 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.914 max_load(range) 0.074-0.275 step 1530/40589 loss 0.4369 acc 0.924 lb 1.910 z 7.165 lr 2.00e-04 | 1457 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1540/40589 loss 0.4698 acc 0.917 lb 1.932 z 7.499 lr 2.00e-04 | 1457 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1550/40589 loss 0.4548 acc 0.921 lb 1.940 z 7.600 lr 2.00e-04 | 1457 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.829 max_load(range) 0.094-0.455 step 1560/40589 loss 0.4469 acc 0.922 lb 1.903 z 6.923 lr 2.00e-04 | 1457 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1570/40589 loss 0.4324 acc 0.924 lb 1.920 z 7.157 lr 2.00e-04 | 1457 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.898 max_load(range) 0.070-0.193 step 1580/40589 loss 0.4637 acc 0.918 lb 1.926 z 7.223 lr 2.00e-04 | 1456 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1590/40589 loss 0.4548 acc 0.921 lb 1.935 z 7.358 lr 2.00e-04 | 1456 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1600/40589 loss 0.4451 acc 0.921 lb 1.904 z 6.783 lr 2.00e-04 | 1456 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.891 max_load(range) 0.066-0.259 step 1610/40589 loss 0.4288 acc 0.925 lb 1.918 z 7.037 lr 2.00e-04 | 1456 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1620/40589 loss 0.4606 acc 0.919 lb 1.931 z 7.189 lr 2.00e-04 | 1456 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.923 max_load(range) 0.063-0.257 step 1630/40589 loss 0.4481 acc 0.921 lb 1.935 z 7.248 lr 2.00e-04 | 1456 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1640/40589 loss 0.4350 acc 0.923 lb 1.905 z 6.652 lr 2.00e-04 | 1455 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1650/40589 loss 0.4294 acc 0.925 lb 1.912 z 6.786 lr 2.00e-04 | 1455 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.919 max_load(range) 0.056-0.300 step 1660/40589 loss 0.4539 acc 0.920 lb 1.921 z 6.965 lr 2.00e-04 | 1455 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1670/40589 loss 0.4456 acc 0.922 lb 1.936 z 7.204 lr 2.00e-04 | 1455 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.827 max_load(range) 0.089-0.459 step 1680/40589 loss 0.4272 acc 0.924 lb 1.912 z 6.663 lr 2.00e-04 | 1455 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1690/40589 loss 0.4251 acc 0.925 lb 1.906 z 6.617 lr 2.00e-04 | 1455 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1700/40589 loss 0.4499 acc 0.920 lb 1.919 z 6.860 lr 2.00e-04 | 1455 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.941 max_load(range) 0.060-0.210 step 1710/40589 loss 0.4395 acc 0.923 lb 1.929 z 6.988 lr 2.00e-04 | 1454 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1720/40589 loss 0.4306 acc 0.924 lb 1.912 z 6.601 lr 2.00e-04 | 1454 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.869 max_load(range) 0.088-0.365 step 1730/40589 loss 0.4183 acc 0.926 lb 1.913 z 6.588 lr 2.00e-04 | 1454 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1740/40589 loss 0.4465 acc 0.920 lb 1.916 z 6.677 lr 2.00e-04 | 1454 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1750/40589 loss 0.4354 acc 0.923 lb 1.922 z 6.822 lr 2.00e-04 | 1454 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.835 max_load(range) 0.085-0.459 step 1760/40589 loss 0.4248 acc 0.924 lb 1.906 z 6.440 lr 2.00e-04 | 1454 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1770/40589 loss 0.4119 acc 0.927 lb 1.907 z 6.491 lr 2.00e-04 | 1454 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.898 max_load(range) 0.073-0.187 step 1780/40589 loss 0.4399 acc 0.922 lb 1.909 z 6.537 lr 2.00e-04 | 1454 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1790/40589 loss 0.4292 acc 0.925 lb 1.929 z 6.812 lr 2.00e-04 | 1453 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1800/40589 loss 0.4243 acc 0.925 lb 1.903 z 6.319 lr 2.00e-04 | 1453 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.896 max_load(range) 0.077-0.212 step 1810/40589 loss 0.4137 acc 0.927 lb 1.911 z 6.444 lr 2.00e-04 | 1453 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1820/40589 loss 0.4413 acc 0.922 lb 1.915 z 6.509 lr 2.00e-04 | 1453 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.931 max_load(range) 0.065-0.227 step 1830/40589 loss 0.4252 acc 0.924 lb 1.923 z 6.654 lr 2.00e-04 | 1453 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1840/40589 loss 0.4179 acc 0.926 lb 1.902 z 6.140 lr 2.00e-04 | 1453 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1850/40589 loss 0.4100 acc 0.928 lb 1.912 z 6.412 lr 2.00e-04 | 1453 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.929 max_load(range) 0.055-0.293 step 1860/40589 loss 0.4375 acc 0.922 lb 1.923 z 6.548 lr 2.00e-04 | 1453 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1870/40589 loss 0.4239 acc 0.925 lb 1.927 z 6.628 lr 2.00e-04 | 1452 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.837 max_load(range) 0.093-0.449 step 1880/40589 loss 0.4175 acc 0.926 lb 1.905 z 6.116 lr 2.00e-04 | 1452 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1890/40589 loss 0.4035 acc 0.929 lb 1.908 z 6.252 lr 2.00e-04 | 1452 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left step 1900/40589 loss 0.4315 acc 0.923 lb 1.922 z 6.453 lr 2.00e-04 | 1452 pairs/s | mem 63.1/79 GiB (min 61.9) | 5.1d left [routing] mean entropy_norm 0.943 max_load(range) 0.052-0.222 step 1910/40589 loss 0.4211 acc 0.926 lb 1.927 z 6.566 lr 2.00e-04 | 1452 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.1d left step 1920/40589 loss 0.4083 acc 0.927 lb 1.904 z 6.045 lr 2.00e-04 | 1452 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.1d left [routing] mean entropy_norm 0.891 max_load(range) 0.074-0.335 step 1930/40589 loss 0.4009 acc 0.929 lb 1.914 z 6.241 lr 2.00e-04 | 1452 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 1940/40589 loss 0.4298 acc 0.923 lb 1.921 z 6.347 lr 2.00e-04 | 1452 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 1950/40589 loss 0.4210 acc 0.926 lb 1.918 z 6.305 lr 2.00e-04 | 1452 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.842 max_load(range) 0.089-0.463 step 1960/40589 loss 0.4107 acc 0.927 lb 1.904 z 5.895 lr 2.00e-04 | 1451 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 1970/40589 loss 0.4000 acc 0.929 lb 1.914 z 6.118 lr 2.00e-04 | 1451 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.901 max_load(range) 0.069-0.193 step 1980/40589 loss 0.4313 acc 0.923 lb 1.920 z 6.195 lr 2.00e-04 | 1451 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 1990/40589 loss 0.4118 acc 0.927 lb 1.921 z 6.323 lr 2.00e-04 | 1451 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2000/40589 loss 0.4060 acc 0.928 lb 1.905 z 5.870 lr 2.00e-04 | 1451 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.890 max_load(range) 0.068-0.234 [save] runs/pretrain_flagship_h100_4x8/step2000 step 2010/40589 loss 0.3944 acc 0.931 lb 1.913 z 6.030 lr 2.00e-04 | 1449 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2020/40589 loss 0.4254 acc 0.924 lb 1.921 z 6.158 lr 2.00e-04 | 1449 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.922 max_load(range) 0.076-0.256 step 2030/40589 loss 0.4127 acc 0.927 lb 1.937 z 6.496 lr 2.00e-04 | 1449 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2040/40589 loss 0.4043 acc 0.928 lb 1.904 z 5.720 lr 2.00e-04 | 1449 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2050/40589 loss 0.3947 acc 0.930 lb 1.915 z 6.025 lr 2.00e-04 | 1449 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.928 max_load(range) 0.056-0.312 step 2060/40589 loss 0.4199 acc 0.925 lb 1.921 z 6.141 lr 2.00e-04 | 1449 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2070/40589 loss 0.4078 acc 0.928 lb 1.918 z 6.056 lr 2.00e-04 | 1448 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.843 max_load(range) 0.087-0.455 step 2080/40589 loss 0.4028 acc 0.928 lb 1.899 z 5.624 lr 2.00e-04 | 1448 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2090/40589 loss 0.3955 acc 0.930 lb 1.919 z 5.974 lr 2.00e-04 | 1448 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2100/40589 loss 0.4251 acc 0.924 lb 1.914 z 5.926 lr 2.00e-04 | 1448 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.940 max_load(range) 0.057-0.219 step 2110/40589 loss 0.4051 acc 0.928 lb 1.925 z 6.119 lr 2.00e-04 | 1448 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2120/40589 loss 0.3947 acc 0.929 lb 1.906 z 5.682 lr 2.00e-04 | 1448 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.898 max_load(range) 0.069-0.327 step 2130/40589 loss 0.3830 acc 0.932 lb 1.910 z 5.777 lr 2.00e-04 | 1448 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2140/40589 loss 0.4172 acc 0.925 lb 1.918 z 5.919 lr 2.00e-04 | 1448 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2150/40589 loss 0.3995 acc 0.929 lb 1.923 z 6.006 lr 2.00e-04 | 1447 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.859 max_load(range) 0.077-0.426 step 2160/40589 loss 0.3907 acc 0.930 lb 1.905 z 5.573 lr 2.00e-04 | 1447 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2170/40589 loss 0.3807 acc 0.933 lb 1.910 z 5.711 lr 2.00e-04 | 1447 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.890 max_load(range) 0.068-0.183 step 2180/40589 loss 0.4085 acc 0.927 lb 1.923 z 5.796 lr 2.00e-04 | 1447 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2190/40589 loss 0.4025 acc 0.929 lb 1.921 z 5.916 lr 2.00e-04 | 1447 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2200/40589 loss 0.3904 acc 0.930 lb 1.904 z 5.454 lr 2.00e-04 | 1447 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.887 max_load(range) 0.073-0.209 step 2210/40589 loss 0.3798 acc 0.933 lb 1.911 z 5.644 lr 2.00e-04 | 1447 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2220/40589 loss 0.4058 acc 0.927 lb 1.910 z 5.630 lr 2.00e-04 | 1447 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.926 max_load(range) 0.061-0.233 step 2230/40589 loss 0.3975 acc 0.929 lb 1.925 z 5.969 lr 2.00e-04 | 1446 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2240/40589 loss 0.3910 acc 0.930 lb 1.902 z 5.392 lr 2.00e-04 | 1446 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2250/40589 loss 0.3806 acc 0.932 lb 1.914 z 5.653 lr 2.00e-04 | 1446 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.926 max_load(range) 0.057-0.315 step 2260/40589 loss 0.4084 acc 0.927 lb 1.917 z 5.737 lr 2.00e-04 | 1446 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2270/40589 loss 0.3991 acc 0.929 lb 1.921 z 5.739 lr 2.00e-04 | 1446 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.842 max_load(range) 0.090-0.449 step 2280/40589 loss 0.3894 acc 0.931 lb 1.897 z 5.157 lr 2.00e-04 | 1446 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2290/40589 loss 0.3740 acc 0.933 lb 1.903 z 5.306 lr 2.00e-04 | 1446 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2300/40589 loss 0.4060 acc 0.928 lb 1.917 z 5.591 lr 2.00e-04 | 1446 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.935 max_load(range) 0.057-0.195 step 2310/40589 loss 0.3910 acc 0.931 lb 1.920 z 5.680 lr 2.00e-04 | 1446 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2320/40589 loss 0.3879 acc 0.931 lb 1.910 z 5.466 lr 2.00e-04 | 1446 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.912 max_load(range) 0.063-0.294 step 2330/40589 loss 0.3756 acc 0.934 lb 1.911 z 5.430 lr 2.00e-04 | 1446 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2340/40589 loss 0.3997 acc 0.928 lb 1.924 z 5.700 lr 2.00e-04 | 1445 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2350/40589 loss 0.3863 acc 0.932 lb 1.920 z 5.603 lr 2.00e-04 | 1445 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.857 max_load(range) 0.073-0.444 step 2360/40589 loss 0.3853 acc 0.931 lb 1.907 z 5.233 lr 2.00e-04 | 1445 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2370/40589 loss 0.3733 acc 0.934 lb 1.909 z 5.340 lr 2.00e-04 | 1445 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.885 max_load(range) 0.062-0.223 step 2380/40589 loss 0.4024 acc 0.928 lb 1.917 z 5.393 lr 2.00e-04 | 1445 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2390/40589 loss 0.3879 acc 0.931 lb 1.926 z 5.726 lr 2.00e-04 | 1445 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2400/40589 loss 0.3841 acc 0.932 lb 1.903 z 5.123 lr 2.00e-04 | 1445 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.896 max_load(range) 0.064-0.205 step 2410/40589 loss 0.3736 acc 0.934 lb 1.903 z 5.115 lr 2.00e-04 | 1445 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2420/40589 loss 0.3989 acc 0.929 lb 1.916 z 5.391 lr 2.00e-04 | 1445 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.921 max_load(range) 0.071-0.239 step 2430/40589 loss 0.3854 acc 0.931 lb 1.927 z 5.587 lr 2.00e-04 | 1445 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2440/40589 loss 0.3772 acc 0.933 lb 1.902 z 5.065 lr 2.00e-04 | 1445 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2450/40589 loss 0.3721 acc 0.934 lb 1.913 z 5.313 lr 2.00e-04 | 1445 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.931 max_load(range) 0.049-0.299 step 2460/40589 loss 0.3995 acc 0.928 lb 1.920 z 5.442 lr 2.00e-04 | 1444 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2470/40589 loss 0.3844 acc 0.932 lb 1.916 z 5.374 lr 2.00e-04 | 1444 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.840 max_load(range) 0.086-0.456 step 2480/40589 loss 0.3765 acc 0.932 lb 1.910 z 5.176 lr 1.99e-04 | 1444 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2490/40589 loss 0.3674 acc 0.934 lb 1.905 z 5.146 lr 1.99e-04 | 1444 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2500/40589 loss 0.3908 acc 0.929 lb 1.916 z 5.268 lr 1.99e-04 | 1444 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.940 max_load(range) 0.055-0.197 step 2510/40589 loss 0.3828 acc 0.932 lb 1.928 z 5.528 lr 1.99e-04 | 1444 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2520/40589 loss 0.3756 acc 0.932 lb 1.892 z 4.843 lr 1.99e-04 | 1444 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.887 max_load(range) 0.074-0.349 step 2530/40589 loss 0.3623 acc 0.935 lb 1.915 z 5.220 lr 1.99e-04 | 1444 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2540/40589 loss 0.3906 acc 0.929 lb 1.916 z 5.305 lr 1.99e-04 | 1444 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2550/40589 loss 0.3796 acc 0.932 lb 1.914 z 5.170 lr 1.99e-04 | 1444 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.861 max_load(range) 0.068-0.456 step 2560/40589 loss 0.3708 acc 0.933 lb 1.890 z 4.712 lr 1.99e-04 | 1444 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2570/40589 loss 0.3586 acc 0.936 lb 1.919 z 5.216 lr 1.99e-04 | 1444 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.896 max_load(range) 0.066-0.202 step 2580/40589 loss 0.3901 acc 0.929 lb 1.911 z 5.054 lr 1.99e-04 | 1443 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2590/40589 loss 0.3815 acc 0.932 lb 1.922 z 5.257 lr 1.99e-04 | 1443 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2600/40589 loss 0.3712 acc 0.933 lb 1.901 z 4.873 lr 1.99e-04 | 1443 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.894 max_load(range) 0.068-0.224 step 2610/40589 loss 0.3600 acc 0.936 lb 1.914 z 5.116 lr 1.99e-04 | 1443 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2620/40589 loss 0.3829 acc 0.931 lb 1.913 z 5.094 lr 1.99e-04 | 1443 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.923 max_load(range) 0.075-0.263 step 2630/40589 loss 0.3767 acc 0.933 lb 1.926 z 5.370 lr 1.99e-04 | 1443 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2640/40589 loss 0.3703 acc 0.934 lb 1.899 z 4.774 lr 1.99e-04 | 1443 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2650/40589 loss 0.3577 acc 0.936 lb 1.908 z 4.889 lr 1.99e-04 | 1443 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.927 max_load(range) 0.054-0.315 step 2660/40589 loss 0.3873 acc 0.930 lb 1.917 z 5.089 lr 1.99e-04 | 1443 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2670/40589 loss 0.3841 acc 0.932 lb 1.921 z 5.197 lr 1.99e-04 | 1443 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.870 max_load(range) 0.082-0.459 step 2680/40589 loss 0.3938 acc 0.932 lb 1.899 z 4.714 lr 1.99e-04 | 1443 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2690/40589 loss 0.3687 acc 0.935 lb 1.917 z 5.035 lr 1.99e-04 | 1443 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left step 2700/40589 loss 0.3928 acc 0.930 lb 1.927 z 5.188 lr 1.99e-04 | 1443 pairs/s | mem 63.1/79 GiB (min 62.2) | 5.0d left [routing] mean entropy_norm 0.933 max_load(range) 0.058-0.219 step 2710/40589 loss 0.3729 acc 0.934 lb 1.920 z 5.128 lr 1.99e-04 | 1443 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left step 2720/40589 loss 0.3692 acc 0.934 lb 1.897 z 4.662 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left [routing] mean entropy_norm 0.877 max_load(range) 0.076-0.359 step 2730/40589 loss 0.3556 acc 0.936 lb 1.912 z 4.872 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left step 2740/40589 loss 0.3875 acc 0.930 lb 1.910 z 4.792 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left step 2750/40589 loss 0.3733 acc 0.933 lb 1.925 z 5.161 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left [routing] mean entropy_norm 0.864 max_load(range) 0.069-0.464 step 2760/40589 loss 0.3653 acc 0.934 lb 1.905 z 4.723 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left step 2770/40589 loss 0.3514 acc 0.938 lb 1.913 z 4.810 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left [routing] mean entropy_norm 0.886 max_load(range) 0.077-0.262 step 2780/40589 loss 0.3804 acc 0.932 lb 1.920 z 4.954 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left step 2790/40589 loss 0.3667 acc 0.934 lb 1.927 z 5.128 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left step 2800/40589 loss 0.3637 acc 0.934 lb 1.899 z 4.652 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left [routing] mean entropy_norm 0.892 max_load(range) 0.065-0.209 step 2810/40589 loss 0.3538 acc 0.936 lb 1.907 z 4.694 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left step 2820/40589 loss 0.3785 acc 0.931 lb 1.921 z 4.902 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left [routing] mean entropy_norm 0.916 max_load(range) 0.053-0.244 step 2830/40589 loss 0.3716 acc 0.934 lb 1.916 z 4.952 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left step 2840/40589 loss 0.3673 acc 0.934 lb 1.904 z 4.578 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left step 2850/40589 loss 0.3506 acc 0.937 lb 1.922 z 4.859 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left [routing] mean entropy_norm 0.925 max_load(range) 0.052-0.325 step 2860/40589 loss 0.3756 acc 0.932 lb 1.921 z 4.886 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left step 2870/40589 loss 0.3701 acc 0.934 lb 1.927 z 5.021 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left [routing] mean entropy_norm 0.870 max_load(range) 0.077-0.441 step 2880/40589 loss 0.3600 acc 0.936 lb 1.900 z 4.436 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left step 2890/40589 loss 0.3472 acc 0.938 lb 1.913 z 4.680 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left step 2900/40589 loss 0.3791 acc 0.931 lb 1.919 z 4.757 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left [routing] mean entropy_norm 0.919 max_load(range) 0.053-0.212 step 2910/40589 loss 0.3666 acc 0.934 lb 1.924 z 4.872 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left step 2920/40589 loss 0.3565 acc 0.936 lb 1.911 z 4.495 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left [routing] mean entropy_norm 0.911 max_load(range) 0.060-0.278 step 2930/40589 loss 0.3471 acc 0.938 lb 1.909 z 4.554 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left step 2940/40589 loss 0.3703 acc 0.933 lb 1.919 z 4.732 lr 1.99e-04 | 1442 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left step 2950/40589 loss 0.3629 acc 0.935 lb 1.914 z 4.688 lr 1.99e-04 | 1441 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left [routing] mean entropy_norm 0.871 max_load(range) 0.071-0.454 step 2960/40589 loss 0.3564 acc 0.935 lb 1.903 z 4.367 lr 1.99e-04 | 1441 pairs/s | mem 63.1/79 GiB (min 62.3) | 5.0d left step 2970/40589 loss 0.3470 acc 0.938 lb 1.912 z 4.622 lr 1.99e-04 | 1441 pairs/s | mem 63.1/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.883 max_load(range) 0.072-0.201 step 2980/40589 loss 0.3709 acc 0.932 lb 1.917 z 4.617 lr 1.99e-04 | 1441 pairs/s | mem 63.1/79 GiB (min 62.3) | 4.9d left step 2990/40589 loss 0.3587 acc 0.936 lb 1.921 z 4.705 lr 1.99e-04 | 1441 pairs/s | mem 63.1/79 GiB (min 62.3) | 4.9d left step 3000/40589 loss 0.3558 acc 0.935 lb 1.910 z 4.397 lr 1.99e-04 | 1441 pairs/s | mem 63.1/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.881 max_load(range) 0.063-0.213 [save] runs/pretrain_flagship_h100_4x8/step3000 step 3010/40589 loss 0.3437 acc 0.938 lb 1.915 z 4.586 lr 1.99e-04 | 1440 pairs/s | mem 63.1/79 GiB (min 62.3) | 4.9d left step 3020/40589 loss 0.3705 acc 0.933 lb 1.925 z 4.673 lr 1.99e-04 | 1440 pairs/s | mem 63.1/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.919 max_load(range) 0.068-0.246 step 3030/40589 loss 0.3588 acc 0.935 lb 1.927 z 4.792 lr 1.99e-04 | 1440 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3040/40589 loss 0.3520 acc 0.937 lb 1.910 z 4.381 lr 1.99e-04 | 1440 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3050/40589 loss 0.3443 acc 0.938 lb 1.919 z 4.606 lr 1.99e-04 | 1440 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.918 max_load(range) 0.052-0.350 step 3060/40589 loss 0.3644 acc 0.934 lb 1.922 z 4.586 lr 1.99e-04 | 1440 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3070/40589 loss 0.3574 acc 0.936 lb 1.925 z 4.645 lr 1.99e-04 | 1440 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.863 max_load(range) 0.074-0.452 step 3080/40589 loss 0.3475 acc 0.938 lb 1.911 z 4.387 lr 1.99e-04 | 1440 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3090/40589 loss 0.3730 acc 0.935 lb 1.924 z 4.532 lr 1.99e-04 | 1440 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3100/40589 loss 0.3950 acc 0.932 lb 1.914 z 4.586 lr 1.99e-04 | 1440 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.926 max_load(range) 0.061-0.195 step 3110/40589 loss 0.3697 acc 0.935 lb 1.922 z 4.687 lr 1.99e-04 | 1440 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3120/40589 loss 0.3618 acc 0.935 lb 1.906 z 4.327 lr 1.99e-04 | 1440 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.901 max_load(range) 0.066-0.304 step 3130/40589 loss 0.3421 acc 0.940 lb 1.915 z 4.452 lr 1.99e-04 | 1440 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3140/40589 loss 0.3683 acc 0.934 lb 1.914 z 4.382 lr 1.99e-04 | 1440 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3150/40589 loss 0.3579 acc 0.936 lb 1.933 z 4.702 lr 1.99e-04 | 1440 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.868 max_load(range) 0.072-0.454 step 3160/40589 loss 0.3485 acc 0.937 lb 1.908 z 4.271 lr 1.99e-04 | 1440 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3170/40589 loss 0.3455 acc 0.938 lb 1.922 z 4.477 lr 1.99e-04 | 1440 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.881 max_load(range) 0.066-0.198 step 3180/40589 loss 0.3640 acc 0.934 lb 1.919 z 4.389 lr 1.99e-04 | 1440 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3190/40589 loss 0.3537 acc 0.936 lb 1.927 z 4.546 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3200/40589 loss 0.3466 acc 0.937 lb 1.917 z 4.290 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.889 max_load(range) 0.058-0.198 step 3210/40589 loss 0.3363 acc 0.939 lb 1.917 z 4.323 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3220/40589 loss 0.3641 acc 0.934 lb 1.922 z 4.426 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.913 max_load(range) 0.063-0.259 step 3230/40589 loss 0.3534 acc 0.936 lb 1.925 z 4.515 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3240/40589 loss 0.3391 acc 0.939 lb 1.913 z 4.333 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3250/40589 loss 0.3382 acc 0.939 lb 1.917 z 4.404 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.930 max_load(range) 0.056-0.330 step 3260/40589 loss 0.3596 acc 0.935 lb 1.914 z 4.270 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3270/40589 loss 0.3538 acc 0.936 lb 1.920 z 4.340 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.869 max_load(range) 0.067-0.453 step 3280/40589 loss 0.3434 acc 0.938 lb 1.913 z 4.204 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3290/40589 loss 0.3328 acc 0.940 lb 1.917 z 4.246 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3300/40589 loss 0.3621 acc 0.934 lb 1.924 z 4.366 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.928 max_load(range) 0.060-0.212 step 3310/40589 loss 0.3475 acc 0.938 lb 1.931 z 4.499 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3320/40589 loss 0.3426 acc 0.938 lb 1.906 z 4.079 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.913 max_load(range) 0.059-0.271 step 3330/40589 loss 0.3339 acc 0.940 lb 1.910 z 4.085 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3340/40589 loss 0.3594 acc 0.935 lb 1.923 z 4.201 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3350/40589 loss 0.3540 acc 0.937 lb 1.922 z 4.244 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.875 max_load(range) 0.066-0.448 step 3360/40589 loss 0.3441 acc 0.938 lb 1.904 z 3.908 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3370/40589 loss 0.3355 acc 0.940 lb 1.920 z 4.203 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.884 max_load(range) 0.076-0.216 step 3380/40589 loss 0.3580 acc 0.935 lb 1.916 z 4.148 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3390/40589 loss 0.3431 acc 0.939 lb 1.925 z 4.321 lr 1.99e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3400/40589 loss 0.3389 acc 0.938 lb 1.913 z 4.058 lr 1.98e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.880 max_load(range) 0.063-0.255 step 3410/40589 loss 0.3336 acc 0.940 lb 1.916 z 4.104 lr 1.98e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3420/40589 loss 0.3574 acc 0.936 lb 1.921 z 4.133 lr 1.98e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.909 max_load(range) 0.052-0.268 step 3430/40589 loss 0.3507 acc 0.936 lb 1.917 z 4.109 lr 1.98e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3440/40589 loss 0.3439 acc 0.937 lb 1.913 z 3.994 lr 1.98e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3450/40589 loss 0.3284 acc 0.940 lb 1.921 z 4.191 lr 1.98e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.924 max_load(range) 0.056-0.343 step 3460/40589 loss 0.3575 acc 0.935 lb 1.912 z 3.930 lr 1.98e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3470/40589 loss 0.3456 acc 0.938 lb 1.928 z 4.245 lr 1.98e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.869 max_load(range) 0.073-0.455 step 3480/40589 loss 0.3344 acc 0.939 lb 1.911 z 3.887 lr 1.98e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3490/40589 loss 0.3293 acc 0.940 lb 1.918 z 3.945 lr 1.98e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3500/40589 loss 0.3563 acc 0.936 lb 1.925 z 4.050 lr 1.98e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.925 max_load(range) 0.059-0.210 step 3510/40589 loss 0.3445 acc 0.939 lb 1.927 z 4.226 lr 1.98e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3520/40589 loss 0.3375 acc 0.939 lb 1.904 z 3.770 lr 1.98e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.918 max_load(range) 0.054-0.276 step 3530/40589 loss 0.3299 acc 0.941 lb 1.914 z 3.919 lr 1.98e-04 | 1439 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3540/40589 loss 0.3529 acc 0.936 lb 1.917 z 3.995 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3550/40589 loss 0.3426 acc 0.938 lb 1.926 z 4.144 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.878 max_load(range) 0.067-0.445 step 3560/40589 loss 0.3363 acc 0.939 lb 1.913 z 3.855 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3570/40589 loss 0.3261 acc 0.942 lb 1.919 z 3.973 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.884 max_load(range) 0.071-0.217 step 3580/40589 loss 0.3483 acc 0.937 lb 1.918 z 3.869 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3590/40589 loss 0.3411 acc 0.938 lb 1.918 z 3.921 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3600/40589 loss 0.3348 acc 0.939 lb 1.907 z 3.771 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.864 max_load(range) 0.081-0.251 step 3610/40589 loss 0.3248 acc 0.942 lb 1.921 z 4.013 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3620/40589 loss 0.3539 acc 0.936 lb 1.924 z 3.892 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.903 max_load(range) 0.070-0.237 step 3630/40589 loss 0.3423 acc 0.939 lb 1.928 z 4.067 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3640/40589 loss 0.3346 acc 0.939 lb 1.920 z 3.839 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3650/40589 loss 0.3233 acc 0.942 lb 1.921 z 3.929 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.924 max_load(range) 0.050-0.335 step 3660/40589 loss 0.3506 acc 0.936 lb 1.928 z 3.945 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3670/40589 loss 0.3422 acc 0.938 lb 1.927 z 3.953 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.872 max_load(range) 0.073-0.457 step 3680/40589 loss 0.3308 acc 0.940 lb 1.911 z 3.740 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3690/40589 loss 0.3289 acc 0.940 lb 1.928 z 3.874 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3700/40589 loss 0.3493 acc 0.936 lb 1.928 z 3.883 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.923 max_load(range) 0.060-0.216 step 3710/40589 loss 0.3405 acc 0.938 lb 1.933 z 4.031 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3720/40589 loss 0.3315 acc 0.940 lb 1.918 z 3.724 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.908 max_load(range) 0.061-0.299 step 3730/40589 loss 0.3238 acc 0.942 lb 1.919 z 3.777 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3740/40589 loss 0.3460 acc 0.937 lb 1.928 z 3.936 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3750/40589 loss 0.3394 acc 0.939 lb 1.928 z 3.913 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.877 max_load(range) 0.065-0.450 step 3760/40589 loss 0.3303 acc 0.940 lb 1.912 z 3.738 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3770/40589 loss 0.3240 acc 0.942 lb 1.926 z 3.915 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.861 max_load(range) 0.081-0.240 step 3780/40589 loss 0.3420 acc 0.937 lb 1.927 z 3.718 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3790/40589 loss 0.3394 acc 0.939 lb 1.934 z 3.976 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left step 3800/40589 loss 0.3298 acc 0.940 lb 1.910 z 3.517 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.9d left [routing] mean entropy_norm 0.871 max_load(range) 0.070-0.231 step 3810/40589 loss 0.3225 acc 0.942 lb 1.927 z 3.860 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 3820/40589 loss 0.3462 acc 0.937 lb 1.925 z 3.777 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.911 max_load(range) 0.063-0.230 step 3830/40589 loss 0.3346 acc 0.940 lb 1.931 z 3.822 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 3840/40589 loss 0.3302 acc 0.940 lb 1.928 z 3.759 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 3850/40589 loss 0.3206 acc 0.942 lb 1.917 z 3.542 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.929 max_load(range) 0.054-0.330 step 3860/40589 loss 0.3450 acc 0.937 lb 1.931 z 3.756 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 3870/40589 loss 0.3381 acc 0.939 lb 1.925 z 3.721 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.874 max_load(range) 0.070-0.458 step 3880/40589 loss 0.3254 acc 0.941 lb 1.916 z 3.578 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 3890/40589 loss 0.3178 acc 0.943 lb 1.921 z 3.627 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 3900/40589 loss 0.3452 acc 0.937 lb 1.929 z 3.763 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.925 max_load(range) 0.060-0.217 step 3910/40589 loss 0.3309 acc 0.940 lb 1.937 z 3.848 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 3920/40589 loss 0.3241 acc 0.941 lb 1.914 z 3.519 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.901 max_load(range) 0.061-0.316 step 3930/40589 loss 0.3142 acc 0.943 lb 1.918 z 3.589 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 3940/40589 loss 0.3399 acc 0.938 lb 1.924 z 3.700 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 3950/40589 loss 0.3325 acc 0.940 lb 1.923 z 3.622 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.890 max_load(range) 0.068-0.444 step 3960/40589 loss 0.3259 acc 0.941 lb 1.900 z 3.238 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 3970/40589 loss 0.3193 acc 0.942 lb 1.915 z 3.534 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.876 max_load(range) 0.071-0.194 step 3980/40589 loss 0.3401 acc 0.938 lb 1.930 z 3.680 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 3990/40589 loss 0.3318 acc 0.941 lb 1.929 z 3.725 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4000/40589 loss 0.3247 acc 0.940 lb 1.915 z 3.441 lr 1.98e-04 | 1438 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.880 max_load(range) 0.065-0.216 [save] runs/pretrain_flagship_h100_4x8/step4000 step 4010/40589 loss 0.3158 acc 0.943 lb 1.923 z 3.587 lr 1.98e-04 | 1437 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4020/40589 loss 0.3365 acc 0.939 lb 1.924 z 3.578 lr 1.98e-04 | 1437 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.904 max_load(range) 0.069-0.236 step 4030/40589 loss 0.3282 acc 0.941 lb 1.925 z 3.610 lr 1.97e-04 | 1437 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4040/40589 loss 0.3236 acc 0.941 lb 1.917 z 3.471 lr 1.97e-04 | 1437 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4050/40589 loss 0.3181 acc 0.943 lb 1.926 z 3.564 lr 1.97e-04 | 1437 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.916 max_load(range) 0.057-0.366 step 4060/40589 loss 0.3379 acc 0.939 lb 1.925 z 3.516 lr 1.97e-04 | 1437 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4070/40589 loss 0.3290 acc 0.941 lb 1.928 z 3.537 lr 1.97e-04 | 1437 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.881 max_load(range) 0.066-0.465 step 4080/40589 loss 0.3274 acc 0.940 lb 1.908 z 3.303 lr 1.97e-04 | 1437 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4090/40589 loss 0.3138 acc 0.943 lb 1.921 z 3.447 lr 1.97e-04 | 1437 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4100/40589 loss 0.3371 acc 0.938 lb 1.926 z 3.578 lr 1.97e-04 | 1437 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.922 max_load(range) 0.063-0.210 step 4110/40589 loss 0.3296 acc 0.940 lb 1.927 z 3.628 lr 1.97e-04 | 1437 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4120/40589 loss 0.3198 acc 0.942 lb 1.899 z 3.173 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.900 max_load(range) 0.069-0.326 step 4130/40589 loss 0.3134 acc 0.943 lb 1.921 z 3.503 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4140/40589 loss 0.3380 acc 0.938 lb 1.923 z 3.437 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4150/40589 loss 0.3281 acc 0.941 lb 1.931 z 3.632 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.883 max_load(range) 0.068-0.453 step 4160/40589 loss 0.3233 acc 0.941 lb 1.911 z 3.313 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4170/40589 loss 0.3107 acc 0.943 lb 1.920 z 3.468 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.872 max_load(range) 0.077-0.241 step 4180/40589 loss 0.3929 acc 0.932 lb 1.925 z 3.401 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4190/40589 loss 0.3586 acc 0.937 lb 1.936 z 3.692 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4200/40589 loss 0.3312 acc 0.941 lb 1.908 z 3.476 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.871 max_load(range) 0.074-0.215 step 4210/40589 loss 0.3231 acc 0.942 lb 1.924 z 3.574 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4220/40589 loss 0.3391 acc 0.939 lb 1.924 z 3.406 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.902 max_load(range) 0.061-0.233 step 4230/40589 loss 0.3331 acc 0.940 lb 1.934 z 3.508 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4240/40589 loss 0.3197 acc 0.942 lb 1.921 z 3.456 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4250/40589 loss 0.3139 acc 0.943 lb 1.926 z 3.528 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.918 max_load(range) 0.056-0.338 step 4260/40589 loss 0.3345 acc 0.939 lb 1.930 z 3.485 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4270/40589 loss 0.3265 acc 0.941 lb 1.930 z 3.400 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.883 max_load(range) 0.067-0.455 step 4280/40589 loss 0.3181 acc 0.942 lb 1.915 z 3.297 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4290/40589 loss 0.3120 acc 0.944 lb 1.924 z 3.337 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4300/40589 loss 0.3335 acc 0.939 lb 1.931 z 3.347 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.911 max_load(range) 0.063-0.201 step 4310/40589 loss 0.3267 acc 0.941 lb 1.932 z 3.429 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4320/40589 loss 0.3148 acc 0.943 lb 1.924 z 3.271 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.890 max_load(range) 0.069-0.343 step 4330/40589 loss 0.3114 acc 0.943 lb 1.925 z 3.319 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4340/40589 loss 0.3298 acc 0.940 lb 1.926 z 3.212 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4350/40589 loss 0.3244 acc 0.942 lb 1.938 z 3.451 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.891 max_load(range) 0.063-0.448 step 4360/40589 loss 0.3178 acc 0.942 lb 1.928 z 3.205 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4370/40589 loss 0.3106 acc 0.944 lb 1.935 z 3.347 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.885 max_load(range) 0.074-0.233 step 4380/40589 loss 0.3281 acc 0.940 lb 1.924 z 3.316 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4390/40589 loss 0.3225 acc 0.941 lb 1.934 z 3.321 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4400/40589 loss 0.3176 acc 0.942 lb 1.921 z 3.220 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.872 max_load(range) 0.068-0.218 step 4410/40589 loss 0.3097 acc 0.944 lb 1.931 z 3.273 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4420/40589 loss 0.3357 acc 0.938 lb 1.930 z 3.193 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.907 max_load(range) 0.062-0.235 step 4430/40589 loss 0.3239 acc 0.942 lb 1.932 z 3.200 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4440/40589 loss 0.3146 acc 0.943 lb 1.923 z 3.107 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4450/40589 loss 0.3093 acc 0.944 lb 1.929 z 3.239 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left [routing] mean entropy_norm 0.927 max_load(range) 0.056-0.328 step 4460/40589 loss 0.3297 acc 0.940 lb 1.933 z 3.194 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.3) | 4.8d left step 4470/40589 loss 0.3194 acc 0.942 lb 1.935 z 3.239 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.8d left [routing] mean entropy_norm 0.880 max_load(range) 0.065-0.460 step 4480/40589 loss 0.3136 acc 0.942 lb 1.921 z 3.036 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.8d left step 4490/40589 loss 0.3057 acc 0.945 lb 1.922 z 3.173 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.8d left step 4500/40589 loss 0.3286 acc 0.940 lb 1.937 z 3.255 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.8d left [routing] mean entropy_norm 0.913 max_load(range) 0.067-0.221 step 4510/40589 loss 0.3242 acc 0.941 lb 1.933 z 3.315 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.8d left step 4520/40589 loss 0.3138 acc 0.942 lb 1.920 z 3.059 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.8d left [routing] mean entropy_norm 0.909 max_load(range) 0.057-0.307 step 4530/40589 loss 0.3079 acc 0.944 lb 1.931 z 3.231 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.8d left step 4540/40589 loss 0.3278 acc 0.940 lb 1.935 z 3.196 lr 1.97e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.8d left step 4550/40589 loss 0.3172 acc 0.943 lb 1.934 z 3.178 lr 1.96e-04 | 1436 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.8d left [routing] mean entropy_norm 0.887 max_load(range) 0.067-0.448 step 4560/40589 loss 0.3094 acc 0.943 lb 1.923 z 3.129 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.8d left step 4570/40589 loss 0.3034 acc 0.945 lb 1.932 z 3.240 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.8d left [routing] mean entropy_norm 0.877 max_load(range) 0.076-0.183 step 4580/40589 loss 0.3273 acc 0.940 lb 1.933 z 3.150 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.8d left step 4590/40589 loss 0.3163 acc 0.943 lb 1.931 z 3.168 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.8d left step 4600/40589 loss 0.3147 acc 0.942 lb 1.923 z 3.102 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.8d left [routing] mean entropy_norm 0.881 max_load(range) 0.073-0.206 step 4610/40589 loss 0.3022 acc 0.945 lb 1.926 z 3.112 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.8d left step 4620/40589 loss 0.3265 acc 0.940 lb 1.932 z 3.042 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.8d left [routing] mean entropy_norm 0.907 max_load(range) 0.069-0.256 step 4630/40589 loss 0.3203 acc 0.942 lb 1.936 z 3.109 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.8d left step 4640/40589 loss 0.3127 acc 0.943 lb 1.924 z 3.087 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4650/40589 loss 0.3049 acc 0.944 lb 1.930 z 3.040 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.932 max_load(range) 0.055-0.335 step 4660/40589 loss 0.3299 acc 0.940 lb 1.927 z 2.964 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4670/40589 loss 0.3155 acc 0.943 lb 1.939 z 3.148 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.882 max_load(range) 0.066-0.450 step 4680/40589 loss 0.3116 acc 0.943 lb 1.912 z 2.763 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4690/40589 loss 0.3016 acc 0.945 lb 1.931 z 3.171 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4700/40589 loss 0.3214 acc 0.941 lb 1.929 z 3.026 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.918 max_load(range) 0.067-0.216 step 4710/40589 loss 0.3173 acc 0.942 lb 1.934 z 3.028 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4720/40589 loss 0.3117 acc 0.944 lb 1.921 z 2.902 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.899 max_load(range) 0.063-0.306 step 4730/40589 loss 0.3016 acc 0.945 lb 1.929 z 2.934 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4740/40589 loss 0.3248 acc 0.941 lb 1.930 z 2.865 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4750/40589 loss 0.3200 acc 0.941 lb 1.939 z 3.086 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.891 max_load(range) 0.064-0.442 step 4760/40589 loss 0.3150 acc 0.942 lb 1.926 z 2.895 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4770/40589 loss 0.3052 acc 0.945 lb 1.929 z 2.972 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.872 max_load(range) 0.082-0.220 step 4780/40589 loss 0.3240 acc 0.940 lb 1.934 z 2.915 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4790/40589 loss 0.3168 acc 0.942 lb 1.932 z 2.901 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4800/40589 loss 0.3064 acc 0.944 lb 1.933 z 2.951 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.862 max_load(range) 0.089-0.206 step 4810/40589 loss 0.3014 acc 0.945 lb 1.923 z 2.938 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4820/40589 loss 0.3214 acc 0.942 lb 1.936 z 2.908 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.899 max_load(range) 0.067-0.258 step 4830/40589 loss 0.3188 acc 0.942 lb 1.935 z 2.941 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4840/40589 loss 0.3079 acc 0.943 lb 1.927 z 2.866 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4850/40589 loss 0.2969 acc 0.946 lb 1.927 z 2.928 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.917 max_load(range) 0.060-0.382 step 4860/40589 loss 0.3242 acc 0.941 lb 1.930 z 2.860 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4870/40589 loss 0.3110 acc 0.944 lb 1.938 z 3.045 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.889 max_load(range) 0.066-0.454 step 4880/40589 loss 0.3090 acc 0.944 lb 1.934 z 3.035 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4890/40589 loss 0.2990 acc 0.945 lb 1.929 z 2.876 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4900/40589 loss 0.3225 acc 0.941 lb 1.931 z 2.877 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.918 max_load(range) 0.061-0.214 step 4910/40589 loss 0.3147 acc 0.943 lb 1.931 z 2.939 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4920/40589 loss 0.3037 acc 0.944 lb 1.920 z 2.816 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.899 max_load(range) 0.067-0.316 step 4930/40589 loss 0.3003 acc 0.946 lb 1.934 z 2.937 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4940/40589 loss 0.3203 acc 0.942 lb 1.932 z 2.899 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4950/40589 loss 0.3112 acc 0.943 lb 1.940 z 2.928 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.895 max_load(range) 0.062-0.447 step 4960/40589 loss 0.3051 acc 0.944 lb 1.927 z 2.779 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4970/40589 loss 0.2970 acc 0.946 lb 1.930 z 2.852 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.868 max_load(range) 0.076-0.208 step 4980/40589 loss 0.3200 acc 0.942 lb 1.931 z 2.751 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 4990/40589 loss 0.3161 acc 0.943 lb 1.935 z 2.898 lr 1.96e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5000/40589 loss 0.3052 acc 0.943 lb 1.927 z 2.818 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.879 max_load(range) 0.067-0.241 [save] runs/pretrain_flagship_h100_4x8/step5000 step 5010/40589 loss 0.2976 acc 0.946 lb 1.918 z 2.787 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5020/40589 loss 0.3204 acc 0.941 lb 1.935 z 2.803 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.913 max_load(range) 0.055-0.247 step 5030/40589 loss 0.3162 acc 0.943 lb 1.936 z 2.875 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5040/40589 loss 0.3058 acc 0.943 lb 1.924 z 2.786 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5050/40589 loss 0.2936 acc 0.947 lb 1.927 z 2.793 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.931 max_load(range) 0.051-0.331 step 5060/40589 loss 0.3196 acc 0.941 lb 1.935 z 2.797 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5070/40589 loss 0.3147 acc 0.943 lb 1.933 z 2.774 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.887 max_load(range) 0.065-0.455 step 5080/40589 loss 0.3044 acc 0.944 lb 1.928 z 2.770 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5090/40589 loss 0.2955 acc 0.946 lb 1.932 z 2.792 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5100/40589 loss 0.3199 acc 0.941 lb 1.935 z 2.735 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.905 max_load(range) 0.066-0.194 step 5110/40589 loss 0.3118 acc 0.943 lb 1.936 z 2.687 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5120/40589 loss 0.3068 acc 0.943 lb 1.923 z 2.636 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.903 max_load(range) 0.058-0.277 step 5130/40589 loss 0.2966 acc 0.946 lb 1.928 z 2.719 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5140/40589 loss 0.3168 acc 0.941 lb 1.934 z 2.731 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5150/40589 loss 0.3100 acc 0.943 lb 1.939 z 2.800 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.904 max_load(range) 0.059-0.434 step 5160/40589 loss 0.3020 acc 0.944 lb 1.922 z 2.655 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5170/40589 loss 0.2963 acc 0.946 lb 1.932 z 2.694 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.876 max_load(range) 0.073-0.181 step 5180/40589 loss 0.3181 acc 0.941 lb 1.931 z 2.655 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5190/40589 loss 0.3127 acc 0.943 lb 1.933 z 2.739 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5200/40589 loss 0.3002 acc 0.945 lb 1.932 z 2.728 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.870 max_load(range) 0.073-0.203 step 5210/40589 loss 0.2954 acc 0.946 lb 1.933 z 2.688 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5220/40589 loss 0.3150 acc 0.942 lb 1.937 z 2.630 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.903 max_load(range) 0.062-0.215 step 5230/40589 loss 0.3093 acc 0.943 lb 1.933 z 2.653 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5240/40589 loss 0.3048 acc 0.944 lb 1.919 z 2.554 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5250/40589 loss 0.2928 acc 0.947 lb 1.924 z 2.681 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.934 max_load(range) 0.055-0.326 step 5260/40589 loss 0.3161 acc 0.942 lb 1.933 z 2.658 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5270/40589 loss 0.3058 acc 0.945 lb 1.931 z 2.653 lr 1.95e-04 | 1435 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.897 max_load(range) 0.063-0.451 step 5280/40589 loss 0.3031 acc 0.945 lb 1.917 z 2.573 lr 1.95e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5290/40589 loss 0.2939 acc 0.946 lb 1.927 z 2.647 lr 1.95e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5300/40589 loss 0.3108 acc 0.942 lb 1.934 z 2.619 lr 1.95e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.921 max_load(range) 0.057-0.200 step 5310/40589 loss 0.3082 acc 0.944 lb 1.931 z 2.647 lr 1.95e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5320/40589 loss 0.3004 acc 0.945 lb 1.919 z 2.579 lr 1.95e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.911 max_load(range) 0.061-0.295 step 5330/40589 loss 0.2902 acc 0.947 lb 1.932 z 2.678 lr 1.95e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5340/40589 loss 0.3166 acc 0.942 lb 1.927 z 2.526 lr 1.95e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5350/40589 loss 0.3094 acc 0.943 lb 1.931 z 2.608 lr 1.95e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.901 max_load(range) 0.064-0.446 step 5360/40589 loss 0.2997 acc 0.945 lb 1.920 z 2.508 lr 1.95e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5370/40589 loss 0.2922 acc 0.947 lb 1.927 z 2.582 lr 1.95e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.861 max_load(range) 0.079-0.222 step 5380/40589 loss 0.3123 acc 0.943 lb 1.936 z 2.546 lr 1.95e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5390/40589 loss 0.3054 acc 0.944 lb 1.937 z 2.562 lr 1.95e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5400/40589 loss 0.3010 acc 0.944 lb 1.929 z 2.538 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left [routing] mean entropy_norm 0.871 max_load(range) 0.075-0.228 step 5410/40589 loss 0.2920 acc 0.947 lb 1.932 z 2.671 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.7d left step 5420/40589 loss 0.3116 acc 0.943 lb 1.925 z 2.507 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.906 max_load(range) 0.059-0.226 step 5430/40589 loss 0.3092 acc 0.943 lb 1.935 z 2.508 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5440/40589 loss 0.2988 acc 0.945 lb 1.932 z 2.550 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5450/40589 loss 0.2903 acc 0.947 lb 1.931 z 2.520 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.933 max_load(range) 0.055-0.341 step 5460/40589 loss 0.3112 acc 0.942 lb 1.929 z 2.476 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5470/40589 loss 0.3085 acc 0.944 lb 1.939 z 2.597 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.896 max_load(range) 0.066-0.443 step 5480/40589 loss 0.2976 acc 0.945 lb 1.922 z 2.405 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5490/40589 loss 0.2910 acc 0.946 lb 1.930 z 2.543 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5500/40589 loss 0.3133 acc 0.942 lb 1.937 z 2.506 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.913 max_load(range) 0.062-0.208 step 5510/40589 loss 0.3072 acc 0.945 lb 1.940 z 2.504 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5520/40589 loss 0.2994 acc 0.945 lb 1.924 z 2.424 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.893 max_load(range) 0.074-0.352 step 5530/40589 loss 0.2876 acc 0.947 lb 1.937 z 2.485 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5540/40589 loss 0.3146 acc 0.942 lb 1.932 z 2.394 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5550/40589 loss 0.3082 acc 0.944 lb 1.941 z 2.578 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.902 max_load(range) 0.060-0.435 step 5560/40589 loss 0.2942 acc 0.946 lb 1.925 z 2.419 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5570/40589 loss 0.2893 acc 0.947 lb 1.934 z 2.438 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.867 max_load(range) 0.076-0.177 step 5580/40589 loss 0.3126 acc 0.942 lb 1.932 z 2.392 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5590/40589 loss 0.3065 acc 0.944 lb 1.940 z 2.510 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5600/40589 loss 0.2979 acc 0.945 lb 1.924 z 2.392 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.875 max_load(range) 0.061-0.230 step 5610/40589 loss 0.2904 acc 0.947 lb 1.933 z 2.489 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5620/40589 loss 0.3132 acc 0.942 lb 1.931 z 2.345 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.913 max_load(range) 0.056-0.257 step 5630/40589 loss 0.3010 acc 0.945 lb 1.939 z 2.459 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5640/40589 loss 0.2981 acc 0.946 lb 1.930 z 2.541 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5650/40589 loss 0.2873 acc 0.947 lb 1.934 z 2.429 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.934 max_load(range) 0.054-0.340 step 5660/40589 loss 0.3104 acc 0.943 lb 1.932 z 2.442 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5670/40589 loss 0.2961 acc 0.946 lb 1.941 z 2.506 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.890 max_load(range) 0.060-0.472 step 5680/40589 loss 0.2965 acc 0.946 lb 1.925 z 2.314 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5690/40589 loss 0.2883 acc 0.947 lb 1.931 z 2.395 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5700/40589 loss 0.3095 acc 0.943 lb 1.931 z 2.288 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.911 max_load(range) 0.059-0.210 step 5710/40589 loss 0.2992 acc 0.945 lb 1.943 z 2.426 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5720/40589 loss 0.2942 acc 0.946 lb 1.926 z 2.361 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.899 max_load(range) 0.065-0.322 step 5730/40589 loss 0.2841 acc 0.948 lb 1.931 z 2.394 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5740/40589 loss 0.3067 acc 0.944 lb 1.935 z 2.310 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5750/40589 loss 0.3023 acc 0.946 lb 1.933 z 2.386 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.895 max_load(range) 0.060-0.450 step 5760/40589 loss 0.2928 acc 0.946 lb 1.921 z 2.363 lr 1.94e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5770/40589 loss 0.2833 acc 0.949 lb 1.935 z 2.318 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.881 max_load(range) 0.068-0.205 step 5780/40589 loss 0.3062 acc 0.944 lb 1.932 z 2.282 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5790/40589 loss 0.3016 acc 0.945 lb 1.939 z 2.342 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5800/40589 loss 0.2980 acc 0.946 lb 1.918 z 2.295 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.865 max_load(range) 0.076-0.220 step 5810/40589 loss 0.2882 acc 0.947 lb 1.929 z 2.337 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5820/40589 loss 0.3071 acc 0.944 lb 1.934 z 2.275 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.908 max_load(range) 0.058-0.232 step 5830/40589 loss 0.3037 acc 0.944 lb 1.938 z 2.356 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5840/40589 loss 0.2969 acc 0.945 lb 1.929 z 2.330 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5850/40589 loss 0.2869 acc 0.948 lb 1.940 z 2.368 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.924 max_load(range) 0.053-0.349 step 5860/40589 loss 0.3089 acc 0.943 lb 1.934 z 2.173 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5870/40589 loss 0.3026 acc 0.945 lb 1.935 z 2.266 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.899 max_load(range) 0.057-0.451 step 5880/40589 loss 0.2956 acc 0.945 lb 1.925 z 2.336 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5890/40589 loss 0.2887 acc 0.947 lb 1.931 z 2.305 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5900/40589 loss 0.3056 acc 0.944 lb 1.924 z 2.192 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.914 max_load(range) 0.070-0.206 step 5910/40589 loss 0.3027 acc 0.945 lb 1.938 z 2.298 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5920/40589 loss 0.2941 acc 0.946 lb 1.931 z 2.340 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.896 max_load(range) 0.066-0.327 step 5930/40589 loss 0.2832 acc 0.948 lb 1.938 z 2.343 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5940/40589 loss 0.3027 acc 0.944 lb 1.935 z 2.170 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5950/40589 loss 0.2994 acc 0.946 lb 1.942 z 2.220 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.904 max_load(range) 0.059-0.441 step 5960/40589 loss 0.2895 acc 0.947 lb 1.936 z 2.295 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5970/40589 loss 0.2828 acc 0.948 lb 1.933 z 2.227 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.857 max_load(range) 0.098-0.184 step 5980/40589 loss 0.3056 acc 0.944 lb 1.940 z 2.244 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 5990/40589 loss 0.2979 acc 0.946 lb 1.937 z 2.218 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 6000/40589 loss 0.2906 acc 0.947 lb 1.933 z 2.294 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.878 max_load(range) 0.077-0.197 [save] runs/pretrain_flagship_h100_4x8/step6000 step 6010/40589 loss 0.2849 acc 0.948 lb 1.937 z 2.259 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 6020/40589 loss 0.3054 acc 0.943 lb 1.937 z 2.175 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.897 max_load(range) 0.059-0.240 step 6030/40589 loss 0.3011 acc 0.945 lb 1.944 z 2.280 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 6040/40589 loss 0.2871 acc 0.947 lb 1.936 z 2.324 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 6050/40589 loss 0.2853 acc 0.947 lb 1.933 z 2.240 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.928 max_load(range) 0.060-0.357 step 6060/40589 loss 0.3075 acc 0.944 lb 1.943 z 2.172 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 6070/40589 loss 0.2975 acc 0.946 lb 1.943 z 2.219 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.902 max_load(range) 0.063-0.448 step 6080/40589 loss 0.2865 acc 0.948 lb 1.932 z 2.259 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 6090/40589 loss 0.2852 acc 0.947 lb 1.936 z 2.144 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 6100/40589 loss 0.3017 acc 0.944 lb 1.942 z 2.202 lr 1.93e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.911 max_load(range) 0.071-0.219 step 6110/40589 loss 0.2968 acc 0.946 lb 1.940 z 2.226 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 6120/40589 loss 0.2871 acc 0.947 lb 1.930 z 2.235 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.905 max_load(range) 0.066-0.315 step 6130/40589 loss 0.2822 acc 0.949 lb 1.940 z 2.235 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 6140/40589 loss 0.3053 acc 0.943 lb 1.935 z 2.066 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 6150/40589 loss 0.2949 acc 0.946 lb 1.934 z 2.126 lr 1.92e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.908 max_load(range) 0.058-0.437 step 6160/40589 loss 0.2903 acc 0.946 lb 1.928 z 2.251 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 6170/40589 loss 0.2830 acc 0.948 lb 1.927 z 2.098 lr 1.92e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left [routing] mean entropy_norm 0.863 max_load(range) 0.081-0.218 step 6180/40589 loss 0.3066 acc 0.944 lb 1.945 z 2.175 lr 1.92e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 6190/40589 loss 0.2971 acc 0.946 lb 1.941 z 2.188 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.6d left step 6200/40589 loss 0.2858 acc 0.947 lb 1.931 z 2.192 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.874 max_load(range) 0.074-0.222 step 6210/40589 loss 0.2778 acc 0.949 lb 1.927 z 2.101 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6220/40589 loss 0.3013 acc 0.944 lb 1.940 z 2.086 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.898 max_load(range) 0.063-0.250 step 6230/40589 loss 0.2938 acc 0.946 lb 1.939 z 2.053 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6240/40589 loss 0.2906 acc 0.946 lb 1.931 z 2.169 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6250/40589 loss 0.2794 acc 0.949 lb 1.940 z 2.153 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.922 max_load(range) 0.049-0.345 step 6260/40589 loss 0.3036 acc 0.944 lb 1.943 z 2.107 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6270/40589 loss 0.2961 acc 0.946 lb 1.945 z 2.056 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.910 max_load(range) 0.059-0.439 step 6280/40589 loss 0.2904 acc 0.946 lb 1.934 z 2.136 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6290/40589 loss 0.2809 acc 0.949 lb 1.937 z 2.153 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6300/40589 loss 0.3041 acc 0.943 lb 1.936 z 2.046 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.921 max_load(range) 0.069-0.213 step 6310/40589 loss 0.2953 acc 0.946 lb 1.936 z 2.033 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6320/40589 loss 0.2872 acc 0.947 lb 1.927 z 2.076 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.901 max_load(range) 0.065-0.311 step 6330/40589 loss 0.2779 acc 0.949 lb 1.935 z 2.149 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6340/40589 loss 0.3022 acc 0.944 lb 1.942 z 2.047 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6350/40589 loss 0.2953 acc 0.947 lb 1.935 z 2.084 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.903 max_load(range) 0.057-0.453 step 6360/40589 loss 0.2867 acc 0.947 lb 1.919 z 2.072 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6370/40589 loss 0.2756 acc 0.949 lb 1.933 z 2.106 lr 1.92e-04 | 1434 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.869 max_load(range) 0.097-0.189 step 6380/40589 loss 0.3032 acc 0.944 lb 1.928 z 1.980 lr 1.92e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6390/40589 loss 0.2932 acc 0.947 lb 1.938 z 2.048 lr 1.92e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6400/40589 loss 0.2850 acc 0.948 lb 1.929 z 2.061 lr 1.92e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.865 max_load(range) 0.092-0.215 step 6410/40589 loss 0.2788 acc 0.948 lb 1.939 z 2.076 lr 1.92e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6420/40589 loss 0.3013 acc 0.944 lb 1.938 z 1.989 lr 1.92e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.911 max_load(range) 0.063-0.224 step 6430/40589 loss 0.2953 acc 0.945 lb 1.944 z 2.041 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6440/40589 loss 0.2866 acc 0.947 lb 1.934 z 2.167 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6450/40589 loss 0.2780 acc 0.949 lb 1.933 z 2.041 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.926 max_load(range) 0.054-0.360 step 6460/40589 loss 0.3009 acc 0.944 lb 1.934 z 1.999 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6470/40589 loss 0.2911 acc 0.946 lb 1.936 z 2.034 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.902 max_load(range) 0.063-0.446 step 6480/40589 loss 0.2901 acc 0.946 lb 1.924 z 1.999 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6490/40589 loss 0.2814 acc 0.949 lb 1.938 z 2.117 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6500/40589 loss 0.3003 acc 0.945 lb 1.939 z 2.010 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.921 max_load(range) 0.060-0.197 step 6510/40589 loss 0.2920 acc 0.947 lb 1.935 z 1.969 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6520/40589 loss 0.2885 acc 0.947 lb 1.932 z 2.026 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.914 max_load(range) 0.066-0.279 step 6530/40589 loss 0.2765 acc 0.949 lb 1.930 z 1.995 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6540/40589 loss 0.3028 acc 0.944 lb 1.939 z 1.978 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6550/40589 loss 0.2910 acc 0.947 lb 1.944 z 1.996 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.907 max_load(range) 0.061-0.449 step 6560/40589 loss 0.2869 acc 0.947 lb 1.927 z 2.025 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6570/40589 loss 0.2786 acc 0.949 lb 1.933 z 1.983 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.871 max_load(range) 0.079-0.215 step 6580/40589 loss 0.2988 acc 0.944 lb 1.927 z 1.844 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6590/40589 loss 0.2933 acc 0.947 lb 1.938 z 1.984 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6600/40589 loss 0.2874 acc 0.947 lb 1.924 z 2.044 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.885 max_load(range) 0.076-0.193 step 6610/40589 loss 0.2763 acc 0.948 lb 1.932 z 1.971 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6620/40589 loss 0.2991 acc 0.945 lb 1.929 z 1.865 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.917 max_load(range) 0.065-0.226 step 6630/40589 loss 0.2919 acc 0.946 lb 1.932 z 1.891 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6640/40589 loss 0.2853 acc 0.947 lb 1.925 z 1.932 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6650/40589 loss 0.2801 acc 0.948 lb 1.940 z 1.953 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.934 max_load(range) 0.054-0.359 step 6660/40589 loss 0.2977 acc 0.944 lb 1.942 z 1.974 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6670/40589 loss 0.2895 acc 0.946 lb 1.944 z 1.922 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.910 max_load(range) 0.057-0.433 step 6680/40589 loss 0.2838 acc 0.947 lb 1.932 z 1.981 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6690/40589 loss 0.2765 acc 0.949 lb 1.937 z 1.960 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6700/40589 loss 0.3015 acc 0.945 lb 1.932 z 1.864 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.926 max_load(range) 0.059-0.219 step 6710/40589 loss 0.2931 acc 0.946 lb 1.936 z 1.890 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6720/40589 loss 0.2834 acc 0.947 lb 1.928 z 1.975 lr 1.91e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.916 max_load(range) 0.061-0.264 step 6730/40589 loss 0.2765 acc 0.949 lb 1.938 z 1.944 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6740/40589 loss 0.3009 acc 0.944 lb 1.939 z 1.822 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6750/40589 loss 0.2925 acc 0.946 lb 1.934 z 1.860 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.909 max_load(range) 0.067-0.443 step 6760/40589 loss 0.2818 acc 0.947 lb 1.925 z 1.984 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6770/40589 loss 0.2752 acc 0.949 lb 1.941 z 2.009 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.882 max_load(range) 0.078-0.212 step 6780/40589 loss 0.2962 acc 0.946 lb 1.939 z 1.886 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6790/40589 loss 0.2918 acc 0.947 lb 1.941 z 1.901 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6800/40589 loss 0.2860 acc 0.947 lb 1.923 z 1.926 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.864 max_load(range) 0.098-0.186 step 6810/40589 loss 0.2768 acc 0.949 lb 1.938 z 1.854 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6820/40589 loss 0.3000 acc 0.944 lb 1.940 z 1.875 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.911 max_load(range) 0.066-0.226 step 6830/40589 loss 0.2902 acc 0.947 lb 1.946 z 1.926 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6840/40589 loss 0.2884 acc 0.946 lb 1.933 z 1.984 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6850/40589 loss 0.2798 acc 0.949 lb 1.939 z 1.897 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.929 max_load(range) 0.051-0.350 step 6860/40589 loss 0.2998 acc 0.945 lb 1.936 z 1.795 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6870/40589 loss 0.2913 acc 0.946 lb 1.937 z 1.772 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.903 max_load(range) 0.057-0.438 step 6880/40589 loss 0.2839 acc 0.947 lb 1.927 z 1.888 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6890/40589 loss 0.2736 acc 0.949 lb 1.936 z 1.870 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6900/40589 loss 0.3017 acc 0.944 lb 1.941 z 1.835 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.914 max_load(range) 0.072-0.203 step 6910/40589 loss 0.2885 acc 0.947 lb 1.940 z 1.828 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6920/40589 loss 0.2843 acc 0.947 lb 1.926 z 1.879 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.912 max_load(range) 0.067-0.287 step 6930/40589 loss 0.2756 acc 0.950 lb 1.931 z 1.898 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6940/40589 loss 0.2981 acc 0.945 lb 1.935 z 1.777 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6950/40589 loss 0.2905 acc 0.947 lb 1.943 z 1.878 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left [routing] mean entropy_norm 0.914 max_load(range) 0.061-0.419 step 6960/40589 loss 0.2803 acc 0.948 lb 1.933 z 1.894 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.5d left step 6970/40589 loss 0.2752 acc 0.950 lb 1.934 z 1.849 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.878 max_load(range) 0.075-0.212 step 6980/40589 loss 0.2958 acc 0.945 lb 1.942 z 1.808 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 6990/40589 loss 0.2892 acc 0.947 lb 1.936 z 1.743 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7000/40589 loss 0.2827 acc 0.948 lb 1.937 z 1.899 lr 1.90e-04 | 1433 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.873 max_load(range) 0.072-0.246 [save] runs/pretrain_flagship_h100_4x8/step7000 step 7010/40589 loss 0.2778 acc 0.948 lb 1.937 z 1.842 lr 1.90e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7020/40589 loss 0.2957 acc 0.944 lb 1.935 z 1.725 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.900 max_load(range) 0.059-0.227 step 7030/40589 loss 0.2915 acc 0.946 lb 1.946 z 1.840 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7040/40589 loss 0.2775 acc 0.948 lb 1.934 z 1.959 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7050/40589 loss 0.2746 acc 0.949 lb 1.941 z 1.802 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.930 max_load(range) 0.052-0.352 step 7060/40589 loss 0.2955 acc 0.945 lb 1.942 z 1.752 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7070/40589 loss 0.2903 acc 0.947 lb 1.942 z 1.816 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.901 max_load(range) 0.057-0.454 step 7080/40589 loss 0.2818 acc 0.948 lb 1.929 z 1.872 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7090/40589 loss 0.2742 acc 0.948 lb 1.936 z 1.817 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7100/40589 loss 0.2994 acc 0.945 lb 1.944 z 1.704 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.913 max_load(range) 0.078-0.208 step 7110/40589 loss 0.2886 acc 0.947 lb 1.941 z 1.723 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7120/40589 loss 0.2838 acc 0.947 lb 1.932 z 1.740 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.885 max_load(range) 0.072-0.353 step 7130/40589 loss 0.2766 acc 0.949 lb 1.940 z 1.847 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7140/40589 loss 0.2965 acc 0.945 lb 1.940 z 1.665 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7150/40589 loss 0.2869 acc 0.948 lb 1.943 z 1.766 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.916 max_load(range) 0.053-0.410 step 7160/40589 loss 0.2818 acc 0.948 lb 1.937 z 1.879 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7170/40589 loss 0.2804 acc 0.948 lb 1.940 z 1.877 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.878 max_load(range) 0.079-0.199 step 7180/40589 loss 0.2981 acc 0.943 lb 1.936 z 1.708 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7190/40589 loss 0.2868 acc 0.948 lb 1.943 z 1.735 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7200/40589 loss 0.2800 acc 0.948 lb 1.933 z 1.786 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.873 max_load(range) 0.087-0.203 step 7210/40589 loss 0.2797 acc 0.948 lb 1.942 z 1.821 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7220/40589 loss 0.2981 acc 0.944 lb 1.950 z 1.760 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.906 max_load(range) 0.063-0.236 step 7230/40589 loss 0.2868 acc 0.948 lb 1.940 z 1.743 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7240/40589 loss 0.2778 acc 0.948 lb 1.929 z 1.808 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7250/40589 loss 0.2735 acc 0.949 lb 1.945 z 1.743 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.936 max_load(range) 0.052-0.360 step 7260/40589 loss 0.2931 acc 0.945 lb 1.937 z 1.762 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7270/40589 loss 0.2874 acc 0.947 lb 1.936 z 1.741 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.903 max_load(range) 0.052-0.431 step 7280/40589 loss 0.2802 acc 0.948 lb 1.924 z 1.721 lr 1.89e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7290/40589 loss 0.2746 acc 0.949 lb 1.930 z 1.810 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7300/40589 loss 0.2957 acc 0.944 lb 1.937 z 1.698 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.933 max_load(range) 0.064-0.208 step 7310/40589 loss 0.2867 acc 0.947 lb 1.948 z 1.722 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7320/40589 loss 0.2773 acc 0.949 lb 1.924 z 1.731 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.909 max_load(range) 0.065-0.327 step 7330/40589 loss 0.2765 acc 0.949 lb 1.936 z 1.801 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7340/40589 loss 0.2957 acc 0.944 lb 1.937 z 1.658 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7350/40589 loss 0.2876 acc 0.947 lb 1.944 z 1.687 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.914 max_load(range) 0.056-0.439 step 7360/40589 loss 0.2801 acc 0.948 lb 1.925 z 1.726 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7370/40589 loss 0.2722 acc 0.949 lb 1.935 z 1.685 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.885 max_load(range) 0.085-0.182 step 7380/40589 loss 0.2958 acc 0.945 lb 1.947 z 1.641 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7390/40589 loss 0.2844 acc 0.948 lb 1.942 z 1.710 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7400/40589 loss 0.2791 acc 0.948 lb 1.929 z 1.737 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.872 max_load(range) 0.085-0.251 step 7410/40589 loss 0.2721 acc 0.950 lb 1.939 z 1.709 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7420/40589 loss 0.2918 acc 0.946 lb 1.936 z 1.692 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.917 max_load(range) 0.063-0.249 step 7430/40589 loss 0.2855 acc 0.947 lb 1.948 z 1.721 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7440/40589 loss 0.2810 acc 0.947 lb 1.927 z 1.780 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7450/40589 loss 0.2702 acc 0.950 lb 1.938 z 1.747 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.936 max_load(range) 0.053-0.368 step 7460/40589 loss 0.2942 acc 0.945 lb 1.936 z 1.630 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7470/40589 loss 0.2842 acc 0.947 lb 1.935 z 1.629 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.908 max_load(range) 0.055-0.440 step 7480/40589 loss 0.2758 acc 0.949 lb 1.923 z 1.697 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7490/40589 loss 0.2712 acc 0.950 lb 1.944 z 1.744 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left step 7500/40589 loss 0.2941 acc 0.946 lb 1.940 z 1.605 lr 1.88e-04 | 1432 pairs/s | mem 63.2/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.931 max_load(range) 0.066-0.199 step 7510/40589 loss 0.2822 acc 0.948 lb 1.939 z 1.657 lr 1.88e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left step 7520/40589 loss 0.2794 acc 0.947 lb 1.930 z 1.720 lr 1.88e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.901 max_load(range) 0.069-0.334 step 7530/40589 loss 0.2717 acc 0.950 lb 1.935 z 1.693 lr 1.88e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left step 7540/40589 loss 0.2910 acc 0.945 lb 1.936 z 1.617 lr 1.88e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left step 7550/40589 loss 0.2852 acc 0.948 lb 1.946 z 1.615 lr 1.88e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.919 max_load(range) 0.057-0.434 step 7560/40589 loss 0.2719 acc 0.949 lb 1.936 z 1.678 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left step 7570/40589 loss 0.2712 acc 0.949 lb 1.932 z 1.675 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.881 max_load(range) 0.081-0.183 step 7580/40589 loss 0.2926 acc 0.945 lb 1.940 z 1.594 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left step 7590/40589 loss 0.2827 acc 0.948 lb 1.946 z 1.592 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left step 7600/40589 loss 0.2780 acc 0.948 lb 1.934 z 1.671 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.878 max_load(range) 0.081-0.220 step 7610/40589 loss 0.2717 acc 0.950 lb 1.939 z 1.665 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left step 7620/40589 loss 0.2878 acc 0.946 lb 1.937 z 1.501 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.906 max_load(range) 0.062-0.224 step 7630/40589 loss 0.2815 acc 0.948 lb 1.940 z 1.530 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left step 7640/40589 loss 0.2767 acc 0.948 lb 1.936 z 1.662 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left step 7650/40589 loss 0.2729 acc 0.949 lb 1.930 z 1.603 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.929 max_load(range) 0.053-0.362 step 7660/40589 loss 0.2892 acc 0.946 lb 1.941 z 1.607 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left step 7670/40589 loss 0.2861 acc 0.948 lb 1.938 z 1.538 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.904 max_load(range) 0.062-0.440 step 7680/40589 loss 0.2768 acc 0.949 lb 1.930 z 1.555 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left step 7690/40589 loss 0.2743 acc 0.949 lb 1.937 z 1.630 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left step 7700/40589 loss 0.2902 acc 0.945 lb 1.940 z 1.589 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.915 max_load(range) 0.065-0.185 step 7710/40589 loss 0.2830 acc 0.948 lb 1.941 z 1.599 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left step 7720/40589 loss 0.2724 acc 0.950 lb 1.927 z 1.635 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left [routing] mean entropy_norm 0.901 max_load(range) 0.070-0.329 step 7730/40589 loss 0.2708 acc 0.949 lb 1.939 z 1.561 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left step 7740/40589 loss 0.2922 acc 0.946 lb 1.934 z 1.551 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.4d left step 7750/40589 loss 0.2831 acc 0.948 lb 1.947 z 1.560 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.917 max_load(range) 0.054-0.449 step 7760/40589 loss 0.2737 acc 0.948 lb 1.931 z 1.643 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 7770/40589 loss 0.2698 acc 0.950 lb 1.938 z 1.610 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.889 max_load(range) 0.068-0.190 step 7780/40589 loss 0.2921 acc 0.945 lb 1.940 z 1.583 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 7790/40589 loss 0.2817 acc 0.948 lb 1.948 z 1.534 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 7800/40589 loss 0.2758 acc 0.949 lb 1.936 z 1.678 lr 1.87e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.874 max_load(range) 0.078-0.274 step 7810/40589 loss 0.2701 acc 0.950 lb 1.943 z 1.609 lr 1.86e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 7820/40589 loss 0.2897 acc 0.946 lb 1.940 z 1.518 lr 1.86e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.921 max_load(range) 0.069-0.230 step 7830/40589 loss 0.2825 acc 0.948 lb 1.946 z 1.519 lr 1.86e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 7840/40589 loss 0.2781 acc 0.948 lb 1.934 z 1.655 lr 1.86e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 7850/40589 loss 0.2705 acc 0.949 lb 1.940 z 1.608 lr 1.86e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.939 max_load(range) 0.052-0.360 step 7860/40589 loss 0.2873 acc 0.946 lb 1.947 z 1.518 lr 1.86e-04 | 1432 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 7870/40589 loss 0.2821 acc 0.948 lb 1.942 z 1.554 lr 1.86e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.911 max_load(range) 0.054-0.440 step 7880/40589 loss 0.2762 acc 0.949 lb 1.927 z 1.607 lr 1.86e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 7890/40589 loss 0.2700 acc 0.950 lb 1.927 z 1.533 lr 1.86e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 7900/40589 loss 0.2888 acc 0.946 lb 1.940 z 1.526 lr 1.86e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.924 max_load(range) 0.067-0.196 step 7910/40589 loss 0.2804 acc 0.948 lb 1.939 z 1.451 lr 1.86e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 7920/40589 loss 0.2734 acc 0.949 lb 1.932 z 1.673 lr 1.86e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.907 max_load(range) 0.071-0.312 step 7930/40589 loss 0.2716 acc 0.949 lb 1.937 z 1.562 lr 1.86e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 7940/40589 loss 0.2919 acc 0.946 lb 1.936 z 1.512 lr 1.86e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 7950/40589 loss 0.2819 acc 0.948 lb 1.942 z 1.520 lr 1.86e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.921 max_load(range) 0.058-0.429 step 7960/40589 loss 0.2773 acc 0.948 lb 1.927 z 1.512 lr 1.86e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 7970/40589 loss 0.2707 acc 0.950 lb 1.937 z 1.548 lr 1.86e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.893 max_load(range) 0.071-0.169 step 7980/40589 loss 0.2869 acc 0.946 lb 1.936 z 1.420 lr 1.86e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 7990/40589 loss 0.2793 acc 0.948 lb 1.938 z 1.501 lr 1.86e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8000/40589 loss 0.2752 acc 0.949 lb 1.926 z 1.588 lr 1.86e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.879 max_load(range) 0.087-0.175 [save] runs/pretrain_flagship_h100_4x8/step8000 step 8010/40589 loss 0.2690 acc 0.949 lb 1.933 z 1.519 lr 1.86e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8020/40589 loss 0.2894 acc 0.946 lb 1.943 z 1.385 lr 1.86e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.920 max_load(range) 0.069-0.220 step 8030/40589 loss 0.2778 acc 0.949 lb 1.942 z 1.438 lr 1.86e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8040/40589 loss 0.2748 acc 0.949 lb 1.935 z 1.593 lr 1.86e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8050/40589 loss 0.2680 acc 0.950 lb 1.935 z 1.472 lr 1.86e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.941 max_load(range) 0.048-0.339 step 8060/40589 loss 0.2915 acc 0.945 lb 1.936 z 1.427 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8070/40589 loss 0.2764 acc 0.948 lb 1.949 z 1.560 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.920 max_load(range) 0.056-0.413 step 8080/40589 loss 0.2737 acc 0.949 lb 1.937 z 1.575 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8090/40589 loss 0.2683 acc 0.949 lb 1.938 z 1.475 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8100/40589 loss 0.2912 acc 0.946 lb 1.946 z 1.401 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.920 max_load(range) 0.062-0.207 step 8110/40589 loss 0.2791 acc 0.948 lb 1.947 z 1.474 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8120/40589 loss 0.2745 acc 0.949 lb 1.933 z 1.496 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.902 max_load(range) 0.063-0.331 step 8130/40589 loss 0.2674 acc 0.950 lb 1.940 z 1.517 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8140/40589 loss 0.2859 acc 0.947 lb 1.942 z 1.460 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8150/40589 loss 0.2810 acc 0.948 lb 1.933 z 1.445 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.917 max_load(range) 0.060-0.446 step 8160/40589 loss 0.2736 acc 0.949 lb 1.928 z 1.628 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8170/40589 loss 0.2699 acc 0.950 lb 1.944 z 1.535 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.891 max_load(range) 0.071-0.185 step 8180/40589 loss 0.2893 acc 0.945 lb 1.937 z 1.464 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8190/40589 loss 0.2756 acc 0.949 lb 1.935 z 1.434 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8200/40589 loss 0.2700 acc 0.950 lb 1.933 z 1.566 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.882 max_load(range) 0.067-0.198 step 8210/40589 loss 0.2684 acc 0.949 lb 1.937 z 1.503 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8220/40589 loss 0.2834 acc 0.947 lb 1.945 z 1.384 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.909 max_load(range) 0.061-0.243 step 8230/40589 loss 0.2762 acc 0.949 lb 1.944 z 1.425 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8240/40589 loss 0.2713 acc 0.949 lb 1.929 z 1.503 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8250/40589 loss 0.2665 acc 0.951 lb 1.941 z 1.467 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.942 max_load(range) 0.052-0.342 step 8260/40589 loss 0.2865 acc 0.946 lb 1.938 z 1.417 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8270/40589 loss 0.2782 acc 0.949 lb 1.941 z 1.453 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.911 max_load(range) 0.059-0.447 step 8280/40589 loss 0.2728 acc 0.949 lb 1.933 z 1.532 lr 1.85e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8290/40589 loss 0.2657 acc 0.950 lb 1.942 z 1.518 lr 1.84e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8300/40589 loss 0.2865 acc 0.947 lb 1.937 z 1.372 lr 1.84e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.912 max_load(range) 0.068-0.189 step 8310/40589 loss 0.2781 acc 0.949 lb 1.944 z 1.450 lr 1.84e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8320/40589 loss 0.2718 acc 0.949 lb 1.936 z 1.443 lr 1.84e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.914 max_load(range) 0.063-0.282 step 8330/40589 loss 0.2690 acc 0.950 lb 1.938 z 1.460 lr 1.84e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8340/40589 loss 0.2883 acc 0.946 lb 1.939 z 1.330 lr 1.84e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8350/40589 loss 0.2748 acc 0.949 lb 1.947 z 1.382 lr 1.84e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.918 max_load(range) 0.062-0.434 step 8360/40589 loss 0.2705 acc 0.949 lb 1.931 z 1.440 lr 1.84e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8370/40589 loss 0.2673 acc 0.950 lb 1.934 z 1.410 lr 1.84e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.888 max_load(range) 0.081-0.191 step 8380/40589 loss 0.2863 acc 0.946 lb 1.940 z 1.435 lr 1.84e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8390/40589 loss 0.2759 acc 0.949 lb 1.944 z 1.400 lr 1.84e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8400/40589 loss 0.2721 acc 0.949 lb 1.940 z 1.501 lr 1.84e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.883 max_load(range) 0.074-0.228 step 8410/40589 loss 0.2617 acc 0.951 lb 1.942 z 1.468 lr 1.84e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8420/40589 loss 0.2862 acc 0.946 lb 1.942 z 1.329 lr 1.84e-04 | 1431 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.920 max_load(range) 0.069-0.219 step 8430/40589 loss 0.2765 acc 0.949 lb 1.940 z 1.362 lr 1.84e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8440/40589 loss 0.2705 acc 0.949 lb 1.930 z 1.534 lr 1.84e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8450/40589 loss 0.2634 acc 0.950 lb 1.933 z 1.421 lr 1.84e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.936 max_load(range) 0.057-0.385 step 8460/40589 loss 0.2857 acc 0.946 lb 1.945 z 1.389 lr 1.84e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8470/40589 loss 0.2776 acc 0.949 lb 1.942 z 1.391 lr 1.84e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.919 max_load(range) 0.062-0.416 step 8480/40589 loss 0.2698 acc 0.949 lb 1.933 z 1.441 lr 1.84e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8490/40589 loss 0.2676 acc 0.950 lb 1.937 z 1.435 lr 1.84e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8500/40589 loss 0.2834 acc 0.947 lb 1.941 z 1.389 lr 1.84e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.925 max_load(range) 0.081-0.201 step 8510/40589 loss 0.2761 acc 0.949 lb 1.944 z 1.423 lr 1.84e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8520/40589 loss 0.2757 acc 0.948 lb 1.931 z 1.405 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left [routing] mean entropy_norm 0.899 max_load(range) 0.065-0.327 step 8530/40589 loss 0.2662 acc 0.950 lb 1.942 z 1.398 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.3d left step 8540/40589 loss 0.2852 acc 0.947 lb 1.943 z 1.297 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8550/40589 loss 0.2791 acc 0.948 lb 1.943 z 1.361 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.917 max_load(range) 0.056-0.434 step 8560/40589 loss 0.2719 acc 0.949 lb 1.920 z 1.479 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8570/40589 loss 0.2646 acc 0.950 lb 1.946 z 1.375 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.879 max_load(range) 0.076-0.197 step 8580/40589 loss 0.2857 acc 0.946 lb 1.940 z 1.322 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8590/40589 loss 0.2779 acc 0.949 lb 1.940 z 1.369 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8600/40589 loss 0.2711 acc 0.949 lb 1.926 z 1.420 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.871 max_load(range) 0.091-0.169 step 8610/40589 loss 0.2639 acc 0.951 lb 1.940 z 1.422 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8620/40589 loss 0.2827 acc 0.947 lb 1.943 z 1.346 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.916 max_load(range) 0.057-0.232 step 8630/40589 loss 0.2772 acc 0.948 lb 1.940 z 1.359 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8640/40589 loss 0.2690 acc 0.950 lb 1.929 z 1.326 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8650/40589 loss 0.2655 acc 0.950 lb 1.949 z 1.391 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.937 max_load(range) 0.050-0.372 step 8660/40589 loss 0.2830 acc 0.946 lb 1.947 z 1.315 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8670/40589 loss 0.2765 acc 0.949 lb 1.941 z 1.315 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.917 max_load(range) 0.057-0.439 step 8680/40589 loss 0.2692 acc 0.949 lb 1.940 z 1.475 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8690/40589 loss 0.2654 acc 0.950 lb 1.936 z 1.391 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8700/40589 loss 0.2855 acc 0.946 lb 1.945 z 1.301 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.927 max_load(range) 0.063-0.207 step 8710/40589 loss 0.2799 acc 0.948 lb 1.942 z 1.368 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8720/40589 loss 0.2690 acc 0.950 lb 1.922 z 1.422 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.920 max_load(range) 0.060-0.294 step 8730/40589 loss 0.2672 acc 0.950 lb 1.938 z 1.395 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8740/40589 loss 0.2812 acc 0.947 lb 1.936 z 1.228 lr 1.83e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8750/40589 loss 0.2769 acc 0.948 lb 1.939 z 1.323 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.916 max_load(range) 0.061-0.438 step 8760/40589 loss 0.2682 acc 0.950 lb 1.924 z 1.321 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8770/40589 loss 0.2638 acc 0.951 lb 1.940 z 1.329 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.895 max_load(range) 0.074-0.180 step 8780/40589 loss 0.2797 acc 0.947 lb 1.942 z 1.313 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8790/40589 loss 0.2766 acc 0.949 lb 1.949 z 1.319 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8800/40589 loss 0.2690 acc 0.949 lb 1.933 z 1.415 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.883 max_load(range) 0.088-0.247 step 8810/40589 loss 0.2643 acc 0.950 lb 1.940 z 1.399 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8820/40589 loss 0.2831 acc 0.947 lb 1.940 z 1.268 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.918 max_load(range) 0.059-0.250 step 8830/40589 loss 0.2722 acc 0.949 lb 1.943 z 1.339 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8840/40589 loss 0.2691 acc 0.950 lb 1.937 z 1.492 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8850/40589 loss 0.2628 acc 0.951 lb 1.939 z 1.340 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.934 max_load(range) 0.056-0.358 step 8860/40589 loss 0.2844 acc 0.947 lb 1.943 z 1.291 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8870/40589 loss 0.2726 acc 0.949 lb 1.940 z 1.262 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.920 max_load(range) 0.063-0.427 step 8880/40589 loss 0.2666 acc 0.949 lb 1.935 z 1.439 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8890/40589 loss 0.2589 acc 0.952 lb 1.931 z 1.338 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8900/40589 loss 0.2825 acc 0.947 lb 1.945 z 1.352 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.929 max_load(range) 0.066-0.203 step 8910/40589 loss 0.2722 acc 0.950 lb 1.946 z 1.332 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8920/40589 loss 0.2650 acc 0.950 lb 1.935 z 1.331 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.896 max_load(range) 0.065-0.343 step 8930/40589 loss 0.2640 acc 0.950 lb 1.943 z 1.354 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8940/40589 loss 0.2790 acc 0.947 lb 1.937 z 1.203 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8950/40589 loss 0.2727 acc 0.950 lb 1.944 z 1.276 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.925 max_load(range) 0.055-0.430 step 8960/40589 loss 0.2698 acc 0.949 lb 1.942 z 1.404 lr 1.82e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8970/40589 loss 0.2645 acc 0.951 lb 1.945 z 1.376 lr 1.81e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.885 max_load(range) 0.073-0.187 step 8980/40589 loss 0.2797 acc 0.947 lb 1.946 z 1.247 lr 1.81e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 8990/40589 loss 0.2706 acc 0.950 lb 1.942 z 1.271 lr 1.81e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 9000/40589 loss 0.2705 acc 0.949 lb 1.937 z 1.307 lr 1.81e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.880 max_load(range) 0.096-0.185 [save] runs/pretrain_flagship_h100_4x8/step9000 step 9010/40589 loss 0.2641 acc 0.950 lb 1.947 z 1.331 lr 1.81e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 9020/40589 loss 0.2821 acc 0.947 lb 1.938 z 1.214 lr 1.81e-04 | 1430 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.916 max_load(range) 0.066-0.233 step 9030/40589 loss 0.2714 acc 0.950 lb 1.942 z 1.212 lr 1.81e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 9040/40589 loss 0.2678 acc 0.949 lb 1.928 z 1.289 lr 1.81e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 9050/40589 loss 0.2651 acc 0.950 lb 1.940 z 1.288 lr 1.81e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.948 max_load(range) 0.053-0.342 step 9060/40589 loss 0.2842 acc 0.946 lb 1.944 z 1.239 lr 1.81e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 9070/40589 loss 0.2756 acc 0.949 lb 1.940 z 1.234 lr 1.81e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.940 max_load(range) 0.053-0.382 step 9080/40589 loss 0.2675 acc 0.950 lb 1.935 z 1.412 lr 1.81e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 9090/40589 loss 0.2617 acc 0.950 lb 1.937 z 1.325 lr 1.81e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 9100/40589 loss 0.2825 acc 0.947 lb 1.945 z 1.278 lr 1.81e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.930 max_load(range) 0.065-0.186 step 9110/40589 loss 0.2712 acc 0.949 lb 1.933 z 1.239 lr 1.81e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 9120/40589 loss 0.2647 acc 0.950 lb 1.934 z 1.354 lr 1.81e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.897 max_load(range) 0.076-0.341 step 9130/40589 loss 0.2659 acc 0.949 lb 1.938 z 1.303 lr 1.81e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 9140/40589 loss 0.2787 acc 0.947 lb 1.943 z 1.157 lr 1.81e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 9150/40589 loss 0.2718 acc 0.950 lb 1.949 z 1.275 lr 1.81e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.917 max_load(range) 0.059-0.428 step 9160/40589 loss 0.2671 acc 0.949 lb 1.942 z 1.366 lr 1.81e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 9170/40589 loss 0.2623 acc 0.951 lb 1.941 z 1.236 lr 1.81e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.876 max_load(range) 0.086-0.195 step 9180/40589 loss 0.2828 acc 0.946 lb 1.942 z 1.159 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 9190/40589 loss 0.2735 acc 0.949 lb 1.949 z 1.266 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 9200/40589 loss 0.2655 acc 0.950 lb 1.942 z 1.289 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.865 max_load(range) 0.078-0.212 step 9210/40589 loss 0.2645 acc 0.951 lb 1.938 z 1.302 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 9220/40589 loss 0.2772 acc 0.948 lb 1.950 z 1.211 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.907 max_load(range) 0.062-0.232 step 9230/40589 loss 0.2698 acc 0.950 lb 1.957 z 1.274 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 9240/40589 loss 0.2647 acc 0.950 lb 1.936 z 1.323 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 9250/40589 loss 0.2624 acc 0.951 lb 1.941 z 1.226 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.936 max_load(range) 0.055-0.350 step 9260/40589 loss 0.2767 acc 0.948 lb 1.950 z 1.168 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 9270/40589 loss 0.2732 acc 0.948 lb 1.952 z 1.212 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.947 max_load(range) 0.058-0.335 step 9280/40589 loss 0.2669 acc 0.950 lb 1.939 z 1.456 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 9290/40589 loss 0.2580 acc 0.952 lb 1.947 z 1.351 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left step 9300/40589 loss 0.2790 acc 0.947 lb 1.946 z 1.167 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.2d left [routing] mean entropy_norm 0.918 max_load(range) 0.079-0.195 step 9310/40589 loss 0.2714 acc 0.949 lb 1.954 z 1.242 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9320/40589 loss 0.2646 acc 0.950 lb 1.935 z 1.299 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.913 max_load(range) 0.059-0.288 step 9330/40589 loss 0.2593 acc 0.951 lb 1.935 z 1.299 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9340/40589 loss 0.2788 acc 0.947 lb 1.940 z 1.150 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9350/40589 loss 0.2704 acc 0.949 lb 1.943 z 1.231 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.923 max_load(range) 0.058-0.420 step 9360/40589 loss 0.2658 acc 0.950 lb 1.932 z 1.301 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9370/40589 loss 0.2552 acc 0.952 lb 1.936 z 1.305 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.892 max_load(range) 0.080-0.169 step 9380/40589 loss 0.2796 acc 0.947 lb 1.938 z 1.116 lr 1.80e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9390/40589 loss 0.2689 acc 0.950 lb 1.952 z 1.254 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9400/40589 loss 0.2639 acc 0.950 lb 1.942 z 1.325 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.886 max_load(range) 0.085-0.222 step 9410/40589 loss 0.2600 acc 0.951 lb 1.939 z 1.267 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9420/40589 loss 0.2783 acc 0.947 lb 1.944 z 1.165 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.917 max_load(range) 0.067-0.246 step 9430/40589 loss 0.2710 acc 0.950 lb 1.942 z 1.158 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9440/40589 loss 0.2613 acc 0.952 lb 1.938 z 1.334 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9450/40589 loss 0.2597 acc 0.951 lb 1.937 z 1.216 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.940 max_load(range) 0.052-0.345 step 9460/40589 loss 0.2774 acc 0.948 lb 1.940 z 1.149 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9470/40589 loss 0.2734 acc 0.948 lb 1.948 z 1.156 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.949 max_load(range) 0.054-0.293 step 9480/40589 loss 0.2646 acc 0.950 lb 1.929 z 1.268 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9490/40589 loss 0.2635 acc 0.951 lb 1.953 z 1.308 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9500/40589 loss 0.2826 acc 0.946 lb 1.946 z 1.178 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.908 max_load(range) 0.078-0.185 step 9510/40589 loss 0.2692 acc 0.949 lb 1.956 z 1.174 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9520/40589 loss 0.2663 acc 0.949 lb 1.941 z 1.378 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.896 max_load(range) 0.065-0.331 step 9530/40589 loss 0.2622 acc 0.951 lb 1.946 z 1.251 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9540/40589 loss 0.2760 acc 0.947 lb 1.952 z 1.105 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9550/40589 loss 0.2669 acc 0.950 lb 1.949 z 1.225 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.917 max_load(range) 0.057-0.451 step 9560/40589 loss 0.2637 acc 0.950 lb 1.934 z 1.251 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9570/40589 loss 0.2553 acc 0.951 lb 1.944 z 1.195 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.867 max_load(range) 0.077-0.187 step 9580/40589 loss 0.2812 acc 0.947 lb 1.946 z 1.146 lr 1.79e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9590/40589 loss 0.2643 acc 0.951 lb 1.944 z 1.104 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9600/40589 loss 0.2634 acc 0.950 lb 1.940 z 1.236 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.874 max_load(range) 0.094-0.206 step 9610/40589 loss 0.2587 acc 0.951 lb 1.940 z 1.199 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9620/40589 loss 0.2802 acc 0.947 lb 1.944 z 1.118 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.920 max_load(range) 0.058-0.248 step 9630/40589 loss 0.2648 acc 0.951 lb 1.940 z 1.137 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9640/40589 loss 0.2650 acc 0.950 lb 1.927 z 1.107 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9650/40589 loss 0.2580 acc 0.951 lb 1.940 z 1.247 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.939 max_load(range) 0.049-0.314 step 9660/40589 loss 0.2787 acc 0.947 lb 1.941 z 1.111 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9670/40589 loss 0.2692 acc 0.949 lb 1.946 z 1.151 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.949 max_load(range) 0.056-0.297 step 9680/40589 loss 0.2660 acc 0.950 lb 1.940 z 1.280 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9690/40589 loss 0.2624 acc 0.950 lb 1.929 z 1.265 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9700/40589 loss 0.2768 acc 0.948 lb 1.939 z 1.129 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.916 max_load(range) 0.076-0.206 step 9710/40589 loss 0.2681 acc 0.950 lb 1.952 z 1.211 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9720/40589 loss 0.2638 acc 0.950 lb 1.933 z 1.293 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.907 max_load(range) 0.068-0.315 step 9730/40589 loss 0.2590 acc 0.952 lb 1.940 z 1.237 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9740/40589 loss 0.2760 acc 0.947 lb 1.947 z 1.093 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9750/40589 loss 0.2684 acc 0.950 lb 1.942 z 1.167 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.924 max_load(range) 0.057-0.423 step 9760/40589 loss 0.2623 acc 0.951 lb 1.931 z 1.208 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9770/40589 loss 0.2575 acc 0.951 lb 1.942 z 1.221 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.894 max_load(range) 0.069-0.185 step 9780/40589 loss 0.2755 acc 0.948 lb 1.954 z 1.137 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9790/40589 loss 0.2665 acc 0.951 lb 1.952 z 1.189 lr 1.78e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9800/40589 loss 0.2632 acc 0.950 lb 1.930 z 1.214 lr 1.77e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.896 max_load(range) 0.065-0.231 step 9810/40589 loss 0.2595 acc 0.951 lb 1.941 z 1.201 lr 1.77e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9820/40589 loss 0.2784 acc 0.947 lb 1.946 z 1.040 lr 1.77e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.923 max_load(range) 0.058-0.237 step 9830/40589 loss 0.2689 acc 0.950 lb 1.941 z 1.131 lr 1.77e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9840/40589 loss 0.2616 acc 0.950 lb 1.939 z 1.249 lr 1.77e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9850/40589 loss 0.2544 acc 0.952 lb 1.940 z 1.189 lr 1.77e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.945 max_load(range) 0.053-0.307 step 9860/40589 loss 0.2784 acc 0.946 lb 1.944 z 1.101 lr 1.77e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9870/40589 loss 0.2715 acc 0.949 lb 1.942 z 1.086 lr 1.77e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.950 max_load(range) 0.059-0.264 step 9880/40589 loss 0.2643 acc 0.951 lb 1.933 z 1.242 lr 1.77e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9890/40589 loss 0.2600 acc 0.951 lb 1.945 z 1.227 lr 1.77e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9900/40589 loss 0.2756 acc 0.948 lb 1.948 z 1.097 lr 1.77e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.926 max_load(range) 0.061-0.214 step 9910/40589 loss 0.2692 acc 0.949 lb 1.943 z 1.128 lr 1.77e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9920/40589 loss 0.2608 acc 0.950 lb 1.932 z 1.199 lr 1.77e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.910 max_load(range) 0.062-0.313 step 9930/40589 loss 0.2569 acc 0.951 lb 1.943 z 1.127 lr 1.77e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9940/40589 loss 0.2764 acc 0.947 lb 1.945 z 1.021 lr 1.77e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9950/40589 loss 0.2687 acc 0.950 lb 1.945 z 1.123 lr 1.77e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.925 max_load(range) 0.058-0.425 step 9960/40589 loss 0.2634 acc 0.949 lb 1.935 z 1.255 lr 1.77e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9970/40589 loss 0.2590 acc 0.950 lb 1.936 z 1.149 lr 1.77e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.888 max_load(range) 0.067-0.176 step 9980/40589 loss 0.2763 acc 0.947 lb 1.950 z 1.055 lr 1.77e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 9990/40589 loss 0.2661 acc 0.951 lb 1.946 z 1.077 lr 1.76e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 10000/40589 loss 0.2624 acc 0.950 lb 1.937 z 1.201 lr 1.76e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.886 max_load(range) 0.082-0.242 [save] runs/pretrain_flagship_h100_4x8/step10000 step 10010/40589 loss 0.2545 acc 0.952 lb 1.945 z 1.164 lr 1.76e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 10020/40589 loss 0.2732 acc 0.948 lb 1.950 z 1.036 lr 1.76e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.912 max_load(range) 0.067-0.224 step 10030/40589 loss 0.2659 acc 0.951 lb 1.951 z 1.076 lr 1.76e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 10040/40589 loss 0.2638 acc 0.950 lb 1.935 z 1.197 lr 1.76e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 10050/40589 loss 0.2550 acc 0.952 lb 1.937 z 1.156 lr 1.76e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.933 max_load(range) 0.051-0.313 step 10060/40589 loss 0.2694 acc 0.949 lb 1.947 z 1.068 lr 1.76e-04 | 1429 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left step 10070/40589 loss 0.2642 acc 0.950 lb 1.950 z 1.111 lr 1.76e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.1d left [routing] mean entropy_norm 0.950 max_load(range) 0.054-0.266 step 10080/40589 loss 0.2631 acc 0.950 lb 1.939 z 1.122 lr 1.76e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10090/40589 loss 0.2579 acc 0.951 lb 1.939 z 1.179 lr 1.76e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10100/40589 loss 0.2764 acc 0.947 lb 1.949 z 1.095 lr 1.76e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.929 max_load(range) 0.074-0.218 step 10110/40589 loss 0.2643 acc 0.951 lb 1.953 z 1.153 lr 1.76e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10120/40589 loss 0.2590 acc 0.951 lb 1.936 z 1.163 lr 1.76e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.912 max_load(range) 0.063-0.321 step 10130/40589 loss 0.2582 acc 0.951 lb 1.945 z 1.146 lr 1.76e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10140/40589 loss 0.2775 acc 0.947 lb 1.950 z 1.078 lr 1.76e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10150/40589 loss 0.2666 acc 0.950 lb 1.946 z 1.086 lr 1.76e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.931 max_load(range) 0.057-0.442 step 10160/40589 loss 0.2627 acc 0.950 lb 1.935 z 1.117 lr 1.76e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10170/40589 loss 0.2572 acc 0.951 lb 1.938 z 1.147 lr 1.76e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.900 max_load(range) 0.084-0.163 step 10180/40589 loss 0.2717 acc 0.948 lb 1.934 z 1.036 lr 1.75e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10190/40589 loss 0.2658 acc 0.950 lb 1.939 z 1.054 lr 1.75e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10200/40589 loss 0.2565 acc 0.951 lb 1.942 z 1.248 lr 1.75e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.893 max_load(range) 0.078-0.187 step 10210/40589 loss 0.2584 acc 0.951 lb 1.938 z 1.121 lr 1.75e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10220/40589 loss 0.2741 acc 0.948 lb 1.948 z 1.001 lr 1.75e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.922 max_load(range) 0.067-0.246 step 10230/40589 loss 0.2653 acc 0.950 lb 1.939 z 1.069 lr 1.75e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10240/40589 loss 0.2618 acc 0.950 lb 1.936 z 1.187 lr 1.75e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10250/40589 loss 0.2537 acc 0.952 lb 1.941 z 1.181 lr 1.75e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.944 max_load(range) 0.052-0.274 step 10260/40589 loss 0.2720 acc 0.948 lb 1.945 z 1.001 lr 1.75e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10270/40589 loss 0.2614 acc 0.951 lb 1.937 z 1.007 lr 1.75e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.942 max_load(range) 0.063-0.227 step 10280/40589 loss 0.2592 acc 0.951 lb 1.929 z 1.208 lr 1.75e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10290/40589 loss 0.2542 acc 0.952 lb 1.948 z 1.158 lr 1.75e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10300/40589 loss 0.2726 acc 0.948 lb 1.943 z 1.025 lr 1.75e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.937 max_load(range) 0.060-0.197 step 10310/40589 loss 0.2644 acc 0.950 lb 1.949 z 1.103 lr 1.75e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10320/40589 loss 0.2627 acc 0.950 lb 1.938 z 1.170 lr 1.75e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.911 max_load(range) 0.068-0.316 step 10330/40589 loss 0.2522 acc 0.952 lb 1.944 z 1.147 lr 1.75e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10340/40589 loss 0.2724 acc 0.948 lb 1.949 z 1.031 lr 1.75e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10350/40589 loss 0.2621 acc 0.951 lb 1.944 z 1.015 lr 1.75e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.929 max_load(range) 0.056-0.437 step 10360/40589 loss 0.2580 acc 0.951 lb 1.938 z 1.163 lr 1.75e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10370/40589 loss 0.2551 acc 0.951 lb 1.939 z 1.101 lr 1.74e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.912 max_load(range) 0.074-0.176 step 10380/40589 loss 0.2675 acc 0.949 lb 1.949 z 0.975 lr 1.74e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10390/40589 loss 0.2634 acc 0.950 lb 1.941 z 1.085 lr 1.74e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10400/40589 loss 0.2576 acc 0.951 lb 1.933 z 1.112 lr 1.74e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.872 max_load(range) 0.085-0.209 step 10410/40589 loss 0.2553 acc 0.952 lb 1.937 z 1.097 lr 1.74e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10420/40589 loss 0.2732 acc 0.948 lb 1.940 z 0.981 lr 1.74e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.929 max_load(range) 0.065-0.225 step 10430/40589 loss 0.2641 acc 0.950 lb 1.946 z 1.044 lr 1.74e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10440/40589 loss 0.2586 acc 0.951 lb 1.935 z 1.163 lr 1.74e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10450/40589 loss 0.2576 acc 0.951 lb 1.940 z 1.072 lr 1.74e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.939 max_load(range) 0.049-0.291 step 10460/40589 loss 0.2738 acc 0.949 lb 1.943 z 0.961 lr 1.74e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10470/40589 loss 0.2597 acc 0.951 lb 1.941 z 1.016 lr 1.74e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.936 max_load(range) 0.075-0.209 step 10480/40589 loss 0.2587 acc 0.951 lb 1.933 z 1.144 lr 1.74e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10490/40589 loss 0.2553 acc 0.952 lb 1.939 z 1.055 lr 1.74e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10500/40589 loss 0.2698 acc 0.948 lb 1.944 z 0.916 lr 1.74e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.925 max_load(range) 0.064-0.192 step 10510/40589 loss 0.2610 acc 0.951 lb 1.943 z 1.016 lr 1.74e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10520/40589 loss 0.2552 acc 0.951 lb 1.940 z 1.070 lr 1.74e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.908 max_load(range) 0.067-0.314 step 10530/40589 loss 0.2569 acc 0.952 lb 1.939 z 1.139 lr 1.74e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10540/40589 loss 0.2724 acc 0.948 lb 1.944 z 0.958 lr 1.74e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10550/40589 loss 0.2660 acc 0.951 lb 1.947 z 1.035 lr 1.74e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.929 max_load(range) 0.055-0.408 step 10560/40589 loss 0.2554 acc 0.951 lb 1.935 z 1.054 lr 1.73e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10570/40589 loss 0.2575 acc 0.951 lb 1.936 z 1.138 lr 1.73e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.893 max_load(range) 0.077-0.179 step 10580/40589 loss 0.2737 acc 0.948 lb 1.941 z 1.019 lr 1.73e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10590/40589 loss 0.2630 acc 0.951 lb 1.948 z 1.054 lr 1.73e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10600/40589 loss 0.2609 acc 0.950 lb 1.949 z 1.201 lr 1.73e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.891 max_load(range) 0.070-0.217 step 10610/40589 loss 0.2554 acc 0.952 lb 1.941 z 1.078 lr 1.73e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10620/40589 loss 0.2741 acc 0.947 lb 1.942 z 0.996 lr 1.73e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.919 max_load(range) 0.058-0.216 step 10630/40589 loss 0.2638 acc 0.950 lb 1.951 z 1.007 lr 1.73e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10640/40589 loss 0.2563 acc 0.951 lb 1.940 z 1.170 lr 1.73e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10650/40589 loss 0.2548 acc 0.951 lb 1.945 z 1.109 lr 1.73e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.933 max_load(range) 0.050-0.283 step 10660/40589 loss 0.2715 acc 0.949 lb 1.949 z 0.940 lr 1.73e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10670/40589 loss 0.2645 acc 0.950 lb 1.956 z 1.060 lr 1.73e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.922 max_load(range) 0.063-0.191 step 10680/40589 loss 0.2565 acc 0.951 lb 1.923 z 1.062 lr 1.73e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10690/40589 loss 0.2538 acc 0.952 lb 1.937 z 1.073 lr 1.73e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10700/40589 loss 0.2710 acc 0.948 lb 1.952 z 1.005 lr 1.73e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.932 max_load(range) 0.071-0.216 step 10710/40589 loss 0.2607 acc 0.951 lb 1.940 z 0.983 lr 1.73e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10720/40589 loss 0.2557 acc 0.952 lb 1.933 z 1.154 lr 1.73e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.909 max_load(range) 0.066-0.325 step 10730/40589 loss 0.2552 acc 0.951 lb 1.941 z 1.085 lr 1.73e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10740/40589 loss 0.2709 acc 0.948 lb 1.945 z 0.966 lr 1.72e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10750/40589 loss 0.2595 acc 0.951 lb 1.946 z 0.952 lr 1.72e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.935 max_load(range) 0.056-0.417 step 10760/40589 loss 0.2569 acc 0.951 lb 1.933 z 1.095 lr 1.72e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10770/40589 loss 0.2542 acc 0.952 lb 1.941 z 1.104 lr 1.72e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.900 max_load(range) 0.077-0.192 step 10780/40589 loss 0.2689 acc 0.949 lb 1.948 z 1.019 lr 1.72e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10790/40589 loss 0.2628 acc 0.950 lb 1.944 z 0.978 lr 1.72e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10800/40589 loss 0.2548 acc 0.952 lb 1.945 z 1.127 lr 1.72e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.890 max_load(range) 0.088-0.243 step 10810/40589 loss 0.2511 acc 0.952 lb 1.944 z 1.051 lr 1.72e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10820/40589 loss 0.2712 acc 0.948 lb 1.953 z 0.959 lr 1.72e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.917 max_load(range) 0.064-0.218 step 10830/40589 loss 0.2623 acc 0.951 lb 1.955 z 1.004 lr 1.72e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10840/40589 loss 0.2570 acc 0.951 lb 1.936 z 1.096 lr 1.72e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left step 10850/40589 loss 0.2534 acc 0.952 lb 1.940 z 1.052 lr 1.72e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 4.0d left [routing] mean entropy_norm 0.937 max_load(range) 0.054-0.280 step 10860/40589 loss 0.2690 acc 0.949 lb 1.945 z 0.978 lr 1.72e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 10870/40589 loss 0.2619 acc 0.951 lb 1.948 z 0.959 lr 1.72e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.919 max_load(range) 0.073-0.183 step 10880/40589 loss 0.2566 acc 0.951 lb 1.942 z 1.052 lr 1.72e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 10890/40589 loss 0.2504 acc 0.952 lb 1.944 z 1.041 lr 1.72e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 10900/40589 loss 0.2721 acc 0.948 lb 1.946 z 0.953 lr 1.72e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.922 max_load(range) 0.048-0.216 step 10910/40589 loss 0.2603 acc 0.951 lb 1.948 z 1.037 lr 1.72e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 10920/40589 loss 0.2553 acc 0.951 lb 1.941 z 1.103 lr 1.71e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.917 max_load(range) 0.061-0.289 step 10930/40589 loss 0.2506 acc 0.952 lb 1.938 z 1.062 lr 1.71e-04 | 1428 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 10940/40589 loss 0.2676 acc 0.949 lb 1.946 z 0.943 lr 1.71e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 10950/40589 loss 0.2627 acc 0.951 lb 1.950 z 0.965 lr 1.71e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.933 max_load(range) 0.059-0.405 step 10960/40589 loss 0.2562 acc 0.952 lb 1.931 z 1.039 lr 1.71e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 10970/40589 loss 0.2513 acc 0.952 lb 1.938 z 1.088 lr 1.71e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.888 max_load(range) 0.079-0.194 step 10980/40589 loss 0.2746 acc 0.947 lb 1.949 z 0.929 lr 1.71e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 10990/40589 loss 0.2659 acc 0.950 lb 1.950 z 0.978 lr 1.71e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11000/40589 loss 0.2595 acc 0.951 lb 1.942 z 1.138 lr 1.71e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.892 max_load(range) 0.076-0.231 [save] runs/pretrain_flagship_h100_4x8/step11000 step 11010/40589 loss 0.2514 acc 0.952 lb 1.937 z 1.042 lr 1.71e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11020/40589 loss 0.2716 acc 0.948 lb 1.946 z 0.907 lr 1.71e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.925 max_load(range) 0.071-0.240 step 11030/40589 loss 0.2593 acc 0.952 lb 1.949 z 1.005 lr 1.71e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11040/40589 loss 0.2488 acc 0.953 lb 1.939 z 1.034 lr 1.71e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11050/40589 loss 0.2502 acc 0.952 lb 1.941 z 1.035 lr 1.71e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.938 max_load(range) 0.059-0.247 step 11060/40589 loss 0.2653 acc 0.950 lb 1.950 z 0.926 lr 1.71e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11070/40589 loss 0.2578 acc 0.952 lb 1.946 z 0.997 lr 1.71e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.919 max_load(range) 0.082-0.180 step 11080/40589 loss 0.2530 acc 0.952 lb 1.936 z 1.068 lr 1.71e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11090/40589 loss 0.2519 acc 0.952 lb 1.951 z 1.038 lr 1.71e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11100/40589 loss 0.2688 acc 0.948 lb 1.956 z 0.928 lr 1.70e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.924 max_load(range) 0.058-0.214 step 11110/40589 loss 0.2608 acc 0.951 lb 1.950 z 0.983 lr 1.70e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11120/40589 loss 0.2547 acc 0.951 lb 1.944 z 1.051 lr 1.70e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.909 max_load(range) 0.060-0.320 step 11130/40589 loss 0.2521 acc 0.952 lb 1.943 z 1.035 lr 1.70e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11140/40589 loss 0.2663 acc 0.949 lb 1.953 z 0.914 lr 1.70e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11150/40589 loss 0.2597 acc 0.951 lb 1.949 z 0.937 lr 1.70e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.931 max_load(range) 0.056-0.433 step 11160/40589 loss 0.2553 acc 0.951 lb 1.944 z 1.073 lr 1.70e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11170/40589 loss 0.2480 acc 0.952 lb 1.943 z 0.992 lr 1.70e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.896 max_load(range) 0.075-0.190 step 11180/40589 loss 0.2688 acc 0.948 lb 1.949 z 0.947 lr 1.70e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11190/40589 loss 0.2602 acc 0.951 lb 1.947 z 0.971 lr 1.70e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11200/40589 loss 0.2548 acc 0.951 lb 1.933 z 1.102 lr 1.70e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.893 max_load(range) 0.077-0.227 step 11210/40589 loss 0.2521 acc 0.952 lb 1.942 z 1.067 lr 1.70e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11220/40589 loss 0.2674 acc 0.949 lb 1.951 z 0.916 lr 1.70e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.926 max_load(range) 0.067-0.220 step 11230/40589 loss 0.2592 acc 0.951 lb 1.952 z 0.965 lr 1.70e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11240/40589 loss 0.2550 acc 0.951 lb 1.945 z 1.029 lr 1.70e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11250/40589 loss 0.2520 acc 0.951 lb 1.943 z 1.034 lr 1.70e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.923 max_load(range) 0.061-0.263 step 11260/40589 loss 0.2696 acc 0.949 lb 1.950 z 0.930 lr 1.70e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11270/40589 loss 0.2573 acc 0.952 lb 1.953 z 0.952 lr 1.70e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.909 max_load(range) 0.084-0.168 step 11280/40589 loss 0.2536 acc 0.952 lb 1.943 z 1.089 lr 1.69e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11290/40589 loss 0.2523 acc 0.952 lb 1.944 z 1.013 lr 1.69e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11300/40589 loss 0.2682 acc 0.949 lb 1.951 z 0.901 lr 1.69e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.929 max_load(range) 0.059-0.216 step 11310/40589 loss 0.2625 acc 0.950 lb 1.947 z 0.985 lr 1.69e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11320/40589 loss 0.2533 acc 0.952 lb 1.940 z 1.012 lr 1.69e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.923 max_load(range) 0.060-0.244 step 11330/40589 loss 0.2502 acc 0.952 lb 1.946 z 0.993 lr 1.69e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11340/40589 loss 0.2657 acc 0.949 lb 1.947 z 0.874 lr 1.69e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11350/40589 loss 0.2556 acc 0.951 lb 1.941 z 0.871 lr 1.69e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.929 max_load(range) 0.050-0.437 step 11360/40589 loss 0.2533 acc 0.952 lb 1.937 z 1.080 lr 1.69e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11370/40589 loss 0.2511 acc 0.951 lb 1.949 z 0.998 lr 1.69e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.893 max_load(range) 0.072-0.144 step 11380/40589 loss 0.2698 acc 0.948 lb 1.945 z 0.920 lr 1.69e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11390/40589 loss 0.2599 acc 0.951 lb 1.953 z 0.929 lr 1.69e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11400/40589 loss 0.2545 acc 0.951 lb 1.939 z 1.102 lr 1.69e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.884 max_load(range) 0.076-0.250 step 11410/40589 loss 0.2552 acc 0.951 lb 1.940 z 1.063 lr 1.69e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11420/40589 loss 0.2719 acc 0.948 lb 1.954 z 0.913 lr 1.69e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.922 max_load(range) 0.056-0.221 step 11430/40589 loss 0.2593 acc 0.951 lb 1.950 z 0.923 lr 1.69e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11440/40589 loss 0.2527 acc 0.952 lb 1.940 z 1.049 lr 1.69e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11450/40589 loss 0.2529 acc 0.952 lb 1.939 z 0.990 lr 1.68e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.935 max_load(range) 0.063-0.257 step 11460/40589 loss 0.2635 acc 0.950 lb 1.957 z 0.871 lr 1.68e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11470/40589 loss 0.2596 acc 0.951 lb 1.946 z 0.928 lr 1.68e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.929 max_load(range) 0.078-0.180 step 11480/40589 loss 0.2542 acc 0.951 lb 1.939 z 1.069 lr 1.68e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11490/40589 loss 0.2497 acc 0.953 lb 1.937 z 1.037 lr 1.68e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11500/40589 loss 0.2673 acc 0.949 lb 1.948 z 0.882 lr 1.68e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.927 max_load(range) 0.062-0.180 step 11510/40589 loss 0.2596 acc 0.951 lb 1.949 z 0.917 lr 1.68e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11520/40589 loss 0.2548 acc 0.951 lb 1.936 z 1.033 lr 1.68e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.914 max_load(range) 0.061-0.296 step 11530/40589 loss 0.2473 acc 0.952 lb 1.942 z 0.965 lr 1.68e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11540/40589 loss 0.2656 acc 0.949 lb 1.949 z 0.879 lr 1.68e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11550/40589 loss 0.2566 acc 0.952 lb 1.949 z 0.910 lr 1.68e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.928 max_load(range) 0.052-0.424 step 11560/40589 loss 0.2540 acc 0.951 lb 1.941 z 1.061 lr 1.68e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11570/40589 loss 0.2503 acc 0.952 lb 1.943 z 1.071 lr 1.68e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.898 max_load(range) 0.074-0.177 step 11580/40589 loss 0.2669 acc 0.949 lb 1.955 z 0.912 lr 1.68e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11590/40589 loss 0.2556 acc 0.952 lb 1.949 z 0.929 lr 1.68e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11600/40589 loss 0.2541 acc 0.952 lb 1.935 z 1.010 lr 1.68e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.888 max_load(range) 0.083-0.247 step 11610/40589 loss 0.2475 acc 0.953 lb 1.935 z 0.953 lr 1.68e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left step 11620/40589 loss 0.2686 acc 0.949 lb 1.947 z 0.830 lr 1.67e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.9d left [routing] mean entropy_norm 0.920 max_load(range) 0.062-0.236 step 11630/40589 loss 0.2561 acc 0.951 lb 1.951 z 0.921 lr 1.67e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11640/40589 loss 0.2531 acc 0.952 lb 1.946 z 1.021 lr 1.67e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11650/40589 loss 0.2474 acc 0.953 lb 1.944 z 0.995 lr 1.67e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.931 max_load(range) 0.052-0.265 step 11660/40589 loss 0.2635 acc 0.950 lb 1.957 z 0.930 lr 1.67e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11670/40589 loss 0.2575 acc 0.952 lb 1.955 z 0.968 lr 1.67e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.920 max_load(range) 0.075-0.161 step 11680/40589 loss 0.2512 acc 0.952 lb 1.948 z 1.036 lr 1.67e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11690/40589 loss 0.2491 acc 0.952 lb 1.941 z 0.990 lr 1.67e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11700/40589 loss 0.2700 acc 0.948 lb 1.956 z 0.877 lr 1.67e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.928 max_load(range) 0.070-0.188 step 11710/40589 loss 0.2578 acc 0.951 lb 1.958 z 0.946 lr 1.67e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11720/40589 loss 0.2562 acc 0.951 lb 1.940 z 1.026 lr 1.67e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.910 max_load(range) 0.070-0.329 step 11730/40589 loss 0.2463 acc 0.953 lb 1.934 z 0.998 lr 1.67e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11740/40589 loss 0.2663 acc 0.949 lb 1.955 z 0.920 lr 1.67e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11750/40589 loss 0.2573 acc 0.951 lb 1.948 z 0.900 lr 1.67e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.938 max_load(range) 0.051-0.388 step 11760/40589 loss 0.2520 acc 0.952 lb 1.934 z 1.036 lr 1.67e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11770/40589 loss 0.2504 acc 0.952 lb 1.941 z 1.013 lr 1.67e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.896 max_load(range) 0.068-0.183 step 11780/40589 loss 0.2676 acc 0.949 lb 1.952 z 0.877 lr 1.67e-04 | 1427 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11790/40589 loss 0.2554 acc 0.951 lb 1.953 z 0.941 lr 1.66e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11800/40589 loss 0.2525 acc 0.951 lb 1.942 z 1.017 lr 1.66e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.873 max_load(range) 0.079-0.257 step 11810/40589 loss 0.2469 acc 0.953 lb 1.943 z 0.964 lr 1.66e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11820/40589 loss 0.2656 acc 0.949 lb 1.958 z 0.916 lr 1.66e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.913 max_load(range) 0.059-0.236 step 11830/40589 loss 0.2580 acc 0.951 lb 1.959 z 0.861 lr 1.66e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11840/40589 loss 0.2533 acc 0.951 lb 1.947 z 1.078 lr 1.66e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11850/40589 loss 0.2472 acc 0.953 lb 1.943 z 0.992 lr 1.66e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.921 max_load(range) 0.058-0.229 step 11860/40589 loss 0.2628 acc 0.949 lb 1.953 z 0.826 lr 1.66e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11870/40589 loss 0.2557 acc 0.952 lb 1.949 z 0.935 lr 1.66e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.913 max_load(range) 0.068-0.164 step 11880/40589 loss 0.2537 acc 0.952 lb 1.929 z 1.022 lr 1.66e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11890/40589 loss 0.2479 acc 0.952 lb 1.943 z 0.981 lr 1.66e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11900/40589 loss 0.2664 acc 0.949 lb 1.948 z 0.873 lr 1.66e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.935 max_load(range) 0.060-0.200 step 11910/40589 loss 0.2535 acc 0.952 lb 1.939 z 0.884 lr 1.66e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11920/40589 loss 0.2514 acc 0.952 lb 1.940 z 0.977 lr 1.66e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.920 max_load(range) 0.059-0.298 step 11930/40589 loss 0.2474 acc 0.953 lb 1.948 z 1.003 lr 1.66e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11940/40589 loss 0.2626 acc 0.949 lb 1.950 z 0.839 lr 1.66e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11950/40589 loss 0.2522 acc 0.952 lb 1.945 z 0.907 lr 1.66e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.934 max_load(range) 0.055-0.413 step 11960/40589 loss 0.2522 acc 0.952 lb 1.937 z 0.986 lr 1.65e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11970/40589 loss 0.2463 acc 0.953 lb 1.943 z 0.966 lr 1.65e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.899 max_load(range) 0.069-0.177 step 11980/40589 loss 0.2664 acc 0.949 lb 1.957 z 0.854 lr 1.65e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 11990/40589 loss 0.2530 acc 0.952 lb 1.948 z 0.880 lr 1.65e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12000/40589 loss 0.2522 acc 0.952 lb 1.939 z 1.028 lr 1.65e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.886 max_load(range) 0.078-0.200 [save] runs/pretrain_flagship_h100_4x8/step12000 step 12010/40589 loss 0.2474 acc 0.953 lb 1.945 z 0.978 lr 1.65e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12020/40589 loss 0.2627 acc 0.949 lb 1.951 z 0.812 lr 1.65e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.923 max_load(range) 0.066-0.218 step 12030/40589 loss 0.2544 acc 0.952 lb 1.949 z 0.883 lr 1.65e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12040/40589 loss 0.2452 acc 0.953 lb 1.941 z 0.931 lr 1.65e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12050/40589 loss 0.2448 acc 0.953 lb 1.942 z 0.926 lr 1.65e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.933 max_load(range) 0.065-0.278 step 12060/40589 loss 0.2634 acc 0.949 lb 1.956 z 0.867 lr 1.65e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12070/40589 loss 0.2543 acc 0.952 lb 1.945 z 0.865 lr 1.65e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.907 max_load(range) 0.075-0.182 step 12080/40589 loss 0.2482 acc 0.952 lb 1.937 z 1.062 lr 1.65e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12090/40589 loss 0.2463 acc 0.952 lb 1.946 z 0.985 lr 1.65e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12100/40589 loss 0.2631 acc 0.950 lb 1.957 z 0.849 lr 1.65e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.920 max_load(range) 0.069-0.211 step 12110/40589 loss 0.2560 acc 0.952 lb 1.955 z 0.914 lr 1.65e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12120/40589 loss 0.2485 acc 0.953 lb 1.944 z 0.927 lr 1.64e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.916 max_load(range) 0.055-0.290 step 12130/40589 loss 0.2487 acc 0.952 lb 1.952 z 1.005 lr 1.64e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12140/40589 loss 0.2630 acc 0.949 lb 1.957 z 0.843 lr 1.64e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12150/40589 loss 0.2555 acc 0.952 lb 1.950 z 0.910 lr 1.64e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.935 max_load(range) 0.059-0.417 step 12160/40589 loss 0.2478 acc 0.953 lb 1.945 z 0.928 lr 1.64e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12170/40589 loss 0.2463 acc 0.953 lb 1.943 z 0.966 lr 1.64e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.887 max_load(range) 0.085-0.168 step 12180/40589 loss 0.2636 acc 0.949 lb 1.954 z 0.833 lr 1.64e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12190/40589 loss 0.2550 acc 0.952 lb 1.949 z 0.867 lr 1.64e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12200/40589 loss 0.2530 acc 0.951 lb 1.942 z 0.978 lr 1.64e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.887 max_load(range) 0.082-0.219 step 12210/40589 loss 0.2504 acc 0.952 lb 1.937 z 0.925 lr 1.64e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12220/40589 loss 0.2647 acc 0.949 lb 1.957 z 0.893 lr 1.64e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.925 max_load(range) 0.065-0.209 step 12230/40589 loss 0.2524 acc 0.952 lb 1.944 z 0.851 lr 1.64e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12240/40589 loss 0.2501 acc 0.952 lb 1.946 z 1.003 lr 1.64e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12250/40589 loss 0.2433 acc 0.954 lb 1.946 z 0.953 lr 1.64e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.923 max_load(range) 0.062-0.261 step 12260/40589 loss 0.2672 acc 0.949 lb 1.964 z 0.821 lr 1.64e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12270/40589 loss 0.2517 acc 0.952 lb 1.963 z 0.836 lr 1.64e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.892 max_load(range) 0.093-0.220 step 12280/40589 loss 0.2468 acc 0.953 lb 1.942 z 0.983 lr 1.64e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12290/40589 loss 0.2473 acc 0.953 lb 1.949 z 1.011 lr 1.63e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12300/40589 loss 0.2630 acc 0.950 lb 1.956 z 0.856 lr 1.63e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.907 max_load(range) 0.074-0.187 step 12310/40589 loss 0.2574 acc 0.951 lb 1.959 z 0.851 lr 1.63e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12320/40589 loss 0.2500 acc 0.953 lb 1.947 z 0.979 lr 1.63e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.902 max_load(range) 0.064-0.320 step 12330/40589 loss 0.2470 acc 0.952 lb 1.948 z 0.922 lr 1.63e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12340/40589 loss 0.2602 acc 0.950 lb 1.958 z 0.843 lr 1.63e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12350/40589 loss 0.2525 acc 0.953 lb 1.947 z 0.883 lr 1.63e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.933 max_load(range) 0.054-0.425 step 12360/40589 loss 0.2472 acc 0.953 lb 1.939 z 0.975 lr 1.63e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12370/40589 loss 0.2483 acc 0.952 lb 1.944 z 0.953 lr 1.63e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left [routing] mean entropy_norm 0.897 max_load(range) 0.076-0.197 step 12380/40589 loss 0.2628 acc 0.949 lb 1.949 z 0.890 lr 1.63e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12390/40589 loss 0.2530 acc 0.952 lb 1.947 z 0.870 lr 1.63e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.8d left step 12400/40589 loss 0.2469 acc 0.953 lb 1.944 z 1.010 lr 1.63e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.891 max_load(range) 0.085-0.226 step 12410/40589 loss 0.2459 acc 0.953 lb 1.943 z 0.966 lr 1.63e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12420/40589 loss 0.2634 acc 0.949 lb 1.957 z 0.859 lr 1.63e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.927 max_load(range) 0.060-0.218 step 12430/40589 loss 0.2545 acc 0.952 lb 1.959 z 0.854 lr 1.63e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12440/40589 loss 0.2484 acc 0.952 lb 1.945 z 0.898 lr 1.63e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12450/40589 loss 0.2476 acc 0.952 lb 1.945 z 0.912 lr 1.62e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.940 max_load(range) 0.055-0.254 step 12460/40589 loss 0.2683 acc 0.948 lb 1.956 z 0.847 lr 1.62e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12470/40589 loss 0.2547 acc 0.951 lb 1.952 z 0.891 lr 1.62e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.889 max_load(range) 0.098-0.207 step 12480/40589 loss 0.2438 acc 0.953 lb 1.945 z 0.957 lr 1.62e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12490/40589 loss 0.2441 acc 0.953 lb 1.941 z 0.888 lr 1.62e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12500/40589 loss 0.2640 acc 0.949 lb 1.953 z 0.813 lr 1.62e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.930 max_load(range) 0.062-0.199 step 12510/40589 loss 0.2498 acc 0.953 lb 1.951 z 0.854 lr 1.62e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12520/40589 loss 0.2444 acc 0.953 lb 1.944 z 0.988 lr 1.62e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.916 max_load(range) 0.055-0.312 step 12530/40589 loss 0.2431 acc 0.953 lb 1.949 z 0.926 lr 1.62e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12540/40589 loss 0.2574 acc 0.950 lb 1.948 z 0.835 lr 1.62e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12550/40589 loss 0.2497 acc 0.953 lb 1.957 z 0.840 lr 1.62e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.937 max_load(range) 0.054-0.409 step 12560/40589 loss 0.2475 acc 0.953 lb 1.944 z 0.987 lr 1.62e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12570/40589 loss 0.2441 acc 0.953 lb 1.936 z 0.894 lr 1.62e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.892 max_load(range) 0.082-0.221 step 12580/40589 loss 0.2627 acc 0.949 lb 1.946 z 0.796 lr 1.62e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12590/40589 loss 0.2488 acc 0.953 lb 1.951 z 0.863 lr 1.62e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12600/40589 loss 0.2496 acc 0.952 lb 1.943 z 0.983 lr 1.62e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.885 max_load(range) 0.070-0.201 step 12610/40589 loss 0.2461 acc 0.953 lb 1.943 z 0.941 lr 1.61e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12620/40589 loss 0.2612 acc 0.949 lb 1.961 z 0.836 lr 1.61e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.927 max_load(range) 0.059-0.224 step 12630/40589 loss 0.2502 acc 0.953 lb 1.946 z 0.818 lr 1.61e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12640/40589 loss 0.2460 acc 0.953 lb 1.939 z 0.982 lr 1.61e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12650/40589 loss 0.2436 acc 0.953 lb 1.946 z 0.886 lr 1.61e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.934 max_load(range) 0.052-0.239 step 12660/40589 loss 0.2640 acc 0.949 lb 1.955 z 0.839 lr 1.61e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12670/40589 loss 0.2533 acc 0.953 lb 1.950 z 0.832 lr 1.61e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.916 max_load(range) 0.080-0.167 step 12680/40589 loss 0.2490 acc 0.952 lb 1.943 z 0.896 lr 1.61e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12690/40589 loss 0.2459 acc 0.953 lb 1.950 z 0.943 lr 1.61e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12700/40589 loss 0.2600 acc 0.951 lb 1.950 z 0.810 lr 1.61e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.938 max_load(range) 0.066-0.213 step 12710/40589 loss 0.2521 acc 0.953 lb 1.949 z 0.861 lr 1.61e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12720/40589 loss 0.2421 acc 0.953 lb 1.936 z 0.947 lr 1.61e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.913 max_load(range) 0.062-0.307 step 12730/40589 loss 0.2423 acc 0.954 lb 1.945 z 0.931 lr 1.61e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12740/40589 loss 0.2614 acc 0.950 lb 1.962 z 0.788 lr 1.61e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12750/40589 loss 0.2489 acc 0.953 lb 1.949 z 0.834 lr 1.61e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.935 max_load(range) 0.050-0.423 step 12760/40589 loss 0.2459 acc 0.953 lb 1.949 z 1.000 lr 1.60e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12770/40589 loss 0.2428 acc 0.953 lb 1.949 z 0.930 lr 1.60e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.896 max_load(range) 0.083-0.160 step 12780/40589 loss 0.2602 acc 0.950 lb 1.950 z 0.784 lr 1.60e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12790/40589 loss 0.2495 acc 0.953 lb 1.950 z 0.838 lr 1.60e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12800/40589 loss 0.2455 acc 0.953 lb 1.938 z 0.868 lr 1.60e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.892 max_load(range) 0.085-0.196 step 12810/40589 loss 0.2428 acc 0.954 lb 1.951 z 0.898 lr 1.60e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12820/40589 loss 0.2582 acc 0.950 lb 1.954 z 0.819 lr 1.60e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.923 max_load(range) 0.062-0.207 step 12830/40589 loss 0.2507 acc 0.953 lb 1.955 z 0.845 lr 1.60e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12840/40589 loss 0.2468 acc 0.953 lb 1.952 z 0.954 lr 1.60e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12850/40589 loss 0.2437 acc 0.953 lb 1.948 z 0.932 lr 1.60e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.921 max_load(range) 0.058-0.254 step 12860/40589 loss 0.2620 acc 0.949 lb 1.958 z 0.863 lr 1.60e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12870/40589 loss 0.2508 acc 0.952 lb 1.957 z 0.828 lr 1.60e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.894 max_load(range) 0.076-0.181 step 12880/40589 loss 0.2452 acc 0.953 lb 1.952 z 0.943 lr 1.60e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12890/40589 loss 0.2477 acc 0.953 lb 1.943 z 0.898 lr 1.60e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12900/40589 loss 0.2574 acc 0.951 lb 1.953 z 0.760 lr 1.60e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.925 max_load(range) 0.075-0.184 step 12910/40589 loss 0.2521 acc 0.953 lb 1.954 z 0.827 lr 1.60e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12920/40589 loss 0.2478 acc 0.952 lb 1.942 z 0.930 lr 1.59e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.912 max_load(range) 0.055-0.298 step 12930/40589 loss 0.2435 acc 0.953 lb 1.948 z 0.946 lr 1.59e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12940/40589 loss 0.2598 acc 0.950 lb 1.957 z 0.794 lr 1.59e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12950/40589 loss 0.2479 acc 0.953 lb 1.956 z 0.808 lr 1.59e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.935 max_load(range) 0.055-0.418 step 12960/40589 loss 0.2472 acc 0.952 lb 1.941 z 0.919 lr 1.59e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12970/40589 loss 0.2421 acc 0.954 lb 1.954 z 0.879 lr 1.59e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.880 max_load(range) 0.083-0.173 step 12980/40589 loss 0.2579 acc 0.951 lb 1.959 z 0.789 lr 1.59e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 12990/40589 loss 0.2493 acc 0.953 lb 1.957 z 0.832 lr 1.59e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 13000/40589 loss 0.2455 acc 0.953 lb 1.950 z 0.985 lr 1.59e-04 | 1426 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.889 max_load(range) 0.074-0.249 [save] runs/pretrain_flagship_h100_4x8/step13000 step 13010/40589 loss 0.2423 acc 0.953 lb 1.951 z 0.901 lr 1.59e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 13020/40589 loss 0.2589 acc 0.950 lb 1.958 z 0.794 lr 1.59e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.924 max_load(range) 0.054-0.212 step 13030/40589 loss 0.2500 acc 0.953 lb 1.950 z 0.795 lr 1.59e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 13040/40589 loss 0.2477 acc 0.952 lb 1.943 z 0.902 lr 1.59e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 13050/40589 loss 0.2421 acc 0.953 lb 1.942 z 0.868 lr 1.59e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.930 max_load(range) 0.061-0.210 step 13060/40589 loss 0.2561 acc 0.951 lb 1.951 z 0.764 lr 1.59e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 13070/40589 loss 0.2479 acc 0.953 lb 1.953 z 0.837 lr 1.59e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.887 max_load(range) 0.084-0.221 step 13080/40589 loss 0.2437 acc 0.953 lb 1.942 z 0.880 lr 1.58e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 13090/40589 loss 0.2403 acc 0.954 lb 1.944 z 0.908 lr 1.58e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 13100/40589 loss 0.2597 acc 0.950 lb 1.954 z 0.815 lr 1.58e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.935 max_load(range) 0.058-0.193 step 13110/40589 loss 0.2469 acc 0.954 lb 1.953 z 0.829 lr 1.58e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 13120/40589 loss 0.2445 acc 0.953 lb 1.938 z 0.894 lr 1.58e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.917 max_load(range) 0.060-0.322 step 13130/40589 loss 0.2396 acc 0.954 lb 1.941 z 0.830 lr 1.58e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 13140/40589 loss 0.2563 acc 0.951 lb 1.946 z 0.784 lr 1.58e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left step 13150/40589 loss 0.2495 acc 0.953 lb 1.951 z 0.793 lr 1.58e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.7d left [routing] mean entropy_norm 0.939 max_load(range) 0.053-0.406 step 13160/40589 loss 0.2488 acc 0.952 lb 1.936 z 0.907 lr 1.58e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13170/40589 loss 0.2398 acc 0.954 lb 1.942 z 0.868 lr 1.58e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.897 max_load(range) 0.082-0.166 step 13180/40589 loss 0.2623 acc 0.949 lb 1.950 z 0.754 lr 1.58e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13190/40589 loss 0.2468 acc 0.953 lb 1.953 z 0.824 lr 1.58e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13200/40589 loss 0.2433 acc 0.953 lb 1.943 z 0.916 lr 1.58e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.889 max_load(range) 0.087-0.214 step 13210/40589 loss 0.2400 acc 0.954 lb 1.946 z 0.821 lr 1.58e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13220/40589 loss 0.2567 acc 0.951 lb 1.953 z 0.778 lr 1.58e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.930 max_load(range) 0.056-0.235 step 13230/40589 loss 0.2501 acc 0.953 lb 1.953 z 0.767 lr 1.57e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13240/40589 loss 0.2445 acc 0.954 lb 1.936 z 0.892 lr 1.57e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13250/40589 loss 0.2445 acc 0.953 lb 1.947 z 0.907 lr 1.57e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.936 max_load(range) 0.063-0.262 step 13260/40589 loss 0.2577 acc 0.950 lb 1.954 z 0.747 lr 1.57e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13270/40589 loss 0.2485 acc 0.953 lb 1.953 z 0.795 lr 1.57e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.894 max_load(range) 0.074-0.187 step 13280/40589 loss 0.2459 acc 0.953 lb 1.947 z 0.921 lr 1.57e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13290/40589 loss 0.2407 acc 0.954 lb 1.942 z 0.886 lr 1.57e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13300/40589 loss 0.2576 acc 0.951 lb 1.963 z 0.752 lr 1.57e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.931 max_load(range) 0.071-0.183 step 13310/40589 loss 0.2494 acc 0.953 lb 1.952 z 0.782 lr 1.57e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13320/40589 loss 0.2445 acc 0.953 lb 1.936 z 0.911 lr 1.57e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.926 max_load(range) 0.063-0.255 step 13330/40589 loss 0.2384 acc 0.954 lb 1.942 z 0.890 lr 1.57e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13340/40589 loss 0.2555 acc 0.951 lb 1.960 z 0.794 lr 1.57e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13350/40589 loss 0.2477 acc 0.953 lb 1.957 z 0.820 lr 1.57e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.942 max_load(range) 0.053-0.404 step 13360/40589 loss 0.2402 acc 0.954 lb 1.948 z 0.904 lr 1.57e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13370/40589 loss 0.2393 acc 0.954 lb 1.949 z 0.824 lr 1.57e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.909 max_load(range) 0.067-0.187 step 13380/40589 loss 0.2574 acc 0.950 lb 1.954 z 0.774 lr 1.56e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13390/40589 loss 0.2461 acc 0.954 lb 1.945 z 0.810 lr 1.56e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13400/40589 loss 0.2458 acc 0.952 lb 1.943 z 0.872 lr 1.56e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.895 max_load(range) 0.070-0.199 step 13410/40589 loss 0.2366 acc 0.954 lb 1.945 z 0.836 lr 1.56e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13420/40589 loss 0.2591 acc 0.949 lb 1.950 z 0.771 lr 1.56e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.934 max_load(range) 0.054-0.244 step 13430/40589 loss 0.2466 acc 0.953 lb 1.949 z 0.798 lr 1.56e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13440/40589 loss 0.2446 acc 0.953 lb 1.940 z 0.935 lr 1.56e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13450/40589 loss 0.2397 acc 0.954 lb 1.944 z 0.834 lr 1.56e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.937 max_load(range) 0.056-0.247 step 13460/40589 loss 0.2564 acc 0.950 lb 1.953 z 0.728 lr 1.56e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13470/40589 loss 0.2443 acc 0.954 lb 1.944 z 0.771 lr 1.56e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.915 max_load(range) 0.079-0.148 step 13480/40589 loss 0.2424 acc 0.953 lb 1.943 z 0.913 lr 1.56e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13490/40589 loss 0.2396 acc 0.954 lb 1.945 z 0.877 lr 1.56e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13500/40589 loss 0.2575 acc 0.950 lb 1.961 z 0.747 lr 1.56e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.939 max_load(range) 0.065-0.202 step 13510/40589 loss 0.2453 acc 0.954 lb 1.949 z 0.805 lr 1.56e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13520/40589 loss 0.2430 acc 0.954 lb 1.939 z 0.896 lr 1.56e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.929 max_load(range) 0.061-0.278 step 13530/40589 loss 0.2348 acc 0.955 lb 1.942 z 0.857 lr 1.55e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13540/40589 loss 0.2599 acc 0.950 lb 1.953 z 0.760 lr 1.55e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13550/40589 loss 0.2473 acc 0.953 lb 1.951 z 0.781 lr 1.55e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.938 max_load(range) 0.048-0.408 step 13560/40589 loss 0.2398 acc 0.954 lb 1.943 z 0.860 lr 1.55e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13570/40589 loss 0.2354 acc 0.955 lb 1.948 z 0.857 lr 1.55e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.902 max_load(range) 0.074-0.167 step 13580/40589 loss 0.2535 acc 0.951 lb 1.953 z 0.768 lr 1.55e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13590/40589 loss 0.2466 acc 0.953 lb 1.950 z 0.796 lr 1.55e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13600/40589 loss 0.2403 acc 0.954 lb 1.937 z 0.887 lr 1.55e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.896 max_load(range) 0.072-0.220 step 13610/40589 loss 0.2370 acc 0.955 lb 1.950 z 0.858 lr 1.55e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13620/40589 loss 0.2552 acc 0.951 lb 1.954 z 0.788 lr 1.55e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.923 max_load(range) 0.053-0.248 step 13630/40589 loss 0.2459 acc 0.953 lb 1.951 z 0.738 lr 1.55e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13640/40589 loss 0.2417 acc 0.954 lb 1.949 z 0.881 lr 1.55e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13650/40589 loss 0.2335 acc 0.955 lb 1.944 z 0.840 lr 1.55e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.930 max_load(range) 0.064-0.237 step 13660/40589 loss 0.2550 acc 0.951 lb 1.955 z 0.728 lr 1.55e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13670/40589 loss 0.2463 acc 0.953 lb 1.954 z 0.774 lr 1.55e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.896 max_load(range) 0.073-0.220 step 13680/40589 loss 0.2422 acc 0.953 lb 1.942 z 0.893 lr 1.54e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13690/40589 loss 0.2361 acc 0.955 lb 1.943 z 0.843 lr 1.54e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13700/40589 loss 0.2541 acc 0.951 lb 1.959 z 0.751 lr 1.54e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.932 max_load(range) 0.069-0.192 step 13710/40589 loss 0.2444 acc 0.954 lb 1.948 z 0.721 lr 1.54e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13720/40589 loss 0.2428 acc 0.953 lb 1.946 z 0.881 lr 1.54e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.914 max_load(range) 0.060-0.300 step 13730/40589 loss 0.2351 acc 0.955 lb 1.948 z 0.814 lr 1.54e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13740/40589 loss 0.2534 acc 0.952 lb 1.952 z 0.711 lr 1.54e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13750/40589 loss 0.2451 acc 0.954 lb 1.959 z 0.792 lr 1.54e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.938 max_load(range) 0.050-0.433 step 13760/40589 loss 0.2385 acc 0.955 lb 1.937 z 0.867 lr 1.54e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13770/40589 loss 0.2358 acc 0.956 lb 1.948 z 0.830 lr 1.54e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.902 max_load(range) 0.070-0.181 step 13780/40589 loss 0.2522 acc 0.952 lb 1.955 z 0.770 lr 1.54e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13790/40589 loss 0.2451 acc 0.954 lb 1.950 z 0.751 lr 1.54e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13800/40589 loss 0.2428 acc 0.953 lb 1.952 z 0.924 lr 1.54e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.885 max_load(range) 0.069-0.298 step 13810/40589 loss 0.2340 acc 0.955 lb 1.940 z 0.823 lr 1.54e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13820/40589 loss 0.2545 acc 0.951 lb 1.957 z 0.741 lr 1.54e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.925 max_load(range) 0.059-0.217 step 13830/40589 loss 0.2462 acc 0.954 lb 1.950 z 0.752 lr 1.53e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13840/40589 loss 0.2447 acc 0.953 lb 1.942 z 0.864 lr 1.53e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13850/40589 loss 0.2352 acc 0.955 lb 1.948 z 0.832 lr 1.53e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.925 max_load(range) 0.055-0.223 step 13860/40589 loss 0.2524 acc 0.952 lb 1.960 z 0.734 lr 1.53e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13870/40589 loss 0.2424 acc 0.955 lb 1.951 z 0.737 lr 1.53e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.881 max_load(range) 0.090-0.199 step 13880/40589 loss 0.2422 acc 0.954 lb 1.945 z 0.861 lr 1.53e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13890/40589 loss 0.2373 acc 0.955 lb 1.945 z 0.827 lr 1.53e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13900/40589 loss 0.2546 acc 0.951 lb 1.955 z 0.734 lr 1.53e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.927 max_load(range) 0.075-0.211 step 13910/40589 loss 0.2436 acc 0.954 lb 1.950 z 0.747 lr 1.53e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left step 13920/40589 loss 0.2424 acc 0.954 lb 1.942 z 0.815 lr 1.53e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.6d left [routing] mean entropy_norm 0.915 max_load(range) 0.057-0.282 step 13930/40589 loss 0.2348 acc 0.955 lb 1.954 z 0.836 lr 1.53e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 13940/40589 loss 0.2592 acc 0.950 lb 1.963 z 0.774 lr 1.53e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 13950/40589 loss 0.2422 acc 0.954 lb 1.956 z 0.747 lr 1.53e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.936 max_load(range) 0.056-0.406 step 13960/40589 loss 0.2412 acc 0.954 lb 1.946 z 0.817 lr 1.53e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 13970/40589 loss 0.2338 acc 0.955 lb 1.951 z 0.841 lr 1.53e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.892 max_load(range) 0.069-0.196 step 13980/40589 loss 0.2532 acc 0.952 lb 1.964 z 0.753 lr 1.52e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 13990/40589 loss 0.2446 acc 0.954 lb 1.958 z 0.763 lr 1.52e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14000/40589 loss 0.2440 acc 0.953 lb 1.942 z 0.833 lr 1.52e-04 | 1425 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.880 max_load(range) 0.072-0.182 [save] runs/pretrain_flagship_h100_4x8/step14000 step 14010/40589 loss 0.2355 acc 0.956 lb 1.947 z 0.830 lr 1.52e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14020/40589 loss 0.2537 acc 0.951 lb 1.962 z 0.761 lr 1.52e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.929 max_load(range) 0.045-0.230 step 14030/40589 loss 0.2447 acc 0.954 lb 1.952 z 0.732 lr 1.52e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14040/40589 loss 0.2446 acc 0.953 lb 1.949 z 0.874 lr 1.52e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14050/40589 loss 0.2357 acc 0.955 lb 1.943 z 0.820 lr 1.52e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.931 max_load(range) 0.048-0.201 step 14060/40589 loss 0.2551 acc 0.951 lb 1.955 z 0.724 lr 1.52e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14070/40589 loss 0.2437 acc 0.954 lb 1.952 z 0.746 lr 1.52e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.892 max_load(range) 0.089-0.173 step 14080/40589 loss 0.2408 acc 0.954 lb 1.945 z 0.814 lr 1.52e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14090/40589 loss 0.2317 acc 0.955 lb 1.951 z 0.846 lr 1.52e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14100/40589 loss 0.2535 acc 0.952 lb 1.953 z 0.749 lr 1.52e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.937 max_load(range) 0.061-0.204 step 14110/40589 loss 0.2419 acc 0.954 lb 1.948 z 0.729 lr 1.52e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14120/40589 loss 0.2402 acc 0.954 lb 1.947 z 0.885 lr 1.52e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.928 max_load(range) 0.060-0.248 step 14130/40589 loss 0.2337 acc 0.956 lb 1.944 z 0.821 lr 1.51e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14140/40589 loss 0.2534 acc 0.951 lb 1.962 z 0.731 lr 1.51e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14150/40589 loss 0.2461 acc 0.954 lb 1.955 z 0.780 lr 1.51e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.943 max_load(range) 0.052-0.402 step 14160/40589 loss 0.2410 acc 0.954 lb 1.947 z 0.861 lr 1.51e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14170/40589 loss 0.2353 acc 0.956 lb 1.950 z 0.798 lr 1.51e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.880 max_load(range) 0.083-0.186 step 14180/40589 loss 0.2526 acc 0.951 lb 1.969 z 0.720 lr 1.51e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14190/40589 loss 0.2460 acc 0.954 lb 1.957 z 0.768 lr 1.51e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14200/40589 loss 0.2418 acc 0.954 lb 1.945 z 0.872 lr 1.51e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.888 max_load(range) 0.093-0.191 step 14210/40589 loss 0.2353 acc 0.955 lb 1.948 z 0.825 lr 1.51e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14220/40589 loss 0.2498 acc 0.952 lb 1.954 z 0.725 lr 1.51e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.932 max_load(range) 0.053-0.230 step 14230/40589 loss 0.2423 acc 0.954 lb 1.960 z 0.716 lr 1.51e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14240/40589 loss 0.2401 acc 0.954 lb 1.949 z 0.862 lr 1.51e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14250/40589 loss 0.2313 acc 0.956 lb 1.950 z 0.862 lr 1.51e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.938 max_load(range) 0.055-0.236 step 14260/40589 loss 0.2520 acc 0.952 lb 1.963 z 0.712 lr 1.51e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14270/40589 loss 0.2451 acc 0.954 lb 1.958 z 0.713 lr 1.50e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.908 max_load(range) 0.081-0.158 step 14280/40589 loss 0.2398 acc 0.954 lb 1.953 z 0.903 lr 1.50e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14290/40589 loss 0.2351 acc 0.955 lb 1.944 z 0.819 lr 1.50e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14300/40589 loss 0.2494 acc 0.952 lb 1.959 z 0.730 lr 1.50e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.945 max_load(range) 0.065-0.209 step 14310/40589 loss 0.2452 acc 0.954 lb 1.954 z 0.749 lr 1.50e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14320/40589 loss 0.2379 acc 0.954 lb 1.948 z 0.885 lr 1.50e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.924 max_load(range) 0.059-0.286 step 14330/40589 loss 0.2342 acc 0.955 lb 1.953 z 0.845 lr 1.50e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14340/40589 loss 0.2493 acc 0.953 lb 1.952 z 0.705 lr 1.50e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14350/40589 loss 0.2421 acc 0.955 lb 1.948 z 0.722 lr 1.50e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.940 max_load(range) 0.051-0.427 step 14360/40589 loss 0.2390 acc 0.954 lb 1.947 z 0.882 lr 1.50e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14370/40589 loss 0.2348 acc 0.955 lb 1.946 z 0.836 lr 1.50e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.904 max_load(range) 0.079-0.192 step 14380/40589 loss 0.2520 acc 0.952 lb 1.956 z 0.724 lr 1.50e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14390/40589 loss 0.2444 acc 0.953 lb 1.955 z 0.742 lr 1.50e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14400/40589 loss 0.2396 acc 0.954 lb 1.949 z 0.855 lr 1.50e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.900 max_load(range) 0.068-0.284 step 14410/40589 loss 0.2365 acc 0.954 lb 1.949 z 0.815 lr 1.50e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14420/40589 loss 0.2541 acc 0.951 lb 1.957 z 0.725 lr 1.49e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.929 max_load(range) 0.057-0.215 step 14430/40589 loss 0.2420 acc 0.954 lb 1.953 z 0.756 lr 1.49e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14440/40589 loss 0.2396 acc 0.954 lb 1.951 z 0.847 lr 1.49e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14450/40589 loss 0.2368 acc 0.955 lb 1.947 z 0.825 lr 1.49e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.929 max_load(range) 0.056-0.236 step 14460/40589 loss 0.2518 acc 0.952 lb 1.961 z 0.700 lr 1.49e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14470/40589 loss 0.2419 acc 0.954 lb 1.953 z 0.732 lr 1.49e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.899 max_load(range) 0.089-0.189 step 14480/40589 loss 0.2378 acc 0.955 lb 1.950 z 0.840 lr 1.49e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14490/40589 loss 0.2314 acc 0.956 lb 1.949 z 0.767 lr 1.49e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14500/40589 loss 0.2487 acc 0.953 lb 1.956 z 0.729 lr 1.49e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.929 max_load(range) 0.066-0.197 step 14510/40589 loss 0.2406 acc 0.955 lb 1.954 z 0.728 lr 1.49e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14520/40589 loss 0.2414 acc 0.954 lb 1.940 z 0.822 lr 1.49e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.922 max_load(range) 0.064-0.280 step 14530/40589 loss 0.2357 acc 0.955 lb 1.951 z 0.843 lr 1.49e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14540/40589 loss 0.2505 acc 0.952 lb 1.956 z 0.688 lr 1.49e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14550/40589 loss 0.2442 acc 0.954 lb 1.949 z 0.677 lr 1.49e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.942 max_load(range) 0.055-0.400 step 14560/40589 loss 0.2396 acc 0.955 lb 1.942 z 0.825 lr 1.48e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14570/40589 loss 0.2368 acc 0.955 lb 1.954 z 0.795 lr 1.48e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.906 max_load(range) 0.067-0.179 step 14580/40589 loss 0.2528 acc 0.952 lb 1.956 z 0.712 lr 1.48e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14590/40589 loss 0.2442 acc 0.954 lb 1.955 z 0.703 lr 1.48e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14600/40589 loss 0.2446 acc 0.953 lb 1.952 z 0.874 lr 1.48e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.893 max_load(range) 0.076-0.179 step 14610/40589 loss 0.2356 acc 0.955 lb 1.948 z 0.798 lr 1.48e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14620/40589 loss 0.2499 acc 0.952 lb 1.963 z 0.669 lr 1.48e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.928 max_load(range) 0.058-0.225 step 14630/40589 loss 0.2404 acc 0.955 lb 1.947 z 0.732 lr 1.48e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14640/40589 loss 0.2358 acc 0.955 lb 1.943 z 0.795 lr 1.48e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14650/40589 loss 0.2335 acc 0.955 lb 1.953 z 0.805 lr 1.48e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.924 max_load(range) 0.059-0.257 step 14660/40589 loss 0.2530 acc 0.952 lb 1.963 z 0.732 lr 1.48e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14670/40589 loss 0.2393 acc 0.955 lb 1.950 z 0.705 lr 1.48e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left [routing] mean entropy_norm 0.883 max_load(range) 0.089-0.234 step 14680/40589 loss 0.2407 acc 0.954 lb 1.945 z 0.773 lr 1.48e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.5d left step 14690/40589 loss 0.2351 acc 0.956 lb 1.953 z 0.813 lr 1.48e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 14700/40589 loss 0.2521 acc 0.952 lb 1.953 z 0.665 lr 1.47e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.928 max_load(range) 0.067-0.213 step 14710/40589 loss 0.2410 acc 0.955 lb 1.959 z 0.700 lr 1.47e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 14720/40589 loss 0.2422 acc 0.954 lb 1.952 z 0.841 lr 1.47e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.923 max_load(range) 0.056-0.260 step 14730/40589 loss 0.2366 acc 0.955 lb 1.952 z 0.794 lr 1.47e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 14740/40589 loss 0.2512 acc 0.952 lb 1.961 z 0.715 lr 1.47e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 14750/40589 loss 0.2380 acc 0.955 lb 1.952 z 0.738 lr 1.47e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.939 max_load(range) 0.051-0.400 step 14760/40589 loss 0.2366 acc 0.955 lb 1.940 z 0.802 lr 1.47e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 14770/40589 loss 0.2334 acc 0.956 lb 1.957 z 0.801 lr 1.47e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.882 max_load(range) 0.069-0.186 step 14780/40589 loss 0.2537 acc 0.951 lb 1.968 z 0.705 lr 1.47e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 14790/40589 loss 0.2445 acc 0.954 lb 1.959 z 0.703 lr 1.47e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 14800/40589 loss 0.2417 acc 0.953 lb 1.950 z 0.851 lr 1.47e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.876 max_load(range) 0.079-0.212 step 14810/40589 loss 0.2365 acc 0.955 lb 1.956 z 0.755 lr 1.47e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 14820/40589 loss 0.2506 acc 0.952 lb 1.962 z 0.653 lr 1.47e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.921 max_load(range) 0.054-0.241 step 14830/40589 loss 0.2400 acc 0.955 lb 1.958 z 0.692 lr 1.47e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 14840/40589 loss 0.2393 acc 0.954 lb 1.945 z 0.791 lr 1.47e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 14850/40589 loss 0.2351 acc 0.955 lb 1.952 z 0.795 lr 1.46e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.928 max_load(range) 0.053-0.242 step 14860/40589 loss 0.2512 acc 0.952 lb 1.959 z 0.675 lr 1.46e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 14870/40589 loss 0.2405 acc 0.955 lb 1.952 z 0.729 lr 1.46e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.881 max_load(range) 0.082-0.202 step 14880/40589 loss 0.2415 acc 0.954 lb 1.947 z 0.840 lr 1.46e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 14890/40589 loss 0.2337 acc 0.955 lb 1.947 z 0.771 lr 1.46e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 14900/40589 loss 0.2500 acc 0.953 lb 1.957 z 0.677 lr 1.46e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.941 max_load(range) 0.062-0.209 step 14910/40589 loss 0.2407 acc 0.955 lb 1.951 z 0.697 lr 1.46e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 14920/40589 loss 0.2401 acc 0.954 lb 1.941 z 0.796 lr 1.46e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.910 max_load(range) 0.060-0.303 step 14930/40589 loss 0.2289 acc 0.957 lb 1.951 z 0.771 lr 1.46e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 14940/40589 loss 0.2488 acc 0.953 lb 1.963 z 0.670 lr 1.46e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 14950/40589 loss 0.2407 acc 0.955 lb 1.962 z 0.713 lr 1.46e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.941 max_load(range) 0.051-0.417 step 14960/40589 loss 0.2372 acc 0.955 lb 1.939 z 0.772 lr 1.46e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 14970/40589 loss 0.2340 acc 0.956 lb 1.952 z 0.795 lr 1.46e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.894 max_load(range) 0.087-0.190 step 14980/40589 loss 0.2480 acc 0.953 lb 1.955 z 0.672 lr 1.46e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 14990/40589 loss 0.2405 acc 0.955 lb 1.954 z 0.669 lr 1.45e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15000/40589 loss 0.2396 acc 0.954 lb 1.945 z 0.814 lr 1.45e-04 | 1424 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.894 max_load(range) 0.069-0.209 [save] runs/pretrain_flagship_h100_4x8/step15000 step 15010/40589 loss 0.2319 acc 0.956 lb 1.951 z 0.789 lr 1.45e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15020/40589 loss 0.2487 acc 0.952 lb 1.960 z 0.699 lr 1.45e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.937 max_load(range) 0.054-0.212 step 15030/40589 loss 0.2425 acc 0.955 lb 1.948 z 0.664 lr 1.45e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15040/40589 loss 0.2390 acc 0.954 lb 1.943 z 0.839 lr 1.45e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15050/40589 loss 0.2311 acc 0.956 lb 1.953 z 0.823 lr 1.45e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.939 max_load(range) 0.059-0.232 step 15060/40589 loss 0.2506 acc 0.952 lb 1.957 z 0.678 lr 1.45e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15070/40589 loss 0.2425 acc 0.955 lb 1.955 z 0.690 lr 1.45e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.889 max_load(range) 0.080-0.201 step 15080/40589 loss 0.2374 acc 0.955 lb 1.950 z 0.768 lr 1.45e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15090/40589 loss 0.2314 acc 0.956 lb 1.951 z 0.740 lr 1.45e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15100/40589 loss 0.2470 acc 0.952 lb 1.947 z 0.678 lr 1.45e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.932 max_load(range) 0.073-0.191 step 15110/40589 loss 0.2385 acc 0.955 lb 1.952 z 0.685 lr 1.45e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15120/40589 loss 0.2398 acc 0.955 lb 1.950 z 0.829 lr 1.45e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.930 max_load(range) 0.051-0.260 step 15130/40589 loss 0.2309 acc 0.956 lb 1.948 z 0.771 lr 1.44e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15140/40589 loss 0.2461 acc 0.954 lb 1.949 z 0.656 lr 1.44e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15150/40589 loss 0.2439 acc 0.954 lb 1.954 z 0.685 lr 1.44e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.941 max_load(range) 0.052-0.414 step 15160/40589 loss 0.2403 acc 0.954 lb 1.947 z 0.842 lr 1.44e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15170/40589 loss 0.2318 acc 0.955 lb 1.951 z 0.803 lr 1.44e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.903 max_load(range) 0.078-0.181 step 15180/40589 loss 0.2477 acc 0.953 lb 1.962 z 0.682 lr 1.44e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15190/40589 loss 0.2404 acc 0.955 lb 1.956 z 0.641 lr 1.44e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15200/40589 loss 0.2387 acc 0.954 lb 1.944 z 0.798 lr 1.44e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.902 max_load(range) 0.067-0.188 step 15210/40589 loss 0.2351 acc 0.956 lb 1.945 z 0.739 lr 1.44e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15220/40589 loss 0.2450 acc 0.953 lb 1.967 z 0.688 lr 1.44e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.927 max_load(range) 0.065-0.195 step 15230/40589 loss 0.2377 acc 0.956 lb 1.947 z 0.659 lr 1.44e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15240/40589 loss 0.2403 acc 0.955 lb 1.952 z 0.819 lr 1.44e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15250/40589 loss 0.2298 acc 0.956 lb 1.956 z 0.785 lr 1.44e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.932 max_load(range) 0.055-0.240 step 15260/40589 loss 0.2492 acc 0.952 lb 1.961 z 0.669 lr 1.44e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15270/40589 loss 0.2396 acc 0.955 lb 1.953 z 0.677 lr 1.43e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.896 max_load(range) 0.074-0.212 step 15280/40589 loss 0.2391 acc 0.954 lb 1.950 z 0.786 lr 1.43e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15290/40589 loss 0.2321 acc 0.956 lb 1.952 z 0.750 lr 1.43e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15300/40589 loss 0.2490 acc 0.952 lb 1.961 z 0.685 lr 1.43e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.938 max_load(range) 0.059-0.216 step 15310/40589 loss 0.2407 acc 0.955 lb 1.959 z 0.698 lr 1.43e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15320/40589 loss 0.2386 acc 0.955 lb 1.943 z 0.796 lr 1.43e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.915 max_load(range) 0.058-0.311 step 15330/40589 loss 0.2321 acc 0.956 lb 1.956 z 0.784 lr 1.43e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15340/40589 loss 0.2468 acc 0.953 lb 1.956 z 0.669 lr 1.43e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15350/40589 loss 0.2356 acc 0.956 lb 1.964 z 0.690 lr 1.43e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.945 max_load(range) 0.050-0.390 step 15360/40589 loss 0.2367 acc 0.955 lb 1.949 z 0.772 lr 1.43e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15370/40589 loss 0.2321 acc 0.956 lb 1.948 z 0.763 lr 1.43e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.894 max_load(range) 0.083-0.185 step 15380/40589 loss 0.2490 acc 0.952 lb 1.960 z 0.653 lr 1.43e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15390/40589 loss 0.2402 acc 0.955 lb 1.956 z 0.697 lr 1.43e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15400/40589 loss 0.2374 acc 0.955 lb 1.944 z 0.764 lr 1.43e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.892 max_load(range) 0.092-0.205 step 15410/40589 loss 0.2326 acc 0.956 lb 1.944 z 0.738 lr 1.42e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15420/40589 loss 0.2447 acc 0.953 lb 1.953 z 0.660 lr 1.42e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left [routing] mean entropy_norm 0.931 max_load(range) 0.054-0.226 step 15430/40589 loss 0.2385 acc 0.955 lb 1.955 z 0.649 lr 1.42e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15440/40589 loss 0.2403 acc 0.954 lb 1.945 z 0.790 lr 1.42e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.4d left step 15450/40589 loss 0.2303 acc 0.956 lb 1.944 z 0.788 lr 1.42e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.939 max_load(range) 0.053-0.234 step 15460/40589 loss 0.2457 acc 0.954 lb 1.969 z 0.685 lr 1.42e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15470/40589 loss 0.2390 acc 0.955 lb 1.961 z 0.671 lr 1.42e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.893 max_load(range) 0.081-0.205 step 15480/40589 loss 0.2407 acc 0.954 lb 1.944 z 0.784 lr 1.42e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15490/40589 loss 0.2300 acc 0.957 lb 1.958 z 0.733 lr 1.42e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15500/40589 loss 0.2454 acc 0.953 lb 1.956 z 0.627 lr 1.42e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.942 max_load(range) 0.060-0.214 step 15510/40589 loss 0.2370 acc 0.955 lb 1.958 z 0.704 lr 1.42e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15520/40589 loss 0.2403 acc 0.954 lb 1.942 z 0.777 lr 1.42e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.919 max_load(range) 0.056-0.301 step 15530/40589 loss 0.2330 acc 0.955 lb 1.955 z 0.774 lr 1.42e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15540/40589 loss 0.2447 acc 0.954 lb 1.963 z 0.667 lr 1.41e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15550/40589 loss 0.2388 acc 0.955 lb 1.955 z 0.690 lr 1.41e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.941 max_load(range) 0.053-0.415 step 15560/40589 loss 0.2401 acc 0.954 lb 1.952 z 0.775 lr 1.41e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15570/40589 loss 0.2324 acc 0.956 lb 1.954 z 0.753 lr 1.41e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.903 max_load(range) 0.075-0.186 step 15580/40589 loss 0.2492 acc 0.953 lb 1.957 z 0.668 lr 1.41e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15590/40589 loss 0.2392 acc 0.955 lb 1.952 z 0.711 lr 1.41e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15600/40589 loss 0.2381 acc 0.954 lb 1.948 z 0.785 lr 1.41e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.895 max_load(range) 0.070-0.186 step 15610/40589 loss 0.2336 acc 0.956 lb 1.951 z 0.738 lr 1.41e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15620/40589 loss 0.2454 acc 0.953 lb 1.963 z 0.666 lr 1.41e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.930 max_load(range) 0.056-0.234 step 15630/40589 loss 0.2383 acc 0.955 lb 1.955 z 0.710 lr 1.41e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15640/40589 loss 0.2394 acc 0.954 lb 1.956 z 0.767 lr 1.41e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15650/40589 loss 0.2313 acc 0.957 lb 1.956 z 0.757 lr 1.41e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.925 max_load(range) 0.058-0.251 step 15660/40589 loss 0.2453 acc 0.953 lb 1.959 z 0.690 lr 1.41e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15670/40589 loss 0.2384 acc 0.955 lb 1.956 z 0.680 lr 1.41e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.890 max_load(range) 0.083-0.186 step 15680/40589 loss 0.2394 acc 0.954 lb 1.949 z 0.754 lr 1.40e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15690/40589 loss 0.2357 acc 0.956 lb 1.952 z 0.747 lr 1.40e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15700/40589 loss 0.2459 acc 0.953 lb 1.962 z 0.676 lr 1.40e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.938 max_load(range) 0.067-0.207 step 15710/40589 loss 0.2411 acc 0.955 lb 1.958 z 0.665 lr 1.40e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15720/40589 loss 0.2389 acc 0.955 lb 1.950 z 0.771 lr 1.40e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.930 max_load(range) 0.059-0.244 step 15730/40589 loss 0.2322 acc 0.956 lb 1.950 z 0.709 lr 1.40e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15740/40589 loss 0.2455 acc 0.953 lb 1.959 z 0.670 lr 1.40e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15750/40589 loss 0.2370 acc 0.955 lb 1.956 z 0.663 lr 1.40e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.943 max_load(range) 0.052-0.418 step 15760/40589 loss 0.2356 acc 0.955 lb 1.943 z 0.786 lr 1.40e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15770/40589 loss 0.2297 acc 0.957 lb 1.952 z 0.769 lr 1.40e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.910 max_load(range) 0.070-0.147 step 15780/40589 loss 0.2463 acc 0.953 lb 1.961 z 0.655 lr 1.40e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15790/40589 loss 0.2372 acc 0.955 lb 1.959 z 0.688 lr 1.40e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15800/40589 loss 0.2383 acc 0.955 lb 1.954 z 0.770 lr 1.40e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.894 max_load(range) 0.079-0.184 step 15810/40589 loss 0.2314 acc 0.956 lb 1.950 z 0.745 lr 1.40e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15820/40589 loss 0.2472 acc 0.953 lb 1.961 z 0.650 lr 1.39e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.929 max_load(range) 0.058-0.247 step 15830/40589 loss 0.2356 acc 0.956 lb 1.962 z 0.675 lr 1.39e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15840/40589 loss 0.2385 acc 0.955 lb 1.951 z 0.781 lr 1.39e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15850/40589 loss 0.2286 acc 0.957 lb 1.953 z 0.775 lr 1.39e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.936 max_load(range) 0.060-0.239 step 15860/40589 loss 0.2428 acc 0.953 lb 1.957 z 0.656 lr 1.39e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15870/40589 loss 0.2371 acc 0.955 lb 1.948 z 0.651 lr 1.39e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.904 max_load(range) 0.072-0.160 step 15880/40589 loss 0.2350 acc 0.955 lb 1.940 z 0.754 lr 1.39e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15890/40589 loss 0.2335 acc 0.955 lb 1.947 z 0.736 lr 1.39e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15900/40589 loss 0.2462 acc 0.953 lb 1.966 z 0.662 lr 1.39e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.941 max_load(range) 0.068-0.206 step 15910/40589 loss 0.2370 acc 0.956 lb 1.961 z 0.654 lr 1.39e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15920/40589 loss 0.2374 acc 0.955 lb 1.947 z 0.781 lr 1.39e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.914 max_load(range) 0.066-0.314 step 15930/40589 loss 0.2352 acc 0.956 lb 1.948 z 0.725 lr 1.39e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15940/40589 loss 0.2466 acc 0.953 lb 1.959 z 0.647 lr 1.39e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15950/40589 loss 0.2348 acc 0.956 lb 1.958 z 0.674 lr 1.39e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.944 max_load(range) 0.051-0.389 step 15960/40589 loss 0.2373 acc 0.955 lb 1.950 z 0.791 lr 1.38e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15970/40589 loss 0.2277 acc 0.957 lb 1.945 z 0.737 lr 1.38e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.897 max_load(range) 0.077-0.174 step 15980/40589 loss 0.2419 acc 0.954 lb 1.958 z 0.650 lr 1.38e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 15990/40589 loss 0.2380 acc 0.955 lb 1.956 z 0.708 lr 1.38e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 16000/40589 loss 0.2379 acc 0.955 lb 1.952 z 0.789 lr 1.38e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.894 max_load(range) 0.074-0.235 [save] runs/pretrain_flagship_h100_4x8/step16000 step 16010/40589 loss 0.2276 acc 0.957 lb 1.951 z 0.703 lr 1.38e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 16020/40589 loss 0.2424 acc 0.954 lb 1.961 z 0.630 lr 1.38e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.933 max_load(range) 0.067-0.226 step 16030/40589 loss 0.2325 acc 0.956 lb 1.956 z 0.650 lr 1.38e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 16040/40589 loss 0.2402 acc 0.954 lb 1.946 z 0.705 lr 1.38e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 16050/40589 loss 0.2315 acc 0.956 lb 1.944 z 0.746 lr 1.38e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.934 max_load(range) 0.057-0.257 step 16060/40589 loss 0.2430 acc 0.953 lb 1.954 z 0.653 lr 1.38e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 16070/40589 loss 0.2363 acc 0.956 lb 1.955 z 0.687 lr 1.38e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.905 max_load(range) 0.070-0.174 step 16080/40589 loss 0.2355 acc 0.955 lb 1.946 z 0.733 lr 1.38e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 16090/40589 loss 0.2300 acc 0.957 lb 1.951 z 0.703 lr 1.37e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 16100/40589 loss 0.2423 acc 0.954 lb 1.961 z 0.659 lr 1.37e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.939 max_load(range) 0.066-0.191 step 16110/40589 loss 0.2377 acc 0.955 lb 1.957 z 0.700 lr 1.37e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 16120/40589 loss 0.2345 acc 0.956 lb 1.939 z 0.762 lr 1.37e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.931 max_load(range) 0.052-0.262 step 16130/40589 loss 0.2326 acc 0.956 lb 1.951 z 0.676 lr 1.37e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 16140/40589 loss 0.2440 acc 0.953 lb 1.962 z 0.642 lr 1.37e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 16150/40589 loss 0.2328 acc 0.956 lb 1.950 z 0.597 lr 1.37e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.944 max_load(range) 0.051-0.384 step 16160/40589 loss 0.2339 acc 0.955 lb 1.962 z 0.745 lr 1.37e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 16170/40589 loss 0.2300 acc 0.956 lb 1.957 z 0.783 lr 1.37e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.907 max_load(range) 0.059-0.161 step 16180/40589 loss 0.2432 acc 0.954 lb 1.960 z 0.640 lr 1.37e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 16190/40589 loss 0.2339 acc 0.957 lb 1.951 z 0.655 lr 1.37e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left step 16200/40589 loss 0.2387 acc 0.955 lb 1.947 z 0.721 lr 1.37e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.3d left [routing] mean entropy_norm 0.895 max_load(range) 0.070-0.169 step 16210/40589 loss 0.2271 acc 0.957 lb 1.956 z 0.686 lr 1.37e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16220/40589 loss 0.2448 acc 0.953 lb 1.958 z 0.654 lr 1.37e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.936 max_load(range) 0.060-0.219 step 16230/40589 loss 0.2357 acc 0.956 lb 1.953 z 0.653 lr 1.36e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16240/40589 loss 0.2364 acc 0.955 lb 1.947 z 0.722 lr 1.36e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16250/40589 loss 0.2286 acc 0.956 lb 1.952 z 0.725 lr 1.36e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.933 max_load(range) 0.058-0.234 step 16260/40589 loss 0.2400 acc 0.954 lb 1.950 z 0.698 lr 1.36e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16270/40589 loss 0.2359 acc 0.956 lb 1.961 z 0.676 lr 1.36e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.904 max_load(range) 0.077-0.206 step 16280/40589 loss 0.2318 acc 0.956 lb 1.946 z 0.737 lr 1.36e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16290/40589 loss 0.2277 acc 0.956 lb 1.953 z 0.673 lr 1.36e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16300/40589 loss 0.2457 acc 0.953 lb 1.961 z 0.618 lr 1.36e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.943 max_load(range) 0.061-0.199 step 16310/40589 loss 0.2301 acc 0.956 lb 1.955 z 0.663 lr 1.36e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16320/40589 loss 0.2366 acc 0.955 lb 1.948 z 0.738 lr 1.36e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.923 max_load(range) 0.056-0.300 step 16330/40589 loss 0.2266 acc 0.957 lb 1.949 z 0.675 lr 1.36e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16340/40589 loss 0.2428 acc 0.954 lb 1.967 z 0.635 lr 1.36e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16350/40589 loss 0.2307 acc 0.957 lb 1.957 z 0.605 lr 1.36e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.944 max_load(range) 0.046-0.402 step 16360/40589 loss 0.2337 acc 0.955 lb 1.948 z 0.739 lr 1.35e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16370/40589 loss 0.2291 acc 0.956 lb 1.954 z 0.693 lr 1.35e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.905 max_load(range) 0.072-0.167 step 16380/40589 loss 0.2419 acc 0.954 lb 1.955 z 0.590 lr 1.35e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16390/40589 loss 0.2318 acc 0.956 lb 1.960 z 0.658 lr 1.35e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16400/40589 loss 0.2353 acc 0.955 lb 1.954 z 0.784 lr 1.35e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.899 max_load(range) 0.071-0.244 step 16410/40589 loss 0.2332 acc 0.956 lb 1.954 z 0.717 lr 1.35e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16420/40589 loss 0.2412 acc 0.954 lb 1.961 z 0.590 lr 1.35e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.939 max_load(range) 0.060-0.208 step 16430/40589 loss 0.2340 acc 0.956 lb 1.948 z 0.624 lr 1.35e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16440/40589 loss 0.2315 acc 0.957 lb 1.941 z 0.731 lr 1.35e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16450/40589 loss 0.2269 acc 0.957 lb 1.950 z 0.691 lr 1.35e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.936 max_load(range) 0.062-0.247 step 16460/40589 loss 0.2436 acc 0.954 lb 1.959 z 0.629 lr 1.35e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16470/40589 loss 0.2351 acc 0.955 lb 1.952 z 0.607 lr 1.35e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.896 max_load(range) 0.087-0.166 step 16480/40589 loss 0.2353 acc 0.956 lb 1.958 z 0.778 lr 1.35e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16490/40589 loss 0.2259 acc 0.958 lb 1.952 z 0.721 lr 1.34e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16500/40589 loss 0.2424 acc 0.954 lb 1.961 z 0.601 lr 1.34e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.943 max_load(range) 0.064-0.200 step 16510/40589 loss 0.2336 acc 0.957 lb 1.954 z 0.637 lr 1.34e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16520/40589 loss 0.2341 acc 0.956 lb 1.947 z 0.692 lr 1.34e-04 | 1423 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.934 max_load(range) 0.051-0.253 step 16530/40589 loss 0.2265 acc 0.958 lb 1.953 z 0.690 lr 1.34e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16540/40589 loss 0.2378 acc 0.954 lb 1.963 z 0.656 lr 1.34e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16550/40589 loss 0.2324 acc 0.956 lb 1.952 z 0.628 lr 1.34e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.944 max_load(range) 0.052-0.404 step 16560/40589 loss 0.2331 acc 0.955 lb 1.954 z 0.712 lr 1.34e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16570/40589 loss 0.2293 acc 0.956 lb 1.951 z 0.728 lr 1.34e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.914 max_load(range) 0.078-0.180 step 16580/40589 loss 0.2429 acc 0.954 lb 1.965 z 0.619 lr 1.34e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16590/40589 loss 0.2308 acc 0.957 lb 1.955 z 0.636 lr 1.34e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16600/40589 loss 0.2312 acc 0.956 lb 1.946 z 0.726 lr 1.34e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.896 max_load(range) 0.081-0.185 step 16610/40589 loss 0.2286 acc 0.957 lb 1.950 z 0.699 lr 1.34e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16620/40589 loss 0.2416 acc 0.954 lb 1.959 z 0.621 lr 1.34e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.936 max_load(range) 0.055-0.212 step 16630/40589 loss 0.2350 acc 0.955 lb 1.956 z 0.641 lr 1.33e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16640/40589 loss 0.2314 acc 0.955 lb 1.949 z 0.786 lr 1.33e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16650/40589 loss 0.2253 acc 0.957 lb 1.954 z 0.719 lr 1.33e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.932 max_load(range) 0.051-0.233 step 16660/40589 loss 0.2409 acc 0.954 lb 1.957 z 0.699 lr 1.33e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16670/40589 loss 0.2315 acc 0.956 lb 1.958 z 0.629 lr 1.33e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.888 max_load(range) 0.090-0.194 step 16680/40589 loss 0.2355 acc 0.955 lb 1.950 z 0.780 lr 1.33e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16690/40589 loss 0.2253 acc 0.957 lb 1.956 z 0.685 lr 1.33e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16700/40589 loss 0.2387 acc 0.954 lb 1.956 z 0.714 lr 1.33e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.937 max_load(range) 0.069-0.192 step 16710/40589 loss 0.2289 acc 0.957 lb 1.953 z 0.616 lr 1.33e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16720/40589 loss 0.2339 acc 0.955 lb 1.948 z 0.729 lr 1.33e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.922 max_load(range) 0.057-0.294 step 16730/40589 loss 0.2223 acc 0.958 lb 1.959 z 0.723 lr 1.33e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16740/40589 loss 0.2401 acc 0.954 lb 1.960 z 0.615 lr 1.33e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16750/40589 loss 0.2319 acc 0.956 lb 1.962 z 0.645 lr 1.33e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.944 max_load(range) 0.050-0.409 step 16760/40589 loss 0.2345 acc 0.956 lb 1.949 z 0.733 lr 1.32e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16770/40589 loss 0.2254 acc 0.957 lb 1.946 z 0.674 lr 1.32e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.908 max_load(range) 0.082-0.191 step 16780/40589 loss 0.2358 acc 0.955 lb 1.962 z 0.656 lr 1.32e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16790/40589 loss 0.2302 acc 0.957 lb 1.959 z 0.634 lr 1.32e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16800/40589 loss 0.2323 acc 0.956 lb 1.959 z 0.761 lr 1.32e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.895 max_load(range) 0.087-0.190 step 16810/40589 loss 0.2280 acc 0.956 lb 1.952 z 0.718 lr 1.32e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16820/40589 loss 0.2395 acc 0.954 lb 1.967 z 0.623 lr 1.32e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.929 max_load(range) 0.058-0.221 step 16830/40589 loss 0.2305 acc 0.957 lb 1.966 z 0.655 lr 1.32e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16840/40589 loss 0.2290 acc 0.956 lb 1.952 z 0.776 lr 1.32e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16850/40589 loss 0.2262 acc 0.957 lb 1.954 z 0.685 lr 1.32e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.932 max_load(range) 0.056-0.230 step 16860/40589 loss 0.2387 acc 0.954 lb 1.968 z 0.628 lr 1.32e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16870/40589 loss 0.2270 acc 0.957 lb 1.963 z 0.641 lr 1.32e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.927 max_load(range) 0.051-0.252 step 16880/40589 loss 0.2334 acc 0.956 lb 1.947 z 0.768 lr 1.32e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16890/40589 loss 0.2258 acc 0.956 lb 1.960 z 0.694 lr 1.31e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16900/40589 loss 0.2370 acc 0.955 lb 1.961 z 0.605 lr 1.31e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.938 max_load(range) 0.058-0.194 step 16910/40589 loss 0.2287 acc 0.956 lb 1.951 z 0.604 lr 1.31e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16920/40589 loss 0.2318 acc 0.956 lb 1.948 z 0.702 lr 1.31e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.931 max_load(range) 0.055-0.285 step 16930/40589 loss 0.2234 acc 0.958 lb 1.946 z 0.681 lr 1.31e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16940/40589 loss 0.2386 acc 0.954 lb 1.957 z 0.746 lr 1.31e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16950/40589 loss 0.2284 acc 0.957 lb 1.961 z 0.635 lr 1.31e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left [routing] mean entropy_norm 0.949 max_load(range) 0.054-0.377 step 16960/40589 loss 0.2307 acc 0.956 lb 1.950 z 0.726 lr 1.31e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.2d left step 16970/40589 loss 0.2239 acc 0.958 lb 1.958 z 0.697 lr 1.31e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.897 max_load(range) 0.080-0.157 step 16980/40589 loss 0.2387 acc 0.955 lb 1.958 z 0.649 lr 1.31e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 16990/40589 loss 0.2294 acc 0.957 lb 1.959 z 0.624 lr 1.31e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17000/40589 loss 0.2317 acc 0.955 lb 1.954 z 0.732 lr 1.31e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.897 max_load(range) 0.083-0.204 [save] runs/pretrain_flagship_h100_4x8/step17000 step 17010/40589 loss 0.2282 acc 0.957 lb 1.960 z 0.655 lr 1.31e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17020/40589 loss 0.2384 acc 0.954 lb 1.961 z 0.597 lr 1.30e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.935 max_load(range) 0.057-0.222 step 17030/40589 loss 0.2315 acc 0.956 lb 1.959 z 0.613 lr 1.30e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17040/40589 loss 0.2328 acc 0.956 lb 1.942 z 0.762 lr 1.30e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17050/40589 loss 0.2237 acc 0.958 lb 1.956 z 0.686 lr 1.30e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.938 max_load(range) 0.055-0.236 step 17060/40589 loss 0.2377 acc 0.954 lb 1.959 z 0.608 lr 1.30e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17070/40589 loss 0.2284 acc 0.957 lb 1.958 z 0.613 lr 1.30e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.904 max_load(range) 0.067-0.178 step 17080/40589 loss 0.2319 acc 0.955 lb 1.955 z 0.760 lr 1.30e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17090/40589 loss 0.2260 acc 0.957 lb 1.954 z 0.673 lr 1.30e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17100/40589 loss 0.2361 acc 0.955 lb 1.968 z 0.600 lr 1.30e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.937 max_load(range) 0.062-0.201 step 17110/40589 loss 0.2265 acc 0.957 lb 1.954 z 0.607 lr 1.30e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17120/40589 loss 0.2302 acc 0.956 lb 1.953 z 0.747 lr 1.30e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.917 max_load(range) 0.058-0.282 step 17130/40589 loss 0.2260 acc 0.958 lb 1.946 z 0.706 lr 1.30e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17140/40589 loss 0.2385 acc 0.954 lb 1.961 z 0.622 lr 1.30e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17150/40589 loss 0.2306 acc 0.956 lb 1.960 z 0.615 lr 1.30e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.944 max_load(range) 0.056-0.393 step 17160/40589 loss 0.2284 acc 0.956 lb 1.950 z 0.702 lr 1.29e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17170/40589 loss 0.2233 acc 0.958 lb 1.961 z 0.683 lr 1.29e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.897 max_load(range) 0.080-0.177 step 17180/40589 loss 0.2368 acc 0.955 lb 1.969 z 0.611 lr 1.29e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17190/40589 loss 0.2273 acc 0.957 lb 1.967 z 0.626 lr 1.29e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17200/40589 loss 0.2317 acc 0.956 lb 1.951 z 0.725 lr 1.29e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.890 max_load(range) 0.086-0.190 step 17210/40589 loss 0.2244 acc 0.957 lb 1.953 z 0.693 lr 1.29e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17220/40589 loss 0.2358 acc 0.955 lb 1.964 z 0.594 lr 1.29e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.936 max_load(range) 0.054-0.222 step 17230/40589 loss 0.2269 acc 0.957 lb 1.958 z 0.606 lr 1.29e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17240/40589 loss 0.2276 acc 0.956 lb 1.953 z 0.693 lr 1.29e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17250/40589 loss 0.2224 acc 0.958 lb 1.957 z 0.669 lr 1.29e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.941 max_load(range) 0.050-0.233 step 17260/40589 loss 0.2378 acc 0.955 lb 1.955 z 0.594 lr 1.29e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17270/40589 loss 0.2272 acc 0.957 lb 1.955 z 0.680 lr 1.29e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.905 max_load(range) 0.076-0.147 step 17280/40589 loss 0.2302 acc 0.956 lb 1.944 z 0.698 lr 1.29e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17290/40589 loss 0.2244 acc 0.958 lb 1.953 z 0.753 lr 1.28e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17300/40589 loss 0.2390 acc 0.954 lb 1.961 z 0.589 lr 1.28e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.941 max_load(range) 0.057-0.192 step 17310/40589 loss 0.2285 acc 0.957 lb 1.953 z 0.593 lr 1.28e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17320/40589 loss 0.2312 acc 0.956 lb 1.952 z 0.744 lr 1.28e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.935 max_load(range) 0.060-0.229 step 17330/40589 loss 0.2221 acc 0.958 lb 1.956 z 0.699 lr 1.28e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17340/40589 loss 0.2345 acc 0.955 lb 1.963 z 0.605 lr 1.28e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17350/40589 loss 0.2261 acc 0.958 lb 1.954 z 0.590 lr 1.28e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.950 max_load(range) 0.053-0.386 step 17360/40589 loss 0.2281 acc 0.957 lb 1.959 z 0.716 lr 1.28e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17370/40589 loss 0.2230 acc 0.958 lb 1.960 z 0.689 lr 1.28e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.899 max_load(range) 0.080-0.153 step 17380/40589 loss 0.2337 acc 0.955 lb 1.963 z 0.555 lr 1.28e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17390/40589 loss 0.2244 acc 0.958 lb 1.963 z 0.607 lr 1.28e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17400/40589 loss 0.2281 acc 0.957 lb 1.955 z 0.785 lr 1.28e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.886 max_load(range) 0.084-0.209 step 17410/40589 loss 0.2207 acc 0.958 lb 1.956 z 0.670 lr 1.28e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17420/40589 loss 0.2372 acc 0.955 lb 1.959 z 0.590 lr 1.27e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.935 max_load(range) 0.058-0.209 step 17430/40589 loss 0.2258 acc 0.957 lb 1.962 z 0.570 lr 1.27e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17440/40589 loss 0.2298 acc 0.956 lb 1.950 z 0.630 lr 1.27e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17450/40589 loss 0.2212 acc 0.958 lb 1.961 z 0.704 lr 1.27e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.942 max_load(range) 0.049-0.255 step 17460/40589 loss 0.2313 acc 0.956 lb 1.969 z 0.621 lr 1.27e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17470/40589 loss 0.2280 acc 0.957 lb 1.956 z 0.589 lr 1.27e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.927 max_load(range) 0.052-0.272 step 17480/40589 loss 0.2235 acc 0.957 lb 1.950 z 0.709 lr 1.27e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17490/40589 loss 0.2214 acc 0.958 lb 1.958 z 0.621 lr 1.27e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17500/40589 loss 0.2352 acc 0.955 lb 1.962 z 0.672 lr 1.27e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.944 max_load(range) 0.059-0.213 step 17510/40589 loss 0.2262 acc 0.957 lb 1.955 z 0.604 lr 1.27e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17520/40589 loss 0.2295 acc 0.957 lb 1.954 z 0.743 lr 1.27e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.930 max_load(range) 0.056-0.253 step 17530/40589 loss 0.2211 acc 0.958 lb 1.961 z 0.663 lr 1.27e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17540/40589 loss 0.2354 acc 0.955 lb 1.969 z 0.563 lr 1.27e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17550/40589 loss 0.2221 acc 0.958 lb 1.953 z 0.575 lr 1.26e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.948 max_load(range) 0.052-0.376 step 17560/40589 loss 0.2273 acc 0.957 lb 1.952 z 0.686 lr 1.26e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17570/40589 loss 0.2204 acc 0.958 lb 1.955 z 0.699 lr 1.26e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.908 max_load(range) 0.079-0.171 step 17580/40589 loss 0.2302 acc 0.956 lb 1.964 z 0.644 lr 1.26e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17590/40589 loss 0.2257 acc 0.958 lb 1.963 z 0.610 lr 1.26e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17600/40589 loss 0.2285 acc 0.957 lb 1.945 z 0.702 lr 1.26e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.887 max_load(range) 0.076-0.171 step 17610/40589 loss 0.2220 acc 0.958 lb 1.955 z 0.674 lr 1.26e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17620/40589 loss 0.2373 acc 0.955 lb 1.964 z 0.599 lr 1.26e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.935 max_load(range) 0.058-0.197 step 17630/40589 loss 0.2222 acc 0.958 lb 1.953 z 0.658 lr 1.26e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17640/40589 loss 0.2306 acc 0.956 lb 1.960 z 0.719 lr 1.26e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17650/40589 loss 0.2241 acc 0.958 lb 1.957 z 0.667 lr 1.26e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.937 max_load(range) 0.058-0.247 step 17660/40589 loss 0.2350 acc 0.956 lb 1.966 z 0.656 lr 1.26e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17670/40589 loss 0.2252 acc 0.958 lb 1.969 z 0.600 lr 1.26e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.921 max_load(range) 0.054-0.312 step 17680/40589 loss 0.2248 acc 0.957 lb 1.963 z 0.768 lr 1.25e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17690/40589 loss 0.2212 acc 0.958 lb 1.964 z 0.664 lr 1.25e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17700/40589 loss 0.2346 acc 0.955 lb 1.964 z 0.684 lr 1.25e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.938 max_load(range) 0.059-0.201 step 17710/40589 loss 0.2246 acc 0.958 lb 1.957 z 0.648 lr 1.25e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left step 17720/40589 loss 0.2293 acc 0.956 lb 1.958 z 0.707 lr 1.25e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.1d left [routing] mean entropy_norm 0.925 max_load(range) 0.056-0.297 step 17730/40589 loss 0.2208 acc 0.959 lb 1.952 z 0.674 lr 1.25e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 17740/40589 loss 0.2348 acc 0.955 lb 1.965 z 0.658 lr 1.25e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 17750/40589 loss 0.2223 acc 0.958 lb 1.959 z 0.577 lr 1.25e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.941 max_load(range) 0.052-0.416 step 17760/40589 loss 0.2247 acc 0.957 lb 1.953 z 0.656 lr 1.25e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 17770/40589 loss 0.2222 acc 0.958 lb 1.960 z 0.640 lr 1.25e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.903 max_load(range) 0.071-0.179 step 17780/40589 loss 0.2337 acc 0.955 lb 1.961 z 0.571 lr 1.25e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 17790/40589 loss 0.2265 acc 0.957 lb 1.960 z 0.653 lr 1.25e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 17800/40589 loss 0.2207 acc 0.958 lb 1.945 z 0.737 lr 1.25e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.888 max_load(range) 0.082-0.195 step 17810/40589 loss 0.2225 acc 0.957 lb 1.955 z 0.655 lr 1.24e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 17820/40589 loss 0.2323 acc 0.955 lb 1.970 z 0.585 lr 1.24e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.936 max_load(range) 0.055-0.218 step 17830/40589 loss 0.2260 acc 0.957 lb 1.965 z 0.566 lr 1.24e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 17840/40589 loss 0.2259 acc 0.957 lb 1.949 z 0.690 lr 1.24e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 17850/40589 loss 0.2233 acc 0.957 lb 1.957 z 0.642 lr 1.24e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.947 max_load(range) 0.054-0.233 step 17860/40589 loss 0.2317 acc 0.956 lb 1.966 z 0.577 lr 1.24e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 17870/40589 loss 0.2220 acc 0.959 lb 1.962 z 0.562 lr 1.24e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.909 max_load(range) 0.073-0.139 step 17880/40589 loss 0.2279 acc 0.956 lb 1.955 z 0.666 lr 1.24e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 17890/40589 loss 0.2227 acc 0.957 lb 1.962 z 0.649 lr 1.24e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 17900/40589 loss 0.2326 acc 0.955 lb 1.965 z 0.571 lr 1.24e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.943 max_load(range) 0.063-0.188 step 17910/40589 loss 0.2257 acc 0.957 lb 1.961 z 0.600 lr 1.24e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 17920/40589 loss 0.2261 acc 0.957 lb 1.955 z 0.684 lr 1.24e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.932 max_load(range) 0.055-0.270 step 17930/40589 loss 0.2175 acc 0.959 lb 1.956 z 0.677 lr 1.24e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 17940/40589 loss 0.2367 acc 0.955 lb 1.965 z 0.544 lr 1.23e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 17950/40589 loss 0.2195 acc 0.958 lb 1.964 z 0.588 lr 1.23e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.948 max_load(range) 0.052-0.397 step 17960/40589 loss 0.2306 acc 0.956 lb 1.953 z 0.667 lr 1.23e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 17970/40589 loss 0.2197 acc 0.958 lb 1.960 z 0.634 lr 1.23e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.905 max_load(range) 0.069-0.147 step 17980/40589 loss 0.2343 acc 0.955 lb 1.965 z 0.547 lr 1.23e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 17990/40589 loss 0.2231 acc 0.958 lb 1.958 z 0.656 lr 1.23e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18000/40589 loss 0.2275 acc 0.956 lb 1.951 z 0.675 lr 1.23e-04 | 1422 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.898 max_load(range) 0.080-0.174 [save] runs/pretrain_flagship_h100_4x8/step18000 step 18010/40589 loss 0.2209 acc 0.958 lb 1.961 z 0.662 lr 1.23e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18020/40589 loss 0.2356 acc 0.955 lb 1.967 z 0.594 lr 1.23e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.939 max_load(range) 0.064-0.242 step 18030/40589 loss 0.2228 acc 0.958 lb 1.959 z 0.624 lr 1.23e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18040/40589 loss 0.2247 acc 0.957 lb 1.951 z 0.720 lr 1.23e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18050/40589 loss 0.2180 acc 0.959 lb 1.961 z 0.630 lr 1.23e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.948 max_load(range) 0.053-0.245 step 18060/40589 loss 0.2344 acc 0.955 lb 1.974 z 0.600 lr 1.22e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18070/40589 loss 0.2242 acc 0.958 lb 1.961 z 0.543 lr 1.22e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.908 max_load(range) 0.070-0.204 step 18080/40589 loss 0.2238 acc 0.957 lb 1.955 z 0.702 lr 1.22e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18090/40589 loss 0.2214 acc 0.959 lb 1.954 z 0.723 lr 1.22e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18100/40589 loss 0.2318 acc 0.955 lb 1.967 z 0.682 lr 1.22e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.946 max_load(range) 0.055-0.195 step 18110/40589 loss 0.2253 acc 0.958 lb 1.963 z 0.603 lr 1.22e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18120/40589 loss 0.2263 acc 0.957 lb 1.953 z 0.731 lr 1.22e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.927 max_load(range) 0.055-0.289 step 18130/40589 loss 0.2216 acc 0.958 lb 1.958 z 0.619 lr 1.22e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18140/40589 loss 0.2336 acc 0.955 lb 1.971 z 0.590 lr 1.22e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18150/40589 loss 0.2204 acc 0.958 lb 1.962 z 0.575 lr 1.22e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.944 max_load(range) 0.049-0.403 step 18160/40589 loss 0.2247 acc 0.957 lb 1.949 z 0.732 lr 1.22e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18170/40589 loss 0.2232 acc 0.958 lb 1.958 z 0.670 lr 1.22e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.909 max_load(range) 0.081-0.149 step 18180/40589 loss 0.2347 acc 0.956 lb 1.965 z 0.601 lr 1.22e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18190/40589 loss 0.2239 acc 0.958 lb 1.952 z 0.660 lr 1.21e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18200/40589 loss 0.2259 acc 0.957 lb 1.953 z 0.696 lr 1.21e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.902 max_load(range) 0.067-0.213 step 18210/40589 loss 0.2186 acc 0.959 lb 1.964 z 0.627 lr 1.21e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18220/40589 loss 0.2357 acc 0.955 lb 1.969 z 0.653 lr 1.21e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.931 max_load(range) 0.066-0.216 step 18230/40589 loss 0.2223 acc 0.958 lb 1.959 z 0.624 lr 1.21e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18240/40589 loss 0.2208 acc 0.958 lb 1.956 z 0.650 lr 1.21e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18250/40589 loss 0.2199 acc 0.959 lb 1.959 z 0.631 lr 1.21e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.947 max_load(range) 0.052-0.238 step 18260/40589 loss 0.2324 acc 0.956 lb 1.967 z 0.671 lr 1.21e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18270/40589 loss 0.2235 acc 0.958 lb 1.957 z 0.614 lr 1.21e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.928 max_load(range) 0.056-0.258 step 18280/40589 loss 0.2268 acc 0.957 lb 1.956 z 0.712 lr 1.21e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18290/40589 loss 0.2199 acc 0.959 lb 1.959 z 0.634 lr 1.21e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18300/40589 loss 0.2341 acc 0.955 lb 1.966 z 0.608 lr 1.21e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.944 max_load(range) 0.069-0.205 step 18310/40589 loss 0.2258 acc 0.958 lb 1.962 z 0.611 lr 1.21e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18320/40589 loss 0.2248 acc 0.957 lb 1.947 z 0.757 lr 1.20e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.932 max_load(range) 0.051-0.234 step 18330/40589 loss 0.2185 acc 0.958 lb 1.959 z 0.613 lr 1.20e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18340/40589 loss 0.2339 acc 0.955 lb 1.968 z 0.563 lr 1.20e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18350/40589 loss 0.2207 acc 0.959 lb 1.971 z 0.566 lr 1.20e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.944 max_load(range) 0.053-0.394 step 18360/40589 loss 0.2259 acc 0.957 lb 1.953 z 0.710 lr 1.20e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18370/40589 loss 0.2178 acc 0.959 lb 1.955 z 0.614 lr 1.20e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.911 max_load(range) 0.069-0.186 step 18380/40589 loss 0.2307 acc 0.956 lb 1.965 z 0.643 lr 1.20e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18390/40589 loss 0.2195 acc 0.958 lb 1.966 z 0.567 lr 1.20e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18400/40589 loss 0.2276 acc 0.956 lb 1.961 z 0.669 lr 1.20e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.888 max_load(range) 0.078-0.275 step 18410/40589 loss 0.2199 acc 0.959 lb 1.957 z 0.598 lr 1.20e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18420/40589 loss 0.2322 acc 0.956 lb 1.960 z 0.618 lr 1.20e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.934 max_load(range) 0.056-0.220 step 18430/40589 loss 0.2224 acc 0.958 lb 1.961 z 0.550 lr 1.20e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18440/40589 loss 0.2242 acc 0.957 lb 1.956 z 0.722 lr 1.20e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18450/40589 loss 0.2202 acc 0.958 lb 1.960 z 0.617 lr 1.19e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.945 max_load(range) 0.055-0.235 step 18460/40589 loss 0.2326 acc 0.956 lb 1.966 z 0.551 lr 1.19e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left step 18470/40589 loss 0.2211 acc 0.959 lb 1.961 z 0.644 lr 1.19e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 3.0d left [routing] mean entropy_norm 0.882 max_load(range) 0.071-0.186 step 18480/40589 loss 0.2233 acc 0.957 lb 1.954 z 0.673 lr 1.19e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18490/40589 loss 0.2213 acc 0.958 lb 1.963 z 0.679 lr 1.19e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18500/40589 loss 0.2313 acc 0.956 lb 1.967 z 0.542 lr 1.19e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.942 max_load(range) 0.048-0.201 step 18510/40589 loss 0.2227 acc 0.958 lb 1.960 z 0.635 lr 1.19e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18520/40589 loss 0.2250 acc 0.957 lb 1.952 z 0.704 lr 1.19e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.932 max_load(range) 0.054-0.236 step 18530/40589 loss 0.2172 acc 0.959 lb 1.964 z 0.621 lr 1.19e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18540/40589 loss 0.2353 acc 0.955 lb 1.963 z 0.540 lr 1.19e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18550/40589 loss 0.2198 acc 0.959 lb 1.964 z 0.636 lr 1.19e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.947 max_load(range) 0.051-0.398 step 18560/40589 loss 0.2218 acc 0.958 lb 1.955 z 0.681 lr 1.19e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18570/40589 loss 0.2156 acc 0.959 lb 1.959 z 0.697 lr 1.19e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.914 max_load(range) 0.077-0.174 step 18580/40589 loss 0.2321 acc 0.955 lb 1.972 z 0.618 lr 1.18e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18590/40589 loss 0.2232 acc 0.959 lb 1.956 z 0.638 lr 1.18e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18600/40589 loss 0.2226 acc 0.958 lb 1.958 z 0.666 lr 1.18e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.904 max_load(range) 0.074-0.214 step 18610/40589 loss 0.2179 acc 0.959 lb 1.962 z 0.693 lr 1.18e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18620/40589 loss 0.2326 acc 0.955 lb 1.968 z 0.616 lr 1.18e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.941 max_load(range) 0.051-0.221 step 18630/40589 loss 0.2231 acc 0.958 lb 1.957 z 0.580 lr 1.18e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18640/40589 loss 0.2230 acc 0.958 lb 1.952 z 0.738 lr 1.18e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18650/40589 loss 0.2182 acc 0.959 lb 1.957 z 0.694 lr 1.18e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.954 max_load(range) 0.049-0.251 step 18660/40589 loss 0.2317 acc 0.956 lb 1.961 z 0.678 lr 1.18e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18670/40589 loss 0.2202 acc 0.958 lb 1.962 z 0.565 lr 1.18e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.911 max_load(range) 0.077-0.147 step 18680/40589 loss 0.2217 acc 0.958 lb 1.955 z 0.627 lr 1.18e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18690/40589 loss 0.2174 acc 0.959 lb 1.953 z 0.659 lr 1.18e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18700/40589 loss 0.2299 acc 0.956 lb 1.962 z 0.591 lr 1.17e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.949 max_load(range) 0.057-0.207 step 18710/40589 loss 0.2202 acc 0.959 lb 1.961 z 0.607 lr 1.17e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18720/40589 loss 0.2215 acc 0.958 lb 1.951 z 0.709 lr 1.17e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.914 max_load(range) 0.066-0.309 step 18730/40589 loss 0.2202 acc 0.958 lb 1.964 z 0.674 lr 1.17e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18740/40589 loss 0.2278 acc 0.956 lb 1.999 z 0.557 lr 1.17e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18750/40589 loss 0.2238 acc 0.958 lb 1.996 z 0.497 lr 1.17e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.945 max_load(range) 0.051-0.397 step 18760/40589 loss 0.2217 acc 0.958 lb 1.979 z 0.443 lr 1.17e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18770/40589 loss 0.2194 acc 0.959 lb 1.992 z 0.472 lr 1.17e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.836 max_load(range) 0.091-0.218 step 18780/40589 loss 0.2294 acc 0.956 lb 2.013 z 0.420 lr 1.17e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18790/40589 loss 0.2170 acc 0.959 lb 1.994 z 0.411 lr 1.17e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18800/40589 loss 0.2209 acc 0.958 lb 1.975 z 0.392 lr 1.17e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.851 max_load(range) 0.103-0.207 step 18810/40589 loss 0.2167 acc 0.959 lb 1.985 z 0.475 lr 1.17e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18820/40589 loss 0.2305 acc 0.956 lb 2.012 z 0.403 lr 1.17e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.934 max_load(range) 0.050-0.211 step 18830/40589 loss 0.2202 acc 0.958 lb 1.995 z 0.406 lr 1.16e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18840/40589 loss 0.2201 acc 0.958 lb 1.984 z 0.373 lr 1.16e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18850/40589 loss 0.2190 acc 0.959 lb 1.989 z 0.495 lr 1.16e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.946 max_load(range) 0.054-0.253 step 18860/40589 loss 0.2269 acc 0.957 lb 2.001 z 0.416 lr 1.16e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18870/40589 loss 0.2153 acc 0.959 lb 1.988 z 0.414 lr 1.16e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.923 max_load(range) 0.070-0.184 step 18880/40589 loss 0.2173 acc 0.959 lb 1.984 z 0.372 lr 1.16e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18890/40589 loss 0.2141 acc 0.959 lb 1.986 z 0.477 lr 1.16e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18900/40589 loss 0.2277 acc 0.956 lb 2.006 z 0.419 lr 1.16e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.952 max_load(range) 0.059-0.186 step 18910/40589 loss 0.2157 acc 0.959 lb 1.988 z 0.410 lr 1.16e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18920/40589 loss 0.2227 acc 0.957 lb 1.978 z 0.380 lr 1.16e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.947 max_load(range) 0.055-0.279 step 18930/40589 loss 0.2156 acc 0.959 lb 1.983 z 0.464 lr 1.16e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18940/40589 loss 0.2279 acc 0.956 lb 2.002 z 0.414 lr 1.16e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18950/40589 loss 0.2180 acc 0.959 lb 1.987 z 0.412 lr 1.16e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.951 max_load(range) 0.050-0.370 step 18960/40589 loss 0.2226 acc 0.958 lb 1.976 z 0.385 lr 1.15e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18970/40589 loss 0.2144 acc 0.959 lb 1.982 z 0.477 lr 1.15e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.873 max_load(range) 0.089-0.198 step 18980/40589 loss 0.2292 acc 0.956 lb 2.001 z 0.412 lr 1.15e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 18990/40589 loss 0.2168 acc 0.959 lb 1.983 z 0.417 lr 1.15e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 19000/40589 loss 0.2184 acc 0.958 lb 1.978 z 0.381 lr 1.15e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.890 max_load(range) 0.078-0.130 [save] runs/pretrain_flagship_h100_4x8/step19000 step 19010/40589 loss 0.2166 acc 0.959 lb 1.976 z 0.473 lr 1.15e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 19020/40589 loss 0.2293 acc 0.955 lb 2.000 z 0.423 lr 1.15e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.950 max_load(range) 0.054-0.195 step 19030/40589 loss 0.2200 acc 0.958 lb 1.979 z 0.427 lr 1.15e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 19040/40589 loss 0.2233 acc 0.957 lb 1.970 z 0.394 lr 1.15e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 19050/40589 loss 0.2166 acc 0.959 lb 1.983 z 0.471 lr 1.15e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.949 max_load(range) 0.054-0.275 step 19060/40589 loss 0.2293 acc 0.956 lb 1.999 z 0.416 lr 1.15e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 19070/40589 loss 0.2226 acc 0.958 lb 1.981 z 0.425 lr 1.15e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.940 max_load(range) 0.070-0.136 step 19080/40589 loss 0.2262 acc 0.956 lb 1.975 z 0.392 lr 1.14e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 19090/40589 loss 0.2160 acc 0.959 lb 1.978 z 0.472 lr 1.14e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 19100/40589 loss 0.2283 acc 0.956 lb 1.991 z 0.420 lr 1.14e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.958 max_load(range) 0.052-0.212 step 19110/40589 loss 0.2211 acc 0.958 lb 1.984 z 0.411 lr 1.14e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 19120/40589 loss 0.2232 acc 0.957 lb 1.975 z 0.391 lr 1.14e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.953 max_load(range) 0.056-0.283 step 19130/40589 loss 0.2155 acc 0.959 lb 1.982 z 0.459 lr 1.14e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 19140/40589 loss 0.2324 acc 0.955 lb 1.994 z 0.421 lr 1.14e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 19150/40589 loss 0.2197 acc 0.958 lb 1.984 z 0.405 lr 1.14e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.953 max_load(range) 0.050-0.360 step 19160/40589 loss 0.2241 acc 0.957 lb 1.972 z 0.381 lr 1.14e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 19170/40589 loss 0.2180 acc 0.958 lb 1.976 z 0.481 lr 1.14e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.909 max_load(range) 0.083-0.165 step 19180/40589 loss 0.2284 acc 0.955 lb 1.992 z 0.423 lr 1.14e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 19190/40589 loss 0.2211 acc 0.957 lb 1.982 z 0.408 lr 1.14e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 19200/40589 loss 0.2257 acc 0.957 lb 1.973 z 0.390 lr 1.14e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.902 max_load(range) 0.075-0.141 step 19210/40589 loss 0.2183 acc 0.958 lb 1.981 z 0.468 lr 1.13e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 19220/40589 loss 0.2287 acc 0.956 lb 1.996 z 0.419 lr 1.13e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left [routing] mean entropy_norm 0.951 max_load(range) 0.049-0.169 step 19230/40589 loss 0.2212 acc 0.957 lb 1.977 z 0.426 lr 1.13e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.9d left step 19240/40589 loss 0.2245 acc 0.956 lb 1.972 z 0.389 lr 1.13e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19250/40589 loss 0.2177 acc 0.958 lb 1.975 z 0.462 lr 1.13e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.951 max_load(range) 0.046-0.268 step 19260/40589 loss 0.2300 acc 0.955 lb 1.991 z 0.421 lr 1.13e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19270/40589 loss 0.2225 acc 0.958 lb 1.981 z 0.412 lr 1.13e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.949 max_load(range) 0.057-0.137 step 19280/40589 loss 0.2215 acc 0.957 lb 1.970 z 0.389 lr 1.13e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19290/40589 loss 0.2225 acc 0.957 lb 1.976 z 0.468 lr 1.13e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19300/40589 loss 0.2339 acc 0.955 lb 1.990 z 0.422 lr 1.13e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.960 max_load(range) 0.051-0.210 step 19310/40589 loss 0.2232 acc 0.957 lb 1.974 z 0.426 lr 1.13e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19320/40589 loss 0.2231 acc 0.957 lb 1.966 z 0.397 lr 1.13e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.954 max_load(range) 0.052-0.296 step 19330/40589 loss 0.2208 acc 0.957 lb 1.976 z 0.464 lr 1.13e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19340/40589 loss 0.2358 acc 0.954 lb 1.993 z 0.420 lr 1.12e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19350/40589 loss 0.2229 acc 0.957 lb 1.979 z 0.416 lr 1.12e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.961 max_load(range) 0.050-0.327 step 19360/40589 loss 0.2276 acc 0.955 lb 1.966 z 0.399 lr 1.12e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19370/40589 loss 0.2169 acc 0.958 lb 1.972 z 0.479 lr 1.12e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.904 max_load(range) 0.072-0.162 step 19380/40589 loss 0.2329 acc 0.955 lb 1.988 z 0.429 lr 1.12e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19390/40589 loss 0.2273 acc 0.956 lb 1.981 z 0.416 lr 1.12e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19400/40589 loss 0.2273 acc 0.956 lb 1.971 z 0.395 lr 1.12e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.913 max_load(range) 0.073-0.157 step 19410/40589 loss 0.2177 acc 0.958 lb 1.973 z 0.471 lr 1.12e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19420/40589 loss 0.2339 acc 0.955 lb 1.990 z 0.429 lr 1.12e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.950 max_load(range) 0.065-0.149 step 19430/40589 loss 0.2268 acc 0.956 lb 1.983 z 0.417 lr 1.12e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19440/40589 loss 0.2283 acc 0.955 lb 1.971 z 0.396 lr 1.12e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19450/40589 loss 0.2187 acc 0.958 lb 1.971 z 0.464 lr 1.12e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.934 max_load(range) 0.050-0.269 step 19460/40589 loss 0.2345 acc 0.954 lb 1.988 z 0.427 lr 1.11e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19470/40589 loss 0.2262 acc 0.956 lb 1.980 z 0.416 lr 1.11e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.947 max_load(range) 0.063-0.142 step 19480/40589 loss 0.2304 acc 0.955 lb 1.971 z 0.390 lr 1.11e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19490/40589 loss 0.2225 acc 0.957 lb 1.974 z 0.464 lr 1.11e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19500/40589 loss 0.2340 acc 0.955 lb 1.987 z 0.421 lr 1.11e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.958 max_load(range) 0.057-0.199 step 19510/40589 loss 0.2322 acc 0.954 lb 1.980 z 0.412 lr 1.11e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19520/40589 loss 0.2277 acc 0.956 lb 1.967 z 0.397 lr 1.11e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.957 max_load(range) 0.052-0.308 step 19530/40589 loss 0.2206 acc 0.957 lb 1.973 z 0.464 lr 1.11e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19540/40589 loss 0.2370 acc 0.954 lb 1.987 z 0.423 lr 1.11e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19550/40589 loss 0.2271 acc 0.956 lb 1.975 z 0.421 lr 1.11e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.956 max_load(range) 0.050-0.347 step 19560/40589 loss 0.2322 acc 0.955 lb 1.972 z 0.386 lr 1.11e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19570/40589 loss 0.2241 acc 0.956 lb 1.972 z 0.450 lr 1.11e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.909 max_load(range) 0.075-0.143 step 19580/40589 loss 0.2325 acc 0.955 lb 1.987 z 0.422 lr 1.11e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19590/40589 loss 0.2277 acc 0.955 lb 1.980 z 0.415 lr 1.10e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19600/40589 loss 0.2320 acc 0.955 lb 1.972 z 0.394 lr 1.10e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.945 max_load(range) 0.056-0.135 step 19610/40589 loss 0.2213 acc 0.956 lb 1.976 z 0.459 lr 1.10e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19620/40589 loss 0.2388 acc 0.954 lb 1.988 z 0.417 lr 1.10e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.952 max_load(range) 0.058-0.125 step 19630/40589 loss 0.2308 acc 0.955 lb 1.977 z 0.414 lr 1.10e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19640/40589 loss 0.2321 acc 0.955 lb 1.968 z 0.396 lr 1.10e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19650/40589 loss 0.2272 acc 0.956 lb 1.974 z 0.461 lr 1.10e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.939 max_load(range) 0.052-0.273 step 19660/40589 loss 0.2381 acc 0.954 lb 1.984 z 0.419 lr 1.10e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19670/40589 loss 0.2337 acc 0.955 lb 1.970 z 0.424 lr 1.10e-04 | 1421 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.955 max_load(range) 0.056-0.161 step 19680/40589 loss 0.2320 acc 0.954 lb 1.964 z 0.402 lr 1.10e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19690/40589 loss 0.2280 acc 0.955 lb 1.973 z 0.451 lr 1.10e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19700/40589 loss 0.2406 acc 0.953 lb 1.987 z 0.418 lr 1.10e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.952 max_load(range) 0.052-0.218 step 19710/40589 loss 0.2345 acc 0.954 lb 1.974 z 0.422 lr 1.10e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19720/40589 loss 0.2398 acc 0.953 lb 1.969 z 0.399 lr 1.09e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.953 max_load(range) 0.053-0.301 step 19730/40589 loss 0.2238 acc 0.956 lb 1.976 z 0.454 lr 1.09e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19740/40589 loss 0.2411 acc 0.954 lb 1.981 z 0.422 lr 1.09e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19750/40589 loss 0.2362 acc 0.954 lb 1.975 z 0.418 lr 1.09e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.962 max_load(range) 0.047-0.325 step 19760/40589 loss 0.2368 acc 0.954 lb 1.964 z 0.399 lr 1.09e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19770/40589 loss 0.2255 acc 0.956 lb 1.975 z 0.467 lr 1.09e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.917 max_load(range) 0.065-0.179 step 19780/40589 loss 0.2382 acc 0.953 lb 1.983 z 0.422 lr 1.09e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19790/40589 loss 0.2337 acc 0.955 lb 1.976 z 0.415 lr 1.09e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19800/40589 loss 0.2321 acc 0.954 lb 1.971 z 0.392 lr 1.09e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.916 max_load(range) 0.065-0.152 step 19810/40589 loss 0.2312 acc 0.954 lb 1.972 z 0.446 lr 1.09e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19820/40589 loss 0.2401 acc 0.953 lb 1.982 z 0.427 lr 1.09e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.939 max_load(range) 0.066-0.128 step 19830/40589 loss 0.2361 acc 0.954 lb 1.978 z 0.408 lr 1.09e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19840/40589 loss 0.2418 acc 0.952 lb 1.967 z 0.400 lr 1.08e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19850/40589 loss 0.2312 acc 0.954 lb 1.969 z 0.449 lr 1.08e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.931 max_load(range) 0.050-0.274 step 19860/40589 loss 0.2460 acc 0.951 lb 1.982 z 0.415 lr 1.08e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19870/40589 loss 0.2381 acc 0.954 lb 1.977 z 0.410 lr 1.08e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.948 max_load(range) 0.060-0.153 step 19880/40589 loss 0.2413 acc 0.952 lb 1.969 z 0.394 lr 1.08e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19890/40589 loss 0.2306 acc 0.955 lb 1.972 z 0.449 lr 1.08e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19900/40589 loss 0.2433 acc 0.952 lb 1.983 z 0.419 lr 1.08e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.945 max_load(range) 0.054-0.234 step 19910/40589 loss 0.2386 acc 0.953 lb 1.974 z 0.415 lr 1.08e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19920/40589 loss 0.2390 acc 0.954 lb 1.969 z 0.398 lr 1.08e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.953 max_load(range) 0.053-0.306 step 19930/40589 loss 0.2314 acc 0.954 lb 1.969 z 0.457 lr 1.08e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19940/40589 loss 0.2442 acc 0.953 lb 1.984 z 0.413 lr 1.08e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19950/40589 loss 0.2394 acc 0.953 lb 1.972 z 0.414 lr 1.08e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.960 max_load(range) 0.049-0.304 step 19960/40589 loss 0.2422 acc 0.952 lb 1.966 z 0.398 lr 1.08e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19970/40589 loss 0.2329 acc 0.954 lb 1.973 z 0.454 lr 1.07e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left [routing] mean entropy_norm 0.919 max_load(range) 0.073-0.142 step 19980/40589 loss 0.2470 acc 0.951 lb 1.981 z 0.436 lr 1.07e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 19990/40589 loss 0.2433 acc 0.952 lb 1.973 z 0.424 lr 1.07e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.8d left step 20000/40589 loss 0.2431 acc 0.952 lb 1.967 z 0.401 lr 1.07e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.934 max_load(range) 0.069-0.137 [save] runs/pretrain_flagship_h100_4x8/step20000 step 20010/40589 loss 0.2310 acc 0.955 lb 1.972 z 0.450 lr 1.07e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20020/40589 loss 0.2457 acc 0.951 lb 1.981 z 0.431 lr 1.07e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.943 max_load(range) 0.065-0.126 step 20030/40589 loss 0.2412 acc 0.952 lb 1.971 z 0.417 lr 1.07e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20040/40589 loss 0.2436 acc 0.952 lb 1.963 z 0.395 lr 1.07e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20050/40589 loss 0.2347 acc 0.954 lb 1.968 z 0.446 lr 1.07e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.918 max_load(range) 0.058-0.293 step 20060/40589 loss 0.2472 acc 0.952 lb 1.977 z 0.427 lr 1.07e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20070/40589 loss 0.2429 acc 0.952 lb 1.973 z 0.409 lr 1.07e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.946 max_load(range) 0.065-0.150 step 20080/40589 loss 0.2415 acc 0.952 lb 1.967 z 0.390 lr 1.07e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20090/40589 loss 0.2337 acc 0.954 lb 1.967 z 0.451 lr 1.06e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20100/40589 loss 0.2503 acc 0.951 lb 1.977 z 0.431 lr 1.06e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.948 max_load(range) 0.054-0.217 step 20110/40589 loss 0.2415 acc 0.952 lb 1.974 z 0.420 lr 1.06e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20120/40589 loss 0.2452 acc 0.952 lb 1.966 z 0.404 lr 1.06e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.952 max_load(range) 0.050-0.349 step 20130/40589 loss 0.2380 acc 0.953 lb 1.969 z 0.455 lr 1.06e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20140/40589 loss 0.2496 acc 0.951 lb 1.979 z 0.426 lr 1.06e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20150/40589 loss 0.2440 acc 0.951 lb 1.971 z 0.425 lr 1.06e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.957 max_load(range) 0.052-0.322 step 20160/40589 loss 0.2447 acc 0.953 lb 1.963 z 0.407 lr 1.06e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20170/40589 loss 0.2367 acc 0.953 lb 1.973 z 0.441 lr 1.06e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.918 max_load(range) 0.068-0.126 step 20180/40589 loss 0.2460 acc 0.952 lb 1.979 z 0.414 lr 1.06e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20190/40589 loss 0.2409 acc 0.952 lb 1.976 z 0.413 lr 1.06e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20200/40589 loss 0.2460 acc 0.951 lb 1.964 z 0.404 lr 1.06e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.927 max_load(range) 0.070-0.157 step 20210/40589 loss 0.2356 acc 0.953 lb 1.970 z 0.452 lr 1.06e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20220/40589 loss 0.2509 acc 0.950 lb 1.977 z 0.426 lr 1.05e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.939 max_load(range) 0.060-0.140 step 20230/40589 loss 0.2448 acc 0.951 lb 1.976 z 0.408 lr 1.05e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20240/40589 loss 0.2506 acc 0.949 lb 1.967 z 0.403 lr 1.05e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20250/40589 loss 0.2344 acc 0.954 lb 1.969 z 0.453 lr 1.05e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.928 max_load(range) 0.050-0.276 step 20260/40589 loss 0.2509 acc 0.951 lb 1.978 z 0.428 lr 1.05e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20270/40589 loss 0.2466 acc 0.951 lb 1.972 z 0.417 lr 1.05e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.953 max_load(range) 0.057-0.162 step 20280/40589 loss 0.2491 acc 0.951 lb 1.965 z 0.401 lr 1.05e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20290/40589 loss 0.2382 acc 0.953 lb 1.973 z 0.440 lr 1.05e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20300/40589 loss 0.2550 acc 0.950 lb 1.980 z 0.426 lr 1.05e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.955 max_load(range) 0.055-0.207 step 20310/40589 loss 0.2495 acc 0.950 lb 1.974 z 0.402 lr 1.05e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20320/40589 loss 0.2517 acc 0.950 lb 1.967 z 0.404 lr 1.05e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.954 max_load(range) 0.051-0.341 step 20330/40589 loss 0.2377 acc 0.954 lb 1.969 z 0.439 lr 1.05e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20340/40589 loss 0.2494 acc 0.951 lb 1.976 z 0.431 lr 1.04e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20350/40589 loss 0.2485 acc 0.951 lb 1.969 z 0.415 lr 1.04e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.962 max_load(range) 0.048-0.314 step 20360/40589 loss 0.2472 acc 0.951 lb 1.958 z 0.401 lr 1.04e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20370/40589 loss 0.2402 acc 0.952 lb 1.970 z 0.435 lr 1.04e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.916 max_load(range) 0.079-0.139 step 20380/40589 loss 0.2511 acc 0.950 lb 1.980 z 0.412 lr 1.04e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20390/40589 loss 0.2457 acc 0.951 lb 1.974 z 0.410 lr 1.04e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20400/40589 loss 0.2515 acc 0.950 lb 1.965 z 0.398 lr 1.04e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.941 max_load(range) 0.067-0.123 step 20410/40589 loss 0.2408 acc 0.952 lb 1.975 z 0.446 lr 1.04e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20420/40589 loss 0.2562 acc 0.950 lb 1.975 z 0.424 lr 1.04e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.920 max_load(range) 0.071-0.165 step 20430/40589 loss 0.2512 acc 0.950 lb 1.972 z 0.415 lr 1.04e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20440/40589 loss 0.2497 acc 0.951 lb 1.968 z 0.405 lr 1.04e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20450/40589 loss 0.2424 acc 0.952 lb 1.969 z 0.452 lr 1.04e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.917 max_load(range) 0.059-0.286 step 20460/40589 loss 0.2522 acc 0.950 lb 1.979 z 0.425 lr 1.04e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20470/40589 loss 0.2544 acc 0.949 lb 1.972 z 0.414 lr 1.03e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.958 max_load(range) 0.052-0.175 step 20480/40589 loss 0.2509 acc 0.950 lb 1.963 z 0.401 lr 1.03e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20490/40589 loss 0.2438 acc 0.951 lb 1.966 z 0.444 lr 1.03e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20500/40589 loss 0.2537 acc 0.949 lb 1.973 z 0.427 lr 1.03e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.949 max_load(range) 0.049-0.213 step 20510/40589 loss 0.2512 acc 0.950 lb 1.970 z 0.418 lr 1.03e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20520/40589 loss 0.2545 acc 0.950 lb 1.963 z 0.409 lr 1.03e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.946 max_load(range) 0.052-0.361 step 20530/40589 loss 0.2450 acc 0.951 lb 1.971 z 0.446 lr 1.03e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20540/40589 loss 0.2572 acc 0.950 lb 1.974 z 0.438 lr 1.03e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20550/40589 loss 0.2508 acc 0.950 lb 1.966 z 0.431 lr 1.03e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.965 max_load(range) 0.045-0.274 step 20560/40589 loss 0.2532 acc 0.949 lb 1.962 z 0.409 lr 1.03e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20570/40589 loss 0.2422 acc 0.952 lb 1.971 z 0.444 lr 1.03e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.932 max_load(range) 0.067-0.131 step 20580/40589 loss 0.2577 acc 0.949 lb 1.978 z 0.426 lr 1.03e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20590/40589 loss 0.2554 acc 0.950 lb 1.973 z 0.407 lr 1.02e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20600/40589 loss 0.2534 acc 0.950 lb 1.964 z 0.402 lr 1.02e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.949 max_load(range) 0.059-0.147 step 20610/40589 loss 0.2450 acc 0.952 lb 1.969 z 0.441 lr 1.02e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20620/40589 loss 0.2564 acc 0.949 lb 1.973 z 0.424 lr 1.02e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.918 max_load(range) 0.074-0.149 step 20630/40589 loss 0.2512 acc 0.949 lb 1.972 z 0.421 lr 1.02e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20640/40589 loss 0.2556 acc 0.949 lb 1.963 z 0.409 lr 1.02e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20650/40589 loss 0.2461 acc 0.952 lb 1.971 z 0.446 lr 1.02e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.908 max_load(range) 0.071-0.309 step 20660/40589 loss 0.2563 acc 0.949 lb 1.978 z 0.416 lr 1.02e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20670/40589 loss 0.2553 acc 0.949 lb 1.970 z 0.408 lr 1.02e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.939 max_load(range) 0.060-0.171 step 20680/40589 loss 0.2595 acc 0.948 lb 1.968 z 0.392 lr 1.02e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20690/40589 loss 0.2446 acc 0.952 lb 1.970 z 0.446 lr 1.02e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20700/40589 loss 0.2614 acc 0.948 lb 1.975 z 0.421 lr 1.02e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.947 max_load(range) 0.064-0.214 step 20710/40589 loss 0.2553 acc 0.949 lb 1.974 z 0.405 lr 1.02e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20720/40589 loss 0.2541 acc 0.949 lb 1.966 z 0.399 lr 1.01e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left [routing] mean entropy_norm 0.953 max_load(range) 0.053-0.355 step 20730/40589 loss 0.2475 acc 0.951 lb 1.967 z 0.441 lr 1.01e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20740/40589 loss 0.2588 acc 0.949 lb 1.974 z 0.425 lr 1.01e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.7d left step 20750/40589 loss 0.2571 acc 0.949 lb 1.970 z 0.413 lr 1.01e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.960 max_load(range) 0.046-0.285 step 20760/40589 loss 0.2585 acc 0.948 lb 1.962 z 0.403 lr 1.01e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 20770/40589 loss 0.2509 acc 0.950 lb 1.970 z 0.447 lr 1.01e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.917 max_load(range) 0.077-0.163 step 20780/40589 loss 0.2608 acc 0.949 lb 1.974 z 0.420 lr 1.01e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 20790/40589 loss 0.2607 acc 0.948 lb 1.975 z 0.410 lr 1.01e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 20800/40589 loss 0.2576 acc 0.949 lb 1.969 z 0.402 lr 1.01e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.940 max_load(range) 0.064-0.172 step 20810/40589 loss 0.2478 acc 0.951 lb 1.973 z 0.439 lr 1.01e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 20820/40589 loss 0.2641 acc 0.948 lb 1.975 z 0.421 lr 1.01e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.911 max_load(range) 0.078-0.205 step 20830/40589 loss 0.2567 acc 0.949 lb 1.972 z 0.410 lr 1.01e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 20840/40589 loss 0.2595 acc 0.949 lb 1.963 z 0.403 lr 1.01e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 20850/40589 loss 0.2456 acc 0.951 lb 1.970 z 0.438 lr 1.00e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.901 max_load(range) 0.066-0.295 step 20860/40589 loss 0.2618 acc 0.947 lb 1.974 z 0.430 lr 1.00e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 20870/40589 loss 0.2595 acc 0.948 lb 1.971 z 0.408 lr 1.00e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.946 max_load(range) 0.061-0.176 step 20880/40589 loss 0.2596 acc 0.948 lb 1.964 z 0.406 lr 1.00e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 20890/40589 loss 0.2504 acc 0.950 lb 1.969 z 0.441 lr 1.00e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 20900/40589 loss 0.2656 acc 0.947 lb 1.976 z 0.422 lr 1.00e-04 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.944 max_load(range) 0.047-0.221 step 20910/40589 loss 0.2630 acc 0.947 lb 1.971 z 0.412 lr 9.99e-05 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 20920/40589 loss 0.2636 acc 0.947 lb 1.964 z 0.398 lr 9.99e-05 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.946 max_load(range) 0.056-0.377 step 20930/40589 loss 0.2505 acc 0.950 lb 1.970 z 0.441 lr 9.98e-05 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 20940/40589 loss 0.2634 acc 0.947 lb 1.975 z 0.425 lr 9.97e-05 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 20950/40589 loss 0.2579 acc 0.949 lb 1.974 z 0.413 lr 9.96e-05 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.957 max_load(range) 0.048-0.290 step 20960/40589 loss 0.2610 acc 0.948 lb 1.966 z 0.413 lr 9.95e-05 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 20970/40589 loss 0.2508 acc 0.950 lb 1.973 z 0.439 lr 9.95e-05 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.918 max_load(range) 0.068-0.147 step 20980/40589 loss 0.2646 acc 0.947 lb 1.975 z 0.425 lr 9.94e-05 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 20990/40589 loss 0.2595 acc 0.948 lb 1.969 z 0.418 lr 9.93e-05 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21000/40589 loss 0.2583 acc 0.948 lb 1.961 z 0.406 lr 9.92e-05 | 1420 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.949 max_load(range) 0.051-0.147 [save] runs/pretrain_flagship_h100_4x8/step21000 step 21010/40589 loss 0.2526 acc 0.950 lb 1.969 z 0.435 lr 9.91e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21020/40589 loss 0.2654 acc 0.947 lb 1.972 z 0.433 lr 9.91e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.909 max_load(range) 0.065-0.169 step 21030/40589 loss 0.2605 acc 0.948 lb 1.968 z 0.414 lr 9.90e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21040/40589 loss 0.2628 acc 0.948 lb 1.965 z 0.404 lr 9.89e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21050/40589 loss 0.2522 acc 0.950 lb 1.967 z 0.442 lr 9.88e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.907 max_load(range) 0.062-0.297 step 21060/40589 loss 0.2646 acc 0.948 lb 1.975 z 0.431 lr 9.87e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21070/40589 loss 0.2619 acc 0.948 lb 1.973 z 0.404 lr 9.87e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.949 max_load(range) 0.058-0.179 step 21080/40589 loss 0.2629 acc 0.947 lb 1.965 z 0.400 lr 9.86e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21090/40589 loss 0.2542 acc 0.949 lb 1.967 z 0.428 lr 9.85e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21100/40589 loss 0.2716 acc 0.946 lb 1.973 z 0.417 lr 9.84e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.946 max_load(range) 0.053-0.223 step 21110/40589 loss 0.2624 acc 0.948 lb 1.970 z 0.403 lr 9.83e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21120/40589 loss 0.2639 acc 0.947 lb 1.964 z 0.404 lr 9.83e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.949 max_load(range) 0.053-0.359 step 21130/40589 loss 0.2520 acc 0.950 lb 1.969 z 0.440 lr 9.82e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21140/40589 loss 0.2694 acc 0.946 lb 1.973 z 0.426 lr 9.81e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21150/40589 loss 0.2641 acc 0.948 lb 1.974 z 0.400 lr 9.80e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.962 max_load(range) 0.045-0.269 step 21160/40589 loss 0.2653 acc 0.947 lb 1.961 z 0.407 lr 9.79e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21170/40589 loss 0.2528 acc 0.950 lb 1.969 z 0.440 lr 9.79e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.926 max_load(range) 0.070-0.145 step 21180/40589 loss 0.2682 acc 0.947 lb 1.968 z 0.438 lr 9.78e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21190/40589 loss 0.2652 acc 0.947 lb 1.969 z 0.412 lr 9.77e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21200/40589 loss 0.2601 acc 0.949 lb 1.958 z 0.401 lr 9.76e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.934 max_load(range) 0.068-0.208 step 21210/40589 loss 0.2554 acc 0.949 lb 1.970 z 0.434 lr 9.76e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21220/40589 loss 0.2650 acc 0.947 lb 1.969 z 0.422 lr 9.75e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.897 max_load(range) 0.069-0.187 step 21230/40589 loss 0.2628 acc 0.947 lb 1.966 z 0.404 lr 9.74e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21240/40589 loss 0.2649 acc 0.947 lb 1.966 z 0.400 lr 9.73e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21250/40589 loss 0.2516 acc 0.950 lb 1.969 z 0.426 lr 9.72e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.914 max_load(range) 0.060-0.267 step 21260/40589 loss 0.2694 acc 0.946 lb 1.972 z 0.418 lr 9.72e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21270/40589 loss 0.2647 acc 0.947 lb 1.972 z 0.408 lr 9.71e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.948 max_load(range) 0.055-0.183 step 21280/40589 loss 0.2640 acc 0.947 lb 1.964 z 0.409 lr 9.70e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21290/40589 loss 0.2522 acc 0.949 lb 1.969 z 0.427 lr 9.69e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21300/40589 loss 0.2696 acc 0.946 lb 1.968 z 0.431 lr 9.68e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.947 max_load(range) 0.053-0.209 step 21310/40589 loss 0.2692 acc 0.946 lb 1.967 z 0.414 lr 9.68e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21320/40589 loss 0.2622 acc 0.947 lb 1.964 z 0.413 lr 9.67e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.944 max_load(range) 0.054-0.371 step 21330/40589 loss 0.2542 acc 0.949 lb 1.971 z 0.437 lr 9.66e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21340/40589 loss 0.2674 acc 0.947 lb 1.973 z 0.419 lr 9.65e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21350/40589 loss 0.2689 acc 0.946 lb 1.972 z 0.404 lr 9.64e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.958 max_load(range) 0.050-0.248 step 21360/40589 loss 0.2643 acc 0.946 lb 1.961 z 0.396 lr 9.64e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21370/40589 loss 0.2555 acc 0.949 lb 1.971 z 0.425 lr 9.63e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.919 max_load(range) 0.079-0.148 step 21380/40589 loss 0.2679 acc 0.947 lb 1.969 z 0.427 lr 9.62e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21390/40589 loss 0.2699 acc 0.946 lb 1.970 z 0.401 lr 9.61e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21400/40589 loss 0.2663 acc 0.946 lb 1.963 z 0.398 lr 9.60e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.933 max_load(range) 0.067-0.155 step 21410/40589 loss 0.2558 acc 0.949 lb 1.966 z 0.435 lr 9.60e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21420/40589 loss 0.2703 acc 0.946 lb 1.970 z 0.427 lr 9.59e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.914 max_load(range) 0.075-0.162 step 21430/40589 loss 0.2691 acc 0.946 lb 1.971 z 0.398 lr 9.58e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21440/40589 loss 0.2676 acc 0.946 lb 1.964 z 0.398 lr 9.57e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21450/40589 loss 0.2553 acc 0.949 lb 1.968 z 0.427 lr 9.56e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.894 max_load(range) 0.065-0.310 step 21460/40589 loss 0.2737 acc 0.945 lb 1.971 z 0.412 lr 9.56e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21470/40589 loss 0.2710 acc 0.946 lb 1.970 z 0.402 lr 9.55e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.948 max_load(range) 0.056-0.210 step 21480/40589 loss 0.2699 acc 0.946 lb 1.967 z 0.403 lr 9.54e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21490/40589 loss 0.2596 acc 0.948 lb 1.970 z 0.436 lr 9.53e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left step 21500/40589 loss 0.2696 acc 0.947 lb 1.968 z 0.424 lr 9.52e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.6d left [routing] mean entropy_norm 0.947 max_load(range) 0.050-0.210 step 21510/40589 loss 0.2665 acc 0.946 lb 1.971 z 0.407 lr 9.52e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21520/40589 loss 0.2688 acc 0.947 lb 1.962 z 0.407 lr 9.51e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.947 max_load(range) 0.054-0.359 step 21530/40589 loss 0.2561 acc 0.949 lb 1.967 z 0.439 lr 9.50e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21540/40589 loss 0.2705 acc 0.946 lb 1.972 z 0.420 lr 9.49e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21550/40589 loss 0.2685 acc 0.946 lb 1.966 z 0.411 lr 9.48e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.961 max_load(range) 0.048-0.254 step 21560/40589 loss 0.2687 acc 0.946 lb 1.961 z 0.408 lr 9.48e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21570/40589 loss 0.2583 acc 0.948 lb 1.970 z 0.433 lr 9.47e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.926 max_load(range) 0.067-0.112 step 21580/40589 loss 0.2748 acc 0.945 lb 1.974 z 0.421 lr 9.46e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21590/40589 loss 0.2726 acc 0.946 lb 1.965 z 0.413 lr 9.45e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21600/40589 loss 0.2701 acc 0.945 lb 1.960 z 0.403 lr 9.44e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.941 max_load(range) 0.071-0.155 step 21610/40589 loss 0.2606 acc 0.948 lb 1.970 z 0.426 lr 9.44e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21620/40589 loss 0.2722 acc 0.945 lb 1.973 z 0.425 lr 9.43e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.908 max_load(range) 0.071-0.188 step 21630/40589 loss 0.2725 acc 0.945 lb 1.972 z 0.401 lr 9.42e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21640/40589 loss 0.2688 acc 0.946 lb 1.966 z 0.396 lr 9.41e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21650/40589 loss 0.2583 acc 0.948 lb 1.974 z 0.419 lr 9.40e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.904 max_load(range) 0.069-0.300 step 21660/40589 loss 0.2725 acc 0.946 lb 1.968 z 0.417 lr 9.40e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21670/40589 loss 0.2700 acc 0.946 lb 1.971 z 0.397 lr 9.39e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.947 max_load(range) 0.050-0.202 step 21680/40589 loss 0.2712 acc 0.945 lb 1.962 z 0.400 lr 9.38e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21690/40589 loss 0.2597 acc 0.949 lb 1.968 z 0.423 lr 9.37e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21700/40589 loss 0.2763 acc 0.946 lb 1.969 z 0.415 lr 9.36e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.944 max_load(range) 0.051-0.220 step 21710/40589 loss 0.2701 acc 0.946 lb 1.967 z 0.398 lr 9.36e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21720/40589 loss 0.2714 acc 0.945 lb 1.964 z 0.397 lr 9.35e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.943 max_load(range) 0.054-0.377 step 21730/40589 loss 0.2567 acc 0.949 lb 1.966 z 0.430 lr 9.34e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21740/40589 loss 0.2718 acc 0.946 lb 1.969 z 0.420 lr 9.33e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21750/40589 loss 0.2733 acc 0.945 lb 1.964 z 0.400 lr 9.32e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.961 max_load(range) 0.050-0.242 step 21760/40589 loss 0.2678 acc 0.947 lb 1.957 z 0.402 lr 9.32e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21770/40589 loss 0.2593 acc 0.948 lb 1.966 z 0.421 lr 9.31e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.923 max_load(range) 0.072-0.134 step 21780/40589 loss 0.2707 acc 0.946 lb 1.970 z 0.425 lr 9.30e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21790/40589 loss 0.2720 acc 0.945 lb 1.969 z 0.405 lr 9.29e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21800/40589 loss 0.2665 acc 0.947 lb 1.960 z 0.404 lr 9.28e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.943 max_load(range) 0.061-0.172 step 21810/40589 loss 0.2555 acc 0.949 lb 1.963 z 0.433 lr 9.28e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21820/40589 loss 0.2751 acc 0.945 lb 1.966 z 0.428 lr 9.27e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.910 max_load(range) 0.080-0.200 step 21830/40589 loss 0.2710 acc 0.945 lb 1.971 z 0.397 lr 9.26e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21840/40589 loss 0.2686 acc 0.946 lb 1.966 z 0.398 lr 9.25e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21850/40589 loss 0.2599 acc 0.948 lb 1.967 z 0.428 lr 9.25e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.894 max_load(range) 0.068-0.317 step 21860/40589 loss 0.2698 acc 0.946 lb 1.972 z 0.423 lr 9.24e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21870/40589 loss 0.2728 acc 0.945 lb 1.971 z 0.400 lr 9.23e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.946 max_load(range) 0.048-0.203 step 21880/40589 loss 0.2682 acc 0.946 lb 1.969 z 0.399 lr 9.22e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21890/40589 loss 0.2630 acc 0.947 lb 1.973 z 0.422 lr 9.21e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21900/40589 loss 0.2708 acc 0.946 lb 1.973 z 0.419 lr 9.21e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.941 max_load(range) 0.057-0.221 step 21910/40589 loss 0.2746 acc 0.944 lb 1.966 z 0.405 lr 9.20e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21920/40589 loss 0.2702 acc 0.945 lb 1.966 z 0.399 lr 9.19e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.942 max_load(range) 0.054-0.389 step 21930/40589 loss 0.2642 acc 0.947 lb 1.971 z 0.424 lr 9.18e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21940/40589 loss 0.2742 acc 0.944 lb 1.971 z 0.419 lr 9.17e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21950/40589 loss 0.2702 acc 0.946 lb 1.971 z 0.402 lr 9.17e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.954 max_load(range) 0.056-0.220 step 21960/40589 loss 0.2698 acc 0.946 lb 1.967 z 0.394 lr 9.16e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21970/40589 loss 0.2599 acc 0.947 lb 1.974 z 0.426 lr 9.15e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.926 max_load(range) 0.067-0.170 step 21980/40589 loss 0.2777 acc 0.945 lb 1.970 z 0.418 lr 9.14e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 21990/40589 loss 0.2714 acc 0.945 lb 1.969 z 0.396 lr 9.13e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 22000/40589 loss 0.2718 acc 0.945 lb 1.963 z 0.400 lr 9.13e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.934 max_load(range) 0.066-0.171 [save] runs/pretrain_flagship_h100_4x8/step22000 step 22010/40589 loss 0.2576 acc 0.948 lb 1.966 z 0.429 lr 9.12e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 22020/40589 loss 0.2771 acc 0.945 lb 1.969 z 0.421 lr 9.11e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.896 max_load(range) 0.075-0.224 step 22030/40589 loss 0.2763 acc 0.945 lb 1.968 z 0.403 lr 9.10e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 22040/40589 loss 0.2703 acc 0.945 lb 1.965 z 0.397 lr 9.09e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 22050/40589 loss 0.2598 acc 0.948 lb 1.970 z 0.414 lr 9.09e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.891 max_load(range) 0.070-0.305 step 22060/40589 loss 0.2766 acc 0.944 lb 1.972 z 0.424 lr 9.08e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 22070/40589 loss 0.2742 acc 0.945 lb 1.972 z 0.400 lr 9.07e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.946 max_load(range) 0.054-0.211 step 22080/40589 loss 0.2744 acc 0.945 lb 1.963 z 0.403 lr 9.06e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 22090/40589 loss 0.2614 acc 0.948 lb 1.968 z 0.429 lr 9.05e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 22100/40589 loss 0.2755 acc 0.945 lb 1.973 z 0.409 lr 9.05e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.940 max_load(range) 0.059-0.222 step 22110/40589 loss 0.2749 acc 0.945 lb 1.967 z 0.394 lr 9.04e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 22120/40589 loss 0.2768 acc 0.944 lb 1.961 z 0.401 lr 9.03e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.940 max_load(range) 0.048-0.380 step 22130/40589 loss 0.2666 acc 0.947 lb 1.967 z 0.420 lr 9.02e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 22140/40589 loss 0.2781 acc 0.944 lb 1.971 z 0.414 lr 9.01e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 22150/40589 loss 0.2722 acc 0.945 lb 1.972 z 0.390 lr 9.01e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.957 max_load(range) 0.053-0.254 step 22160/40589 loss 0.2697 acc 0.946 lb 1.961 z 0.391 lr 9.00e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 22170/40589 loss 0.2605 acc 0.948 lb 1.968 z 0.429 lr 8.99e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.915 max_load(range) 0.083-0.220 step 22180/40589 loss 0.2810 acc 0.943 lb 1.967 z 0.411 lr 8.98e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 22190/40589 loss 0.2741 acc 0.944 lb 1.970 z 0.396 lr 8.97e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 22200/40589 loss 0.2693 acc 0.945 lb 1.960 z 0.395 lr 8.97e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.939 max_load(range) 0.067-0.163 step 22210/40589 loss 0.2627 acc 0.947 lb 1.969 z 0.425 lr 8.96e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 22220/40589 loss 0.2806 acc 0.944 lb 1.968 z 0.421 lr 8.95e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.892 max_load(range) 0.069-0.212 step 22230/40589 loss 0.2763 acc 0.944 lb 1.968 z 0.403 lr 8.94e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 22240/40589 loss 0.2748 acc 0.945 lb 1.965 z 0.395 lr 8.94e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left step 22250/40589 loss 0.2593 acc 0.948 lb 1.970 z 0.417 lr 8.93e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.5d left [routing] mean entropy_norm 0.884 max_load(range) 0.068-0.315 step 22260/40589 loss 0.2716 acc 0.945 lb 1.969 z 0.416 lr 8.92e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22270/40589 loss 0.2703 acc 0.945 lb 1.966 z 0.401 lr 8.91e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.949 max_load(range) 0.049-0.189 step 22280/40589 loss 0.2709 acc 0.946 lb 1.964 z 0.396 lr 8.90e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22290/40589 loss 0.2633 acc 0.948 lb 1.969 z 0.410 lr 8.90e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22300/40589 loss 0.2775 acc 0.944 lb 1.968 z 0.413 lr 8.89e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.941 max_load(range) 0.059-0.231 step 22310/40589 loss 0.2757 acc 0.945 lb 1.963 z 0.405 lr 8.88e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22320/40589 loss 0.2758 acc 0.944 lb 1.962 z 0.409 lr 8.87e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.936 max_load(range) 0.050-0.393 step 22330/40589 loss 0.2606 acc 0.948 lb 1.974 z 0.422 lr 8.86e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22340/40589 loss 0.2750 acc 0.945 lb 1.967 z 0.414 lr 8.86e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22350/40589 loss 0.2778 acc 0.944 lb 1.967 z 0.404 lr 8.85e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.957 max_load(range) 0.053-0.242 step 22360/40589 loss 0.2716 acc 0.945 lb 1.959 z 0.407 lr 8.84e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22370/40589 loss 0.2570 acc 0.948 lb 1.970 z 0.418 lr 8.83e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.929 max_load(range) 0.058-0.169 step 22380/40589 loss 0.2728 acc 0.945 lb 1.970 z 0.413 lr 8.82e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22390/40589 loss 0.2762 acc 0.945 lb 1.966 z 0.396 lr 8.82e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22400/40589 loss 0.2770 acc 0.944 lb 1.964 z 0.411 lr 8.81e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.943 max_load(range) 0.062-0.175 step 22410/40589 loss 0.2586 acc 0.948 lb 1.969 z 0.413 lr 8.80e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22420/40589 loss 0.2766 acc 0.944 lb 1.969 z 0.407 lr 8.79e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.894 max_load(range) 0.080-0.197 step 22430/40589 loss 0.2761 acc 0.944 lb 1.969 z 0.389 lr 8.78e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22440/40589 loss 0.2756 acc 0.945 lb 1.961 z 0.396 lr 8.78e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22450/40589 loss 0.2624 acc 0.947 lb 1.972 z 0.412 lr 8.77e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.895 max_load(range) 0.068-0.304 step 22460/40589 loss 0.2740 acc 0.945 lb 1.970 z 0.404 lr 8.76e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22470/40589 loss 0.2785 acc 0.943 lb 1.967 z 0.389 lr 8.75e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.942 max_load(range) 0.061-0.202 step 22480/40589 loss 0.2733 acc 0.945 lb 1.963 z 0.396 lr 8.74e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22490/40589 loss 0.2622 acc 0.947 lb 1.972 z 0.410 lr 8.74e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22500/40589 loss 0.2818 acc 0.944 lb 1.970 z 0.407 lr 8.73e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.943 max_load(range) 0.053-0.213 step 22510/40589 loss 0.2777 acc 0.944 lb 1.970 z 0.389 lr 8.72e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22520/40589 loss 0.2745 acc 0.944 lb 1.960 z 0.392 lr 8.71e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.937 max_load(range) 0.058-0.386 step 22530/40589 loss 0.2654 acc 0.946 lb 1.970 z 0.411 lr 8.71e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22540/40589 loss 0.2773 acc 0.944 lb 1.968 z 0.406 lr 8.70e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22550/40589 loss 0.2760 acc 0.944 lb 1.971 z 0.390 lr 8.69e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.949 max_load(range) 0.050-0.298 step 22560/40589 loss 0.2776 acc 0.944 lb 1.965 z 0.385 lr 8.68e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22570/40589 loss 0.2642 acc 0.947 lb 1.972 z 0.407 lr 8.67e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.920 max_load(range) 0.072-0.174 step 22580/40589 loss 0.2784 acc 0.944 lb 1.972 z 0.402 lr 8.67e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22590/40589 loss 0.2753 acc 0.944 lb 1.968 z 0.387 lr 8.66e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22600/40589 loss 0.2781 acc 0.943 lb 1.961 z 0.391 lr 8.65e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.943 max_load(range) 0.060-0.169 step 22610/40589 loss 0.2587 acc 0.948 lb 1.967 z 0.409 lr 8.64e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22620/40589 loss 0.2723 acc 0.945 lb 1.970 z 0.394 lr 8.63e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.875 max_load(range) 0.071-0.161 step 22630/40589 loss 0.2777 acc 0.943 lb 1.970 z 0.379 lr 8.63e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22640/40589 loss 0.2745 acc 0.945 lb 1.962 z 0.388 lr 8.62e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22650/40589 loss 0.2600 acc 0.947 lb 1.968 z 0.418 lr 8.61e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.896 max_load(range) 0.064-0.296 step 22660/40589 loss 0.2753 acc 0.945 lb 1.970 z 0.406 lr 8.60e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22670/40589 loss 0.2739 acc 0.945 lb 1.968 z 0.389 lr 8.59e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.935 max_load(range) 0.058-0.218 step 22680/40589 loss 0.2757 acc 0.944 lb 1.963 z 0.392 lr 8.59e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22690/40589 loss 0.2595 acc 0.948 lb 1.967 z 0.409 lr 8.58e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22700/40589 loss 0.2799 acc 0.944 lb 1.970 z 0.401 lr 8.57e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.936 max_load(range) 0.059-0.224 step 22710/40589 loss 0.2759 acc 0.945 lb 1.972 z 0.380 lr 8.56e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22720/40589 loss 0.2756 acc 0.945 lb 1.966 z 0.388 lr 8.56e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.938 max_load(range) 0.052-0.390 step 22730/40589 loss 0.2648 acc 0.947 lb 1.970 z 0.407 lr 8.55e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22740/40589 loss 0.2777 acc 0.945 lb 1.967 z 0.406 lr 8.54e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22750/40589 loss 0.2763 acc 0.944 lb 1.970 z 0.390 lr 8.53e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.952 max_load(range) 0.051-0.251 step 22760/40589 loss 0.2734 acc 0.945 lb 1.963 z 0.384 lr 8.52e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22770/40589 loss 0.2581 acc 0.948 lb 1.970 z 0.407 lr 8.52e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.923 max_load(range) 0.072-0.159 step 22780/40589 loss 0.2772 acc 0.945 lb 1.968 z 0.401 lr 8.51e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22790/40589 loss 0.2798 acc 0.943 lb 1.962 z 0.389 lr 8.50e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22800/40589 loss 0.2682 acc 0.946 lb 1.962 z 0.387 lr 8.49e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.939 max_load(range) 0.063-0.170 step 22810/40589 loss 0.2626 acc 0.946 lb 1.971 z 0.405 lr 8.48e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22820/40589 loss 0.2794 acc 0.944 lb 1.973 z 0.403 lr 8.48e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.889 max_load(range) 0.071-0.196 step 22830/40589 loss 0.2768 acc 0.944 lb 1.969 z 0.387 lr 8.47e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22840/40589 loss 0.2714 acc 0.945 lb 1.965 z 0.395 lr 8.46e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22850/40589 loss 0.2635 acc 0.946 lb 1.968 z 0.415 lr 8.45e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.905 max_load(range) 0.065-0.286 step 22860/40589 loss 0.2739 acc 0.945 lb 1.968 z 0.412 lr 8.44e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22870/40589 loss 0.2763 acc 0.944 lb 1.962 z 0.397 lr 8.44e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.943 max_load(range) 0.051-0.220 step 22880/40589 loss 0.2751 acc 0.944 lb 1.962 z 0.393 lr 8.43e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22890/40589 loss 0.2674 acc 0.946 lb 1.973 z 0.400 lr 8.42e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22900/40589 loss 0.2767 acc 0.944 lb 1.968 z 0.402 lr 8.41e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.940 max_load(range) 0.053-0.224 step 22910/40589 loss 0.2741 acc 0.943 lb 1.969 z 0.390 lr 8.41e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22920/40589 loss 0.2746 acc 0.944 lb 1.960 z 0.392 lr 8.40e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.938 max_load(range) 0.052-0.390 step 22930/40589 loss 0.2625 acc 0.947 lb 1.969 z 0.406 lr 8.39e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22940/40589 loss 0.2811 acc 0.944 lb 1.969 z 0.405 lr 8.38e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22950/40589 loss 0.2790 acc 0.944 lb 1.972 z 0.384 lr 8.37e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.958 max_load(range) 0.047-0.252 step 22960/40589 loss 0.2743 acc 0.945 lb 1.963 z 0.395 lr 8.37e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22970/40589 loss 0.2612 acc 0.947 lb 1.973 z 0.408 lr 8.36e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.926 max_load(range) 0.076-0.147 step 22980/40589 loss 0.2817 acc 0.944 lb 1.969 z 0.406 lr 8.35e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 22990/40589 loss 0.2792 acc 0.944 lb 1.969 z 0.378 lr 8.34e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 23000/40589 loss 0.2784 acc 0.944 lb 1.965 z 0.387 lr 8.33e-05 | 1419 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left [routing] mean entropy_norm 0.928 max_load(range) 0.061-0.170 [save] runs/pretrain_flagship_h100_4x8/step23000 step 23010/40589 loss 0.2636 acc 0.947 lb 1.972 z 0.405 lr 8.33e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.4d left step 23020/40589 loss 0.2801 acc 0.944 lb 1.970 z 0.399 lr 8.32e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.895 max_load(range) 0.073-0.181 step 23030/40589 loss 0.2785 acc 0.944 lb 1.970 z 0.389 lr 8.31e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23040/40589 loss 0.2755 acc 0.944 lb 1.962 z 0.383 lr 8.30e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23050/40589 loss 0.2641 acc 0.947 lb 1.968 z 0.411 lr 8.30e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.890 max_load(range) 0.067-0.309 step 23060/40589 loss 0.2808 acc 0.944 lb 1.969 z 0.402 lr 8.29e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23070/40589 loss 0.2810 acc 0.942 lb 1.969 z 0.392 lr 8.28e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.948 max_load(range) 0.054-0.213 step 23080/40589 loss 0.2757 acc 0.944 lb 1.962 z 0.389 lr 8.27e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23090/40589 loss 0.2626 acc 0.947 lb 1.967 z 0.399 lr 8.26e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23100/40589 loss 0.2792 acc 0.944 lb 1.968 z 0.399 lr 8.26e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.939 max_load(range) 0.052-0.225 step 23110/40589 loss 0.2786 acc 0.944 lb 1.969 z 0.380 lr 8.25e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23120/40589 loss 0.2802 acc 0.943 lb 1.958 z 0.386 lr 8.24e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.942 max_load(range) 0.055-0.377 step 23130/40589 loss 0.2655 acc 0.947 lb 1.972 z 0.396 lr 8.23e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23140/40589 loss 0.2790 acc 0.944 lb 1.971 z 0.395 lr 8.22e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23150/40589 loss 0.2788 acc 0.943 lb 1.967 z 0.387 lr 8.22e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.950 max_load(range) 0.065-0.200 step 23160/40589 loss 0.2763 acc 0.944 lb 1.963 z 0.389 lr 8.21e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23170/40589 loss 0.2641 acc 0.947 lb 1.968 z 0.400 lr 8.20e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.913 max_load(range) 0.083-0.173 step 23180/40589 loss 0.2757 acc 0.944 lb 1.970 z 0.401 lr 8.19e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23190/40589 loss 0.2746 acc 0.944 lb 1.966 z 0.392 lr 8.19e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23200/40589 loss 0.2767 acc 0.944 lb 1.962 z 0.387 lr 8.18e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.940 max_load(range) 0.065-0.166 step 23210/40589 loss 0.2616 acc 0.947 lb 1.970 z 0.402 lr 8.17e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23220/40589 loss 0.2796 acc 0.944 lb 1.965 z 0.402 lr 8.16e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.880 max_load(range) 0.068-0.198 step 23230/40589 loss 0.2801 acc 0.943 lb 1.972 z 0.385 lr 8.15e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23240/40589 loss 0.2791 acc 0.944 lb 1.963 z 0.382 lr 8.15e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23250/40589 loss 0.2644 acc 0.947 lb 1.969 z 0.412 lr 8.14e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.885 max_load(range) 0.067-0.312 step 23260/40589 loss 0.2799 acc 0.943 lb 1.969 z 0.411 lr 8.13e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23270/40589 loss 0.2822 acc 0.942 lb 1.965 z 0.386 lr 8.12e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.936 max_load(range) 0.050-0.226 step 23280/40589 loss 0.2791 acc 0.943 lb 1.958 z 0.390 lr 8.11e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23290/40589 loss 0.2594 acc 0.947 lb 1.969 z 0.410 lr 8.11e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23300/40589 loss 0.2788 acc 0.944 lb 1.967 z 0.401 lr 8.10e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.934 max_load(range) 0.056-0.213 step 23310/40589 loss 0.2763 acc 0.944 lb 1.968 z 0.382 lr 8.09e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23320/40589 loss 0.2764 acc 0.944 lb 1.963 z 0.390 lr 8.08e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.932 max_load(range) 0.054-0.391 step 23330/40589 loss 0.2672 acc 0.947 lb 1.968 z 0.397 lr 8.08e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23340/40589 loss 0.2804 acc 0.943 lb 1.967 z 0.402 lr 8.07e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23350/40589 loss 0.2792 acc 0.944 lb 1.966 z 0.375 lr 8.06e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.956 max_load(range) 0.055-0.243 step 23360/40589 loss 0.2760 acc 0.944 lb 1.958 z 0.388 lr 8.05e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23370/40589 loss 0.2654 acc 0.947 lb 1.966 z 0.409 lr 8.04e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.924 max_load(range) 0.070-0.151 step 23380/40589 loss 0.2780 acc 0.944 lb 1.968 z 0.403 lr 8.04e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23390/40589 loss 0.2772 acc 0.943 lb 1.966 z 0.381 lr 8.03e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23400/40589 loss 0.2785 acc 0.944 lb 1.962 z 0.387 lr 8.02e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.933 max_load(range) 0.062-0.161 step 23410/40589 loss 0.2694 acc 0.945 lb 1.972 z 0.394 lr 8.01e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23420/40589 loss 0.2758 acc 0.945 lb 1.967 z 0.400 lr 8.01e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.891 max_load(range) 0.078-0.190 step 23430/40589 loss 0.2799 acc 0.944 lb 1.968 z 0.388 lr 8.00e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23440/40589 loss 0.2736 acc 0.945 lb 1.965 z 0.380 lr 7.99e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23450/40589 loss 0.2613 acc 0.947 lb 1.970 z 0.394 lr 7.98e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.883 max_load(range) 0.069-0.318 step 23460/40589 loss 0.2786 acc 0.944 lb 1.968 z 0.393 lr 7.97e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23470/40589 loss 0.2764 acc 0.944 lb 1.967 z 0.378 lr 7.97e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.946 max_load(range) 0.050-0.206 step 23480/40589 loss 0.2751 acc 0.944 lb 1.964 z 0.384 lr 7.96e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23490/40589 loss 0.2681 acc 0.946 lb 1.970 z 0.401 lr 7.95e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23500/40589 loss 0.2784 acc 0.944 lb 1.966 z 0.400 lr 7.94e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.938 max_load(range) 0.056-0.218 step 23510/40589 loss 0.2805 acc 0.943 lb 1.968 z 0.377 lr 7.93e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23520/40589 loss 0.2756 acc 0.945 lb 1.962 z 0.383 lr 7.93e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.936 max_load(range) 0.057-0.404 step 23530/40589 loss 0.2656 acc 0.946 lb 1.969 z 0.397 lr 7.92e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23540/40589 loss 0.2837 acc 0.943 lb 1.966 z 0.393 lr 7.91e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23550/40589 loss 0.2804 acc 0.943 lb 1.967 z 0.377 lr 7.90e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.955 max_load(range) 0.055-0.237 step 23560/40589 loss 0.2776 acc 0.944 lb 1.961 z 0.381 lr 7.90e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23570/40589 loss 0.2617 acc 0.947 lb 1.966 z 0.401 lr 7.89e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.924 max_load(range) 0.080-0.159 step 23580/40589 loss 0.2808 acc 0.943 lb 1.969 z 0.397 lr 7.88e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23590/40589 loss 0.2751 acc 0.944 lb 1.965 z 0.387 lr 7.87e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23600/40589 loss 0.2792 acc 0.944 lb 1.962 z 0.383 lr 7.86e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.944 max_load(range) 0.058-0.156 step 23610/40589 loss 0.2637 acc 0.946 lb 1.971 z 0.398 lr 7.86e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23620/40589 loss 0.2834 acc 0.942 lb 1.970 z 0.399 lr 7.85e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.891 max_load(range) 0.068-0.179 step 23630/40589 loss 0.2811 acc 0.944 lb 1.967 z 0.375 lr 7.84e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23640/40589 loss 0.2755 acc 0.944 lb 1.964 z 0.379 lr 7.83e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23650/40589 loss 0.2627 acc 0.947 lb 1.970 z 0.391 lr 7.83e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.893 max_load(range) 0.069-0.310 step 23660/40589 loss 0.2793 acc 0.944 lb 1.968 z 0.390 lr 7.82e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23670/40589 loss 0.2828 acc 0.942 lb 1.968 z 0.386 lr 7.81e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.944 max_load(range) 0.059-0.217 step 23680/40589 loss 0.2752 acc 0.945 lb 1.962 z 0.376 lr 7.80e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23690/40589 loss 0.2617 acc 0.947 lb 1.972 z 0.395 lr 7.79e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23700/40589 loss 0.2808 acc 0.944 lb 1.970 z 0.394 lr 7.79e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.935 max_load(range) 0.058-0.227 step 23710/40589 loss 0.2779 acc 0.944 lb 1.968 z 0.383 lr 7.78e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23720/40589 loss 0.2732 acc 0.945 lb 1.964 z 0.386 lr 7.77e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.934 max_load(range) 0.057-0.390 step 23730/40589 loss 0.2614 acc 0.948 lb 1.973 z 0.399 lr 7.76e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23740/40589 loss 0.2817 acc 0.943 lb 1.965 z 0.394 lr 7.76e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23750/40589 loss 0.2787 acc 0.943 lb 1.970 z 0.378 lr 7.75e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left [routing] mean entropy_norm 0.951 max_load(range) 0.060-0.251 step 23760/40589 loss 0.2710 acc 0.945 lb 1.961 z 0.387 lr 7.74e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.3d left step 23770/40589 loss 0.2675 acc 0.946 lb 1.968 z 0.404 lr 7.73e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.921 max_load(range) 0.070-0.142 step 23780/40589 loss 0.2784 acc 0.944 lb 1.964 z 0.397 lr 7.72e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 23790/40589 loss 0.2781 acc 0.944 lb 1.968 z 0.377 lr 7.72e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 23800/40589 loss 0.2761 acc 0.944 lb 1.964 z 0.372 lr 7.71e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.930 max_load(range) 0.063-0.171 step 23810/40589 loss 0.2628 acc 0.947 lb 1.967 z 0.402 lr 7.70e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 23820/40589 loss 0.2786 acc 0.944 lb 1.967 z 0.407 lr 7.69e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.878 max_load(range) 0.068-0.178 step 23830/40589 loss 0.2792 acc 0.943 lb 1.968 z 0.379 lr 7.69e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 23840/40589 loss 0.2772 acc 0.943 lb 1.964 z 0.378 lr 7.68e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 23850/40589 loss 0.2646 acc 0.947 lb 1.967 z 0.398 lr 7.67e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.878 max_load(range) 0.072-0.325 step 23860/40589 loss 0.2758 acc 0.944 lb 1.970 z 0.393 lr 7.66e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 23870/40589 loss 0.2802 acc 0.944 lb 1.967 z 0.366 lr 7.65e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.934 max_load(range) 0.056-0.212 step 23880/40589 loss 0.2725 acc 0.944 lb 1.961 z 0.377 lr 7.65e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 23890/40589 loss 0.2650 acc 0.947 lb 1.972 z 0.397 lr 7.64e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 23900/40589 loss 0.2818 acc 0.943 lb 1.965 z 0.389 lr 7.63e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.930 max_load(range) 0.056-0.226 step 23910/40589 loss 0.2767 acc 0.943 lb 1.969 z 0.380 lr 7.62e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 23920/40589 loss 0.2759 acc 0.944 lb 1.963 z 0.382 lr 7.62e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.931 max_load(range) 0.057-0.380 step 23930/40589 loss 0.2627 acc 0.947 lb 1.972 z 0.397 lr 7.61e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 23940/40589 loss 0.2812 acc 0.943 lb 1.968 z 0.401 lr 7.60e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 23950/40589 loss 0.2788 acc 0.943 lb 1.970 z 0.373 lr 7.59e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.947 max_load(range) 0.055-0.242 step 23960/40589 loss 0.2784 acc 0.943 lb 1.965 z 0.386 lr 7.58e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 23970/40589 loss 0.2658 acc 0.947 lb 1.971 z 0.392 lr 7.58e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.915 max_load(range) 0.081-0.155 step 23980/40589 loss 0.2781 acc 0.944 lb 1.967 z 0.394 lr 7.57e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 23990/40589 loss 0.2768 acc 0.943 lb 1.965 z 0.373 lr 7.56e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24000/40589 loss 0.2754 acc 0.944 lb 1.959 z 0.373 lr 7.55e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.935 max_load(range) 0.064-0.181 [save] runs/pretrain_flagship_h100_4x8/step24000 step 24010/40589 loss 0.2664 acc 0.946 lb 1.964 z 0.393 lr 7.55e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24020/40589 loss 0.2792 acc 0.944 lb 1.966 z 0.393 lr 7.54e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.891 max_load(range) 0.077-0.198 step 24030/40589 loss 0.2798 acc 0.943 lb 1.969 z 0.373 lr 7.53e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24040/40589 loss 0.2771 acc 0.944 lb 1.967 z 0.381 lr 7.52e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24050/40589 loss 0.2649 acc 0.947 lb 1.972 z 0.391 lr 7.52e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.877 max_load(range) 0.074-0.321 step 24060/40589 loss 0.2807 acc 0.944 lb 1.967 z 0.398 lr 7.51e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24070/40589 loss 0.2772 acc 0.943 lb 1.967 z 0.381 lr 7.50e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.941 max_load(range) 0.059-0.222 step 24080/40589 loss 0.2765 acc 0.944 lb 1.962 z 0.379 lr 7.49e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24090/40589 loss 0.2643 acc 0.947 lb 1.966 z 0.387 lr 7.48e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24100/40589 loss 0.2830 acc 0.943 lb 1.970 z 0.385 lr 7.48e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.940 max_load(range) 0.056-0.230 step 24110/40589 loss 0.2768 acc 0.944 lb 1.973 z 0.366 lr 7.47e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24120/40589 loss 0.2736 acc 0.944 lb 1.964 z 0.373 lr 7.46e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.935 max_load(range) 0.056-0.385 step 24130/40589 loss 0.2655 acc 0.946 lb 1.969 z 0.391 lr 7.45e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24140/40589 loss 0.2770 acc 0.944 lb 1.966 z 0.381 lr 7.45e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24150/40589 loss 0.2783 acc 0.944 lb 1.966 z 0.363 lr 7.44e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.958 max_load(range) 0.048-0.264 step 24160/40589 loss 0.2754 acc 0.944 lb 1.958 z 0.375 lr 7.43e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24170/40589 loss 0.2641 acc 0.946 lb 1.968 z 0.393 lr 7.42e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.922 max_load(range) 0.075-0.115 step 24180/40589 loss 0.2806 acc 0.943 lb 1.970 z 0.386 lr 7.41e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24190/40589 loss 0.2773 acc 0.944 lb 1.969 z 0.366 lr 7.41e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24200/40589 loss 0.2757 acc 0.944 lb 1.966 z 0.372 lr 7.40e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.954 max_load(range) 0.059-0.168 step 24210/40589 loss 0.2643 acc 0.946 lb 1.972 z 0.379 lr 7.39e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24220/40589 loss 0.2816 acc 0.943 lb 1.967 z 0.378 lr 7.38e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.895 max_load(range) 0.079-0.209 step 24230/40589 loss 0.2773 acc 0.944 lb 1.965 z 0.368 lr 7.38e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24240/40589 loss 0.2778 acc 0.944 lb 1.964 z 0.377 lr 7.37e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24250/40589 loss 0.2682 acc 0.946 lb 1.968 z 0.390 lr 7.36e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.886 max_load(range) 0.068-0.320 step 24260/40589 loss 0.2813 acc 0.944 lb 1.966 z 0.398 lr 7.35e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24270/40589 loss 0.2778 acc 0.944 lb 1.964 z 0.371 lr 7.35e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.947 max_load(range) 0.052-0.220 step 24280/40589 loss 0.2791 acc 0.943 lb 1.965 z 0.388 lr 7.34e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24290/40589 loss 0.2602 acc 0.947 lb 1.972 z 0.392 lr 7.33e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24300/40589 loss 0.2798 acc 0.944 lb 1.965 z 0.390 lr 7.32e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.942 max_load(range) 0.051-0.221 step 24310/40589 loss 0.2811 acc 0.943 lb 1.971 z 0.367 lr 7.31e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24320/40589 loss 0.2780 acc 0.943 lb 1.965 z 0.372 lr 7.31e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.935 max_load(range) 0.055-0.370 step 24330/40589 loss 0.2643 acc 0.947 lb 1.970 z 0.381 lr 7.30e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24340/40589 loss 0.2773 acc 0.945 lb 1.963 z 0.389 lr 7.29e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24350/40589 loss 0.2778 acc 0.943 lb 1.967 z 0.367 lr 7.28e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.957 max_load(range) 0.056-0.231 step 24360/40589 loss 0.2747 acc 0.944 lb 1.959 z 0.386 lr 7.28e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24370/40589 loss 0.2628 acc 0.947 lb 1.961 z 0.387 lr 7.27e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.927 max_load(range) 0.074-0.162 step 24380/40589 loss 0.2800 acc 0.944 lb 1.971 z 0.382 lr 7.26e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24390/40589 loss 0.2771 acc 0.944 lb 1.968 z 0.366 lr 7.25e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24400/40589 loss 0.2732 acc 0.945 lb 1.960 z 0.367 lr 7.25e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.950 max_load(range) 0.058-0.165 step 24410/40589 loss 0.2671 acc 0.946 lb 1.966 z 0.383 lr 7.24e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24420/40589 loss 0.2780 acc 0.944 lb 1.969 z 0.393 lr 7.23e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.887 max_load(range) 0.075-0.197 step 24430/40589 loss 0.2793 acc 0.944 lb 1.965 z 0.374 lr 7.22e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24440/40589 loss 0.2776 acc 0.944 lb 1.959 z 0.369 lr 7.22e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24450/40589 loss 0.2620 acc 0.947 lb 1.968 z 0.381 lr 7.21e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.891 max_load(range) 0.066-0.310 step 24460/40589 loss 0.2814 acc 0.943 lb 1.964 z 0.384 lr 7.20e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24470/40589 loss 0.2770 acc 0.944 lb 1.969 z 0.359 lr 7.19e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.942 max_load(range) 0.063-0.217 step 24480/40589 loss 0.2745 acc 0.944 lb 1.962 z 0.371 lr 7.18e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24490/40589 loss 0.2627 acc 0.946 lb 1.969 z 0.391 lr 7.18e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24500/40589 loss 0.2800 acc 0.943 lb 1.966 z 0.382 lr 7.17e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left [routing] mean entropy_norm 0.934 max_load(range) 0.058-0.226 step 24510/40589 loss 0.2775 acc 0.944 lb 1.966 z 0.363 lr 7.16e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.2d left step 24520/40589 loss 0.2743 acc 0.944 lb 1.963 z 0.369 lr 7.15e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.935 max_load(range) 0.056-0.396 step 24530/40589 loss 0.2617 acc 0.946 lb 1.971 z 0.380 lr 7.15e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.1d left step 24540/40589 loss 0.2755 acc 0.945 lb 1.967 z 0.382 lr 7.14e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.1d left step 24550/40589 loss 0.2819 acc 0.943 lb 1.964 z 0.362 lr 7.13e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.951 max_load(range) 0.056-0.238 step 24560/40589 loss 0.2755 acc 0.943 lb 1.961 z 0.365 lr 7.12e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.1d left step 24570/40589 loss 0.2650 acc 0.946 lb 1.961 z 0.389 lr 7.12e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.925 max_load(range) 0.066-0.143 step 24580/40589 loss 0.2794 acc 0.944 lb 1.966 z 0.395 lr 7.11e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.1d left step 24590/40589 loss 0.2763 acc 0.944 lb 1.967 z 0.372 lr 7.10e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.1d left step 24600/40589 loss 0.2726 acc 0.945 lb 1.965 z 0.366 lr 7.09e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.948 max_load(range) 0.064-0.171 step 24610/40589 loss 0.2636 acc 0.947 lb 1.969 z 0.379 lr 7.09e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.1d left step 24620/40589 loss 0.2776 acc 0.944 lb 1.962 z 0.375 lr 7.08e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.890 max_load(range) 0.074-0.205 step 24630/40589 loss 0.2772 acc 0.943 lb 1.969 z 0.369 lr 7.07e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.1d left step 24640/40589 loss 0.2778 acc 0.944 lb 1.960 z 0.367 lr 7.06e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.1d left step 24650/40589 loss 0.2624 acc 0.947 lb 1.969 z 0.384 lr 7.05e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.888 max_load(range) 0.070-0.320 step 24660/40589 loss 0.2814 acc 0.943 lb 1.962 z 0.378 lr 7.05e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.1d left step 24670/40589 loss 0.2798 acc 0.943 lb 1.965 z 0.363 lr 7.04e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.947 max_load(range) 0.058-0.206 step 24680/40589 loss 0.2780 acc 0.943 lb 1.957 z 0.375 lr 7.03e-05 | 1418 pairs/s | mem 63.3/79 GiB (min 62.4) | 2.1d left step 24690/40589 loss 0.2642 acc 0.947 lb 1.971 z 0.383 lr 7.02e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 24700/40589 loss 0.2806 acc 0.944 lb 1.968 z 0.375 lr 7.02e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.938 max_load(range) 0.063-0.209 step 24710/40589 loss 0.2801 acc 0.943 lb 1.965 z 0.371 lr 7.01e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 24720/40589 loss 0.2773 acc 0.944 lb 1.959 z 0.374 lr 7.00e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.931 max_load(range) 0.057-0.389 step 24730/40589 loss 0.2642 acc 0.946 lb 1.965 z 0.393 lr 6.99e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 24740/40589 loss 0.2762 acc 0.944 lb 1.966 z 0.394 lr 6.99e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 24750/40589 loss 0.2772 acc 0.944 lb 1.969 z 0.364 lr 6.98e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.958 max_load(range) 0.053-0.222 step 24760/40589 loss 0.2749 acc 0.944 lb 1.961 z 0.362 lr 6.97e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 24770/40589 loss 0.2648 acc 0.947 lb 1.967 z 0.387 lr 6.96e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.928 max_load(range) 0.061-0.142 step 24780/40589 loss 0.2818 acc 0.943 lb 1.968 z 0.383 lr 6.96e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 24790/40589 loss 0.2794 acc 0.943 lb 1.969 z 0.362 lr 6.95e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 24800/40589 loss 0.2768 acc 0.944 lb 1.961 z 0.373 lr 6.94e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.935 max_load(range) 0.065-0.166 step 24810/40589 loss 0.2656 acc 0.946 lb 1.969 z 0.382 lr 6.93e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 24820/40589 loss 0.2831 acc 0.943 lb 1.966 z 0.376 lr 6.93e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.891 max_load(range) 0.076-0.166 step 24830/40589 loss 0.2764 acc 0.944 lb 1.969 z 0.370 lr 6.92e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 24840/40589 loss 0.2753 acc 0.944 lb 1.960 z 0.369 lr 6.91e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 24850/40589 loss 0.2656 acc 0.946 lb 1.967 z 0.386 lr 6.90e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.883 max_load(range) 0.071-0.310 step 24860/40589 loss 0.2770 acc 0.945 lb 1.965 z 0.389 lr 6.89e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 24870/40589 loss 0.2781 acc 0.943 lb 1.965 z 0.363 lr 6.89e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.945 max_load(range) 0.057-0.221 step 24880/40589 loss 0.2753 acc 0.944 lb 1.962 z 0.372 lr 6.88e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 24890/40589 loss 0.2641 acc 0.947 lb 1.965 z 0.381 lr 6.87e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 24900/40589 loss 0.2818 acc 0.944 lb 1.963 z 0.374 lr 6.86e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.943 max_load(range) 0.054-0.224 step 24910/40589 loss 0.2772 acc 0.944 lb 1.966 z 0.373 lr 6.86e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 24920/40589 loss 0.2801 acc 0.943 lb 1.960 z 0.377 lr 6.85e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.931 max_load(range) 0.062-0.388 step 24930/40589 loss 0.2627 acc 0.947 lb 1.967 z 0.378 lr 6.84e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 24940/40589 loss 0.2808 acc 0.943 lb 1.967 z 0.375 lr 6.83e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 24950/40589 loss 0.2801 acc 0.943 lb 1.962 z 0.360 lr 6.83e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.956 max_load(range) 0.052-0.223 step 24960/40589 loss 0.2772 acc 0.943 lb 1.959 z 0.364 lr 6.82e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 24970/40589 loss 0.2650 acc 0.947 lb 1.967 z 0.387 lr 6.81e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.931 max_load(range) 0.069-0.137 step 24980/40589 loss 0.2824 acc 0.943 lb 1.965 z 0.398 lr 6.80e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 24990/40589 loss 0.2771 acc 0.944 lb 1.966 z 0.372 lr 6.80e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 25000/40589 loss 0.2754 acc 0.944 lb 1.956 z 0.384 lr 6.79e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.943 max_load(range) 0.057-0.143 [save] runs/pretrain_flagship_h100_4x8/step25000 step 25010/40589 loss 0.2643 acc 0.946 lb 1.964 z 0.388 lr 6.78e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 25020/40589 loss 0.2836 acc 0.943 lb 1.966 z 0.377 lr 6.77e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.901 max_load(range) 0.076-0.175 step 25030/40589 loss 0.2809 acc 0.943 lb 1.964 z 0.360 lr 6.77e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 25040/40589 loss 0.2752 acc 0.944 lb 1.957 z 0.379 lr 6.76e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 25050/40589 loss 0.2635 acc 0.946 lb 1.965 z 0.381 lr 6.75e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.887 max_load(range) 0.064-0.324 step 25060/40589 loss 0.2818 acc 0.943 lb 1.967 z 0.377 lr 6.74e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 25070/40589 loss 0.2771 acc 0.944 lb 1.966 z 0.361 lr 6.74e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.949 max_load(range) 0.061-0.206 step 25080/40589 loss 0.2723 acc 0.944 lb 1.962 z 0.371 lr 6.73e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 25090/40589 loss 0.2638 acc 0.947 lb 1.968 z 0.372 lr 6.72e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 25100/40589 loss 0.2807 acc 0.944 lb 1.965 z 0.385 lr 6.71e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.922 max_load(range) 0.063-0.275 step 25110/40589 loss 0.2768 acc 0.944 lb 1.967 z 0.361 lr 6.71e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 25120/40589 loss 0.2749 acc 0.944 lb 1.966 z 0.353 lr 6.70e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.936 max_load(range) 0.052-0.379 step 25130/40589 loss 0.2627 acc 0.947 lb 1.972 z 0.370 lr 6.69e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 25140/40589 loss 0.2761 acc 0.943 lb 1.968 z 0.374 lr 6.68e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 25150/40589 loss 0.2791 acc 0.943 lb 1.970 z 0.354 lr 6.68e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.953 max_load(range) 0.051-0.235 step 25160/40589 loss 0.2736 acc 0.945 lb 1.962 z 0.359 lr 6.67e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 25170/40589 loss 0.2649 acc 0.946 lb 1.961 z 0.378 lr 6.66e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.924 max_load(range) 0.072-0.126 step 25180/40589 loss 0.2777 acc 0.943 lb 1.958 z 0.409 lr 6.65e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 25190/40589 loss 0.2790 acc 0.943 lb 1.964 z 0.363 lr 6.65e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 25200/40589 loss 0.2750 acc 0.945 lb 1.958 z 0.384 lr 6.64e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.944 max_load(range) 0.061-0.180 step 25210/40589 loss 0.2633 acc 0.947 lb 1.969 z 0.384 lr 6.63e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 25220/40589 loss 0.2790 acc 0.943 lb 1.966 z 0.381 lr 6.62e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.885 max_load(range) 0.072-0.188 step 25230/40589 loss 0.2802 acc 0.943 lb 1.967 z 0.357 lr 6.62e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 25240/40589 loss 0.2735 acc 0.945 lb 1.965 z 0.363 lr 6.61e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 25250/40589 loss 0.2663 acc 0.946 lb 1.968 z 0.375 lr 6.60e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left [routing] mean entropy_norm 0.886 max_load(range) 0.063-0.309 step 25260/40589 loss 0.2777 acc 0.943 lb 1.962 z 0.374 lr 6.59e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.1d left step 25270/40589 loss 0.2770 acc 0.944 lb 1.966 z 0.362 lr 6.59e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left [routing] mean entropy_norm 0.937 max_load(range) 0.059-0.198 step 25280/40589 loss 0.2726 acc 0.944 lb 1.960 z 0.371 lr 6.58e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left step 25290/40589 loss 0.2592 acc 0.947 lb 1.966 z 0.392 lr 6.57e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left step 25300/40589 loss 0.2795 acc 0.944 lb 1.968 z 0.384 lr 6.56e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left [routing] mean entropy_norm 0.926 max_load(range) 0.059-0.243 step 25310/40589 loss 0.2806 acc 0.943 lb 1.967 z 0.371 lr 6.56e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left step 25320/40589 loss 0.2767 acc 0.944 lb 1.965 z 0.367 lr 6.55e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left [routing] mean entropy_norm 0.935 max_load(range) 0.055-0.382 step 25330/40589 loss 0.2620 acc 0.947 lb 1.969 z 0.382 lr 6.54e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left step 25340/40589 loss 0.2779 acc 0.944 lb 1.965 z 0.416 lr 6.53e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left step 25350/40589 loss 0.2780 acc 0.944 lb 1.969 z 0.359 lr 6.53e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left [routing] mean entropy_norm 0.953 max_load(range) 0.054-0.224 step 25360/40589 loss 0.2785 acc 0.943 lb 1.965 z 0.368 lr 6.52e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left step 25370/40589 loss 0.2643 acc 0.947 lb 1.972 z 0.373 lr 6.51e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left [routing] mean entropy_norm 0.930 max_load(range) 0.071-0.159 step 25380/40589 loss 0.2766 acc 0.944 lb 1.968 z 0.379 lr 6.50e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left step 25390/40589 loss 0.2785 acc 0.943 lb 1.965 z 0.360 lr 6.50e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left step 25400/40589 loss 0.2764 acc 0.943 lb 1.962 z 0.362 lr 6.49e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left [routing] mean entropy_norm 0.940 max_load(range) 0.057-0.161 step 25410/40589 loss 0.2604 acc 0.947 lb 1.971 z 0.394 lr 6.48e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left step 25420/40589 loss 0.2816 acc 0.943 lb 1.972 z 0.380 lr 6.47e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left [routing] mean entropy_norm 0.875 max_load(range) 0.065-0.186 step 25430/40589 loss 0.2741 acc 0.944 lb 1.967 z 0.362 lr 6.47e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left step 25440/40589 loss 0.2771 acc 0.943 lb 1.957 z 0.369 lr 6.46e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left step 25450/40589 loss 0.2651 acc 0.946 lb 1.971 z 0.393 lr 6.45e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left [routing] mean entropy_norm 0.882 max_load(range) 0.070-0.320 step 25460/40589 loss 0.2801 acc 0.944 lb 1.964 z 0.385 lr 6.44e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left step 25470/40589 loss 0.2794 acc 0.943 lb 1.965 z 0.384 lr 6.44e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left [routing] mean entropy_norm 0.944 max_load(range) 0.058-0.221 step 25480/40589 loss 0.2761 acc 0.943 lb 1.961 z 0.369 lr 6.43e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left step 25490/40589 loss 0.2610 acc 0.947 lb 1.963 z 0.375 lr 6.42e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.4) | 2.0d left step 25500/40589 loss 0.2781 acc 0.944 lb 1.963 z 0.366 lr 6.41e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.934 max_load(range) 0.056-0.237 step 25510/40589 loss 0.2764 acc 0.943 lb 1.965 z 0.352 lr 6.41e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.5) | 2.0d left step 25520/40589 loss 0.2761 acc 0.943 lb 1.962 z 0.377 lr 6.40e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.929 max_load(range) 0.060-0.403 step 25530/40589 loss 0.2644 acc 0.947 lb 1.967 z 0.399 lr 6.39e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.5) | 2.0d left step 25540/40589 loss 0.2759 acc 0.944 lb 1.964 z 0.392 lr 6.38e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.5) | 2.0d left step 25550/40589 loss 0.2722 acc 0.945 lb 1.965 z 0.365 lr 6.38e-05 | 1418 pairs/s | mem 63.4/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.954 max_load(range) 0.056-0.250 step 25560/40589 loss 0.2720 acc 0.945 lb 1.962 z 0.365 lr 6.37e-05 | 1418 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25570/40589 loss 0.2641 acc 0.946 lb 1.968 z 0.385 lr 6.36e-05 | 1418 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.927 max_load(range) 0.078-0.139 step 25580/40589 loss 0.2795 acc 0.944 lb 1.970 z 0.375 lr 6.35e-05 | 1418 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25590/40589 loss 0.2777 acc 0.943 lb 1.964 z 0.353 lr 6.35e-05 | 1418 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25600/40589 loss 0.2745 acc 0.944 lb 1.962 z 0.363 lr 6.34e-05 | 1418 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.946 max_load(range) 0.059-0.173 step 25610/40589 loss 0.2639 acc 0.946 lb 1.973 z 0.369 lr 6.33e-05 | 1418 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25620/40589 loss 0.2778 acc 0.944 lb 1.963 z 0.376 lr 6.32e-05 | 1418 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.889 max_load(range) 0.068-0.203 step 25630/40589 loss 0.2772 acc 0.943 lb 1.956 z 0.356 lr 6.32e-05 | 1418 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25640/40589 loss 0.2751 acc 0.944 lb 1.956 z 0.365 lr 6.31e-05 | 1418 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25650/40589 loss 0.2643 acc 0.946 lb 1.965 z 0.402 lr 6.30e-05 | 1418 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.873 max_load(range) 0.073-0.335 step 25660/40589 loss 0.2780 acc 0.943 lb 1.966 z 0.373 lr 6.29e-05 | 1418 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25670/40589 loss 0.2762 acc 0.944 lb 1.966 z 0.434 lr 6.29e-05 | 1418 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.947 max_load(range) 0.055-0.204 step 25680/40589 loss 0.2749 acc 0.944 lb 1.964 z 0.363 lr 6.28e-05 | 1418 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25690/40589 loss 0.2644 acc 0.946 lb 1.965 z 0.376 lr 6.27e-05 | 1418 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25700/40589 loss 0.2814 acc 0.943 lb 1.967 z 0.366 lr 6.27e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.937 max_load(range) 0.054-0.207 step 25710/40589 loss 0.2796 acc 0.943 lb 1.961 z 0.352 lr 6.26e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25720/40589 loss 0.2749 acc 0.943 lb 1.959 z 0.363 lr 6.25e-05 | 1418 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.932 max_load(range) 0.056-0.378 step 25730/40589 loss 0.2618 acc 0.946 lb 1.968 z 0.372 lr 6.24e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25740/40589 loss 0.2770 acc 0.944 lb 1.964 z 0.365 lr 6.24e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25750/40589 loss 0.2790 acc 0.942 lb 1.968 z 0.352 lr 6.23e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.952 max_load(range) 0.054-0.241 step 25760/40589 loss 0.2705 acc 0.945 lb 1.960 z 0.369 lr 6.22e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25770/40589 loss 0.2612 acc 0.947 lb 1.965 z 0.372 lr 6.21e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.921 max_load(range) 0.078-0.135 step 25780/40589 loss 0.2765 acc 0.944 lb 1.964 z 0.371 lr 6.21e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25790/40589 loss 0.2780 acc 0.943 lb 1.967 z 0.351 lr 6.20e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25800/40589 loss 0.2743 acc 0.943 lb 1.958 z 0.361 lr 6.19e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.940 max_load(range) 0.062-0.184 step 25810/40589 loss 0.2632 acc 0.946 lb 1.965 z 0.382 lr 6.18e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25820/40589 loss 0.2795 acc 0.943 lb 1.964 z 0.410 lr 6.18e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.878 max_load(range) 0.078-0.184 step 25830/40589 loss 0.2785 acc 0.943 lb 1.969 z 0.359 lr 6.17e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25840/40589 loss 0.2743 acc 0.944 lb 1.962 z 0.361 lr 6.16e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25850/40589 loss 0.2641 acc 0.946 lb 1.964 z 0.394 lr 6.15e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.889 max_load(range) 0.067-0.307 step 25860/40589 loss 0.2767 acc 0.944 lb 1.962 z 0.373 lr 6.15e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25870/40589 loss 0.2805 acc 0.943 lb 1.967 z 0.363 lr 6.14e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.938 max_load(range) 0.064-0.213 step 25880/40589 loss 0.2766 acc 0.944 lb 1.960 z 0.424 lr 6.13e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25890/40589 loss 0.2651 acc 0.946 lb 1.963 z 0.394 lr 6.12e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25900/40589 loss 0.2745 acc 0.944 lb 1.961 z 0.384 lr 6.12e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.934 max_load(range) 0.056-0.215 step 25910/40589 loss 0.2775 acc 0.943 lb 1.965 z 0.379 lr 6.11e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25920/40589 loss 0.2742 acc 0.944 lb 1.964 z 0.381 lr 6.10e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.933 max_load(range) 0.054-0.387 step 25930/40589 loss 0.2668 acc 0.945 lb 1.972 z 0.414 lr 6.10e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25940/40589 loss 0.2822 acc 0.942 lb 1.964 z 0.395 lr 6.09e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25950/40589 loss 0.2790 acc 0.943 lb 1.963 z 0.351 lr 6.08e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.955 max_load(range) 0.053-0.238 step 25960/40589 loss 0.2749 acc 0.944 lb 1.957 z 0.356 lr 6.07e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25970/40589 loss 0.2583 acc 0.947 lb 1.966 z 0.374 lr 6.07e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.923 max_load(range) 0.076-0.156 step 25980/40589 loss 0.2769 acc 0.943 lb 1.964 z 0.385 lr 6.06e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 25990/40589 loss 0.2794 acc 0.943 lb 1.966 z 0.395 lr 6.05e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 26000/40589 loss 0.2731 acc 0.944 lb 1.959 z 0.360 lr 6.04e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left [routing] mean entropy_norm 0.950 max_load(range) 0.062-0.163 [save] runs/pretrain_flagship_h100_4x8/step26000 step 26010/40589 loss 0.2606 acc 0.946 lb 1.966 z 0.408 lr 6.04e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 2.0d left step 26020/40589 loss 0.2774 acc 0.943 lb 1.963 z 0.434 lr 6.03e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.882 max_load(range) 0.074-0.156 step 26030/40589 loss 0.2750 acc 0.944 lb 1.964 z 0.347 lr 6.02e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26040/40589 loss 0.2691 acc 0.945 lb 1.962 z 0.369 lr 6.01e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26050/40589 loss 0.2591 acc 0.948 lb 1.970 z 0.373 lr 6.01e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.880 max_load(range) 0.075-0.337 step 26060/40589 loss 0.2756 acc 0.944 lb 1.957 z 0.439 lr 6.00e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26070/40589 loss 0.2817 acc 0.942 lb 1.968 z 0.349 lr 5.99e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.944 max_load(range) 0.052-0.233 step 26080/40589 loss 0.2762 acc 0.943 lb 1.964 z 0.352 lr 5.99e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26090/40589 loss 0.2614 acc 0.946 lb 1.971 z 0.368 lr 5.98e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26100/40589 loss 0.2793 acc 0.943 lb 1.965 z 0.373 lr 5.97e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.934 max_load(range) 0.056-0.241 step 26110/40589 loss 0.2728 acc 0.944 lb 1.965 z 0.357 lr 5.96e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26120/40589 loss 0.2713 acc 0.945 lb 1.960 z 0.356 lr 5.96e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.934 max_load(range) 0.058-0.398 step 26130/40589 loss 0.2643 acc 0.945 lb 1.969 z 0.368 lr 5.95e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26140/40589 loss 0.2787 acc 0.944 lb 1.964 z 0.384 lr 5.94e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26150/40589 loss 0.2794 acc 0.943 lb 1.967 z 0.347 lr 5.93e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.955 max_load(range) 0.056-0.248 step 26160/40589 loss 0.2710 acc 0.945 lb 1.965 z 0.366 lr 5.93e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26170/40589 loss 0.2645 acc 0.947 lb 1.971 z 0.377 lr 5.92e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.919 max_load(range) 0.070-0.177 step 26180/40589 loss 0.2759 acc 0.944 lb 1.961 z 0.364 lr 5.91e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26190/40589 loss 0.2792 acc 0.943 lb 1.962 z 0.353 lr 5.91e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26200/40589 loss 0.2709 acc 0.945 lb 1.961 z 0.361 lr 5.90e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.949 max_load(range) 0.056-0.181 step 26210/40589 loss 0.2641 acc 0.946 lb 1.966 z 0.371 lr 5.89e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26220/40589 loss 0.2765 acc 0.944 lb 1.965 z 0.362 lr 5.88e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.884 max_load(range) 0.069-0.201 step 26230/40589 loss 0.2781 acc 0.943 lb 1.967 z 0.352 lr 5.88e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26240/40589 loss 0.2754 acc 0.943 lb 1.958 z 0.347 lr 5.87e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26250/40589 loss 0.2598 acc 0.947 lb 1.966 z 0.365 lr 5.86e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.889 max_load(range) 0.066-0.305 step 26260/40589 loss 0.2784 acc 0.944 lb 1.963 z 0.370 lr 5.85e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26270/40589 loss 0.2766 acc 0.943 lb 1.966 z 0.346 lr 5.85e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.944 max_load(range) 0.059-0.234 step 26280/40589 loss 0.2742 acc 0.943 lb 1.958 z 0.357 lr 5.84e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26290/40589 loss 0.2648 acc 0.946 lb 1.961 z 0.374 lr 5.83e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26300/40589 loss 0.2801 acc 0.943 lb 1.964 z 0.443 lr 5.83e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.918 max_load(range) 0.060-0.266 step 26310/40589 loss 0.2753 acc 0.944 lb 1.967 z 0.351 lr 5.82e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26320/40589 loss 0.2747 acc 0.943 lb 1.964 z 0.357 lr 5.81e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.937 max_load(range) 0.055-0.372 step 26330/40589 loss 0.2578 acc 0.948 lb 1.966 z 0.412 lr 5.80e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26340/40589 loss 0.2769 acc 0.944 lb 1.963 z 0.395 lr 5.80e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26350/40589 loss 0.2724 acc 0.944 lb 1.962 z 0.351 lr 5.79e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.953 max_load(range) 0.055-0.228 step 26360/40589 loss 0.2746 acc 0.944 lb 1.959 z 0.352 lr 5.78e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26370/40589 loss 0.2626 acc 0.946 lb 1.966 z 0.412 lr 5.77e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.923 max_load(range) 0.079-0.144 step 26380/40589 loss 0.2806 acc 0.943 lb 1.960 z 0.371 lr 5.77e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26390/40589 loss 0.2780 acc 0.943 lb 1.966 z 0.348 lr 5.76e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26400/40589 loss 0.2743 acc 0.944 lb 1.956 z 0.352 lr 5.75e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.944 max_load(range) 0.063-0.172 step 26410/40589 loss 0.2631 acc 0.946 lb 1.969 z 0.368 lr 5.75e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26420/40589 loss 0.2796 acc 0.944 lb 1.968 z 0.371 lr 5.74e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.881 max_load(range) 0.074-0.214 step 26430/40589 loss 0.2830 acc 0.942 lb 1.972 z 0.342 lr 5.73e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26440/40589 loss 0.2762 acc 0.944 lb 1.959 z 0.354 lr 5.72e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26450/40589 loss 0.2647 acc 0.946 lb 1.963 z 0.405 lr 5.72e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.876 max_load(range) 0.069-0.339 step 26460/40589 loss 0.2805 acc 0.943 lb 1.967 z 0.364 lr 5.71e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26470/40589 loss 0.2799 acc 0.942 lb 1.964 z 0.352 lr 5.70e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.945 max_load(range) 0.054-0.197 step 26480/40589 loss 0.2749 acc 0.943 lb 1.963 z 0.356 lr 5.70e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26490/40589 loss 0.2643 acc 0.947 lb 1.965 z 0.355 lr 5.69e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26500/40589 loss 0.2805 acc 0.943 lb 1.963 z 0.372 lr 5.68e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.928 max_load(range) 0.059-0.248 step 26510/40589 loss 0.2738 acc 0.944 lb 1.964 z 0.350 lr 5.67e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26520/40589 loss 0.2714 acc 0.944 lb 1.961 z 0.349 lr 5.67e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.930 max_load(range) 0.061-0.378 step 26530/40589 loss 0.2605 acc 0.947 lb 1.963 z 0.376 lr 5.66e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26540/40589 loss 0.2747 acc 0.944 lb 1.961 z 0.359 lr 5.65e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26550/40589 loss 0.2794 acc 0.943 lb 1.965 z 0.389 lr 5.65e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.958 max_load(range) 0.052-0.248 step 26560/40589 loss 0.2716 acc 0.945 lb 1.957 z 0.352 lr 5.64e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26570/40589 loss 0.2600 acc 0.947 lb 1.962 z 0.403 lr 5.63e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.926 max_load(range) 0.066-0.154 step 26580/40589 loss 0.2791 acc 0.944 lb 1.959 z 0.402 lr 5.62e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26590/40589 loss 0.2758 acc 0.944 lb 1.964 z 0.349 lr 5.62e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26600/40589 loss 0.2753 acc 0.944 lb 1.961 z 0.371 lr 5.61e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.941 max_load(range) 0.062-0.149 step 26610/40589 loss 0.2608 acc 0.948 lb 1.969 z 0.359 lr 5.60e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26620/40589 loss 0.2756 acc 0.944 lb 1.961 z 0.365 lr 5.59e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.903 max_load(range) 0.071-0.154 step 26630/40589 loss 0.2823 acc 0.942 lb 1.965 z 0.360 lr 5.59e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26640/40589 loss 0.2723 acc 0.945 lb 1.954 z 0.344 lr 5.58e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26650/40589 loss 0.2626 acc 0.947 lb 1.965 z 0.424 lr 5.57e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.895 max_load(range) 0.064-0.304 step 26660/40589 loss 0.2751 acc 0.944 lb 1.966 z 0.371 lr 5.57e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26670/40589 loss 0.2776 acc 0.944 lb 1.962 z 0.341 lr 5.56e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.941 max_load(range) 0.056-0.204 step 26680/40589 loss 0.2726 acc 0.944 lb 1.961 z 0.346 lr 5.55e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26690/40589 loss 0.2582 acc 0.947 lb 1.964 z 0.376 lr 5.54e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26700/40589 loss 0.2788 acc 0.943 lb 1.965 z 0.384 lr 5.54e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.915 max_load(range) 0.062-0.267 step 26710/40589 loss 0.2784 acc 0.943 lb 1.966 z 0.340 lr 5.53e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26720/40589 loss 0.2751 acc 0.944 lb 1.958 z 0.389 lr 5.52e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.938 max_load(range) 0.057-0.374 step 26730/40589 loss 0.2618 acc 0.947 lb 1.962 z 0.371 lr 5.52e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26740/40589 loss 0.2758 acc 0.945 lb 1.966 z 0.440 lr 5.51e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26750/40589 loss 0.2768 acc 0.943 lb 1.961 z 0.343 lr 5.50e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left [routing] mean entropy_norm 0.958 max_load(range) 0.053-0.236 step 26760/40589 loss 0.2698 acc 0.944 lb 1.958 z 0.351 lr 5.49e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.9d left step 26770/40589 loss 0.2605 acc 0.947 lb 1.964 z 0.463 lr 5.49e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.8d left [routing] mean entropy_norm 0.932 max_load(range) 0.066-0.140 step 26780/40589 loss 0.2762 acc 0.944 lb 1.961 z 0.441 lr 5.48e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.8d left step 26790/40589 loss 0.2793 acc 0.943 lb 1.963 z 0.343 lr 5.47e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.8d left step 26800/40589 loss 0.2742 acc 0.944 lb 1.953 z 0.347 lr 5.47e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.8d left [routing] mean entropy_norm 0.945 max_load(range) 0.056-0.165 step 26810/40589 loss 0.2555 acc 0.948 lb 1.965 z 0.433 lr 5.46e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.8d left step 26820/40589 loss 0.2775 acc 0.943 lb 1.964 z 0.352 lr 5.45e-05 | 1417 pairs/s | mem 63.5/79 GiB (min 62.5) | 1.8d left [routing] mean entropy_norm 0.890 max_load(range) 0.069-0.185 step 26830/40589 loss 0.2780 acc 0.943 lb 1.967 z 0.343 lr 5.45e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.5) | 1.8d left step 26840/40589 loss 0.2717 acc 0.944 lb 1.957 z 0.346 lr 5.44e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.5) | 1.8d left step 26850/40589 loss 0.2625 acc 0.947 lb 1.957 z 0.361 lr 5.43e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.5) | 1.8d left [routing] mean entropy_norm 0.912 max_load(range) 0.058-0.278 step 26860/40589 loss 0.2799 acc 0.943 lb 1.968 z 0.375 lr 5.42e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.5) | 1.8d left step 26870/40589 loss 0.2751 acc 0.944 lb 1.963 z 0.363 lr 5.42e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.5) | 1.8d left [routing] mean entropy_norm 0.948 max_load(range) 0.057-0.190 step 26880/40589 loss 0.2708 acc 0.945 lb 1.955 z 0.374 lr 5.41e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.5) | 1.8d left step 26890/40589 loss 0.2604 acc 0.948 lb 1.963 z 0.381 lr 5.40e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.5) | 1.8d left step 26900/40589 loss 0.2756 acc 0.944 lb 1.964 z 0.430 lr 5.40e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.5) | 1.8d left [routing] mean entropy_norm 0.939 max_load(range) 0.054-0.230 step 26910/40589 loss 0.2785 acc 0.944 lb 1.959 z 0.340 lr 5.39e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.5) | 1.8d left step 26920/40589 loss 0.2733 acc 0.944 lb 1.963 z 0.348 lr 5.38e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.5) | 1.8d left [routing] mean entropy_norm 0.939 max_load(range) 0.057-0.378 step 26930/40589 loss 0.2625 acc 0.946 lb 1.965 z 0.421 lr 5.37e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.5) | 1.8d left step 26940/40589 loss 0.2776 acc 0.943 lb 1.964 z 0.350 lr 5.37e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.5) | 1.8d left step 26950/40589 loss 0.2757 acc 0.944 lb 1.971 z 0.401 lr 5.36e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.5) | 1.8d left [routing] mean entropy_norm 0.958 max_load(range) 0.050-0.243 step 26960/40589 loss 0.2734 acc 0.944 lb 1.962 z 0.363 lr 5.35e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.5) | 1.8d left step 26970/40589 loss 0.2599 acc 0.947 lb 1.966 z 0.366 lr 5.35e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.5) | 1.8d left [routing] mean entropy_norm 0.926 max_load(range) 0.072-0.127 step 26980/40589 loss 0.2754 acc 0.944 lb 1.967 z 0.433 lr 5.34e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.5) | 1.8d left step 26990/40589 loss 0.2796 acc 0.943 lb 1.961 z 0.411 lr 5.33e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.5) | 1.8d left step 27000/40589 loss 0.2703 acc 0.944 lb 1.960 z 0.342 lr 5.32e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.5) | 1.8d left [routing] mean entropy_norm 0.947 max_load(range) 0.061-0.191 [save] runs/pretrain_flagship_h100_4x8/step27000 step 27010/40589 loss 0.2622 acc 0.947 lb 1.964 z 0.441 lr 5.32e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27020/40589 loss 0.2796 acc 0.943 lb 1.964 z 0.373 lr 5.31e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.905 max_load(range) 0.074-0.186 step 27030/40589 loss 0.2801 acc 0.943 lb 1.966 z 0.350 lr 5.30e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27040/40589 loss 0.2718 acc 0.944 lb 1.949 z 0.343 lr 5.30e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27050/40589 loss 0.2660 acc 0.945 lb 1.965 z 0.357 lr 5.29e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.890 max_load(range) 0.065-0.320 step 27060/40589 loss 0.2782 acc 0.943 lb 1.966 z 0.386 lr 5.28e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27070/40589 loss 0.2794 acc 0.943 lb 1.962 z 0.353 lr 5.28e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.949 max_load(range) 0.058-0.223 step 27080/40589 loss 0.2748 acc 0.943 lb 1.955 z 0.417 lr 5.27e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27090/40589 loss 0.2594 acc 0.947 lb 1.963 z 0.373 lr 5.26e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27100/40589 loss 0.2784 acc 0.944 lb 1.961 z 0.358 lr 5.25e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.941 max_load(range) 0.053-0.215 step 27110/40589 loss 0.2796 acc 0.943 lb 1.959 z 0.349 lr 5.25e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27120/40589 loss 0.2723 acc 0.944 lb 1.962 z 0.365 lr 5.24e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.937 max_load(range) 0.058-0.391 step 27130/40589 loss 0.2603 acc 0.947 lb 1.960 z 0.402 lr 5.23e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27140/40589 loss 0.2786 acc 0.943 lb 1.961 z 0.358 lr 5.23e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27150/40589 loss 0.2793 acc 0.942 lb 1.970 z 0.347 lr 5.22e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.961 max_load(range) 0.053-0.237 step 27160/40589 loss 0.2678 acc 0.946 lb 1.962 z 0.342 lr 5.21e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27170/40589 loss 0.2662 acc 0.946 lb 1.962 z 0.451 lr 5.21e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.931 max_load(range) 0.071-0.140 step 27180/40589 loss 0.2752 acc 0.944 lb 1.959 z 0.346 lr 5.20e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27190/40589 loss 0.2788 acc 0.943 lb 1.960 z 0.337 lr 5.19e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27200/40589 loss 0.2723 acc 0.944 lb 1.957 z 0.347 lr 5.18e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.952 max_load(range) 0.060-0.166 step 27210/40589 loss 0.2599 acc 0.947 lb 1.966 z 0.476 lr 5.18e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27220/40589 loss 0.2768 acc 0.944 lb 1.960 z 0.400 lr 5.17e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.886 max_load(range) 0.076-0.179 step 27230/40589 loss 0.2776 acc 0.942 lb 1.961 z 0.342 lr 5.16e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27240/40589 loss 0.2719 acc 0.944 lb 1.954 z 0.350 lr 5.16e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27250/40589 loss 0.2610 acc 0.947 lb 1.961 z 0.374 lr 5.15e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.879 max_load(range) 0.072-0.336 step 27260/40589 loss 0.2738 acc 0.944 lb 1.964 z 0.351 lr 5.14e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27270/40589 loss 0.2768 acc 0.943 lb 1.960 z 0.366 lr 5.14e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.947 max_load(range) 0.055-0.218 step 27280/40589 loss 0.2707 acc 0.945 lb 1.961 z 0.344 lr 5.13e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27290/40589 loss 0.2624 acc 0.946 lb 1.964 z 0.359 lr 5.12e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27300/40589 loss 0.2787 acc 0.943 lb 1.957 z 0.371 lr 5.11e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.934 max_load(range) 0.057-0.238 step 27310/40589 loss 0.2762 acc 0.944 lb 1.964 z 0.350 lr 5.11e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27320/40589 loss 0.2745 acc 0.944 lb 1.962 z 0.412 lr 5.10e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.936 max_load(range) 0.052-0.379 step 27330/40589 loss 0.2598 acc 0.947 lb 1.965 z 0.353 lr 5.09e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27340/40589 loss 0.2773 acc 0.944 lb 1.965 z 0.385 lr 5.09e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27350/40589 loss 0.2748 acc 0.944 lb 1.961 z 0.411 lr 5.08e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.958 max_load(range) 0.055-0.234 step 27360/40589 loss 0.2723 acc 0.944 lb 1.958 z 0.354 lr 5.07e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27370/40589 loss 0.2602 acc 0.947 lb 1.964 z 0.414 lr 5.07e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.926 max_load(range) 0.073-0.130 step 27380/40589 loss 0.2814 acc 0.943 lb 1.965 z 0.363 lr 5.06e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27390/40589 loss 0.2790 acc 0.942 lb 1.959 z 0.415 lr 5.05e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27400/40589 loss 0.2718 acc 0.944 lb 1.952 z 0.346 lr 5.04e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.953 max_load(range) 0.058-0.189 step 27410/40589 loss 0.2632 acc 0.946 lb 1.963 z 0.463 lr 5.04e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27420/40589 loss 0.2727 acc 0.945 lb 1.961 z 0.358 lr 5.03e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.923 max_load(range) 0.071-0.143 step 27430/40589 loss 0.2768 acc 0.943 lb 1.965 z 0.366 lr 5.02e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27440/40589 loss 0.2712 acc 0.944 lb 1.963 z 0.345 lr 5.02e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27450/40589 loss 0.2640 acc 0.946 lb 1.958 z 0.364 lr 5.01e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.875 max_load(range) 0.073-0.345 step 27460/40589 loss 0.2759 acc 0.944 lb 1.963 z 0.349 lr 5.00e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27470/40589 loss 0.2754 acc 0.944 lb 1.956 z 0.381 lr 5.00e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.947 max_load(range) 0.054-0.211 step 27480/40589 loss 0.2757 acc 0.944 lb 1.958 z 0.425 lr 4.99e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27490/40589 loss 0.2613 acc 0.947 lb 1.964 z 0.392 lr 4.98e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27500/40589 loss 0.2759 acc 0.944 lb 1.952 z 0.408 lr 4.98e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left [routing] mean entropy_norm 0.941 max_load(range) 0.057-0.215 step 27510/40589 loss 0.2786 acc 0.943 lb 1.960 z 0.334 lr 4.97e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.8d left step 27520/40589 loss 0.2707 acc 0.945 lb 1.957 z 0.360 lr 4.96e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.938 max_load(range) 0.058-0.369 step 27530/40589 loss 0.2583 acc 0.947 lb 1.964 z 0.395 lr 4.96e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.7d left step 27540/40589 loss 0.2742 acc 0.945 lb 1.967 z 0.355 lr 4.95e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.7d left step 27550/40589 loss 0.2806 acc 0.942 lb 1.960 z 0.339 lr 4.94e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.962 max_load(range) 0.050-0.232 step 27560/40589 loss 0.2718 acc 0.944 lb 1.955 z 0.392 lr 4.93e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.7d left step 27570/40589 loss 0.2613 acc 0.947 lb 1.959 z 0.424 lr 4.93e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.925 max_load(range) 0.066-0.120 step 27580/40589 loss 0.2777 acc 0.943 lb 1.957 z 0.351 lr 4.92e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.7d left step 27590/40589 loss 0.2801 acc 0.942 lb 1.964 z 0.382 lr 4.91e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.7d left step 27600/40589 loss 0.2738 acc 0.944 lb 1.956 z 0.348 lr 4.91e-05 | 1417 pairs/s | mem 63.6/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.952 max_load(range) 0.057-0.169 step 27610/40589 loss 0.2616 acc 0.947 lb 1.966 z 0.353 lr 4.90e-05 | 1417 pairs/s | mem 63.7/79 GiB (min 62.6) | 1.7d left step 27620/40589 loss 0.2749 acc 0.944 lb 1.957 z 0.369 lr 4.89e-05 | 1417 pairs/s | mem 63.7/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.919 max_load(range) 0.073-0.170 step 27630/40589 loss 0.2756 acc 0.943 lb 1.965 z 0.349 lr 4.89e-05 | 1417 pairs/s | mem 63.7/79 GiB (min 62.6) | 1.7d left step 27640/40589 loss 0.2700 acc 0.945 lb 1.956 z 0.380 lr 4.88e-05 | 1417 pairs/s | mem 63.7/79 GiB (min 62.6) | 1.7d left step 27650/40589 loss 0.2575 acc 0.948 lb 1.959 z 0.425 lr 4.87e-05 | 1417 pairs/s | mem 63.7/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.895 max_load(range) 0.065-0.317 step 27660/40589 loss 0.2762 acc 0.944 lb 1.959 z 0.364 lr 4.87e-05 | 1417 pairs/s | mem 63.7/79 GiB (min 62.6) | 1.7d left step 27670/40589 loss 0.2784 acc 0.943 lb 1.962 z 0.360 lr 4.86e-05 | 1417 pairs/s | mem 63.7/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.953 max_load(range) 0.055-0.217 step 27680/40589 loss 0.2675 acc 0.945 lb 1.961 z 0.332 lr 4.85e-05 | 1417 pairs/s | mem 63.7/79 GiB (min 62.6) | 1.7d left step 27690/40589 loss 0.2637 acc 0.946 lb 1.964 z 0.411 lr 4.85e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 27700/40589 loss 0.2789 acc 0.943 lb 1.955 z 0.423 lr 4.84e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.941 max_load(range) 0.056-0.221 step 27710/40589 loss 0.2716 acc 0.944 lb 1.961 z 0.346 lr 4.83e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 27720/40589 loss 0.2745 acc 0.943 lb 1.952 z 0.336 lr 4.82e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.933 max_load(range) 0.058-0.395 step 27730/40589 loss 0.2610 acc 0.947 lb 1.959 z 0.362 lr 4.82e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 27740/40589 loss 0.2719 acc 0.945 lb 1.961 z 0.349 lr 4.81e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 27750/40589 loss 0.2793 acc 0.942 lb 1.958 z 0.342 lr 4.80e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.959 max_load(range) 0.055-0.235 step 27760/40589 loss 0.2693 acc 0.945 lb 1.956 z 0.335 lr 4.80e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 27770/40589 loss 0.2599 acc 0.947 lb 1.963 z 0.354 lr 4.79e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.940 max_load(range) 0.058-0.149 step 27780/40589 loss 0.2755 acc 0.944 lb 1.961 z 0.369 lr 4.78e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 27790/40589 loss 0.2805 acc 0.942 lb 1.965 z 0.325 lr 4.78e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 27800/40589 loss 0.2703 acc 0.945 lb 1.953 z 0.398 lr 4.77e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.952 max_load(range) 0.060-0.201 step 27810/40589 loss 0.2586 acc 0.947 lb 1.958 z 0.393 lr 4.76e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 27820/40589 loss 0.2769 acc 0.944 lb 1.961 z 0.396 lr 4.76e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.886 max_load(range) 0.071-0.198 step 27830/40589 loss 0.2766 acc 0.943 lb 1.961 z 0.336 lr 4.75e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 27840/40589 loss 0.2705 acc 0.945 lb 1.956 z 0.355 lr 4.74e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 27850/40589 loss 0.2581 acc 0.948 lb 1.964 z 0.470 lr 4.74e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.896 max_load(range) 0.064-0.306 step 27860/40589 loss 0.2767 acc 0.943 lb 1.961 z 0.400 lr 4.73e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 27870/40589 loss 0.2733 acc 0.944 lb 1.965 z 0.393 lr 4.72e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.943 max_load(range) 0.059-0.230 step 27880/40589 loss 0.2697 acc 0.945 lb 1.956 z 0.354 lr 4.72e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 27890/40589 loss 0.2626 acc 0.947 lb 1.964 z 0.429 lr 4.71e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 27900/40589 loss 0.2754 acc 0.944 lb 1.961 z 0.348 lr 4.70e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.920 max_load(range) 0.059-0.259 step 27910/40589 loss 0.2792 acc 0.943 lb 1.963 z 0.337 lr 4.70e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 27920/40589 loss 0.2678 acc 0.945 lb 1.960 z 0.340 lr 4.69e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.934 max_load(range) 0.057-0.373 step 27930/40589 loss 0.2611 acc 0.947 lb 1.966 z 0.410 lr 4.68e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 27940/40589 loss 0.2747 acc 0.945 lb 1.961 z 0.348 lr 4.68e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 27950/40589 loss 0.2742 acc 0.943 lb 1.961 z 0.394 lr 4.67e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.957 max_load(range) 0.052-0.229 step 27960/40589 loss 0.2705 acc 0.945 lb 1.952 z 0.373 lr 4.66e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 27970/40589 loss 0.2619 acc 0.946 lb 1.962 z 0.402 lr 4.66e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.931 max_load(range) 0.056-0.149 step 27980/40589 loss 0.2743 acc 0.944 lb 1.961 z 0.354 lr 4.65e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 27990/40589 loss 0.2770 acc 0.943 lb 1.960 z 0.325 lr 4.64e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 28000/40589 loss 0.2705 acc 0.945 lb 1.952 z 0.370 lr 4.63e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.950 max_load(range) 0.056-0.168 [save] runs/pretrain_flagship_h100_4x8/step28000 step 28010/40589 loss 0.2595 acc 0.947 lb 1.962 z 0.351 lr 4.63e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 28020/40589 loss 0.2756 acc 0.944 lb 1.965 z 0.398 lr 4.62e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.919 max_load(range) 0.072-0.148 step 28030/40589 loss 0.2740 acc 0.944 lb 1.962 z 0.348 lr 4.61e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 28040/40589 loss 0.2685 acc 0.945 lb 1.957 z 0.398 lr 4.61e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 28050/40589 loss 0.2600 acc 0.947 lb 1.965 z 0.397 lr 4.60e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.883 max_load(range) 0.068-0.327 step 28060/40589 loss 0.2810 acc 0.943 lb 1.960 z 0.412 lr 4.59e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 28070/40589 loss 0.2748 acc 0.943 lb 1.957 z 0.365 lr 4.59e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.946 max_load(range) 0.054-0.228 step 28080/40589 loss 0.2691 acc 0.945 lb 1.953 z 0.395 lr 4.58e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 28090/40589 loss 0.2611 acc 0.946 lb 1.965 z 0.355 lr 4.57e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 28100/40589 loss 0.2768 acc 0.944 lb 1.962 z 0.413 lr 4.57e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.930 max_load(range) 0.056-0.244 step 28110/40589 loss 0.2757 acc 0.944 lb 1.958 z 0.331 lr 4.56e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 28120/40589 loss 0.2700 acc 0.945 lb 1.951 z 0.328 lr 4.55e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.941 max_load(range) 0.052-0.364 step 28130/40589 loss 0.2585 acc 0.947 lb 1.958 z 0.457 lr 4.55e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 28140/40589 loss 0.2750 acc 0.944 lb 1.964 z 0.395 lr 4.54e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 28150/40589 loss 0.2769 acc 0.944 lb 1.963 z 0.397 lr 4.53e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.959 max_load(range) 0.051-0.236 step 28160/40589 loss 0.2730 acc 0.944 lb 1.956 z 0.375 lr 4.53e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left step 28170/40589 loss 0.2615 acc 0.947 lb 1.959 z 0.366 lr 4.52e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.6) | 1.7d left [routing] mean entropy_norm 0.932 max_load(range) 0.070-0.150 step 28180/40589 loss 0.2734 acc 0.944 lb 1.960 z 0.403 lr 4.51e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.7d left step 28190/40589 loss 0.2749 acc 0.943 lb 1.959 z 0.392 lr 4.51e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.7d left step 28200/40589 loss 0.2714 acc 0.944 lb 1.957 z 0.328 lr 4.50e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.7d left [routing] mean entropy_norm 0.948 max_load(range) 0.063-0.173 step 28210/40589 loss 0.2589 acc 0.947 lb 1.962 z 0.379 lr 4.49e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.7d left step 28220/40589 loss 0.2730 acc 0.944 lb 1.962 z 0.410 lr 4.49e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.7d left [routing] mean entropy_norm 0.886 max_load(range) 0.079-0.190 step 28230/40589 loss 0.2794 acc 0.943 lb 1.963 z 0.323 lr 4.48e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.7d left step 28240/40589 loss 0.2677 acc 0.945 lb 1.951 z 0.379 lr 4.47e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.7d left step 28250/40589 loss 0.2583 acc 0.948 lb 1.961 z 0.379 lr 4.47e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.7d left [routing] mean entropy_norm 0.891 max_load(range) 0.065-0.318 step 28260/40589 loss 0.2746 acc 0.944 lb 1.956 z 0.398 lr 4.46e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.7d left step 28270/40589 loss 0.2766 acc 0.943 lb 1.963 z 0.334 lr 4.45e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.950 max_load(range) 0.054-0.220 step 28280/40589 loss 0.2638 acc 0.946 lb 1.952 z 0.402 lr 4.45e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28290/40589 loss 0.2580 acc 0.947 lb 1.959 z 0.426 lr 4.44e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28300/40589 loss 0.2739 acc 0.944 lb 1.963 z 0.412 lr 4.43e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.930 max_load(range) 0.058-0.251 step 28310/40589 loss 0.2765 acc 0.943 lb 1.962 z 0.326 lr 4.43e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28320/40589 loss 0.2682 acc 0.945 lb 1.957 z 0.387 lr 4.42e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.934 max_load(range) 0.059-0.380 step 28330/40589 loss 0.2621 acc 0.947 lb 1.966 z 0.402 lr 4.41e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28340/40589 loss 0.2720 acc 0.945 lb 1.960 z 0.350 lr 4.41e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28350/40589 loss 0.2732 acc 0.944 lb 1.963 z 0.383 lr 4.40e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.956 max_load(range) 0.056-0.232 step 28360/40589 loss 0.2713 acc 0.944 lb 1.951 z 0.348 lr 4.39e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28370/40589 loss 0.2585 acc 0.947 lb 1.962 z 0.447 lr 4.39e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.931 max_load(range) 0.068-0.126 step 28380/40589 loss 0.2742 acc 0.944 lb 1.961 z 0.391 lr 4.38e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28390/40589 loss 0.2785 acc 0.942 lb 1.966 z 0.373 lr 4.37e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28400/40589 loss 0.2717 acc 0.945 lb 1.957 z 0.368 lr 4.37e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.952 max_load(range) 0.055-0.181 step 28410/40589 loss 0.2605 acc 0.946 lb 1.967 z 0.467 lr 4.36e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28420/40589 loss 0.2713 acc 0.945 lb 1.961 z 0.345 lr 4.36e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.925 max_load(range) 0.074-0.159 step 28430/40589 loss 0.2760 acc 0.943 lb 1.962 z 0.378 lr 4.35e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28440/40589 loss 0.2699 acc 0.944 lb 1.953 z 0.364 lr 4.34e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28450/40589 loss 0.2592 acc 0.947 lb 1.961 z 0.479 lr 4.34e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.889 max_load(range) 0.068-0.326 step 28460/40589 loss 0.2720 acc 0.945 lb 1.959 z 0.414 lr 4.33e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28470/40589 loss 0.2766 acc 0.943 lb 1.961 z 0.337 lr 4.32e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.949 max_load(range) 0.054-0.219 step 28480/40589 loss 0.2697 acc 0.944 lb 1.960 z 0.368 lr 4.32e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28490/40589 loss 0.2581 acc 0.947 lb 1.966 z 0.431 lr 4.31e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28500/40589 loss 0.2745 acc 0.944 lb 1.960 z 0.338 lr 4.30e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.943 max_load(range) 0.058-0.206 step 28510/40589 loss 0.2743 acc 0.944 lb 1.964 z 0.326 lr 4.30e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28520/40589 loss 0.2693 acc 0.945 lb 1.953 z 0.398 lr 4.29e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.939 max_load(range) 0.054-0.376 step 28530/40589 loss 0.2611 acc 0.946 lb 1.960 z 0.426 lr 4.28e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28540/40589 loss 0.2723 acc 0.945 lb 1.964 z 0.338 lr 4.28e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28550/40589 loss 0.2712 acc 0.944 lb 1.959 z 0.388 lr 4.27e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.956 max_load(range) 0.055-0.229 step 28560/40589 loss 0.2726 acc 0.944 lb 1.951 z 0.392 lr 4.26e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28570/40589 loss 0.2599 acc 0.947 lb 1.957 z 0.469 lr 4.26e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.923 max_load(range) 0.069-0.126 step 28580/40589 loss 0.2734 acc 0.944 lb 1.963 z 0.352 lr 4.25e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28590/40589 loss 0.2739 acc 0.944 lb 1.958 z 0.393 lr 4.24e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28600/40589 loss 0.2690 acc 0.945 lb 1.953 z 0.401 lr 4.24e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.944 max_load(range) 0.073-0.166 step 28610/40589 loss 0.2581 acc 0.948 lb 1.961 z 0.401 lr 4.23e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28620/40589 loss 0.2727 acc 0.944 lb 1.958 z 0.412 lr 4.22e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.928 max_load(range) 0.067-0.158 step 28630/40589 loss 0.2775 acc 0.943 lb 1.962 z 0.386 lr 4.22e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28640/40589 loss 0.2696 acc 0.944 lb 1.954 z 0.335 lr 4.21e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28650/40589 loss 0.2569 acc 0.948 lb 1.959 z 0.374 lr 4.20e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.896 max_load(range) 0.064-0.308 step 28660/40589 loss 0.2749 acc 0.944 lb 1.960 z 0.384 lr 4.20e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28670/40589 loss 0.2734 acc 0.943 lb 1.960 z 0.322 lr 4.19e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.952 max_load(range) 0.060-0.209 step 28680/40589 loss 0.2674 acc 0.945 lb 1.956 z 0.327 lr 4.19e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28690/40589 loss 0.2588 acc 0.947 lb 1.960 z 0.396 lr 4.18e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28700/40589 loss 0.2710 acc 0.944 lb 1.960 z 0.394 lr 4.17e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.940 max_load(range) 0.052-0.229 step 28710/40589 loss 0.2745 acc 0.943 lb 1.962 z 0.338 lr 4.17e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28720/40589 loss 0.2683 acc 0.945 lb 1.954 z 0.354 lr 4.16e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.941 max_load(range) 0.057-0.371 step 28730/40589 loss 0.2602 acc 0.947 lb 1.960 z 0.469 lr 4.15e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28740/40589 loss 0.2727 acc 0.945 lb 1.958 z 0.335 lr 4.15e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28750/40589 loss 0.2720 acc 0.944 lb 1.961 z 0.368 lr 4.14e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.961 max_load(range) 0.051-0.239 step 28760/40589 loss 0.2676 acc 0.945 lb 1.953 z 0.387 lr 4.13e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28770/40589 loss 0.2564 acc 0.947 lb 1.960 z 0.409 lr 4.13e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.936 max_load(range) 0.055-0.162 step 28780/40589 loss 0.2738 acc 0.944 lb 1.959 z 0.394 lr 4.12e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28790/40589 loss 0.2778 acc 0.942 lb 1.956 z 0.387 lr 4.11e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28800/40589 loss 0.2664 acc 0.945 lb 1.950 z 0.354 lr 4.11e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.953 max_load(range) 0.057-0.177 step 28810/40589 loss 0.2566 acc 0.948 lb 1.961 z 0.411 lr 4.10e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28820/40589 loss 0.2757 acc 0.943 lb 1.967 z 0.336 lr 4.09e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.925 max_load(range) 0.074-0.147 step 28830/40589 loss 0.2785 acc 0.943 lb 1.958 z 0.365 lr 4.09e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28840/40589 loss 0.2650 acc 0.946 lb 1.952 z 0.324 lr 4.08e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28850/40589 loss 0.2583 acc 0.947 lb 1.962 z 0.390 lr 4.08e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.905 max_load(range) 0.057-0.288 step 28860/40589 loss 0.2727 acc 0.944 lb 1.963 z 0.415 lr 4.07e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28870/40589 loss 0.2755 acc 0.943 lb 1.960 z 0.320 lr 4.06e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.946 max_load(range) 0.059-0.227 step 28880/40589 loss 0.2667 acc 0.945 lb 1.952 z 0.392 lr 4.06e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28890/40589 loss 0.2579 acc 0.947 lb 1.962 z 0.444 lr 4.05e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28900/40589 loss 0.2747 acc 0.944 lb 1.964 z 0.358 lr 4.04e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.921 max_load(range) 0.056-0.264 step 28910/40589 loss 0.2749 acc 0.944 lb 1.961 z 0.380 lr 4.04e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28920/40589 loss 0.2685 acc 0.945 lb 1.955 z 0.387 lr 4.03e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.940 max_load(range) 0.058-0.373 step 28930/40589 loss 0.2535 acc 0.948 lb 1.960 z 0.477 lr 4.02e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28940/40589 loss 0.2739 acc 0.944 lb 1.958 z 0.410 lr 4.02e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28950/40589 loss 0.2750 acc 0.943 lb 1.961 z 0.349 lr 4.01e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.958 max_load(range) 0.055-0.240 step 28960/40589 loss 0.2642 acc 0.946 lb 1.961 z 0.380 lr 4.00e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28970/40589 loss 0.2586 acc 0.947 lb 1.966 z 0.329 lr 4.00e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.926 max_load(range) 0.072-0.152 step 28980/40589 loss 0.2734 acc 0.945 lb 1.962 z 0.341 lr 3.99e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 28990/40589 loss 0.2729 acc 0.944 lb 1.956 z 0.376 lr 3.99e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left step 29000/40589 loss 0.2699 acc 0.945 lb 1.951 z 0.318 lr 3.98e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.6d left [routing] mean entropy_norm 0.947 max_load(range) 0.059-0.188 [save] runs/pretrain_flagship_h100_4x8/step29000 step 29010/40589 loss 0.2572 acc 0.947 lb 1.964 z 0.392 lr 3.97e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29020/40589 loss 0.2731 acc 0.944 lb 1.961 z 0.393 lr 3.97e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.932 max_load(range) 0.064-0.171 step 29030/40589 loss 0.2725 acc 0.944 lb 1.960 z 0.385 lr 3.96e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29040/40589 loss 0.2650 acc 0.946 lb 1.960 z 0.394 lr 3.95e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29050/40589 loss 0.2567 acc 0.947 lb 1.963 z 0.379 lr 3.95e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.894 max_load(range) 0.064-0.315 step 29060/40589 loss 0.2764 acc 0.944 lb 1.961 z 0.399 lr 3.94e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29070/40589 loss 0.2763 acc 0.944 lb 1.962 z 0.328 lr 3.93e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.948 max_load(range) 0.061-0.198 step 29080/40589 loss 0.2688 acc 0.945 lb 1.957 z 0.406 lr 3.93e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29090/40589 loss 0.2566 acc 0.949 lb 1.961 z 0.423 lr 3.92e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29100/40589 loss 0.2754 acc 0.944 lb 1.964 z 0.333 lr 3.92e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.943 max_load(range) 0.054-0.207 step 29110/40589 loss 0.2734 acc 0.944 lb 1.960 z 0.359 lr 3.91e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29120/40589 loss 0.2723 acc 0.944 lb 1.950 z 0.383 lr 3.90e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.940 max_load(range) 0.057-0.360 step 29130/40589 loss 0.2578 acc 0.947 lb 1.961 z 0.393 lr 3.90e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29140/40589 loss 0.2754 acc 0.944 lb 1.965 z 0.398 lr 3.89e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29150/40589 loss 0.2754 acc 0.943 lb 1.965 z 0.383 lr 3.88e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.961 max_load(range) 0.055-0.237 step 29160/40589 loss 0.2684 acc 0.945 lb 1.952 z 0.389 lr 3.88e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29170/40589 loss 0.2551 acc 0.948 lb 1.963 z 0.437 lr 3.87e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.928 max_load(range) 0.072-0.132 step 29180/40589 loss 0.2728 acc 0.944 lb 1.963 z 0.398 lr 3.87e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29190/40589 loss 0.2743 acc 0.943 lb 1.959 z 0.382 lr 3.86e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29200/40589 loss 0.2694 acc 0.945 lb 1.950 z 0.346 lr 3.85e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.951 max_load(range) 0.055-0.182 step 29210/40589 loss 0.2642 acc 0.946 lb 1.959 z 0.437 lr 3.85e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29220/40589 loss 0.2765 acc 0.944 lb 1.964 z 0.364 lr 3.84e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.928 max_load(range) 0.064-0.194 step 29230/40589 loss 0.2763 acc 0.943 lb 1.964 z 0.400 lr 3.83e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29240/40589 loss 0.2707 acc 0.944 lb 1.952 z 0.366 lr 3.83e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29250/40589 loss 0.2500 acc 0.949 lb 1.959 z 0.463 lr 3.82e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.880 max_load(range) 0.072-0.341 step 29260/40589 loss 0.2725 acc 0.944 lb 1.961 z 0.383 lr 3.82e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29270/40589 loss 0.2742 acc 0.943 lb 1.957 z 0.314 lr 3.81e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.946 max_load(range) 0.058-0.215 step 29280/40589 loss 0.2678 acc 0.946 lb 1.947 z 0.341 lr 3.80e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29290/40589 loss 0.2603 acc 0.947 lb 1.961 z 0.348 lr 3.80e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29300/40589 loss 0.2753 acc 0.944 lb 1.964 z 0.416 lr 3.79e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.921 max_load(range) 0.061-0.269 step 29310/40589 loss 0.2771 acc 0.943 lb 1.957 z 0.338 lr 3.78e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29320/40589 loss 0.2655 acc 0.945 lb 1.957 z 0.325 lr 3.78e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.938 max_load(range) 0.054-0.375 step 29330/40589 loss 0.2615 acc 0.947 lb 1.960 z 0.413 lr 3.77e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29340/40589 loss 0.2743 acc 0.944 lb 1.961 z 0.346 lr 3.77e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29350/40589 loss 0.2751 acc 0.943 lb 1.957 z 0.387 lr 3.76e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.957 max_load(range) 0.057-0.227 step 29360/40589 loss 0.2701 acc 0.945 lb 1.952 z 0.384 lr 3.75e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29370/40589 loss 0.2594 acc 0.947 lb 1.960 z 0.462 lr 3.75e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.931 max_load(range) 0.061-0.139 step 29380/40589 loss 0.2721 acc 0.945 lb 1.962 z 0.400 lr 3.74e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29390/40589 loss 0.2740 acc 0.944 lb 1.958 z 0.386 lr 3.73e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29400/40589 loss 0.2687 acc 0.945 lb 1.953 z 0.378 lr 3.73e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.952 max_load(range) 0.057-0.178 step 29410/40589 loss 0.2549 acc 0.948 lb 1.962 z 0.402 lr 3.72e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29420/40589 loss 0.2738 acc 0.944 lb 1.956 z 0.390 lr 3.72e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.932 max_load(range) 0.068-0.188 step 29430/40589 loss 0.2727 acc 0.944 lb 1.960 z 0.386 lr 3.71e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29440/40589 loss 0.2683 acc 0.945 lb 1.951 z 0.340 lr 3.70e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29450/40589 loss 0.2620 acc 0.946 lb 1.962 z 0.456 lr 3.70e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.888 max_load(range) 0.070-0.321 step 29460/40589 loss 0.2732 acc 0.944 lb 1.957 z 0.348 lr 3.69e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29470/40589 loss 0.2753 acc 0.943 lb 1.963 z 0.381 lr 3.68e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.952 max_load(range) 0.052-0.211 step 29480/40589 loss 0.2671 acc 0.946 lb 1.954 z 0.380 lr 3.68e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29490/40589 loss 0.2607 acc 0.947 lb 1.962 z 0.432 lr 3.67e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29500/40589 loss 0.2729 acc 0.944 lb 1.957 z 0.361 lr 3.67e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.947 max_load(range) 0.049-0.207 step 29510/40589 loss 0.2723 acc 0.944 lb 1.955 z 0.337 lr 3.66e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29520/40589 loss 0.2655 acc 0.945 lb 1.949 z 0.361 lr 3.65e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.941 max_load(range) 0.053-0.376 step 29530/40589 loss 0.2610 acc 0.947 lb 1.958 z 0.400 lr 3.65e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29540/40589 loss 0.2741 acc 0.944 lb 1.960 z 0.384 lr 3.64e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29550/40589 loss 0.2759 acc 0.943 lb 1.959 z 0.402 lr 3.63e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.958 max_load(range) 0.052-0.221 step 29560/40589 loss 0.2672 acc 0.946 lb 1.954 z 0.377 lr 3.63e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29570/40589 loss 0.2596 acc 0.947 lb 1.961 z 0.411 lr 3.62e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.937 max_load(range) 0.062-0.138 step 29580/40589 loss 0.2727 acc 0.944 lb 1.959 z 0.370 lr 3.62e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29590/40589 loss 0.2744 acc 0.943 lb 1.964 z 0.375 lr 3.61e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29600/40589 loss 0.2675 acc 0.945 lb 1.954 z 0.390 lr 3.60e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.954 max_load(range) 0.056-0.181 step 29610/40589 loss 0.2596 acc 0.947 lb 1.961 z 0.447 lr 3.60e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29620/40589 loss 0.2711 acc 0.945 lb 1.961 z 0.395 lr 3.59e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.914 max_load(range) 0.068-0.178 step 29630/40589 loss 0.2752 acc 0.943 lb 1.953 z 0.327 lr 3.59e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29640/40589 loss 0.2712 acc 0.944 lb 1.951 z 0.377 lr 3.58e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29650/40589 loss 0.2579 acc 0.947 lb 1.961 z 0.394 lr 3.57e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.893 max_load(range) 0.064-0.312 step 29660/40589 loss 0.2754 acc 0.944 lb 1.957 z 0.342 lr 3.57e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29670/40589 loss 0.2769 acc 0.942 lb 1.957 z 0.371 lr 3.56e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.950 max_load(range) 0.050-0.226 step 29680/40589 loss 0.2665 acc 0.946 lb 1.956 z 0.383 lr 3.56e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29690/40589 loss 0.2578 acc 0.948 lb 1.958 z 0.406 lr 3.55e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29700/40589 loss 0.2697 acc 0.946 lb 1.958 z 0.394 lr 3.54e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.939 max_load(range) 0.056-0.230 step 29710/40589 loss 0.2738 acc 0.943 lb 1.956 z 0.385 lr 3.54e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29720/40589 loss 0.2720 acc 0.944 lb 1.954 z 0.383 lr 3.53e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.938 max_load(range) 0.053-0.376 step 29730/40589 loss 0.2573 acc 0.947 lb 1.963 z 0.485 lr 3.52e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29740/40589 loss 0.2739 acc 0.944 lb 1.958 z 0.323 lr 3.52e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left step 29750/40589 loss 0.2744 acc 0.943 lb 1.958 z 0.367 lr 3.51e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.5d left [routing] mean entropy_norm 0.959 max_load(range) 0.051-0.228 step 29760/40589 loss 0.2635 acc 0.946 lb 1.953 z 0.384 lr 3.51e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 29770/40589 loss 0.2573 acc 0.947 lb 1.957 z 0.433 lr 3.50e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.936 max_load(range) 0.064-0.143 step 29780/40589 loss 0.2711 acc 0.945 lb 1.957 z 0.379 lr 3.49e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 29790/40589 loss 0.2774 acc 0.943 lb 1.958 z 0.362 lr 3.49e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 29800/40589 loss 0.2657 acc 0.946 lb 1.952 z 0.375 lr 3.48e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.952 max_load(range) 0.060-0.177 step 29810/40589 loss 0.2611 acc 0.947 lb 1.956 z 0.473 lr 3.48e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 29820/40589 loss 0.2731 acc 0.944 lb 1.961 z 0.327 lr 3.47e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.936 max_load(range) 0.063-0.176 step 29830/40589 loss 0.2742 acc 0.943 lb 1.960 z 0.379 lr 3.46e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 29840/40589 loss 0.2688 acc 0.946 lb 1.948 z 0.362 lr 3.46e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 29850/40589 loss 0.2533 acc 0.948 lb 1.959 z 0.437 lr 3.45e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.886 max_load(range) 0.069-0.327 step 29860/40589 loss 0.2759 acc 0.944 lb 1.958 z 0.390 lr 3.45e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 29870/40589 loss 0.2744 acc 0.944 lb 1.962 z 0.370 lr 3.44e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.951 max_load(range) 0.056-0.210 step 29880/40589 loss 0.2687 acc 0.945 lb 1.948 z 0.357 lr 3.43e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 29890/40589 loss 0.2586 acc 0.947 lb 1.961 z 0.408 lr 3.43e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 29900/40589 loss 0.2709 acc 0.945 lb 1.957 z 0.388 lr 3.42e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.947 max_load(range) 0.051-0.200 step 29910/40589 loss 0.2739 acc 0.944 lb 1.961 z 0.317 lr 3.42e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 29920/40589 loss 0.2673 acc 0.945 lb 1.951 z 0.381 lr 3.41e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.937 max_load(range) 0.054-0.385 step 29930/40589 loss 0.2596 acc 0.947 lb 1.961 z 0.452 lr 3.40e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 29940/40589 loss 0.2714 acc 0.944 lb 1.957 z 0.347 lr 3.40e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 29950/40589 loss 0.2757 acc 0.943 lb 1.959 z 0.373 lr 3.39e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.960 max_load(range) 0.052-0.244 step 29960/40589 loss 0.2676 acc 0.945 lb 1.955 z 0.381 lr 3.39e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 29970/40589 loss 0.2563 acc 0.948 lb 1.960 z 0.416 lr 3.38e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.933 max_load(range) 0.062-0.148 step 29980/40589 loss 0.2740 acc 0.944 lb 1.961 z 0.398 lr 3.37e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 29990/40589 loss 0.2730 acc 0.944 lb 1.959 z 0.375 lr 3.37e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30000/40589 loss 0.2650 acc 0.945 lb 1.953 z 0.364 lr 3.36e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.944 max_load(range) 0.061-0.167 [save] runs/pretrain_flagship_h100_4x8/step30000 step 30010/40589 loss 0.2590 acc 0.947 lb 1.959 z 0.397 lr 3.36e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30020/40589 loss 0.2696 acc 0.945 lb 1.959 z 0.367 lr 3.35e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.936 max_load(range) 0.066-0.162 step 30030/40589 loss 0.2739 acc 0.943 lb 1.961 z 0.363 lr 3.34e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30040/40589 loss 0.2681 acc 0.945 lb 1.951 z 0.365 lr 3.34e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30050/40589 loss 0.2558 acc 0.948 lb 1.959 z 0.447 lr 3.33e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.895 max_load(range) 0.068-0.318 step 30060/40589 loss 0.2721 acc 0.944 lb 1.960 z 0.381 lr 3.33e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30070/40589 loss 0.2712 acc 0.944 lb 1.958 z 0.314 lr 3.32e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.952 max_load(range) 0.052-0.211 step 30080/40589 loss 0.2696 acc 0.945 lb 1.950 z 0.377 lr 3.31e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30090/40589 loss 0.2531 acc 0.948 lb 1.959 z 0.431 lr 3.31e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30100/40589 loss 0.2713 acc 0.945 lb 1.962 z 0.393 lr 3.30e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.942 max_load(range) 0.053-0.223 step 30110/40589 loss 0.2766 acc 0.943 lb 1.959 z 0.311 lr 3.30e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30120/40589 loss 0.2676 acc 0.945 lb 1.956 z 0.382 lr 3.29e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.939 max_load(range) 0.054-0.372 step 30130/40589 loss 0.2589 acc 0.947 lb 1.956 z 0.450 lr 3.28e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30140/40589 loss 0.2754 acc 0.944 lb 1.959 z 0.388 lr 3.28e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30150/40589 loss 0.2743 acc 0.943 lb 1.958 z 0.369 lr 3.27e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.961 max_load(range) 0.052-0.227 step 30160/40589 loss 0.2638 acc 0.946 lb 1.954 z 0.371 lr 3.27e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30170/40589 loss 0.2527 acc 0.949 lb 1.958 z 0.383 lr 3.26e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.931 max_load(range) 0.071-0.136 step 30180/40589 loss 0.2722 acc 0.945 lb 1.961 z 0.374 lr 3.26e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30190/40589 loss 0.2746 acc 0.943 lb 1.959 z 0.322 lr 3.25e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30200/40589 loss 0.2647 acc 0.945 lb 1.953 z 0.328 lr 3.24e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.954 max_load(range) 0.058-0.159 step 30210/40589 loss 0.2554 acc 0.947 lb 1.958 z 0.478 lr 3.24e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30220/40589 loss 0.2742 acc 0.944 lb 1.964 z 0.372 lr 3.23e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.934 max_load(range) 0.068-0.154 step 30230/40589 loss 0.2770 acc 0.943 lb 1.960 z 0.361 lr 3.23e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30240/40589 loss 0.2658 acc 0.945 lb 1.957 z 0.316 lr 3.22e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30250/40589 loss 0.2571 acc 0.948 lb 1.963 z 0.403 lr 3.21e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.897 max_load(range) 0.067-0.321 step 30260/40589 loss 0.2755 acc 0.944 lb 1.960 z 0.375 lr 3.21e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30270/40589 loss 0.2762 acc 0.942 lb 1.960 z 0.367 lr 3.20e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.954 max_load(range) 0.057-0.199 step 30280/40589 loss 0.2680 acc 0.945 lb 1.952 z 0.314 lr 3.20e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30290/40589 loss 0.2569 acc 0.948 lb 1.958 z 0.435 lr 3.19e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30300/40589 loss 0.2745 acc 0.944 lb 1.962 z 0.380 lr 3.18e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.949 max_load(range) 0.048-0.201 step 30310/40589 loss 0.2733 acc 0.943 lb 1.959 z 0.356 lr 3.18e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30320/40589 loss 0.2668 acc 0.945 lb 1.955 z 0.389 lr 3.17e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.943 max_load(range) 0.055-0.369 step 30330/40589 loss 0.2556 acc 0.948 lb 1.964 z 0.313 lr 3.17e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30340/40589 loss 0.2718 acc 0.944 lb 1.961 z 0.380 lr 3.16e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30350/40589 loss 0.2729 acc 0.944 lb 1.956 z 0.367 lr 3.16e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.962 max_load(range) 0.052-0.218 step 30360/40589 loss 0.2693 acc 0.944 lb 1.953 z 0.358 lr 3.15e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30370/40589 loss 0.2545 acc 0.948 lb 1.959 z 0.427 lr 3.14e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.937 max_load(range) 0.062-0.147 step 30380/40589 loss 0.2691 acc 0.946 lb 1.960 z 0.322 lr 3.14e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30390/40589 loss 0.2728 acc 0.944 lb 1.960 z 0.359 lr 3.13e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30400/40589 loss 0.2655 acc 0.945 lb 1.957 z 0.357 lr 3.13e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.956 max_load(range) 0.060-0.191 step 30410/40589 loss 0.2524 acc 0.948 lb 1.955 z 0.462 lr 3.12e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30420/40589 loss 0.2742 acc 0.944 lb 1.962 z 0.385 lr 3.12e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.938 max_load(range) 0.066-0.166 step 30430/40589 loss 0.2724 acc 0.944 lb 1.959 z 0.364 lr 3.11e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30440/40589 loss 0.2672 acc 0.946 lb 1.949 z 0.371 lr 3.10e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30450/40589 loss 0.2544 acc 0.949 lb 1.959 z 0.429 lr 3.10e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.900 max_load(range) 0.064-0.315 step 30460/40589 loss 0.2710 acc 0.945 lb 1.961 z 0.367 lr 3.09e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30470/40589 loss 0.2743 acc 0.943 lb 1.957 z 0.367 lr 3.09e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.952 max_load(range) 0.055-0.222 step 30480/40589 loss 0.2664 acc 0.945 lb 1.959 z 0.351 lr 3.08e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30490/40589 loss 0.2526 acc 0.948 lb 1.962 z 0.367 lr 3.07e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left step 30500/40589 loss 0.2728 acc 0.944 lb 1.970 z 0.381 lr 3.07e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.4d left [routing] mean entropy_norm 0.941 max_load(range) 0.052-0.230 step 30510/40589 loss 0.2749 acc 0.943 lb 1.960 z 0.364 lr 3.06e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30520/40589 loss 0.2658 acc 0.945 lb 1.959 z 0.375 lr 3.06e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.939 max_load(range) 0.052-0.378 step 30530/40589 loss 0.2556 acc 0.948 lb 1.957 z 0.410 lr 3.05e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30540/40589 loss 0.2717 acc 0.944 lb 1.959 z 0.350 lr 3.05e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30550/40589 loss 0.2704 acc 0.944 lb 1.958 z 0.365 lr 3.04e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.961 max_load(range) 0.048-0.228 step 30560/40589 loss 0.2687 acc 0.945 lb 1.952 z 0.315 lr 3.03e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30570/40589 loss 0.2539 acc 0.948 lb 1.960 z 0.450 lr 3.03e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.937 max_load(range) 0.061-0.145 step 30580/40589 loss 0.2726 acc 0.944 lb 1.959 z 0.385 lr 3.02e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30590/40589 loss 0.2735 acc 0.943 lb 1.957 z 0.346 lr 3.02e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30600/40589 loss 0.2668 acc 0.946 lb 1.957 z 0.371 lr 3.01e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.954 max_load(range) 0.061-0.162 step 30610/40589 loss 0.2551 acc 0.948 lb 1.960 z 0.392 lr 3.01e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30620/40589 loss 0.2727 acc 0.944 lb 1.962 z 0.381 lr 3.00e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.937 max_load(range) 0.059-0.175 step 30630/40589 loss 0.2715 acc 0.943 lb 1.958 z 0.375 lr 2.99e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30640/40589 loss 0.2650 acc 0.945 lb 1.953 z 0.356 lr 2.99e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30650/40589 loss 0.2530 acc 0.948 lb 1.961 z 0.406 lr 2.98e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.888 max_load(range) 0.072-0.333 step 30660/40589 loss 0.2703 acc 0.945 lb 1.962 z 0.385 lr 2.98e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30670/40589 loss 0.2719 acc 0.944 lb 1.957 z 0.353 lr 2.97e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.953 max_load(range) 0.053-0.225 step 30680/40589 loss 0.2621 acc 0.946 lb 1.956 z 0.368 lr 2.97e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30690/40589 loss 0.2563 acc 0.948 lb 1.960 z 0.434 lr 2.96e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30700/40589 loss 0.2682 acc 0.945 lb 1.960 z 0.382 lr 2.95e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.940 max_load(range) 0.055-0.242 step 30710/40589 loss 0.2717 acc 0.944 lb 1.957 z 0.321 lr 2.95e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30720/40589 loss 0.2649 acc 0.946 lb 1.957 z 0.369 lr 2.94e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.940 max_load(range) 0.058-0.373 step 30730/40589 loss 0.2570 acc 0.948 lb 1.957 z 0.437 lr 2.94e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30740/40589 loss 0.2719 acc 0.944 lb 1.959 z 0.358 lr 2.93e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30750/40589 loss 0.2724 acc 0.944 lb 1.955 z 0.383 lr 2.93e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.961 max_load(range) 0.048-0.250 step 30760/40589 loss 0.2699 acc 0.945 lb 1.948 z 0.370 lr 2.92e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30770/40589 loss 0.2521 acc 0.949 lb 1.957 z 0.429 lr 2.92e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.933 max_load(range) 0.063-0.133 step 30780/40589 loss 0.2747 acc 0.944 lb 1.960 z 0.392 lr 2.91e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30790/40589 loss 0.2777 acc 0.943 lb 1.964 z 0.389 lr 2.90e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30800/40589 loss 0.2672 acc 0.945 lb 1.951 z 0.373 lr 2.90e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.956 max_load(range) 0.057-0.174 step 30810/40589 loss 0.2531 acc 0.948 lb 1.962 z 0.433 lr 2.89e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30820/40589 loss 0.2713 acc 0.944 lb 1.967 z 0.363 lr 2.89e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.938 max_load(range) 0.063-0.164 step 30830/40589 loss 0.2722 acc 0.944 lb 1.958 z 0.357 lr 2.88e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30840/40589 loss 0.2647 acc 0.946 lb 1.949 z 0.361 lr 2.88e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30850/40589 loss 0.2567 acc 0.948 lb 1.956 z 0.373 lr 2.87e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.895 max_load(range) 0.065-0.324 step 30860/40589 loss 0.2719 acc 0.945 lb 1.962 z 0.382 lr 2.86e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30870/40589 loss 0.2748 acc 0.943 lb 1.957 z 0.361 lr 2.86e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.954 max_load(range) 0.054-0.224 step 30880/40589 loss 0.2686 acc 0.946 lb 1.949 z 0.355 lr 2.85e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30890/40589 loss 0.2586 acc 0.947 lb 1.957 z 0.453 lr 2.85e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30900/40589 loss 0.2722 acc 0.944 lb 1.959 z 0.374 lr 2.84e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.940 max_load(range) 0.052-0.227 step 30910/40589 loss 0.2747 acc 0.943 lb 1.959 z 0.350 lr 2.84e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30920/40589 loss 0.2691 acc 0.945 lb 1.954 z 0.363 lr 2.83e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.944 max_load(range) 0.056-0.360 step 30930/40589 loss 0.2555 acc 0.948 lb 1.959 z 0.438 lr 2.83e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30940/40589 loss 0.2717 acc 0.945 lb 1.962 z 0.382 lr 2.82e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30950/40589 loss 0.2768 acc 0.943 lb 1.955 z 0.363 lr 2.81e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.963 max_load(range) 0.051-0.218 step 30960/40589 loss 0.2657 acc 0.945 lb 1.954 z 0.369 lr 2.81e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30970/40589 loss 0.2515 acc 0.948 lb 1.957 z 0.446 lr 2.80e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.935 max_load(range) 0.060-0.137 step 30980/40589 loss 0.2689 acc 0.946 lb 1.960 z 0.380 lr 2.80e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 30990/40589 loss 0.2765 acc 0.943 lb 1.955 z 0.295 lr 2.79e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 31000/40589 loss 0.2665 acc 0.945 lb 1.956 z 0.363 lr 2.79e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.955 max_load(range) 0.060-0.175 [save] runs/pretrain_flagship_h100_4x8/step31000 step 31010/40589 loss 0.2539 acc 0.947 lb 1.957 z 0.426 lr 2.78e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 31020/40589 loss 0.2720 acc 0.945 lb 1.963 z 0.387 lr 2.78e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.936 max_load(range) 0.067-0.159 step 31030/40589 loss 0.2778 acc 0.943 lb 1.958 z 0.359 lr 2.77e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 31040/40589 loss 0.2669 acc 0.944 lb 1.949 z 0.361 lr 2.77e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 31050/40589 loss 0.2544 acc 0.948 lb 1.959 z 0.433 lr 2.76e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.878 max_load(range) 0.073-0.342 step 31060/40589 loss 0.2695 acc 0.945 lb 1.959 z 0.327 lr 2.75e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 31070/40589 loss 0.2733 acc 0.944 lb 1.956 z 0.374 lr 2.75e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.951 max_load(range) 0.055-0.230 step 31080/40589 loss 0.2684 acc 0.945 lb 1.953 z 0.364 lr 2.74e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 31090/40589 loss 0.2526 acc 0.949 lb 1.959 z 0.436 lr 2.74e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 31100/40589 loss 0.2723 acc 0.944 lb 1.959 z 0.381 lr 2.73e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.948 max_load(range) 0.050-0.202 step 31110/40589 loss 0.2716 acc 0.944 lb 1.956 z 0.369 lr 2.73e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 31120/40589 loss 0.2661 acc 0.945 lb 1.952 z 0.369 lr 2.72e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.943 max_load(range) 0.053-0.375 step 31130/40589 loss 0.2537 acc 0.948 lb 1.958 z 0.445 lr 2.72e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 31140/40589 loss 0.2733 acc 0.945 lb 1.961 z 0.329 lr 2.71e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 31150/40589 loss 0.2737 acc 0.943 lb 1.958 z 0.361 lr 2.70e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.964 max_load(range) 0.045-0.254 step 31160/40589 loss 0.2653 acc 0.946 lb 1.952 z 0.328 lr 2.70e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 31170/40589 loss 0.2562 acc 0.948 lb 1.956 z 0.421 lr 2.69e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.939 max_load(range) 0.059-0.149 step 31180/40589 loss 0.2729 acc 0.945 lb 1.961 z 0.306 lr 2.69e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 31190/40589 loss 0.2797 acc 0.943 lb 1.959 z 0.367 lr 2.68e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 31200/40589 loss 0.2681 acc 0.945 lb 1.949 z 0.367 lr 2.68e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.953 max_load(range) 0.061-0.175 step 31210/40589 loss 0.2541 acc 0.948 lb 1.961 z 0.426 lr 2.67e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 31220/40589 loss 0.2710 acc 0.944 lb 1.956 z 0.346 lr 2.67e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left [routing] mean entropy_norm 0.914 max_load(range) 0.069-0.204 step 31230/40589 loss 0.2680 acc 0.945 lb 1.962 z 0.318 lr 2.66e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 31240/40589 loss 0.2631 acc 0.946 lb 1.958 z 0.369 lr 2.66e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.3d left step 31250/40589 loss 0.2527 acc 0.948 lb 1.961 z 0.377 lr 2.65e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left [routing] mean entropy_norm 0.901 max_load(range) 0.067-0.309 step 31260/40589 loss 0.2701 acc 0.945 lb 1.962 z 0.380 lr 2.64e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left step 31270/40589 loss 0.2707 acc 0.944 lb 1.962 z 0.361 lr 2.64e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left [routing] mean entropy_norm 0.953 max_load(range) 0.055-0.200 step 31280/40589 loss 0.2638 acc 0.946 lb 1.950 z 0.308 lr 2.63e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left step 31290/40589 loss 0.2543 acc 0.948 lb 1.960 z 0.411 lr 2.63e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left step 31300/40589 loss 0.2712 acc 0.944 lb 1.960 z 0.364 lr 2.62e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left [routing] mean entropy_norm 0.938 max_load(range) 0.051-0.228 step 31310/40589 loss 0.2753 acc 0.943 lb 1.959 z 0.352 lr 2.62e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left step 31320/40589 loss 0.2664 acc 0.945 lb 1.954 z 0.376 lr 2.61e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left [routing] mean entropy_norm 0.941 max_load(range) 0.057-0.373 step 31330/40589 loss 0.2584 acc 0.947 lb 1.960 z 0.428 lr 2.61e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left step 31340/40589 loss 0.2718 acc 0.945 lb 1.961 z 0.330 lr 2.60e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left step 31350/40589 loss 0.2784 acc 0.943 lb 1.957 z 0.358 lr 2.60e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left [routing] mean entropy_norm 0.961 max_load(range) 0.054-0.211 step 31360/40589 loss 0.2643 acc 0.946 lb 1.956 z 0.354 lr 2.59e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left step 31370/40589 loss 0.2580 acc 0.947 lb 1.960 z 0.350 lr 2.59e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left [routing] mean entropy_norm 0.935 max_load(range) 0.057-0.146 step 31380/40589 loss 0.2717 acc 0.944 lb 1.960 z 0.360 lr 2.58e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left step 31390/40589 loss 0.2737 acc 0.944 lb 1.957 z 0.373 lr 2.58e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left step 31400/40589 loss 0.2651 acc 0.946 lb 1.954 z 0.375 lr 2.57e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left [routing] mean entropy_norm 0.951 max_load(range) 0.058-0.172 step 31410/40589 loss 0.2539 acc 0.948 lb 1.957 z 0.398 lr 2.56e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left step 31420/40589 loss 0.2748 acc 0.944 lb 1.961 z 0.372 lr 2.56e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left [routing] mean entropy_norm 0.934 max_load(range) 0.056-0.182 step 31430/40589 loss 0.2751 acc 0.944 lb 1.959 z 0.361 lr 2.55e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left step 31440/40589 loss 0.2676 acc 0.946 lb 1.954 z 0.354 lr 2.55e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.7) | 1.2d left step 31450/40589 loss 0.2555 acc 0.947 lb 1.958 z 0.439 lr 2.54e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.897 max_load(range) 0.067-0.313 step 31460/40589 loss 0.2710 acc 0.945 lb 1.959 z 0.321 lr 2.54e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31470/40589 loss 0.2732 acc 0.943 lb 1.959 z 0.348 lr 2.53e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.955 max_load(range) 0.053-0.182 step 31480/40589 loss 0.2663 acc 0.946 lb 1.952 z 0.369 lr 2.53e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31490/40589 loss 0.2563 acc 0.948 lb 1.962 z 0.435 lr 2.52e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31500/40589 loss 0.2712 acc 0.945 lb 1.962 z 0.363 lr 2.52e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.945 max_load(range) 0.052-0.208 step 31510/40589 loss 0.2739 acc 0.944 lb 1.956 z 0.357 lr 2.51e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31520/40589 loss 0.2674 acc 0.945 lb 1.952 z 0.364 lr 2.51e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.941 max_load(range) 0.055-0.371 step 31530/40589 loss 0.2514 acc 0.948 lb 1.959 z 0.450 lr 2.50e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31540/40589 loss 0.2686 acc 0.945 lb 1.962 z 0.381 lr 2.50e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31550/40589 loss 0.2695 acc 0.944 lb 1.958 z 0.323 lr 2.49e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.962 max_load(range) 0.048-0.231 step 31560/40589 loss 0.2676 acc 0.945 lb 1.951 z 0.359 lr 2.48e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31570/40589 loss 0.2559 acc 0.947 lb 1.960 z 0.443 lr 2.48e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.936 max_load(range) 0.064-0.135 step 31580/40589 loss 0.2736 acc 0.945 lb 1.961 z 0.373 lr 2.47e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31590/40589 loss 0.2739 acc 0.944 lb 1.960 z 0.364 lr 2.47e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31600/40589 loss 0.2712 acc 0.944 lb 1.955 z 0.366 lr 2.46e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.956 max_load(range) 0.057-0.166 step 31610/40589 loss 0.2562 acc 0.948 lb 1.961 z 0.419 lr 2.46e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31620/40589 loss 0.2735 acc 0.945 lb 1.962 z 0.371 lr 2.45e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.936 max_load(range) 0.064-0.162 step 31630/40589 loss 0.2777 acc 0.942 lb 1.959 z 0.359 lr 2.45e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31640/40589 loss 0.2656 acc 0.945 lb 1.954 z 0.371 lr 2.44e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31650/40589 loss 0.2616 acc 0.947 lb 1.957 z 0.443 lr 2.44e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.899 max_load(range) 0.067-0.316 step 31660/40589 loss 0.2708 acc 0.944 lb 1.961 z 0.352 lr 2.43e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31670/40589 loss 0.2728 acc 0.943 lb 1.956 z 0.353 lr 2.43e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.955 max_load(range) 0.052-0.207 step 31680/40589 loss 0.2672 acc 0.945 lb 1.953 z 0.346 lr 2.42e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31690/40589 loss 0.2545 acc 0.949 lb 1.961 z 0.436 lr 2.42e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31700/40589 loss 0.2712 acc 0.945 lb 1.959 z 0.366 lr 2.41e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.948 max_load(range) 0.051-0.202 step 31710/40589 loss 0.2747 acc 0.943 lb 1.956 z 0.363 lr 2.41e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31720/40589 loss 0.2645 acc 0.946 lb 1.954 z 0.357 lr 2.40e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.941 max_load(range) 0.056-0.371 step 31730/40589 loss 0.2567 acc 0.948 lb 1.963 z 0.441 lr 2.40e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31740/40589 loss 0.2707 acc 0.945 lb 1.960 z 0.367 lr 2.39e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31750/40589 loss 0.2773 acc 0.943 lb 1.960 z 0.349 lr 2.39e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.961 max_load(range) 0.048-0.221 step 31760/40589 loss 0.2675 acc 0.945 lb 1.949 z 0.346 lr 2.38e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31770/40589 loss 0.2562 acc 0.947 lb 1.959 z 0.433 lr 2.38e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.932 max_load(range) 0.060-0.135 step 31780/40589 loss 0.2701 acc 0.945 lb 1.961 z 0.372 lr 2.37e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31790/40589 loss 0.2738 acc 0.944 lb 1.960 z 0.374 lr 2.37e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31800/40589 loss 0.2660 acc 0.945 lb 1.947 z 0.340 lr 2.36e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.954 max_load(range) 0.056-0.191 step 31810/40589 loss 0.2558 acc 0.948 lb 1.958 z 0.424 lr 2.35e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31820/40589 loss 0.2709 acc 0.945 lb 1.958 z 0.373 lr 2.35e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.933 max_load(range) 0.063-0.157 step 31830/40589 loss 0.2729 acc 0.944 lb 1.957 z 0.351 lr 2.34e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31840/40589 loss 0.2613 acc 0.946 lb 1.950 z 0.363 lr 2.34e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31850/40589 loss 0.2551 acc 0.948 lb 1.959 z 0.441 lr 2.33e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.903 max_load(range) 0.063-0.299 step 31860/40589 loss 0.2687 acc 0.945 lb 1.960 z 0.370 lr 2.33e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31870/40589 loss 0.2715 acc 0.944 lb 1.957 z 0.370 lr 2.32e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.953 max_load(range) 0.052-0.210 step 31880/40589 loss 0.2692 acc 0.945 lb 1.959 z 0.360 lr 2.32e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31890/40589 loss 0.2548 acc 0.948 lb 1.963 z 0.415 lr 2.31e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31900/40589 loss 0.2751 acc 0.944 lb 1.961 z 0.305 lr 2.31e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.946 max_load(range) 0.051-0.201 step 31910/40589 loss 0.2741 acc 0.943 lb 1.959 z 0.351 lr 2.30e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31920/40589 loss 0.2650 acc 0.946 lb 1.952 z 0.349 lr 2.30e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.942 max_load(range) 0.057-0.356 step 31930/40589 loss 0.2557 acc 0.948 lb 1.961 z 0.426 lr 2.29e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31940/40589 loss 0.2720 acc 0.944 lb 1.962 z 0.368 lr 2.29e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31950/40589 loss 0.2757 acc 0.943 lb 1.957 z 0.337 lr 2.28e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.960 max_load(range) 0.047-0.221 step 31960/40589 loss 0.2653 acc 0.945 lb 1.956 z 0.370 lr 2.28e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31970/40589 loss 0.2560 acc 0.947 lb 1.959 z 0.411 lr 2.27e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left [routing] mean entropy_norm 0.935 max_load(range) 0.062-0.142 step 31980/40589 loss 0.2761 acc 0.944 lb 1.961 z 0.352 lr 2.27e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 31990/40589 loss 0.2749 acc 0.943 lb 1.964 z 0.369 lr 2.26e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.2d left step 32000/40589 loss 0.2678 acc 0.945 lb 1.953 z 0.365 lr 2.26e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.952 max_load(range) 0.060-0.167 [save] runs/pretrain_flagship_h100_4x8/step32000 step 32010/40589 loss 0.2572 acc 0.947 lb 1.962 z 0.401 lr 2.25e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32020/40589 loss 0.2752 acc 0.944 lb 1.962 z 0.381 lr 2.25e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.934 max_load(range) 0.065-0.158 step 32030/40589 loss 0.2774 acc 0.942 lb 1.960 z 0.354 lr 2.24e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32040/40589 loss 0.2686 acc 0.945 lb 1.954 z 0.356 lr 2.24e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32050/40589 loss 0.2550 acc 0.948 lb 1.964 z 0.430 lr 2.23e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.893 max_load(range) 0.066-0.320 step 32060/40589 loss 0.2721 acc 0.944 lb 1.964 z 0.374 lr 2.23e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32070/40589 loss 0.2708 acc 0.944 lb 1.962 z 0.360 lr 2.22e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.955 max_load(range) 0.054-0.213 step 32080/40589 loss 0.2634 acc 0.946 lb 1.955 z 0.351 lr 2.22e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32090/40589 loss 0.2530 acc 0.949 lb 1.961 z 0.424 lr 2.21e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32100/40589 loss 0.2714 acc 0.945 lb 1.965 z 0.368 lr 2.21e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.947 max_load(range) 0.050-0.208 step 32110/40589 loss 0.2735 acc 0.943 lb 1.958 z 0.353 lr 2.20e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32120/40589 loss 0.2636 acc 0.945 lb 1.958 z 0.347 lr 2.20e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.943 max_load(range) 0.054-0.366 step 32130/40589 loss 0.2536 acc 0.948 lb 1.959 z 0.407 lr 2.19e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32140/40589 loss 0.2709 acc 0.945 lb 1.963 z 0.375 lr 2.19e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32150/40589 loss 0.2766 acc 0.943 lb 1.956 z 0.339 lr 2.18e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.961 max_load(range) 0.049-0.244 step 32160/40589 loss 0.2642 acc 0.946 lb 1.955 z 0.347 lr 2.18e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32170/40589 loss 0.2564 acc 0.947 lb 1.961 z 0.422 lr 2.17e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.935 max_load(range) 0.065-0.131 step 32180/40589 loss 0.2724 acc 0.944 lb 1.962 z 0.366 lr 2.17e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32190/40589 loss 0.2766 acc 0.942 lb 1.956 z 0.358 lr 2.16e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32200/40589 loss 0.2676 acc 0.945 lb 1.955 z 0.372 lr 2.16e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.955 max_load(range) 0.057-0.180 step 32210/40589 loss 0.2550 acc 0.948 lb 1.961 z 0.411 lr 2.15e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32220/40589 loss 0.2749 acc 0.944 lb 1.960 z 0.377 lr 2.15e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.928 max_load(range) 0.065-0.145 step 32230/40589 loss 0.2749 acc 0.943 lb 1.955 z 0.341 lr 2.14e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32240/40589 loss 0.2647 acc 0.946 lb 1.949 z 0.358 lr 2.14e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32250/40589 loss 0.2536 acc 0.948 lb 1.960 z 0.443 lr 2.13e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.891 max_load(range) 0.066-0.326 step 32260/40589 loss 0.2729 acc 0.944 lb 1.963 z 0.352 lr 2.13e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32270/40589 loss 0.2770 acc 0.943 lb 1.957 z 0.357 lr 2.12e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.953 max_load(range) 0.051-0.228 step 32280/40589 loss 0.2670 acc 0.945 lb 1.954 z 0.355 lr 2.12e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32290/40589 loss 0.2518 acc 0.948 lb 1.961 z 0.450 lr 2.11e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32300/40589 loss 0.2722 acc 0.944 lb 1.959 z 0.368 lr 2.11e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.946 max_load(range) 0.050-0.192 step 32310/40589 loss 0.2737 acc 0.943 lb 1.956 z 0.351 lr 2.10e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32320/40589 loss 0.2653 acc 0.945 lb 1.954 z 0.323 lr 2.10e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.945 max_load(range) 0.053-0.367 step 32330/40589 loss 0.2552 acc 0.947 lb 1.960 z 0.398 lr 2.09e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32340/40589 loss 0.2712 acc 0.945 lb 1.960 z 0.369 lr 2.09e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32350/40589 loss 0.2754 acc 0.943 lb 1.957 z 0.361 lr 2.08e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.963 max_load(range) 0.046-0.239 step 32360/40589 loss 0.2635 acc 0.946 lb 1.953 z 0.357 lr 2.08e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32370/40589 loss 0.2600 acc 0.947 lb 1.961 z 0.439 lr 2.07e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.935 max_load(range) 0.062-0.140 step 32380/40589 loss 0.2705 acc 0.945 lb 1.961 z 0.368 lr 2.07e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32390/40589 loss 0.2752 acc 0.943 lb 1.960 z 0.361 lr 2.06e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32400/40589 loss 0.2667 acc 0.945 lb 1.952 z 0.342 lr 2.06e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.950 max_load(range) 0.060-0.160 step 32410/40589 loss 0.2596 acc 0.947 lb 1.962 z 0.435 lr 2.06e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32420/40589 loss 0.2712 acc 0.944 lb 1.960 z 0.361 lr 2.05e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.937 max_load(range) 0.062-0.138 step 32430/40589 loss 0.2760 acc 0.943 lb 1.957 z 0.336 lr 2.05e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32440/40589 loss 0.2666 acc 0.945 lb 1.952 z 0.369 lr 2.04e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32450/40589 loss 0.2589 acc 0.946 lb 1.960 z 0.424 lr 2.04e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.891 max_load(range) 0.068-0.328 step 32460/40589 loss 0.2758 acc 0.944 lb 1.964 z 0.370 lr 2.03e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32470/40589 loss 0.2783 acc 0.942 lb 1.956 z 0.332 lr 2.03e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.956 max_load(range) 0.055-0.200 step 32480/40589 loss 0.2678 acc 0.945 lb 1.954 z 0.349 lr 2.02e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32490/40589 loss 0.2558 acc 0.948 lb 1.961 z 0.423 lr 2.02e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32500/40589 loss 0.2689 acc 0.945 lb 1.962 z 0.358 lr 2.01e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.940 max_load(range) 0.055-0.224 step 32510/40589 loss 0.2720 acc 0.944 lb 1.957 z 0.343 lr 2.01e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32520/40589 loss 0.2663 acc 0.946 lb 1.953 z 0.363 lr 2.00e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.946 max_load(range) 0.055-0.363 step 32530/40589 loss 0.2560 acc 0.947 lb 1.958 z 0.423 lr 2.00e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32540/40589 loss 0.2716 acc 0.944 lb 1.962 z 0.346 lr 1.99e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32550/40589 loss 0.2736 acc 0.944 lb 1.962 z 0.296 lr 1.99e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.962 max_load(range) 0.047-0.234 step 32560/40589 loss 0.2653 acc 0.945 lb 1.952 z 0.359 lr 1.98e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32570/40589 loss 0.2589 acc 0.947 lb 1.962 z 0.373 lr 1.98e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.938 max_load(range) 0.063-0.158 step 32580/40589 loss 0.2718 acc 0.944 lb 1.960 z 0.356 lr 1.97e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32590/40589 loss 0.2749 acc 0.943 lb 1.955 z 0.354 lr 1.97e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32600/40589 loss 0.2669 acc 0.945 lb 1.954 z 0.356 lr 1.96e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.954 max_load(range) 0.055-0.176 step 32610/40589 loss 0.2565 acc 0.947 lb 1.961 z 0.413 lr 1.96e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32620/40589 loss 0.2727 acc 0.944 lb 1.962 z 0.364 lr 1.95e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.931 max_load(range) 0.069-0.145 step 32630/40589 loss 0.2733 acc 0.943 lb 1.957 z 0.331 lr 1.95e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32640/40589 loss 0.2688 acc 0.945 lb 1.953 z 0.356 lr 1.94e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32650/40589 loss 0.2589 acc 0.947 lb 1.961 z 0.407 lr 1.94e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.892 max_load(range) 0.066-0.322 step 32660/40589 loss 0.2684 acc 0.945 lb 1.962 z 0.364 lr 1.94e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32670/40589 loss 0.2741 acc 0.944 lb 1.959 z 0.351 lr 1.93e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.954 max_load(range) 0.048-0.217 step 32680/40589 loss 0.2645 acc 0.946 lb 1.952 z 0.357 lr 1.93e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32690/40589 loss 0.2524 acc 0.949 lb 1.960 z 0.417 lr 1.92e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32700/40589 loss 0.2743 acc 0.944 lb 1.965 z 0.367 lr 1.92e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.927 max_load(range) 0.057-0.252 step 32710/40589 loss 0.2717 acc 0.944 lb 1.959 z 0.347 lr 1.91e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32720/40589 loss 0.2655 acc 0.945 lb 1.952 z 0.359 lr 1.91e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left [routing] mean entropy_norm 0.944 max_load(range) 0.055-0.374 step 32730/40589 loss 0.2571 acc 0.947 lb 1.959 z 0.419 lr 1.90e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32740/40589 loss 0.2733 acc 0.944 lb 1.960 z 0.330 lr 1.90e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.1d left step 32750/40589 loss 0.2720 acc 0.944 lb 1.959 z 0.359 lr 1.89e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.0d left [routing] mean entropy_norm 0.963 max_load(range) 0.046-0.244 step 32760/40589 loss 0.2639 acc 0.946 lb 1.952 z 0.352 lr 1.89e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.0d left step 32770/40589 loss 0.2558 acc 0.948 lb 1.963 z 0.424 lr 1.88e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.0d left [routing] mean entropy_norm 0.939 max_load(range) 0.057-0.148 step 32780/40589 loss 0.2735 acc 0.944 lb 1.961 z 0.366 lr 1.88e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.0d left step 32790/40589 loss 0.2704 acc 0.944 lb 1.960 z 0.284 lr 1.87e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.0d left step 32800/40589 loss 0.2669 acc 0.945 lb 1.953 z 0.349 lr 1.87e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.0d left [routing] mean entropy_norm 0.954 max_load(range) 0.061-0.183 step 32810/40589 loss 0.2551 acc 0.948 lb 1.968 z 0.415 lr 1.87e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.0d left step 32820/40589 loss 0.2699 acc 0.945 lb 1.964 z 0.300 lr 1.86e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.0d left [routing] mean entropy_norm 0.940 max_load(range) 0.055-0.181 step 32830/40589 loss 0.2725 acc 0.943 lb 1.958 z 0.349 lr 1.86e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.0d left step 32840/40589 loss 0.2677 acc 0.945 lb 1.950 z 0.357 lr 1.85e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.0d left step 32850/40589 loss 0.2585 acc 0.947 lb 1.959 z 0.400 lr 1.85e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.0d left [routing] mean entropy_norm 0.899 max_load(range) 0.065-0.316 step 32860/40589 loss 0.2717 acc 0.944 lb 1.960 z 0.364 lr 1.84e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.0d left step 32870/40589 loss 0.2739 acc 0.943 lb 1.960 z 0.340 lr 1.84e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.0d left [routing] mean entropy_norm 0.949 max_load(range) 0.051-0.225 step 32880/40589 loss 0.2674 acc 0.945 lb 1.955 z 0.357 lr 1.83e-05 | 1417 pairs/s | mem 63.8/79 GiB (min 62.8) | 1.0d left step 32890/40589 loss 0.2575 acc 0.947 lb 1.962 z 0.404 lr 1.83e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left step 32900/40589 loss 0.2746 acc 0.944 lb 1.961 z 0.359 lr 1.82e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left [routing] mean entropy_norm 0.938 max_load(range) 0.052-0.233 step 32910/40589 loss 0.2741 acc 0.943 lb 1.959 z 0.355 lr 1.82e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left step 32920/40589 loss 0.2644 acc 0.946 lb 1.954 z 0.359 lr 1.81e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left [routing] mean entropy_norm 0.943 max_load(range) 0.056-0.345 step 32930/40589 loss 0.2567 acc 0.948 lb 1.960 z 0.429 lr 1.81e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left step 32940/40589 loss 0.2752 acc 0.944 lb 1.960 z 0.354 lr 1.81e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left step 32950/40589 loss 0.2771 acc 0.943 lb 1.958 z 0.342 lr 1.80e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left [routing] mean entropy_norm 0.962 max_load(range) 0.049-0.240 step 32960/40589 loss 0.2667 acc 0.945 lb 1.953 z 0.339 lr 1.80e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left step 32970/40589 loss 0.2568 acc 0.948 lb 1.960 z 0.439 lr 1.79e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left [routing] mean entropy_norm 0.936 max_load(range) 0.061-0.138 step 32980/40589 loss 0.2722 acc 0.944 lb 1.963 z 0.361 lr 1.79e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left step 32990/40589 loss 0.2773 acc 0.942 lb 1.957 z 0.324 lr 1.78e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left step 33000/40589 loss 0.2700 acc 0.945 lb 1.957 z 0.356 lr 1.78e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left [routing] mean entropy_norm 0.949 max_load(range) 0.058-0.161 [save] runs/pretrain_flagship_h100_4x8/step33000 step 33010/40589 loss 0.2553 acc 0.947 lb 1.960 z 0.427 lr 1.77e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left step 33020/40589 loss 0.2715 acc 0.944 lb 1.962 z 0.358 lr 1.77e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left [routing] mean entropy_norm 0.933 max_load(range) 0.059-0.149 step 33030/40589 loss 0.2736 acc 0.943 lb 1.958 z 0.328 lr 1.76e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left step 33040/40589 loss 0.2648 acc 0.946 lb 1.952 z 0.358 lr 1.76e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left step 33050/40589 loss 0.2533 acc 0.948 lb 1.960 z 0.447 lr 1.76e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left [routing] mean entropy_norm 0.895 max_load(range) 0.066-0.320 step 33060/40589 loss 0.2735 acc 0.944 lb 1.959 z 0.358 lr 1.75e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left step 33070/40589 loss 0.2771 acc 0.943 lb 1.955 z 0.346 lr 1.75e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left [routing] mean entropy_norm 0.954 max_load(range) 0.054-0.223 step 33080/40589 loss 0.2662 acc 0.945 lb 1.955 z 0.350 lr 1.74e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left step 33090/40589 loss 0.2586 acc 0.947 lb 1.961 z 0.400 lr 1.74e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left step 33100/40589 loss 0.2707 acc 0.944 lb 1.960 z 0.366 lr 1.73e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left [routing] mean entropy_norm 0.938 max_load(range) 0.054-0.234 step 33110/40589 loss 0.2721 acc 0.944 lb 1.958 z 0.332 lr 1.73e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 1.0d left step 33120/40589 loss 0.2638 acc 0.946 lb 1.956 z 0.291 lr 1.72e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 24.0h left [routing] mean entropy_norm 0.941 max_load(range) 0.053-0.358 step 33130/40589 loss 0.2575 acc 0.947 lb 1.961 z 0.422 lr 1.72e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 24.0h left step 33140/40589 loss 0.2746 acc 0.944 lb 1.963 z 0.359 lr 1.72e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.9h left step 33150/40589 loss 0.2755 acc 0.943 lb 1.959 z 0.285 lr 1.71e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.9h left [routing] mean entropy_norm 0.961 max_load(range) 0.047-0.236 step 33160/40589 loss 0.2647 acc 0.946 lb 1.953 z 0.343 lr 1.71e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.9h left step 33170/40589 loss 0.2561 acc 0.947 lb 1.961 z 0.392 lr 1.70e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.8h left [routing] mean entropy_norm 0.934 max_load(range) 0.061-0.131 step 33180/40589 loss 0.2741 acc 0.944 lb 1.966 z 0.362 lr 1.70e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.8h left step 33190/40589 loss 0.2749 acc 0.943 lb 1.957 z 0.351 lr 1.69e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.8h left step 33200/40589 loss 0.2631 acc 0.946 lb 1.954 z 0.344 lr 1.69e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.7h left [routing] mean entropy_norm 0.946 max_load(range) 0.064-0.154 step 33210/40589 loss 0.2568 acc 0.948 lb 1.962 z 0.432 lr 1.68e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.7h left step 33220/40589 loss 0.2737 acc 0.944 lb 1.963 z 0.370 lr 1.68e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.7h left [routing] mean entropy_norm 0.937 max_load(range) 0.063-0.188 step 33230/40589 loss 0.2766 acc 0.943 lb 1.962 z 0.356 lr 1.68e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.6h left step 33240/40589 loss 0.2662 acc 0.946 lb 1.954 z 0.349 lr 1.67e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.6h left step 33250/40589 loss 0.2528 acc 0.948 lb 1.961 z 0.438 lr 1.67e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.6h left [routing] mean entropy_norm 0.893 max_load(range) 0.070-0.320 step 33260/40589 loss 0.2731 acc 0.944 lb 1.962 z 0.364 lr 1.66e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.5h left step 33270/40589 loss 0.2780 acc 0.943 lb 1.963 z 0.336 lr 1.66e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.5h left [routing] mean entropy_norm 0.955 max_load(range) 0.049-0.209 step 33280/40589 loss 0.2657 acc 0.945 lb 1.953 z 0.343 lr 1.65e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.5h left step 33290/40589 loss 0.2562 acc 0.947 lb 1.963 z 0.416 lr 1.65e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.4h left step 33300/40589 loss 0.2715 acc 0.945 lb 1.965 z 0.299 lr 1.64e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.4h left [routing] mean entropy_norm 0.942 max_load(range) 0.056-0.221 step 33310/40589 loss 0.2719 acc 0.944 lb 1.963 z 0.344 lr 1.64e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.4h left step 33320/40589 loss 0.2643 acc 0.946 lb 1.953 z 0.339 lr 1.64e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.3h left [routing] mean entropy_norm 0.943 max_load(range) 0.057-0.360 step 33330/40589 loss 0.2557 acc 0.948 lb 1.959 z 0.420 lr 1.63e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.3h left step 33340/40589 loss 0.2701 acc 0.944 lb 1.960 z 0.358 lr 1.63e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.3h left step 33350/40589 loss 0.2759 acc 0.943 lb 1.956 z 0.339 lr 1.62e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.3h left [routing] mean entropy_norm 0.963 max_load(range) 0.048-0.223 step 33360/40589 loss 0.2651 acc 0.946 lb 1.954 z 0.352 lr 1.62e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.2h left step 33370/40589 loss 0.2546 acc 0.948 lb 1.964 z 0.437 lr 1.61e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.2h left [routing] mean entropy_norm 0.936 max_load(range) 0.062-0.136 step 33380/40589 loss 0.2717 acc 0.944 lb 1.961 z 0.358 lr 1.61e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.2h left step 33390/40589 loss 0.2743 acc 0.943 lb 1.958 z 0.284 lr 1.60e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.1h left step 33400/40589 loss 0.2668 acc 0.945 lb 1.954 z 0.335 lr 1.60e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.1h left [routing] mean entropy_norm 0.952 max_load(range) 0.057-0.181 step 33410/40589 loss 0.2585 acc 0.947 lb 1.962 z 0.424 lr 1.60e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.1h left step 33420/40589 loss 0.2712 acc 0.944 lb 1.962 z 0.328 lr 1.59e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.0h left [routing] mean entropy_norm 0.936 max_load(range) 0.063-0.164 step 33430/40589 loss 0.2711 acc 0.944 lb 1.957 z 0.340 lr 1.59e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.0h left step 33440/40589 loss 0.2683 acc 0.945 lb 1.953 z 0.333 lr 1.58e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 23.0h left step 33450/40589 loss 0.2547 acc 0.948 lb 1.962 z 0.395 lr 1.58e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.9h left [routing] mean entropy_norm 0.896 max_load(range) 0.064-0.316 step 33460/40589 loss 0.2733 acc 0.944 lb 1.960 z 0.345 lr 1.57e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.9h left step 33470/40589 loss 0.2744 acc 0.943 lb 1.958 z 0.353 lr 1.57e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.9h left [routing] mean entropy_norm 0.956 max_load(range) 0.054-0.199 step 33480/40589 loss 0.2677 acc 0.945 lb 1.953 z 0.318 lr 1.57e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.8h left step 33490/40589 loss 0.2550 acc 0.948 lb 1.959 z 0.430 lr 1.56e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.8h left step 33500/40589 loss 0.2712 acc 0.944 lb 1.961 z 0.358 lr 1.56e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.8h left [routing] mean entropy_norm 0.942 max_load(range) 0.053-0.215 step 33510/40589 loss 0.2707 acc 0.944 lb 1.962 z 0.350 lr 1.55e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.7h left step 33520/40589 loss 0.2656 acc 0.945 lb 1.951 z 0.345 lr 1.55e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.7h left [routing] mean entropy_norm 0.945 max_load(range) 0.056-0.357 step 33530/40589 loss 0.2556 acc 0.948 lb 1.961 z 0.398 lr 1.54e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.7h left step 33540/40589 loss 0.2733 acc 0.944 lb 1.961 z 0.342 lr 1.54e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.6h left step 33550/40589 loss 0.2720 acc 0.943 lb 1.961 z 0.329 lr 1.54e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.6h left [routing] mean entropy_norm 0.963 max_load(range) 0.049-0.231 step 33560/40589 loss 0.2623 acc 0.946 lb 1.958 z 0.348 lr 1.53e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.6h left step 33570/40589 loss 0.2578 acc 0.947 lb 1.963 z 0.421 lr 1.53e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.5h left [routing] mean entropy_norm 0.935 max_load(range) 0.060-0.135 step 33580/40589 loss 0.2713 acc 0.944 lb 1.965 z 0.346 lr 1.52e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.5h left step 33590/40589 loss 0.2727 acc 0.944 lb 1.962 z 0.335 lr 1.52e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.5h left step 33600/40589 loss 0.2651 acc 0.946 lb 1.958 z 0.355 lr 1.52e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.4h left [routing] mean entropy_norm 0.953 max_load(range) 0.052-0.195 step 33610/40589 loss 0.2579 acc 0.947 lb 1.962 z 0.426 lr 1.51e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.4h left step 33620/40589 loss 0.2743 acc 0.944 lb 1.961 z 0.358 lr 1.51e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.4h left [routing] mean entropy_norm 0.940 max_load(range) 0.061-0.158 step 33630/40589 loss 0.2745 acc 0.943 lb 1.960 z 0.342 lr 1.50e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.4h left step 33640/40589 loss 0.2650 acc 0.946 lb 1.955 z 0.336 lr 1.50e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.3h left step 33650/40589 loss 0.2544 acc 0.948 lb 1.963 z 0.415 lr 1.49e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.3h left [routing] mean entropy_norm 0.894 max_load(range) 0.068-0.322 step 33660/40589 loss 0.2721 acc 0.944 lb 1.962 z 0.370 lr 1.49e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.3h left step 33670/40589 loss 0.2750 acc 0.943 lb 1.961 z 0.355 lr 1.49e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.2h left [routing] mean entropy_norm 0.953 max_load(range) 0.052-0.226 step 33680/40589 loss 0.2664 acc 0.945 lb 1.957 z 0.357 lr 1.48e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.2h left step 33690/40589 loss 0.2569 acc 0.948 lb 1.963 z 0.411 lr 1.48e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.2h left step 33700/40589 loss 0.2716 acc 0.944 lb 1.964 z 0.354 lr 1.47e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.1h left [routing] mean entropy_norm 0.940 max_load(range) 0.054-0.230 step 33710/40589 loss 0.2756 acc 0.943 lb 1.960 z 0.343 lr 1.47e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.1h left step 33720/40589 loss 0.2621 acc 0.946 lb 1.954 z 0.335 lr 1.46e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.1h left [routing] mean entropy_norm 0.944 max_load(range) 0.055-0.371 step 33730/40589 loss 0.2581 acc 0.947 lb 1.964 z 0.395 lr 1.46e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.0h left step 33740/40589 loss 0.2734 acc 0.944 lb 1.965 z 0.357 lr 1.46e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.0h left step 33750/40589 loss 0.2753 acc 0.943 lb 1.958 z 0.325 lr 1.45e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 22.0h left [routing] mean entropy_norm 0.959 max_load(range) 0.052-0.187 step 33760/40589 loss 0.2701 acc 0.945 lb 1.953 z 0.340 lr 1.45e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.9h left step 33770/40589 loss 0.2549 acc 0.948 lb 1.962 z 0.405 lr 1.44e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.9h left [routing] mean entropy_norm 0.935 max_load(range) 0.065-0.129 step 33780/40589 loss 0.2717 acc 0.945 lb 1.965 z 0.361 lr 1.44e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.9h left step 33790/40589 loss 0.2751 acc 0.943 lb 1.959 z 0.331 lr 1.44e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.8h left step 33800/40589 loss 0.2670 acc 0.945 lb 1.957 z 0.345 lr 1.43e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.8h left [routing] mean entropy_norm 0.956 max_load(range) 0.060-0.164 step 33810/40589 loss 0.2562 acc 0.947 lb 1.963 z 0.427 lr 1.43e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.8h left step 33820/40589 loss 0.2734 acc 0.944 lb 1.965 z 0.355 lr 1.42e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.7h left [routing] mean entropy_norm 0.939 max_load(range) 0.059-0.180 step 33830/40589 loss 0.2744 acc 0.944 lb 1.959 z 0.345 lr 1.42e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.7h left step 33840/40589 loss 0.2657 acc 0.946 lb 1.952 z 0.343 lr 1.42e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.7h left step 33850/40589 loss 0.2581 acc 0.947 lb 1.963 z 0.429 lr 1.41e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.6h left [routing] mean entropy_norm 0.896 max_load(range) 0.066-0.316 step 33860/40589 loss 0.2707 acc 0.946 lb 1.963 z 0.358 lr 1.41e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.6h left step 33870/40589 loss 0.2748 acc 0.943 lb 1.960 z 0.341 lr 1.40e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.6h left [routing] mean entropy_norm 0.955 max_load(range) 0.050-0.215 step 33880/40589 loss 0.2652 acc 0.946 lb 1.956 z 0.338 lr 1.40e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.5h left step 33890/40589 loss 0.2547 acc 0.948 lb 1.960 z 0.405 lr 1.39e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.5h left step 33900/40589 loss 0.2691 acc 0.944 lb 1.961 z 0.366 lr 1.39e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.5h left [routing] mean entropy_norm 0.926 max_load(range) 0.055-0.257 step 33910/40589 loss 0.2709 acc 0.944 lb 1.957 z 0.335 lr 1.39e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.5h left step 33920/40589 loss 0.2636 acc 0.946 lb 1.951 z 0.339 lr 1.38e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.4h left [routing] mean entropy_norm 0.943 max_load(range) 0.055-0.373 step 33930/40589 loss 0.2553 acc 0.947 lb 1.961 z 0.388 lr 1.38e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.4h left step 33940/40589 loss 0.2685 acc 0.945 lb 1.960 z 0.334 lr 1.37e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.4h left step 33950/40589 loss 0.2718 acc 0.943 lb 1.958 z 0.353 lr 1.37e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.3h left [routing] mean entropy_norm 0.960 max_load(range) 0.050-0.220 step 33960/40589 loss 0.2671 acc 0.945 lb 1.958 z 0.346 lr 1.37e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.3h left step 33970/40589 loss 0.2560 acc 0.947 lb 1.956 z 0.405 lr 1.36e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.3h left [routing] mean entropy_norm 0.938 max_load(range) 0.060-0.142 step 33980/40589 loss 0.2684 acc 0.945 lb 1.961 z 0.347 lr 1.36e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.2h left step 33990/40589 loss 0.2714 acc 0.944 lb 1.958 z 0.354 lr 1.35e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.2h left step 34000/40589 loss 0.2681 acc 0.945 lb 1.953 z 0.356 lr 1.35e-05 | 1417 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.2h left [routing] mean entropy_norm 0.955 max_load(range) 0.052-0.167 [save] runs/pretrain_flagship_h100_4x8/step34000 step 34010/40589 loss 0.2548 acc 0.948 lb 1.961 z 0.355 lr 1.35e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.1h left step 34020/40589 loss 0.2711 acc 0.945 lb 1.963 z 0.366 lr 1.34e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.1h left [routing] mean entropy_norm 0.937 max_load(range) 0.064-0.157 step 34030/40589 loss 0.2720 acc 0.944 lb 1.960 z 0.332 lr 1.34e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.1h left step 34040/40589 loss 0.2608 acc 0.946 lb 1.957 z 0.327 lr 1.33e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.0h left step 34050/40589 loss 0.2536 acc 0.948 lb 1.962 z 0.428 lr 1.33e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.0h left [routing] mean entropy_norm 0.902 max_load(range) 0.067-0.313 step 34060/40589 loss 0.2761 acc 0.944 lb 1.963 z 0.353 lr 1.33e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 21.0h left step 34070/40589 loss 0.2758 acc 0.943 lb 1.956 z 0.332 lr 1.32e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.9h left [routing] mean entropy_norm 0.953 max_load(range) 0.049-0.222 step 34080/40589 loss 0.2665 acc 0.945 lb 1.954 z 0.336 lr 1.32e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.9h left step 34090/40589 loss 0.2559 acc 0.947 lb 1.959 z 0.420 lr 1.31e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.9h left step 34100/40589 loss 0.2689 acc 0.945 lb 1.961 z 0.343 lr 1.31e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.8h left [routing] mean entropy_norm 0.941 max_load(range) 0.052-0.231 step 34110/40589 loss 0.2729 acc 0.944 lb 1.960 z 0.326 lr 1.31e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.8h left step 34120/40589 loss 0.2644 acc 0.946 lb 1.954 z 0.353 lr 1.30e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.8h left [routing] mean entropy_norm 0.943 max_load(range) 0.055-0.356 step 34130/40589 loss 0.2556 acc 0.947 lb 1.962 z 0.414 lr 1.30e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.8h left step 34140/40589 loss 0.2687 acc 0.945 lb 1.963 z 0.366 lr 1.30e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.7h left step 34150/40589 loss 0.2754 acc 0.944 lb 1.959 z 0.346 lr 1.29e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.7h left [routing] mean entropy_norm 0.965 max_load(range) 0.050-0.214 step 34160/40589 loss 0.2662 acc 0.945 lb 1.951 z 0.346 lr 1.29e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.7h left step 34170/40589 loss 0.2561 acc 0.948 lb 1.960 z 0.365 lr 1.28e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.6h left [routing] mean entropy_norm 0.943 max_load(range) 0.058-0.150 step 34180/40589 loss 0.2702 acc 0.945 lb 1.964 z 0.345 lr 1.28e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.6h left step 34190/40589 loss 0.2722 acc 0.945 lb 1.963 z 0.345 lr 1.28e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.6h left step 34200/40589 loss 0.2665 acc 0.945 lb 1.957 z 0.336 lr 1.27e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.5h left [routing] mean entropy_norm 0.951 max_load(range) 0.061-0.159 step 34210/40589 loss 0.2541 acc 0.948 lb 1.965 z 0.412 lr 1.27e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.5h left step 34220/40589 loss 0.2715 acc 0.944 lb 1.961 z 0.365 lr 1.26e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.5h left [routing] mean entropy_norm 0.940 max_load(range) 0.057-0.180 step 34230/40589 loss 0.2741 acc 0.943 lb 1.963 z 0.342 lr 1.26e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.4h left step 34240/40589 loss 0.2639 acc 0.946 lb 1.953 z 0.338 lr 1.26e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.4h left step 34250/40589 loss 0.2546 acc 0.947 lb 1.960 z 0.400 lr 1.25e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.4h left [routing] mean entropy_norm 0.898 max_load(range) 0.066-0.319 step 34260/40589 loss 0.2722 acc 0.944 lb 1.966 z 0.372 lr 1.25e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.3h left step 34270/40589 loss 0.2706 acc 0.944 lb 1.960 z 0.332 lr 1.24e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.3h left [routing] mean entropy_norm 0.955 max_load(range) 0.054-0.208 step 34280/40589 loss 0.2686 acc 0.945 lb 1.956 z 0.343 lr 1.24e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.3h left step 34290/40589 loss 0.2554 acc 0.948 lb 1.967 z 0.407 lr 1.24e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.2h left step 34300/40589 loss 0.2697 acc 0.945 lb 1.961 z 0.335 lr 1.23e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.2h left [routing] mean entropy_norm 0.947 max_load(range) 0.052-0.204 step 34310/40589 loss 0.2737 acc 0.943 lb 1.962 z 0.344 lr 1.23e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.2h left step 34320/40589 loss 0.2658 acc 0.945 lb 1.956 z 0.339 lr 1.23e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.1h left [routing] mean entropy_norm 0.944 max_load(range) 0.054-0.369 step 34330/40589 loss 0.2529 acc 0.948 lb 1.962 z 0.405 lr 1.22e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.1h left step 34340/40589 loss 0.2701 acc 0.945 lb 1.964 z 0.342 lr 1.22e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.1h left step 34350/40589 loss 0.2730 acc 0.944 lb 1.961 z 0.343 lr 1.21e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.0h left [routing] mean entropy_norm 0.960 max_load(range) 0.049-0.192 step 34360/40589 loss 0.2667 acc 0.945 lb 1.957 z 0.355 lr 1.21e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.0h left step 34370/40589 loss 0.2556 acc 0.948 lb 1.959 z 0.409 lr 1.21e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 20.0h left [routing] mean entropy_norm 0.935 max_load(range) 0.060-0.127 step 34380/40589 loss 0.2715 acc 0.945 lb 1.962 z 0.353 lr 1.20e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.9h left step 34390/40589 loss 0.2710 acc 0.944 lb 1.959 z 0.330 lr 1.20e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.9h left step 34400/40589 loss 0.2673 acc 0.946 lb 1.957 z 0.351 lr 1.19e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.9h left [routing] mean entropy_norm 0.953 max_load(range) 0.059-0.172 step 34410/40589 loss 0.2570 acc 0.948 lb 1.962 z 0.388 lr 1.19e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.9h left step 34420/40589 loss 0.2716 acc 0.944 lb 1.961 z 0.357 lr 1.19e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.8h left [routing] mean entropy_norm 0.939 max_load(range) 0.065-0.161 step 34430/40589 loss 0.2744 acc 0.944 lb 1.962 z 0.336 lr 1.18e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.8h left step 34440/40589 loss 0.2624 acc 0.946 lb 1.957 z 0.341 lr 1.18e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.8h left step 34450/40589 loss 0.2529 acc 0.948 lb 1.962 z 0.422 lr 1.18e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.7h left [routing] mean entropy_norm 0.901 max_load(range) 0.068-0.312 step 34460/40589 loss 0.2701 acc 0.945 lb 1.962 z 0.360 lr 1.17e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.7h left step 34470/40589 loss 0.2724 acc 0.943 lb 1.958 z 0.331 lr 1.17e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.7h left [routing] mean entropy_norm 0.956 max_load(range) 0.050-0.208 step 34480/40589 loss 0.2625 acc 0.947 lb 1.954 z 0.344 lr 1.16e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.6h left step 34490/40589 loss 0.2563 acc 0.948 lb 1.961 z 0.404 lr 1.16e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.6h left step 34500/40589 loss 0.2669 acc 0.946 lb 1.961 z 0.354 lr 1.16e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.6h left [routing] mean entropy_norm 0.935 max_load(range) 0.053-0.242 step 34510/40589 loss 0.2737 acc 0.943 lb 1.959 z 0.345 lr 1.15e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.5h left step 34520/40589 loss 0.2668 acc 0.946 lb 1.957 z 0.346 lr 1.15e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.5h left [routing] mean entropy_norm 0.949 max_load(range) 0.053-0.345 step 34530/40589 loss 0.2534 acc 0.948 lb 1.965 z 0.352 lr 1.15e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.5h left step 34540/40589 loss 0.2696 acc 0.945 lb 1.964 z 0.375 lr 1.14e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.4h left step 34550/40589 loss 0.2725 acc 0.943 lb 1.960 z 0.346 lr 1.14e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.4h left [routing] mean entropy_norm 0.964 max_load(range) 0.047-0.246 step 34560/40589 loss 0.2677 acc 0.945 lb 1.953 z 0.348 lr 1.13e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.4h left step 34570/40589 loss 0.2596 acc 0.946 lb 1.962 z 0.431 lr 1.13e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.3h left [routing] mean entropy_norm 0.936 max_load(range) 0.060-0.127 step 34580/40589 loss 0.2696 acc 0.945 lb 1.958 z 0.364 lr 1.13e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.3h left step 34590/40589 loss 0.2737 acc 0.943 lb 1.958 z 0.343 lr 1.12e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.3h left step 34600/40589 loss 0.2673 acc 0.945 lb 1.956 z 0.348 lr 1.12e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.2h left [routing] mean entropy_norm 0.953 max_load(range) 0.061-0.179 step 34610/40589 loss 0.2529 acc 0.948 lb 1.961 z 0.423 lr 1.12e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.2h left step 34620/40589 loss 0.2677 acc 0.945 lb 1.961 z 0.355 lr 1.11e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.2h left [routing] mean entropy_norm 0.941 max_load(range) 0.057-0.177 step 34630/40589 loss 0.2766 acc 0.942 lb 1.957 z 0.341 lr 1.11e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.1h left step 34640/40589 loss 0.2646 acc 0.946 lb 1.955 z 0.327 lr 1.11e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.1h left step 34650/40589 loss 0.2578 acc 0.947 lb 1.961 z 0.410 lr 1.10e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.1h left [routing] mean entropy_norm 0.896 max_load(range) 0.072-0.322 step 34660/40589 loss 0.2698 acc 0.945 lb 1.962 z 0.350 lr 1.10e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.0h left step 34670/40589 loss 0.2713 acc 0.943 lb 1.959 z 0.338 lr 1.09e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.0h left [routing] mean entropy_norm 0.956 max_load(range) 0.047-0.196 step 34680/40589 loss 0.2667 acc 0.945 lb 1.956 z 0.350 lr 1.09e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.0h left step 34690/40589 loss 0.2562 acc 0.947 lb 1.961 z 0.385 lr 1.09e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 19.0h left step 34700/40589 loss 0.2729 acc 0.944 lb 1.958 z 0.332 lr 1.08e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.9h left [routing] mean entropy_norm 0.948 max_load(range) 0.049-0.199 step 34710/40589 loss 0.2690 acc 0.944 lb 1.961 z 0.339 lr 1.08e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.9h left step 34720/40589 loss 0.2621 acc 0.945 lb 1.957 z 0.332 lr 1.08e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.9h left [routing] mean entropy_norm 0.946 max_load(range) 0.052-0.354 step 34730/40589 loss 0.2537 acc 0.948 lb 1.961 z 0.421 lr 1.07e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.8h left step 34740/40589 loss 0.2748 acc 0.944 lb 1.962 z 0.349 lr 1.07e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.8h left step 34750/40589 loss 0.2711 acc 0.943 lb 1.959 z 0.334 lr 1.07e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.8h left [routing] mean entropy_norm 0.960 max_load(range) 0.050-0.203 step 34760/40589 loss 0.2673 acc 0.945 lb 1.953 z 0.349 lr 1.06e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.7h left step 34770/40589 loss 0.2567 acc 0.947 lb 1.961 z 0.405 lr 1.06e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.7h left [routing] mean entropy_norm 0.936 max_load(range) 0.060-0.135 step 34780/40589 loss 0.2708 acc 0.944 lb 1.959 z 0.357 lr 1.06e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.7h left step 34790/40589 loss 0.2702 acc 0.943 lb 1.959 z 0.339 lr 1.05e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.6h left step 34800/40589 loss 0.2682 acc 0.945 lb 1.955 z 0.343 lr 1.05e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.6h left [routing] mean entropy_norm 0.948 max_load(range) 0.065-0.163 step 34810/40589 loss 0.2587 acc 0.946 lb 1.958 z 0.375 lr 1.04e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.6h left step 34820/40589 loss 0.2709 acc 0.944 lb 1.961 z 0.365 lr 1.04e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.5h left [routing] mean entropy_norm 0.940 max_load(range) 0.055-0.191 step 34830/40589 loss 0.2763 acc 0.943 lb 1.960 z 0.348 lr 1.04e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.5h left step 34840/40589 loss 0.2683 acc 0.945 lb 1.954 z 0.334 lr 1.03e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.5h left step 34850/40589 loss 0.2576 acc 0.947 lb 1.963 z 0.414 lr 1.03e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.4h left [routing] mean entropy_norm 0.898 max_load(range) 0.065-0.314 step 34860/40589 loss 0.2708 acc 0.944 lb 1.964 z 0.357 lr 1.03e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.4h left step 34870/40589 loss 0.2734 acc 0.943 lb 1.957 z 0.323 lr 1.02e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.4h left [routing] mean entropy_norm 0.955 max_load(range) 0.048-0.219 step 34880/40589 loss 0.2623 acc 0.946 lb 1.952 z 0.332 lr 1.02e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.3h left step 34890/40589 loss 0.2574 acc 0.947 lb 1.961 z 0.416 lr 1.02e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.3h left step 34900/40589 loss 0.2711 acc 0.944 lb 1.961 z 0.345 lr 1.01e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.3h left [routing] mean entropy_norm 0.937 max_load(range) 0.056-0.234 step 34910/40589 loss 0.2715 acc 0.944 lb 1.958 z 0.342 lr 1.01e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.2h left step 34920/40589 loss 0.2661 acc 0.945 lb 1.952 z 0.349 lr 1.01e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.2h left [routing] mean entropy_norm 0.946 max_load(range) 0.056-0.349 step 34930/40589 loss 0.2574 acc 0.947 lb 1.963 z 0.411 lr 1.00e-05 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.2h left step 34940/40589 loss 0.2716 acc 0.945 lb 1.960 z 0.348 lr 9.99e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.1h left step 34950/40589 loss 0.2720 acc 0.944 lb 1.958 z 0.333 lr 9.95e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.1h left [routing] mean entropy_norm 0.962 max_load(range) 0.048-0.242 step 34960/40589 loss 0.2670 acc 0.945 lb 1.955 z 0.345 lr 9.92e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.1h left step 34970/40589 loss 0.2555 acc 0.948 lb 1.963 z 0.426 lr 9.88e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.1h left [routing] mean entropy_norm 0.934 max_load(range) 0.058-0.130 step 34980/40589 loss 0.2709 acc 0.944 lb 1.964 z 0.354 lr 9.85e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.0h left step 34990/40589 loss 0.2726 acc 0.943 lb 1.960 z 0.319 lr 9.81e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.0h left step 35000/40589 loss 0.2678 acc 0.945 lb 1.954 z 0.352 lr 9.78e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 18.0h left [routing] mean entropy_norm 0.946 max_load(range) 0.063-0.150 [save] runs/pretrain_flagship_h100_4x8/step35000 step 35010/40589 loss 0.2563 acc 0.947 lb 1.964 z 0.426 lr 9.75e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.9h left step 35020/40589 loss 0.2688 acc 0.945 lb 1.963 z 0.352 lr 9.71e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.9h left [routing] mean entropy_norm 0.941 max_load(range) 0.053-0.167 step 35030/40589 loss 0.2691 acc 0.944 lb 1.959 z 0.333 lr 9.68e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.9h left step 35040/40589 loss 0.2667 acc 0.945 lb 1.953 z 0.344 lr 9.64e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.8h left step 35050/40589 loss 0.2533 acc 0.948 lb 1.963 z 0.418 lr 9.61e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.8h left [routing] mean entropy_norm 0.908 max_load(range) 0.062-0.299 step 35060/40589 loss 0.2701 acc 0.944 lb 1.965 z 0.363 lr 9.57e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.8h left step 35070/40589 loss 0.2671 acc 0.945 lb 1.962 z 0.339 lr 9.54e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.7h left [routing] mean entropy_norm 0.956 max_load(range) 0.050-0.205 step 35080/40589 loss 0.2651 acc 0.946 lb 1.959 z 0.338 lr 9.51e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.7h left step 35090/40589 loss 0.2539 acc 0.948 lb 1.962 z 0.406 lr 9.47e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.7h left step 35100/40589 loss 0.2705 acc 0.944 lb 1.964 z 0.348 lr 9.44e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.6h left [routing] mean entropy_norm 0.948 max_load(range) 0.049-0.197 step 35110/40589 loss 0.2731 acc 0.944 lb 1.961 z 0.341 lr 9.41e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.6h left step 35120/40589 loss 0.2685 acc 0.945 lb 1.957 z 0.336 lr 9.37e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.6h left [routing] mean entropy_norm 0.942 max_load(range) 0.050-0.359 step 35130/40589 loss 0.2569 acc 0.947 lb 1.963 z 0.396 lr 9.34e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.5h left step 35140/40589 loss 0.2676 acc 0.945 lb 1.963 z 0.359 lr 9.30e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.5h left step 35150/40589 loss 0.2715 acc 0.944 lb 1.960 z 0.336 lr 9.27e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.5h left [routing] mean entropy_norm 0.961 max_load(range) 0.050-0.230 step 35160/40589 loss 0.2639 acc 0.946 lb 1.954 z 0.334 lr 9.24e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.4h left step 35170/40589 loss 0.2528 acc 0.949 lb 1.962 z 0.425 lr 9.20e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.4h left [routing] mean entropy_norm 0.937 max_load(range) 0.059-0.140 step 35180/40589 loss 0.2684 acc 0.945 lb 1.962 z 0.360 lr 9.17e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.4h left step 35190/40589 loss 0.2711 acc 0.944 lb 1.960 z 0.326 lr 9.14e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.3h left step 35200/40589 loss 0.2696 acc 0.945 lb 1.955 z 0.354 lr 9.10e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.3h left [routing] mean entropy_norm 0.954 max_load(range) 0.061-0.179 step 35210/40589 loss 0.2544 acc 0.948 lb 1.964 z 0.416 lr 9.07e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.3h left step 35220/40589 loss 0.2710 acc 0.944 lb 1.964 z 0.350 lr 9.04e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.3h left [routing] mean entropy_norm 0.940 max_load(range) 0.052-0.184 step 35230/40589 loss 0.2684 acc 0.944 lb 1.962 z 0.342 lr 9.00e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.2h left step 35240/40589 loss 0.2637 acc 0.946 lb 1.957 z 0.336 lr 8.97e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.2h left step 35250/40589 loss 0.2569 acc 0.948 lb 1.964 z 0.328 lr 8.94e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.2h left [routing] mean entropy_norm 0.903 max_load(range) 0.067-0.305 step 35260/40589 loss 0.2682 acc 0.945 lb 1.962 z 0.360 lr 8.90e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.1h left step 35270/40589 loss 0.2682 acc 0.945 lb 1.959 z 0.343 lr 8.87e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.1h left [routing] mean entropy_norm 0.957 max_load(range) 0.051-0.191 step 35280/40589 loss 0.2611 acc 0.945 lb 1.956 z 0.333 lr 8.84e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.1h left step 35290/40589 loss 0.2563 acc 0.947 lb 1.963 z 0.389 lr 8.81e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.0h left step 35300/40589 loss 0.2674 acc 0.945 lb 1.962 z 0.357 lr 8.77e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.0h left [routing] mean entropy_norm 0.943 max_load(range) 0.051-0.221 step 35310/40589 loss 0.2715 acc 0.944 lb 1.961 z 0.341 lr 8.74e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 17.0h left step 35320/40589 loss 0.2633 acc 0.946 lb 1.954 z 0.339 lr 8.71e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.9h left [routing] mean entropy_norm 0.942 max_load(range) 0.052-0.367 step 35330/40589 loss 0.2578 acc 0.947 lb 1.963 z 0.396 lr 8.68e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.9h left step 35340/40589 loss 0.2668 acc 0.945 lb 1.962 z 0.355 lr 8.64e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.9h left step 35350/40589 loss 0.2684 acc 0.944 lb 1.957 z 0.281 lr 8.61e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.8h left [routing] mean entropy_norm 0.964 max_load(range) 0.046-0.252 step 35360/40589 loss 0.2635 acc 0.946 lb 1.952 z 0.316 lr 8.58e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.8h left step 35370/40589 loss 0.2554 acc 0.948 lb 1.960 z 0.400 lr 8.55e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.8h left [routing] mean entropy_norm 0.935 max_load(range) 0.059-0.124 step 35380/40589 loss 0.2734 acc 0.944 lb 1.963 z 0.340 lr 8.51e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.7h left step 35390/40589 loss 0.2710 acc 0.944 lb 1.962 z 0.346 lr 8.48e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.7h left step 35400/40589 loss 0.2618 acc 0.946 lb 1.955 z 0.329 lr 8.45e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.7h left [routing] mean entropy_norm 0.955 max_load(range) 0.060-0.165 step 35410/40589 loss 0.2544 acc 0.948 lb 1.962 z 0.405 lr 8.42e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.6h left step 35420/40589 loss 0.2703 acc 0.944 lb 1.965 z 0.358 lr 8.39e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.6h left [routing] mean entropy_norm 0.940 max_load(range) 0.056-0.182 step 35430/40589 loss 0.2705 acc 0.944 lb 1.961 z 0.346 lr 8.35e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.6h left step 35440/40589 loss 0.2645 acc 0.945 lb 1.955 z 0.339 lr 8.32e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.5h left step 35450/40589 loss 0.2523 acc 0.948 lb 1.964 z 0.398 lr 8.29e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.5h left [routing] mean entropy_norm 0.910 max_load(range) 0.062-0.294 step 35460/40589 loss 0.2716 acc 0.944 lb 1.963 z 0.344 lr 8.26e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.5h left step 35470/40589 loss 0.2694 acc 0.944 lb 1.960 z 0.314 lr 8.23e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.4h left [routing] mean entropy_norm 0.957 max_load(range) 0.048-0.203 step 35480/40589 loss 0.2667 acc 0.945 lb 1.953 z 0.335 lr 8.19e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.4h left step 35490/40589 loss 0.2546 acc 0.947 lb 1.960 z 0.406 lr 8.16e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.4h left step 35500/40589 loss 0.2703 acc 0.945 lb 1.962 z 0.331 lr 8.13e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.4h left [routing] mean entropy_norm 0.942 max_load(range) 0.055-0.218 step 35510/40589 loss 0.2706 acc 0.945 lb 1.958 z 0.338 lr 8.10e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.3h left step 35520/40589 loss 0.2616 acc 0.947 lb 1.955 z 0.343 lr 8.07e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.3h left [routing] mean entropy_norm 0.946 max_load(range) 0.051-0.349 step 35530/40589 loss 0.2535 acc 0.948 lb 1.961 z 0.388 lr 8.04e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.3h left step 35540/40589 loss 0.2733 acc 0.944 lb 1.960 z 0.355 lr 8.01e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.2h left step 35550/40589 loss 0.2730 acc 0.943 lb 1.958 z 0.347 lr 7.97e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.2h left [routing] mean entropy_norm 0.965 max_load(range) 0.047-0.214 step 35560/40589 loss 0.2635 acc 0.946 lb 1.955 z 0.343 lr 7.94e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.2h left step 35570/40589 loss 0.2552 acc 0.948 lb 1.961 z 0.415 lr 7.91e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.1h left [routing] mean entropy_norm 0.938 max_load(range) 0.058-0.132 step 35580/40589 loss 0.2697 acc 0.945 lb 1.963 z 0.350 lr 7.88e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.1h left step 35590/40589 loss 0.2690 acc 0.944 lb 1.959 z 0.340 lr 7.85e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.1h left step 35600/40589 loss 0.2675 acc 0.945 lb 1.956 z 0.338 lr 7.82e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.0h left [routing] mean entropy_norm 0.954 max_load(range) 0.062-0.174 step 35610/40589 loss 0.2560 acc 0.948 lb 1.964 z 0.427 lr 7.79e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.0h left step 35620/40589 loss 0.2706 acc 0.944 lb 1.964 z 0.356 lr 7.76e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 16.0h left [routing] mean entropy_norm 0.940 max_load(range) 0.062-0.182 step 35630/40589 loss 0.2720 acc 0.944 lb 1.958 z 0.341 lr 7.73e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.9h left step 35640/40589 loss 0.2633 acc 0.946 lb 1.955 z 0.324 lr 7.70e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.9h left step 35650/40589 loss 0.2573 acc 0.947 lb 1.962 z 0.403 lr 7.67e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.9h left [routing] mean entropy_norm 0.886 max_load(range) 0.072-0.337 step 35660/40589 loss 0.2695 acc 0.945 lb 1.964 z 0.355 lr 7.63e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.8h left step 35670/40589 loss 0.2699 acc 0.944 lb 1.960 z 0.331 lr 7.60e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.8h left [routing] mean entropy_norm 0.955 max_load(range) 0.049-0.218 step 35680/40589 loss 0.2613 acc 0.946 lb 1.953 z 0.327 lr 7.57e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.8h left step 35690/40589 loss 0.2527 acc 0.948 lb 1.965 z 0.418 lr 7.54e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.7h left step 35700/40589 loss 0.2690 acc 0.945 lb 1.963 z 0.350 lr 7.51e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.7h left [routing] mean entropy_norm 0.935 max_load(range) 0.055-0.231 step 35710/40589 loss 0.2687 acc 0.945 lb 1.958 z 0.343 lr 7.48e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.7h left step 35720/40589 loss 0.2658 acc 0.945 lb 1.954 z 0.341 lr 7.45e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.6h left [routing] mean entropy_norm 0.943 max_load(range) 0.052-0.366 step 35730/40589 loss 0.2521 acc 0.949 lb 1.963 z 0.417 lr 7.42e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.6h left step 35740/40589 loss 0.2685 acc 0.945 lb 1.962 z 0.343 lr 7.39e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.6h left step 35750/40589 loss 0.2687 acc 0.944 lb 1.961 z 0.349 lr 7.36e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.6h left [routing] mean entropy_norm 0.963 max_load(range) 0.047-0.245 step 35760/40589 loss 0.2627 acc 0.946 lb 1.956 z 0.347 lr 7.33e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.5h left step 35770/40589 loss 0.2537 acc 0.948 lb 1.962 z 0.396 lr 7.30e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.5h left [routing] mean entropy_norm 0.935 max_load(range) 0.059-0.131 step 35780/40589 loss 0.2697 acc 0.945 lb 1.964 z 0.336 lr 7.27e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.5h left step 35790/40589 loss 0.2714 acc 0.944 lb 1.958 z 0.326 lr 7.24e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.4h left step 35800/40589 loss 0.2614 acc 0.947 lb 1.954 z 0.333 lr 7.21e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.4h left [routing] mean entropy_norm 0.957 max_load(range) 0.056-0.165 step 35810/40589 loss 0.2559 acc 0.948 lb 1.960 z 0.405 lr 7.18e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.4h left step 35820/40589 loss 0.2712 acc 0.945 lb 1.964 z 0.334 lr 7.15e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.3h left [routing] mean entropy_norm 0.941 max_load(range) 0.062-0.158 step 35830/40589 loss 0.2713 acc 0.944 lb 1.962 z 0.329 lr 7.12e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.3h left step 35840/40589 loss 0.2667 acc 0.945 lb 1.954 z 0.339 lr 7.09e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.3h left step 35850/40589 loss 0.2516 acc 0.949 lb 1.964 z 0.408 lr 7.06e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.2h left [routing] mean entropy_norm 0.898 max_load(range) 0.071-0.313 step 35860/40589 loss 0.2699 acc 0.945 lb 1.966 z 0.340 lr 7.04e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.2h left step 35870/40589 loss 0.2729 acc 0.944 lb 1.960 z 0.325 lr 7.01e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.2h left [routing] mean entropy_norm 0.954 max_load(range) 0.052-0.227 step 35880/40589 loss 0.2648 acc 0.946 lb 1.956 z 0.347 lr 6.98e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.1h left step 35890/40589 loss 0.2558 acc 0.948 lb 1.961 z 0.401 lr 6.95e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.1h left step 35900/40589 loss 0.2680 acc 0.945 lb 1.962 z 0.342 lr 6.92e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.1h left [routing] mean entropy_norm 0.948 max_load(range) 0.050-0.206 step 35910/40589 loss 0.2666 acc 0.945 lb 1.960 z 0.339 lr 6.89e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.0h left step 35920/40589 loss 0.2702 acc 0.944 lb 1.958 z 0.337 lr 6.86e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.0h left [routing] mean entropy_norm 0.942 max_load(range) 0.054-0.363 step 35930/40589 loss 0.2576 acc 0.947 lb 1.961 z 0.417 lr 6.83e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 15.0h left step 35940/40589 loss 0.2693 acc 0.945 lb 1.962 z 0.339 lr 6.80e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.9h left step 35950/40589 loss 0.2717 acc 0.943 lb 1.958 z 0.330 lr 6.77e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.9h left [routing] mean entropy_norm 0.962 max_load(range) 0.045-0.216 step 35960/40589 loss 0.2638 acc 0.946 lb 1.955 z 0.331 lr 6.74e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.9h left step 35970/40589 loss 0.2541 acc 0.948 lb 1.961 z 0.411 lr 6.72e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.8h left [routing] mean entropy_norm 0.941 max_load(range) 0.061-0.146 step 35980/40589 loss 0.2700 acc 0.945 lb 1.962 z 0.351 lr 6.69e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.8h left step 35990/40589 loss 0.2702 acc 0.944 lb 1.959 z 0.330 lr 6.66e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.8h left step 36000/40589 loss 0.2646 acc 0.945 lb 1.957 z 0.343 lr 6.63e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.7h left [routing] mean entropy_norm 0.953 max_load(range) 0.061-0.174 [save] runs/pretrain_flagship_h100_4x8/step36000 step 36010/40589 loss 0.2508 acc 0.949 lb 1.961 z 0.404 lr 6.60e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.7h left step 36020/40589 loss 0.2710 acc 0.945 lb 1.963 z 0.355 lr 6.57e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.7h left [routing] mean entropy_norm 0.938 max_load(range) 0.057-0.201 step 36030/40589 loss 0.2657 acc 0.945 lb 1.961 z 0.345 lr 6.54e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.7h left step 36040/40589 loss 0.2630 acc 0.946 lb 1.956 z 0.333 lr 6.52e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.6h left step 36050/40589 loss 0.2525 acc 0.948 lb 1.963 z 0.403 lr 6.49e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.6h left [routing] mean entropy_norm 0.901 max_load(range) 0.069-0.313 step 36060/40589 loss 0.2683 acc 0.945 lb 1.964 z 0.356 lr 6.46e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.6h left step 36070/40589 loss 0.2697 acc 0.944 lb 1.961 z 0.335 lr 6.43e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.5h left [routing] mean entropy_norm 0.955 max_load(range) 0.049-0.216 step 36080/40589 loss 0.2613 acc 0.947 lb 1.955 z 0.326 lr 6.40e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.5h left step 36090/40589 loss 0.2509 acc 0.949 lb 1.966 z 0.392 lr 6.37e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.5h left step 36100/40589 loss 0.2680 acc 0.945 lb 1.963 z 0.345 lr 6.35e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.4h left [routing] mean entropy_norm 0.935 max_load(range) 0.057-0.241 step 36110/40589 loss 0.2672 acc 0.945 lb 1.962 z 0.313 lr 6.32e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.4h left step 36120/40589 loss 0.2646 acc 0.946 lb 1.956 z 0.333 lr 6.29e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.4h left [routing] mean entropy_norm 0.945 max_load(range) 0.051-0.356 step 36130/40589 loss 0.2543 acc 0.948 lb 1.963 z 0.412 lr 6.26e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.3h left step 36140/40589 loss 0.2683 acc 0.945 lb 1.964 z 0.348 lr 6.24e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.3h left step 36150/40589 loss 0.2689 acc 0.944 lb 1.960 z 0.318 lr 6.21e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.3h left [routing] mean entropy_norm 0.961 max_load(range) 0.048-0.200 step 36160/40589 loss 0.2637 acc 0.946 lb 1.953 z 0.323 lr 6.18e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.2h left step 36170/40589 loss 0.2546 acc 0.948 lb 1.959 z 0.396 lr 6.15e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.2h left [routing] mean entropy_norm 0.932 max_load(range) 0.061-0.126 step 36180/40589 loss 0.2647 acc 0.947 lb 1.961 z 0.347 lr 6.12e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.2h left step 36190/40589 loss 0.2683 acc 0.945 lb 1.961 z 0.335 lr 6.10e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.1h left step 36200/40589 loss 0.2643 acc 0.946 lb 1.956 z 0.329 lr 6.07e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.1h left [routing] mean entropy_norm 0.953 max_load(range) 0.057-0.166 step 36210/40589 loss 0.2538 acc 0.948 lb 1.962 z 0.430 lr 6.04e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.1h left step 36220/40589 loss 0.2679 acc 0.945 lb 1.963 z 0.359 lr 6.02e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.0h left [routing] mean entropy_norm 0.937 max_load(range) 0.056-0.192 step 36230/40589 loss 0.2682 acc 0.944 lb 1.960 z 0.342 lr 5.99e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.0h left step 36240/40589 loss 0.2627 acc 0.946 lb 1.955 z 0.321 lr 5.96e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 14.0h left step 36250/40589 loss 0.2569 acc 0.947 lb 1.964 z 0.420 lr 5.93e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.9h left [routing] mean entropy_norm 0.891 max_load(range) 0.070-0.328 step 36260/40589 loss 0.2677 acc 0.945 lb 1.961 z 0.325 lr 5.91e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.9h left step 36270/40589 loss 0.2677 acc 0.945 lb 1.961 z 0.344 lr 5.88e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.9h left [routing] mean entropy_norm 0.954 max_load(range) 0.049-0.227 step 36280/40589 loss 0.2631 acc 0.947 lb 1.957 z 0.333 lr 5.85e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.8h left step 36290/40589 loss 0.2549 acc 0.948 lb 1.962 z 0.392 lr 5.83e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.8h left step 36300/40589 loss 0.2710 acc 0.945 lb 1.964 z 0.340 lr 5.80e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.8h left [routing] mean entropy_norm 0.950 max_load(range) 0.047-0.199 step 36310/40589 loss 0.2667 acc 0.945 lb 1.958 z 0.321 lr 5.77e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.8h left step 36320/40589 loss 0.2623 acc 0.946 lb 1.955 z 0.347 lr 5.75e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.7h left [routing] mean entropy_norm 0.945 max_load(range) 0.054-0.362 step 36330/40589 loss 0.2541 acc 0.948 lb 1.964 z 0.409 lr 5.72e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.7h left step 36340/40589 loss 0.2672 acc 0.946 lb 1.962 z 0.348 lr 5.69e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.7h left step 36350/40589 loss 0.2709 acc 0.944 lb 1.962 z 0.333 lr 5.67e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.6h left [routing] mean entropy_norm 0.959 max_load(range) 0.049-0.211 step 36360/40589 loss 0.2613 acc 0.947 lb 1.956 z 0.336 lr 5.64e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.6h left step 36370/40589 loss 0.2542 acc 0.948 lb 1.963 z 0.385 lr 5.61e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.6h left [routing] mean entropy_norm 0.939 max_load(range) 0.059-0.141 step 36380/40589 loss 0.2688 acc 0.945 lb 1.963 z 0.338 lr 5.59e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.5h left step 36390/40589 loss 0.2669 acc 0.945 lb 1.962 z 0.322 lr 5.56e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.5h left step 36400/40589 loss 0.2646 acc 0.946 lb 1.957 z 0.325 lr 5.53e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.5h left [routing] mean entropy_norm 0.952 max_load(range) 0.058-0.154 step 36410/40589 loss 0.2561 acc 0.947 lb 1.963 z 0.413 lr 5.51e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.4h left step 36420/40589 loss 0.2668 acc 0.946 lb 1.965 z 0.347 lr 5.48e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.4h left [routing] mean entropy_norm 0.941 max_load(range) 0.056-0.168 step 36430/40589 loss 0.2716 acc 0.944 lb 1.962 z 0.330 lr 5.46e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.4h left step 36440/40589 loss 0.2607 acc 0.946 lb 1.957 z 0.332 lr 5.43e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.3h left step 36450/40589 loss 0.2527 acc 0.948 lb 1.963 z 0.404 lr 5.40e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.3h left [routing] mean entropy_norm 0.895 max_load(range) 0.070-0.323 step 36460/40589 loss 0.2670 acc 0.946 lb 1.965 z 0.357 lr 5.38e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.3h left step 36470/40589 loss 0.2692 acc 0.944 lb 1.962 z 0.330 lr 5.35e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.2h left [routing] mean entropy_norm 0.959 max_load(range) 0.053-0.219 step 36480/40589 loss 0.2620 acc 0.946 lb 1.955 z 0.329 lr 5.33e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.2h left step 36490/40589 loss 0.2511 acc 0.949 lb 1.962 z 0.418 lr 5.30e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.2h left step 36500/40589 loss 0.2727 acc 0.944 lb 1.966 z 0.363 lr 5.28e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.1h left [routing] mean entropy_norm 0.927 max_load(range) 0.053-0.252 step 36510/40589 loss 0.2683 acc 0.944 lb 1.960 z 0.316 lr 5.25e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.1h left step 36520/40589 loss 0.2611 acc 0.946 lb 1.953 z 0.330 lr 5.22e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.1h left [routing] mean entropy_norm 0.945 max_load(range) 0.054-0.352 step 36530/40589 loss 0.2542 acc 0.948 lb 1.962 z 0.400 lr 5.20e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.0h left step 36540/40589 loss 0.2656 acc 0.946 lb 1.963 z 0.333 lr 5.17e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.0h left step 36550/40589 loss 0.2704 acc 0.944 lb 1.958 z 0.328 lr 5.15e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 13.0h left [routing] mean entropy_norm 0.962 max_load(range) 0.051-0.212 step 36560/40589 loss 0.2583 acc 0.947 lb 1.956 z 0.341 lr 5.12e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.9h left step 36570/40589 loss 0.2524 acc 0.948 lb 1.962 z 0.411 lr 5.10e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.9h left [routing] mean entropy_norm 0.937 max_load(range) 0.061-0.140 step 36580/40589 loss 0.2676 acc 0.945 lb 1.964 z 0.344 lr 5.07e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.9h left step 36590/40589 loss 0.2708 acc 0.944 lb 1.961 z 0.331 lr 5.05e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.9h left step 36600/40589 loss 0.2622 acc 0.946 lb 1.957 z 0.343 lr 5.02e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.8h left [routing] mean entropy_norm 0.958 max_load(range) 0.057-0.171 step 36610/40589 loss 0.2564 acc 0.948 lb 1.963 z 0.409 lr 5.00e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.8h left step 36620/40589 loss 0.2670 acc 0.945 lb 1.962 z 0.347 lr 4.97e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.8h left [routing] mean entropy_norm 0.937 max_load(range) 0.061-0.134 step 36630/40589 loss 0.2640 acc 0.945 lb 1.961 z 0.315 lr 4.95e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.7h left step 36640/40589 loss 0.2632 acc 0.946 lb 1.957 z 0.328 lr 4.92e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.7h left step 36650/40589 loss 0.2510 acc 0.949 lb 1.961 z 0.399 lr 4.90e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.7h left [routing] mean entropy_norm 0.894 max_load(range) 0.070-0.317 step 36660/40589 loss 0.2702 acc 0.945 lb 1.963 z 0.344 lr 4.87e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.6h left step 36670/40589 loss 0.2689 acc 0.945 lb 1.961 z 0.362 lr 4.85e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.6h left [routing] mean entropy_norm 0.959 max_load(range) 0.048-0.200 step 36680/40589 loss 0.2647 acc 0.946 lb 1.957 z 0.334 lr 4.83e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.6h left step 36690/40589 loss 0.2524 acc 0.949 lb 1.963 z 0.417 lr 4.80e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.5h left step 36700/40589 loss 0.2711 acc 0.945 lb 1.965 z 0.340 lr 4.78e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.5h left [routing] mean entropy_norm 0.941 max_load(range) 0.052-0.235 step 36710/40589 loss 0.2697 acc 0.944 lb 1.958 z 0.312 lr 4.75e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.5h left step 36720/40589 loss 0.2633 acc 0.945 lb 1.955 z 0.349 lr 4.73e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.4h left [routing] mean entropy_norm 0.945 max_load(range) 0.055-0.348 step 36730/40589 loss 0.2536 acc 0.948 lb 1.962 z 0.390 lr 4.70e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.4h left step 36740/40589 loss 0.2679 acc 0.945 lb 1.965 z 0.328 lr 4.68e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.4h left step 36750/40589 loss 0.2688 acc 0.944 lb 1.958 z 0.325 lr 4.66e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.3h left [routing] mean entropy_norm 0.963 max_load(range) 0.050-0.224 step 36760/40589 loss 0.2650 acc 0.945 lb 1.956 z 0.336 lr 4.63e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.3h left step 36770/40589 loss 0.2509 acc 0.949 lb 1.964 z 0.422 lr 4.61e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.3h left [routing] mean entropy_norm 0.937 max_load(range) 0.058-0.128 step 36780/40589 loss 0.2686 acc 0.945 lb 1.962 z 0.348 lr 4.58e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.2h left step 36790/40589 loss 0.2708 acc 0.945 lb 1.960 z 0.330 lr 4.56e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.2h left step 36800/40589 loss 0.2636 acc 0.946 lb 1.956 z 0.337 lr 4.54e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.2h left [routing] mean entropy_norm 0.957 max_load(range) 0.057-0.182 step 36810/40589 loss 0.2539 acc 0.948 lb 1.964 z 0.391 lr 4.51e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.1h left step 36820/40589 loss 0.2657 acc 0.945 lb 1.963 z 0.348 lr 4.49e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.1h left [routing] mean entropy_norm 0.944 max_load(range) 0.052-0.175 step 36830/40589 loss 0.2688 acc 0.945 lb 1.961 z 0.336 lr 4.46e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.1h left step 36840/40589 loss 0.2623 acc 0.947 lb 1.958 z 0.339 lr 4.44e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.0h left step 36850/40589 loss 0.2531 acc 0.948 lb 1.965 z 0.422 lr 4.42e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.0h left [routing] mean entropy_norm 0.908 max_load(range) 0.068-0.302 step 36860/40589 loss 0.2656 acc 0.946 lb 1.964 z 0.345 lr 4.39e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.0h left step 36870/40589 loss 0.2658 acc 0.945 lb 1.961 z 0.341 lr 4.37e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 12.0h left [routing] mean entropy_norm 0.958 max_load(range) 0.052-0.196 step 36880/40589 loss 0.2656 acc 0.945 lb 1.957 z 0.333 lr 4.35e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.9h left step 36890/40589 loss 0.2553 acc 0.948 lb 1.961 z 0.412 lr 4.32e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.9h left step 36900/40589 loss 0.2669 acc 0.945 lb 1.962 z 0.328 lr 4.30e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.9h left [routing] mean entropy_norm 0.950 max_load(range) 0.047-0.195 step 36910/40589 loss 0.2697 acc 0.944 lb 1.960 z 0.329 lr 4.28e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.8h left step 36920/40589 loss 0.2672 acc 0.945 lb 1.955 z 0.339 lr 4.25e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.8h left [routing] mean entropy_norm 0.950 max_load(range) 0.051-0.340 step 36930/40589 loss 0.2501 acc 0.949 lb 1.961 z 0.389 lr 4.23e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.8h left step 36940/40589 loss 0.2695 acc 0.944 lb 1.962 z 0.358 lr 4.21e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.7h left step 36950/40589 loss 0.2624 acc 0.946 lb 1.960 z 0.332 lr 4.19e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.7h left [routing] mean entropy_norm 0.961 max_load(range) 0.050-0.256 step 36960/40589 loss 0.2598 acc 0.947 lb 1.954 z 0.333 lr 4.16e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.7h left step 36970/40589 loss 0.2524 acc 0.948 lb 1.963 z 0.412 lr 4.14e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.6h left [routing] mean entropy_norm 0.941 max_load(range) 0.058-0.142 step 36980/40589 loss 0.2665 acc 0.946 lb 1.962 z 0.334 lr 4.12e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.6h left step 36990/40589 loss 0.2656 acc 0.946 lb 1.960 z 0.337 lr 4.10e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.6h left step 37000/40589 loss 0.2611 acc 0.947 lb 1.956 z 0.337 lr 4.07e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.5h left [routing] mean entropy_norm 0.957 max_load(range) 0.061-0.161 [save] runs/pretrain_flagship_h100_4x8/step37000 step 37010/40589 loss 0.2531 acc 0.948 lb 1.963 z 0.397 lr 4.05e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.5h left step 37020/40589 loss 0.2655 acc 0.946 lb 1.962 z 0.341 lr 4.03e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.5h left [routing] mean entropy_norm 0.943 max_load(range) 0.058-0.170 step 37030/40589 loss 0.2673 acc 0.945 lb 1.958 z 0.332 lr 4.01e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.4h left step 37040/40589 loss 0.2634 acc 0.946 lb 1.955 z 0.321 lr 3.98e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.4h left step 37050/40589 loss 0.2558 acc 0.948 lb 1.963 z 0.407 lr 3.96e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.4h left [routing] mean entropy_norm 0.899 max_load(range) 0.068-0.315 step 37060/40589 loss 0.2685 acc 0.945 lb 1.965 z 0.355 lr 3.94e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.3h left step 37070/40589 loss 0.2687 acc 0.945 lb 1.961 z 0.320 lr 3.92e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.3h left [routing] mean entropy_norm 0.959 max_load(range) 0.053-0.223 step 37080/40589 loss 0.2625 acc 0.946 lb 1.957 z 0.334 lr 3.89e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.3h left step 37090/40589 loss 0.2542 acc 0.948 lb 1.964 z 0.405 lr 3.87e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.2h left step 37100/40589 loss 0.2654 acc 0.946 lb 1.964 z 0.331 lr 3.85e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.2h left [routing] mean entropy_norm 0.947 max_load(range) 0.050-0.217 step 37110/40589 loss 0.2698 acc 0.944 lb 1.960 z 0.333 lr 3.83e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.2h left step 37120/40589 loss 0.2619 acc 0.947 lb 1.958 z 0.333 lr 3.81e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.1h left [routing] mean entropy_norm 0.947 max_load(range) 0.055-0.351 step 37130/40589 loss 0.2560 acc 0.948 lb 1.963 z 0.402 lr 3.78e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.1h left step 37140/40589 loss 0.2683 acc 0.945 lb 1.964 z 0.343 lr 3.76e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.1h left step 37150/40589 loss 0.2677 acc 0.945 lb 1.962 z 0.325 lr 3.74e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.1h left [routing] mean entropy_norm 0.960 max_load(range) 0.050-0.204 step 37160/40589 loss 0.2610 acc 0.947 lb 1.958 z 0.334 lr 3.72e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.0h left step 37170/40589 loss 0.2512 acc 0.949 lb 1.962 z 0.374 lr 3.70e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.0h left [routing] mean entropy_norm 0.937 max_load(range) 0.060-0.130 step 37180/40589 loss 0.2704 acc 0.945 lb 1.964 z 0.325 lr 3.68e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 11.0h left step 37190/40589 loss 0.2712 acc 0.943 lb 1.964 z 0.330 lr 3.66e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.9h left step 37200/40589 loss 0.2660 acc 0.945 lb 1.958 z 0.336 lr 3.63e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.9h left [routing] mean entropy_norm 0.958 max_load(range) 0.058-0.181 step 37210/40589 loss 0.2556 acc 0.948 lb 1.964 z 0.400 lr 3.61e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.9h left step 37220/40589 loss 0.2707 acc 0.945 lb 1.964 z 0.343 lr 3.59e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.8h left [routing] mean entropy_norm 0.941 max_load(range) 0.061-0.153 step 37230/40589 loss 0.2655 acc 0.945 lb 1.963 z 0.327 lr 3.57e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.8h left step 37240/40589 loss 0.2620 acc 0.946 lb 1.959 z 0.346 lr 3.55e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.8h left step 37250/40589 loss 0.2500 acc 0.949 lb 1.964 z 0.389 lr 3.53e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.7h left [routing] mean entropy_norm 0.896 max_load(range) 0.064-0.322 step 37260/40589 loss 0.2700 acc 0.945 lb 1.963 z 0.351 lr 3.51e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.7h left step 37270/40589 loss 0.2694 acc 0.944 lb 1.962 z 0.333 lr 3.49e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.7h left [routing] mean entropy_norm 0.954 max_load(range) 0.052-0.228 step 37280/40589 loss 0.2647 acc 0.946 lb 1.956 z 0.339 lr 3.47e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.6h left step 37290/40589 loss 0.2531 acc 0.948 lb 1.964 z 0.408 lr 3.44e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.6h left step 37300/40589 loss 0.2673 acc 0.945 lb 1.964 z 0.348 lr 3.42e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.6h left [routing] mean entropy_norm 0.935 max_load(range) 0.056-0.243 step 37310/40589 loss 0.2689 acc 0.944 lb 1.959 z 0.331 lr 3.40e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.5h left step 37320/40589 loss 0.2614 acc 0.946 lb 1.957 z 0.326 lr 3.38e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.5h left [routing] mean entropy_norm 0.944 max_load(range) 0.053-0.357 step 37330/40589 loss 0.2557 acc 0.948 lb 1.963 z 0.411 lr 3.36e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.5h left step 37340/40589 loss 0.2690 acc 0.946 lb 1.964 z 0.349 lr 3.34e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.4h left step 37350/40589 loss 0.2647 acc 0.945 lb 1.964 z 0.330 lr 3.32e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.4h left [routing] mean entropy_norm 0.964 max_load(range) 0.046-0.239 step 37360/40589 loss 0.2613 acc 0.946 lb 1.958 z 0.347 lr 3.30e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.4h left step 37370/40589 loss 0.2508 acc 0.949 lb 1.964 z 0.416 lr 3.28e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.3h left [routing] mean entropy_norm 0.938 max_load(range) 0.056-0.139 step 37380/40589 loss 0.2693 acc 0.945 lb 1.964 z 0.341 lr 3.26e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.3h left step 37390/40589 loss 0.2658 acc 0.945 lb 1.962 z 0.346 lr 3.24e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.3h left step 37400/40589 loss 0.2607 acc 0.946 lb 1.958 z 0.332 lr 3.22e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.2h left [routing] mean entropy_norm 0.954 max_load(range) 0.058-0.160 step 37410/40589 loss 0.2522 acc 0.949 lb 1.963 z 0.411 lr 3.20e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.2h left step 37420/40589 loss 0.2685 acc 0.946 lb 1.964 z 0.345 lr 3.18e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.2h left [routing] mean entropy_norm 0.942 max_load(range) 0.058-0.164 step 37430/40589 loss 0.2633 acc 0.946 lb 1.960 z 0.329 lr 3.16e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.2h left step 37440/40589 loss 0.2583 acc 0.947 lb 1.957 z 0.325 lr 3.14e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.1h left step 37450/40589 loss 0.2551 acc 0.948 lb 1.962 z 0.402 lr 3.12e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.1h left [routing] mean entropy_norm 0.901 max_load(range) 0.067-0.314 step 37460/40589 loss 0.2679 acc 0.945 lb 1.966 z 0.348 lr 3.10e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.1h left step 37470/40589 loss 0.2651 acc 0.946 lb 1.963 z 0.331 lr 3.08e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.0h left [routing] mean entropy_norm 0.958 max_load(range) 0.050-0.210 step 37480/40589 loss 0.2645 acc 0.946 lb 1.955 z 0.331 lr 3.06e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.0h left step 37490/40589 loss 0.2522 acc 0.949 lb 1.965 z 0.406 lr 3.04e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 10.0h left step 37500/40589 loss 0.2675 acc 0.945 lb 1.964 z 0.352 lr 3.02e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.9h left [routing] mean entropy_norm 0.933 max_load(range) 0.054-0.242 step 37510/40589 loss 0.2608 acc 0.946 lb 1.959 z 0.325 lr 3.00e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.9h left step 37520/40589 loss 0.2636 acc 0.946 lb 1.956 z 0.318 lr 2.98e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.9h left [routing] mean entropy_norm 0.945 max_load(range) 0.052-0.346 step 37530/40589 loss 0.2505 acc 0.949 lb 1.963 z 0.414 lr 2.96e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.8h left step 37540/40589 loss 0.2695 acc 0.945 lb 1.964 z 0.329 lr 2.94e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.8h left step 37550/40589 loss 0.2672 acc 0.945 lb 1.960 z 0.324 lr 2.93e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.8h left [routing] mean entropy_norm 0.964 max_load(range) 0.047-0.234 step 37560/40589 loss 0.2618 acc 0.946 lb 1.957 z 0.329 lr 2.91e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.7h left step 37570/40589 loss 0.2525 acc 0.948 lb 1.962 z 0.402 lr 2.89e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.7h left [routing] mean entropy_norm 0.940 max_load(range) 0.061-0.138 step 37580/40589 loss 0.2670 acc 0.946 lb 1.963 z 0.347 lr 2.87e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.7h left step 37590/40589 loss 0.2694 acc 0.945 lb 1.962 z 0.329 lr 2.85e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.6h left step 37600/40589 loss 0.2631 acc 0.946 lb 1.954 z 0.326 lr 2.83e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.6h left [routing] mean entropy_norm 0.956 max_load(range) 0.058-0.156 step 37610/40589 loss 0.2517 acc 0.949 lb 1.962 z 0.405 lr 2.81e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.6h left step 37620/40589 loss 0.2685 acc 0.945 lb 1.965 z 0.342 lr 2.79e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.5h left [routing] mean entropy_norm 0.939 max_load(range) 0.058-0.149 step 37630/40589 loss 0.2648 acc 0.946 lb 1.961 z 0.324 lr 2.77e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.5h left step 37640/40589 loss 0.2623 acc 0.946 lb 1.956 z 0.330 lr 2.76e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.5h left step 37650/40589 loss 0.2482 acc 0.950 lb 1.965 z 0.400 lr 2.74e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.4h left [routing] mean entropy_norm 0.895 max_load(range) 0.068-0.321 step 37660/40589 loss 0.2680 acc 0.945 lb 1.964 z 0.341 lr 2.72e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.4h left step 37670/40589 loss 0.2691 acc 0.945 lb 1.959 z 0.336 lr 2.70e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.4h left [routing] mean entropy_norm 0.956 max_load(range) 0.051-0.222 step 37680/40589 loss 0.2638 acc 0.946 lb 1.958 z 0.334 lr 2.68e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.3h left step 37690/40589 loss 0.2503 acc 0.949 lb 1.963 z 0.402 lr 2.66e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.3h left step 37700/40589 loss 0.2665 acc 0.946 lb 1.965 z 0.351 lr 2.65e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.3h left [routing] mean entropy_norm 0.930 max_load(range) 0.054-0.255 step 37710/40589 loss 0.2673 acc 0.945 lb 1.959 z 0.325 lr 2.63e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.3h left step 37720/40589 loss 0.2631 acc 0.946 lb 1.956 z 0.334 lr 2.61e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.2h left [routing] mean entropy_norm 0.946 max_load(range) 0.057-0.356 step 37730/40589 loss 0.2529 acc 0.948 lb 1.960 z 0.387 lr 2.59e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.2h left step 37740/40589 loss 0.2678 acc 0.945 lb 1.963 z 0.350 lr 2.57e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.2h left step 37750/40589 loss 0.2655 acc 0.946 lb 1.960 z 0.322 lr 2.55e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.1h left [routing] mean entropy_norm 0.963 max_load(range) 0.044-0.252 step 37760/40589 loss 0.2644 acc 0.946 lb 1.953 z 0.346 lr 2.54e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.1h left step 37770/40589 loss 0.2490 acc 0.949 lb 1.962 z 0.414 lr 2.52e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.1h left [routing] mean entropy_norm 0.941 max_load(range) 0.058-0.139 step 37780/40589 loss 0.2725 acc 0.944 lb 1.961 z 0.341 lr 2.50e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.0h left step 37790/40589 loss 0.2619 acc 0.946 lb 1.963 z 0.336 lr 2.48e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.0h left step 37800/40589 loss 0.2590 acc 0.947 lb 1.955 z 0.342 lr 2.47e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 9.0h left [routing] mean entropy_norm 0.955 max_load(range) 0.058-0.183 step 37810/40589 loss 0.2503 acc 0.949 lb 1.964 z 0.420 lr 2.45e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.9h left step 37820/40589 loss 0.2638 acc 0.946 lb 1.965 z 0.360 lr 2.43e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.9h left [routing] mean entropy_norm 0.942 max_load(range) 0.056-0.159 step 37830/40589 loss 0.2682 acc 0.945 lb 1.960 z 0.329 lr 2.41e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.9h left step 37840/40589 loss 0.2622 acc 0.947 lb 1.955 z 0.345 lr 2.40e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.8h left step 37850/40589 loss 0.2499 acc 0.949 lb 1.963 z 0.418 lr 2.38e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.8h left [routing] mean entropy_norm 0.884 max_load(range) 0.072-0.336 step 37860/40589 loss 0.2688 acc 0.945 lb 1.965 z 0.320 lr 2.36e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.8h left step 37870/40589 loss 0.2671 acc 0.945 lb 1.962 z 0.323 lr 2.34e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.7h left [routing] mean entropy_norm 0.957 max_load(range) 0.051-0.202 step 37880/40589 loss 0.2625 acc 0.947 lb 1.958 z 0.348 lr 2.33e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.7h left step 37890/40589 loss 0.2482 acc 0.950 lb 1.963 z 0.387 lr 2.31e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.7h left step 37900/40589 loss 0.2671 acc 0.946 lb 1.963 z 0.348 lr 2.29e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.6h left [routing] mean entropy_norm 0.951 max_load(range) 0.050-0.197 step 37910/40589 loss 0.2644 acc 0.946 lb 1.958 z 0.321 lr 2.28e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.6h left step 37920/40589 loss 0.2624 acc 0.946 lb 1.957 z 0.345 lr 2.26e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.6h left [routing] mean entropy_norm 0.946 max_load(range) 0.052-0.351 step 37930/40589 loss 0.2548 acc 0.947 lb 1.962 z 0.384 lr 2.24e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.5h left step 37940/40589 loss 0.2641 acc 0.946 lb 1.961 z 0.336 lr 2.23e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.5h left step 37950/40589 loss 0.2657 acc 0.946 lb 1.961 z 0.322 lr 2.21e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.5h left [routing] mean entropy_norm 0.963 max_load(range) 0.049-0.222 step 37960/40589 loss 0.2624 acc 0.946 lb 1.956 z 0.330 lr 2.19e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.4h left step 37970/40589 loss 0.2507 acc 0.949 lb 1.961 z 0.393 lr 2.18e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.4h left [routing] mean entropy_norm 0.938 max_load(range) 0.058-0.131 step 37980/40589 loss 0.2652 acc 0.946 lb 1.963 z 0.336 lr 2.16e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.4h left step 37990/40589 loss 0.2642 acc 0.945 lb 1.961 z 0.328 lr 2.14e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.4h left step 38000/40589 loss 0.2620 acc 0.946 lb 1.958 z 0.334 lr 2.13e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.3h left [routing] mean entropy_norm 0.955 max_load(range) 0.061-0.169 [save] runs/pretrain_flagship_h100_4x8/step38000 step 38010/40589 loss 0.2483 acc 0.949 lb 1.963 z 0.399 lr 2.11e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.3h left step 38020/40589 loss 0.2674 acc 0.945 lb 1.964 z 0.350 lr 2.09e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.3h left [routing] mean entropy_norm 0.941 max_load(range) 0.066-0.163 step 38030/40589 loss 0.2639 acc 0.946 lb 1.961 z 0.330 lr 2.08e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.2h left step 38040/40589 loss 0.2572 acc 0.947 lb 1.958 z 0.335 lr 2.06e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.2h left step 38050/40589 loss 0.2487 acc 0.949 lb 1.962 z 0.402 lr 2.05e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.2h left [routing] mean entropy_norm 0.905 max_load(range) 0.066-0.307 step 38060/40589 loss 0.2650 acc 0.946 lb 1.963 z 0.349 lr 2.03e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.1h left step 38070/40589 loss 0.2674 acc 0.945 lb 1.960 z 0.322 lr 2.01e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.1h left [routing] mean entropy_norm 0.960 max_load(range) 0.047-0.195 step 38080/40589 loss 0.2607 acc 0.947 lb 1.957 z 0.329 lr 2.00e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.1h left step 38090/40589 loss 0.2509 acc 0.949 lb 1.961 z 0.369 lr 1.98e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.0h left step 38100/40589 loss 0.2644 acc 0.946 lb 1.965 z 0.353 lr 1.97e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.0h left [routing] mean entropy_norm 0.938 max_load(range) 0.055-0.235 step 38110/40589 loss 0.2684 acc 0.945 lb 1.960 z 0.337 lr 1.95e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 8.0h left step 38120/40589 loss 0.2620 acc 0.946 lb 1.956 z 0.350 lr 1.93e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 7.9h left [routing] mean entropy_norm 0.946 max_load(range) 0.056-0.354 step 38130/40589 loss 0.2488 acc 0.950 lb 1.964 z 0.404 lr 1.92e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 7.9h left step 38140/40589 loss 0.2674 acc 0.946 lb 1.963 z 0.343 lr 1.90e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 7.9h left step 38150/40589 loss 0.2647 acc 0.945 lb 1.961 z 0.329 lr 1.89e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 7.8h left [routing] mean entropy_norm 0.964 max_load(range) 0.050-0.243 step 38160/40589 loss 0.2614 acc 0.947 lb 1.956 z 0.328 lr 1.87e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 7.8h left step 38170/40589 loss 0.2488 acc 0.950 lb 1.965 z 0.400 lr 1.86e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 7.8h left [routing] mean entropy_norm 0.937 max_load(range) 0.058-0.133 step 38180/40589 loss 0.2695 acc 0.945 lb 1.964 z 0.330 lr 1.84e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 7.7h left step 38190/40589 loss 0.2636 acc 0.945 lb 1.959 z 0.332 lr 1.83e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 7.7h left step 38200/40589 loss 0.2622 acc 0.947 lb 1.957 z 0.330 lr 1.81e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 7.7h left [routing] mean entropy_norm 0.952 max_load(range) 0.064-0.151 step 38210/40589 loss 0.2515 acc 0.948 lb 1.965 z 0.401 lr 1.80e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 7.6h left step 38220/40589 loss 0.2672 acc 0.945 lb 1.964 z 0.352 lr 1.78e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 7.6h left [routing] mean entropy_norm 0.944 max_load(range) 0.058-0.187 step 38230/40589 loss 0.2668 acc 0.945 lb 1.963 z 0.341 lr 1.77e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 7.6h left step 38240/40589 loss 0.2601 acc 0.947 lb 1.960 z 0.338 lr 1.75e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 7.5h left step 38250/40589 loss 0.2483 acc 0.950 lb 1.961 z 0.391 lr 1.74e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 7.5h left [routing] mean entropy_norm 0.892 max_load(range) 0.069-0.326 step 38260/40589 loss 0.2685 acc 0.945 lb 1.960 z 0.335 lr 1.72e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 7.5h left step 38270/40589 loss 0.2612 acc 0.946 lb 1.962 z 0.323 lr 1.71e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 7.5h left [routing] mean entropy_norm 0.960 max_load(range) 0.056-0.194 step 38280/40589 loss 0.2642 acc 0.947 lb 1.957 z 0.329 lr 1.69e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 7.4h left step 38290/40589 loss 0.2471 acc 0.950 lb 1.965 z 0.394 lr 1.68e-06 | 1416 pairs/s | mem 63.9/79 GiB (min 62.8) | 7.4h left step 38300/40589 loss 0.2677 acc 0.946 lb 1.965 z 0.346 lr 1.66e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 7.4h left [routing] mean entropy_norm 0.950 max_load(range) 0.046-0.197 step 38310/40589 loss 0.2637 acc 0.946 lb 1.961 z 0.338 lr 1.65e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 7.3h left step 38320/40589 loss 0.2603 acc 0.947 lb 1.956 z 0.318 lr 1.63e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 7.3h left [routing] mean entropy_norm 0.948 max_load(range) 0.052-0.348 step 38330/40589 loss 0.2508 acc 0.949 lb 1.962 z 0.406 lr 1.62e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 7.3h left step 38340/40589 loss 0.2647 acc 0.946 lb 1.963 z 0.338 lr 1.61e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 7.2h left step 38350/40589 loss 0.2676 acc 0.945 lb 1.960 z 0.320 lr 1.59e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 7.2h left [routing] mean entropy_norm 0.963 max_load(range) 0.048-0.251 step 38360/40589 loss 0.2608 acc 0.946 lb 1.956 z 0.340 lr 1.58e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 7.2h left step 38370/40589 loss 0.2461 acc 0.949 lb 1.963 z 0.404 lr 1.56e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 7.1h left [routing] mean entropy_norm 0.940 max_load(range) 0.058-0.138 step 38380/40589 loss 0.2668 acc 0.946 lb 1.964 z 0.341 lr 1.55e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 7.1h left step 38390/40589 loss 0.2605 acc 0.947 lb 1.959 z 0.331 lr 1.54e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 7.1h left step 38400/40589 loss 0.2606 acc 0.946 lb 1.956 z 0.342 lr 1.52e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 7.0h left [routing] mean entropy_norm 0.955 max_load(range) 0.061-0.192 step 38410/40589 loss 0.2530 acc 0.949 lb 1.962 z 0.383 lr 1.51e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 7.0h left step 38420/40589 loss 0.2627 acc 0.946 lb 1.963 z 0.337 lr 1.49e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 7.0h left [routing] mean entropy_norm 0.944 max_load(range) 0.053-0.177 step 38430/40589 loss 0.2658 acc 0.945 lb 1.962 z 0.335 lr 1.48e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.9h left step 38440/40589 loss 0.2600 acc 0.947 lb 1.958 z 0.329 lr 1.47e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.9h left step 38450/40589 loss 0.2489 acc 0.949 lb 1.961 z 0.396 lr 1.45e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.9h left [routing] mean entropy_norm 0.895 max_load(range) 0.070-0.320 step 38460/40589 loss 0.2632 acc 0.947 lb 1.965 z 0.356 lr 1.44e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.8h left step 38470/40589 loss 0.2666 acc 0.946 lb 1.961 z 0.333 lr 1.43e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.8h left [routing] mean entropy_norm 0.958 max_load(range) 0.049-0.206 step 38480/40589 loss 0.2600 acc 0.946 lb 1.954 z 0.326 lr 1.41e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.8h left step 38490/40589 loss 0.2468 acc 0.950 lb 1.964 z 0.402 lr 1.40e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.7h left step 38500/40589 loss 0.2651 acc 0.946 lb 1.963 z 0.349 lr 1.39e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.7h left [routing] mean entropy_norm 0.941 max_load(range) 0.051-0.237 step 38510/40589 loss 0.2630 acc 0.946 lb 1.960 z 0.327 lr 1.37e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.7h left step 38520/40589 loss 0.2610 acc 0.947 lb 1.959 z 0.329 lr 1.36e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.7h left [routing] mean entropy_norm 0.952 max_load(range) 0.055-0.341 step 38530/40589 loss 0.2501 acc 0.949 lb 1.962 z 0.400 lr 1.35e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.6h left step 38540/40589 loss 0.2650 acc 0.946 lb 1.965 z 0.342 lr 1.33e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.6h left step 38550/40589 loss 0.2667 acc 0.945 lb 1.959 z 0.333 lr 1.32e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.6h left [routing] mean entropy_norm 0.964 max_load(range) 0.046-0.235 step 38560/40589 loss 0.2628 acc 0.946 lb 1.955 z 0.327 lr 1.31e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.5h left step 38570/40589 loss 0.2476 acc 0.950 lb 1.962 z 0.407 lr 1.29e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.5h left [routing] mean entropy_norm 0.937 max_load(range) 0.060-0.135 step 38580/40589 loss 0.2653 acc 0.946 lb 1.963 z 0.330 lr 1.28e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.5h left step 38590/40589 loss 0.2646 acc 0.946 lb 1.959 z 0.320 lr 1.27e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.4h left step 38600/40589 loss 0.2615 acc 0.947 lb 1.956 z 0.317 lr 1.26e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.4h left [routing] mean entropy_norm 0.955 max_load(range) 0.060-0.177 step 38610/40589 loss 0.2494 acc 0.949 lb 1.959 z 0.413 lr 1.24e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.4h left step 38620/40589 loss 0.2705 acc 0.945 lb 1.964 z 0.330 lr 1.23e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.3h left [routing] mean entropy_norm 0.937 max_load(range) 0.060-0.138 step 38630/40589 loss 0.2590 acc 0.947 lb 1.959 z 0.317 lr 1.22e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.3h left step 38640/40589 loss 0.2633 acc 0.946 lb 1.957 z 0.347 lr 1.21e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.3h left step 38650/40589 loss 0.2500 acc 0.949 lb 1.961 z 0.385 lr 1.19e-06 | 1416 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.2h left [routing] mean entropy_norm 0.905 max_load(range) 0.066-0.298 step 38660/40589 loss 0.2700 acc 0.945 lb 1.961 z 0.342 lr 1.18e-06 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.2h left step 38670/40589 loss 0.2629 acc 0.946 lb 1.958 z 0.319 lr 1.17e-06 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.2h left [routing] mean entropy_norm 0.957 max_load(range) 0.051-0.207 step 38680/40589 loss 0.2590 acc 0.947 lb 1.956 z 0.341 lr 1.16e-06 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.1h left step 38690/40589 loss 0.2462 acc 0.950 lb 1.962 z 0.424 lr 1.15e-06 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.1h left step 38700/40589 loss 0.2656 acc 0.946 lb 1.964 z 0.330 lr 1.13e-06 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.1h left [routing] mean entropy_norm 0.945 max_load(range) 0.055-0.219 step 38710/40589 loss 0.2662 acc 0.945 lb 1.962 z 0.329 lr 1.12e-06 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.0h left step 38720/40589 loss 0.2619 acc 0.946 lb 1.957 z 0.324 lr 1.11e-06 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.0h left [routing] mean entropy_norm 0.947 max_load(range) 0.055-0.354 step 38730/40589 loss 0.2510 acc 0.949 lb 1.962 z 0.395 lr 1.10e-06 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 6.0h left step 38740/40589 loss 0.2711 acc 0.945 lb 1.963 z 0.340 lr 1.09e-06 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.9h left step 38750/40589 loss 0.2653 acc 0.945 lb 1.959 z 0.345 lr 1.07e-06 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.9h left [routing] mean entropy_norm 0.963 max_load(range) 0.046-0.247 step 38760/40589 loss 0.2635 acc 0.946 lb 1.959 z 0.344 lr 1.06e-06 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.9h left step 38770/40589 loss 0.2534 acc 0.948 lb 1.961 z 0.385 lr 1.05e-06 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.8h left [routing] mean entropy_norm 0.935 max_load(range) 0.061-0.121 step 38780/40589 loss 0.2654 acc 0.946 lb 1.962 z 0.348 lr 1.04e-06 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.8h left step 38790/40589 loss 0.2652 acc 0.945 lb 1.960 z 0.333 lr 1.03e-06 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.8h left step 38800/40589 loss 0.2604 acc 0.947 lb 1.957 z 0.345 lr 1.02e-06 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.8h left [routing] mean entropy_norm 0.955 max_load(range) 0.059-0.184 step 38810/40589 loss 0.2482 acc 0.950 lb 1.960 z 0.401 lr 1.01e-06 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.7h left step 38820/40589 loss 0.2638 acc 0.946 lb 1.963 z 0.330 lr 9.95e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.7h left [routing] mean entropy_norm 0.943 max_load(range) 0.052-0.192 step 38830/40589 loss 0.2644 acc 0.945 lb 1.963 z 0.342 lr 9.83e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.7h left step 38840/40589 loss 0.2626 acc 0.947 lb 1.957 z 0.338 lr 9.72e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.6h left step 38850/40589 loss 0.2489 acc 0.949 lb 1.962 z 0.403 lr 9.61e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.6h left [routing] mean entropy_norm 0.901 max_load(range) 0.066-0.312 step 38860/40589 loss 0.2630 acc 0.946 lb 1.964 z 0.352 lr 9.50e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.6h left step 38870/40589 loss 0.2683 acc 0.945 lb 1.960 z 0.324 lr 9.39e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.5h left [routing] mean entropy_norm 0.960 max_load(range) 0.046-0.216 step 38880/40589 loss 0.2594 acc 0.947 lb 1.957 z 0.355 lr 9.28e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.5h left step 38890/40589 loss 0.2539 acc 0.948 lb 1.963 z 0.408 lr 9.18e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.5h left step 38900/40589 loss 0.2642 acc 0.946 lb 1.964 z 0.349 lr 9.07e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.4h left [routing] mean entropy_norm 0.948 max_load(range) 0.052-0.215 step 38910/40589 loss 0.2669 acc 0.945 lb 1.961 z 0.345 lr 8.96e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.4h left step 38920/40589 loss 0.2629 acc 0.946 lb 1.957 z 0.333 lr 8.85e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.4h left [routing] mean entropy_norm 0.945 max_load(range) 0.055-0.364 step 38930/40589 loss 0.2522 acc 0.949 lb 1.960 z 0.407 lr 8.75e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.3h left step 38940/40589 loss 0.2651 acc 0.946 lb 1.963 z 0.351 lr 8.64e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.3h left step 38950/40589 loss 0.2682 acc 0.945 lb 1.959 z 0.329 lr 8.54e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.3h left [routing] mean entropy_norm 0.963 max_load(range) 0.045-0.232 step 38960/40589 loss 0.2606 acc 0.947 lb 1.957 z 0.341 lr 8.44e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.2h left step 38970/40589 loss 0.2519 acc 0.949 lb 1.963 z 0.396 lr 8.33e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.2h left [routing] mean entropy_norm 0.941 max_load(range) 0.058-0.142 step 38980/40589 loss 0.2674 acc 0.946 lb 1.963 z 0.332 lr 8.23e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.2h left step 38990/40589 loss 0.2612 acc 0.946 lb 1.959 z 0.328 lr 8.13e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.1h left step 39000/40589 loss 0.2622 acc 0.946 lb 1.954 z 0.333 lr 8.03e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.1h left [routing] mean entropy_norm 0.959 max_load(range) 0.059-0.177 [save] runs/pretrain_flagship_h100_4x8/step39000 step 39010/40589 loss 0.2509 acc 0.949 lb 1.960 z 0.391 lr 7.93e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.1h left step 39020/40589 loss 0.2659 acc 0.945 lb 1.964 z 0.331 lr 7.83e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.0h left [routing] mean entropy_norm 0.944 max_load(range) 0.057-0.161 step 39030/40589 loss 0.2662 acc 0.945 lb 1.957 z 0.325 lr 7.73e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.0h left step 39040/40589 loss 0.2660 acc 0.946 lb 1.958 z 0.339 lr 7.63e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 5.0h left step 39050/40589 loss 0.2478 acc 0.949 lb 1.964 z 0.400 lr 7.53e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.9h left [routing] mean entropy_norm 0.899 max_load(range) 0.070-0.314 step 39060/40589 loss 0.2636 acc 0.946 lb 1.961 z 0.339 lr 7.43e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.9h left step 39070/40589 loss 0.2652 acc 0.946 lb 1.961 z 0.320 lr 7.34e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.9h left [routing] mean entropy_norm 0.959 max_load(range) 0.048-0.206 step 39080/40589 loss 0.2596 acc 0.947 lb 1.956 z 0.343 lr 7.24e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.9h left step 39090/40589 loss 0.2475 acc 0.950 lb 1.960 z 0.384 lr 7.14e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.8h left step 39100/40589 loss 0.2618 acc 0.947 lb 1.963 z 0.332 lr 7.05e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.8h left [routing] mean entropy_norm 0.921 max_load(range) 0.056-0.269 step 39110/40589 loss 0.2651 acc 0.945 lb 1.960 z 0.335 lr 6.96e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.8h left step 39120/40589 loss 0.2638 acc 0.946 lb 1.954 z 0.326 lr 6.86e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.7h left [routing] mean entropy_norm 0.945 max_load(range) 0.056-0.357 step 39130/40589 loss 0.2486 acc 0.949 lb 1.961 z 0.392 lr 6.77e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.7h left step 39140/40589 loss 0.2655 acc 0.946 lb 1.963 z 0.323 lr 6.68e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.7h left step 39150/40589 loss 0.2647 acc 0.946 lb 1.959 z 0.336 lr 6.58e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.6h left [routing] mean entropy_norm 0.963 max_load(range) 0.048-0.217 step 39160/40589 loss 0.2615 acc 0.946 lb 1.956 z 0.320 lr 6.49e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.6h left step 39170/40589 loss 0.2496 acc 0.950 lb 1.960 z 0.413 lr 6.40e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.6h left [routing] mean entropy_norm 0.938 max_load(range) 0.060-0.132 step 39180/40589 loss 0.2681 acc 0.945 lb 1.963 z 0.333 lr 6.31e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.5h left step 39190/40589 loss 0.2651 acc 0.945 lb 1.958 z 0.332 lr 6.22e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.5h left step 39200/40589 loss 0.2601 acc 0.946 lb 1.958 z 0.317 lr 6.14e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.5h left [routing] mean entropy_norm 0.956 max_load(range) 0.056-0.172 step 39210/40589 loss 0.2478 acc 0.950 lb 1.960 z 0.403 lr 6.05e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.4h left step 39220/40589 loss 0.2667 acc 0.945 lb 1.962 z 0.354 lr 5.96e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.4h left [routing] mean entropy_norm 0.942 max_load(range) 0.053-0.191 step 39230/40589 loss 0.2657 acc 0.946 lb 1.960 z 0.343 lr 5.87e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.4h left step 39240/40589 loss 0.2567 acc 0.947 lb 1.958 z 0.329 lr 5.79e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.3h left step 39250/40589 loss 0.2528 acc 0.949 lb 1.963 z 0.409 lr 5.70e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.3h left [routing] mean entropy_norm 0.902 max_load(range) 0.065-0.308 step 39260/40589 loss 0.2651 acc 0.946 lb 1.963 z 0.344 lr 5.62e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.3h left step 39270/40589 loss 0.2663 acc 0.945 lb 1.960 z 0.328 lr 5.53e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.2h left [routing] mean entropy_norm 0.957 max_load(range) 0.048-0.220 step 39280/40589 loss 0.2604 acc 0.947 lb 1.957 z 0.330 lr 5.45e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.2h left step 39290/40589 loss 0.2513 acc 0.949 lb 1.962 z 0.416 lr 5.37e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.2h left step 39300/40589 loss 0.2640 acc 0.946 lb 1.963 z 0.349 lr 5.28e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.1h left [routing] mean entropy_norm 0.946 max_load(range) 0.052-0.229 step 39310/40589 loss 0.2680 acc 0.945 lb 1.961 z 0.310 lr 5.20e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.1h left step 39320/40589 loss 0.2593 acc 0.947 lb 1.955 z 0.342 lr 5.12e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.1h left [routing] mean entropy_norm 0.944 max_load(range) 0.055-0.358 step 39330/40589 loss 0.2491 acc 0.949 lb 1.963 z 0.388 lr 5.04e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.0h left step 39340/40589 loss 0.2641 acc 0.946 lb 1.964 z 0.328 lr 4.96e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.0h left step 39350/40589 loss 0.2649 acc 0.946 lb 1.960 z 0.324 lr 4.88e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.0h left [routing] mean entropy_norm 0.965 max_load(range) 0.045-0.229 step 39360/40589 loss 0.2648 acc 0.946 lb 1.957 z 0.327 lr 4.80e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4.0h left step 39370/40589 loss 0.2500 acc 0.949 lb 1.964 z 0.421 lr 4.73e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.9h left [routing] mean entropy_norm 0.940 max_load(range) 0.059-0.137 step 39380/40589 loss 0.2645 acc 0.946 lb 1.964 z 0.344 lr 4.65e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.9h left step 39390/40589 loss 0.2672 acc 0.945 lb 1.961 z 0.321 lr 4.57e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.9h left step 39400/40589 loss 0.2639 acc 0.946 lb 1.956 z 0.326 lr 4.50e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.8h left [routing] mean entropy_norm 0.955 max_load(range) 0.059-0.180 step 39410/40589 loss 0.2504 acc 0.949 lb 1.962 z 0.416 lr 4.42e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.8h left step 39420/40589 loss 0.2649 acc 0.946 lb 1.963 z 0.336 lr 4.35e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.8h left [routing] mean entropy_norm 0.937 max_load(range) 0.059-0.151 step 39430/40589 loss 0.2655 acc 0.946 lb 1.959 z 0.317 lr 4.27e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.7h left step 39440/40589 loss 0.2633 acc 0.946 lb 1.957 z 0.341 lr 4.20e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.7h left step 39450/40589 loss 0.2509 acc 0.949 lb 1.962 z 0.397 lr 4.13e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.7h left [routing] mean entropy_norm 0.906 max_load(range) 0.060-0.300 step 39460/40589 loss 0.2666 acc 0.946 lb 1.964 z 0.330 lr 4.05e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.6h left step 39470/40589 loss 0.2677 acc 0.944 lb 1.960 z 0.329 lr 3.98e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.6h left [routing] mean entropy_norm 0.961 max_load(range) 0.046-0.208 step 39480/40589 loss 0.2596 acc 0.947 lb 1.953 z 0.336 lr 3.91e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.6h left step 39490/40589 loss 0.2492 acc 0.949 lb 1.962 z 0.405 lr 3.84e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.5h left step 39500/40589 loss 0.2645 acc 0.946 lb 1.959 z 0.334 lr 3.77e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.5h left [routing] mean entropy_norm 0.943 max_load(range) 0.051-0.227 step 39510/40589 loss 0.2645 acc 0.946 lb 1.960 z 0.313 lr 3.70e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.5h left step 39520/40589 loss 0.2619 acc 0.946 lb 1.956 z 0.336 lr 3.64e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.4h left [routing] mean entropy_norm 0.945 max_load(range) 0.054-0.363 step 39530/40589 loss 0.2505 acc 0.949 lb 1.963 z 0.386 lr 3.57e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.4h left step 39540/40589 loss 0.2692 acc 0.945 lb 1.962 z 0.360 lr 3.50e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.4h left step 39550/40589 loss 0.2618 acc 0.946 lb 1.958 z 0.307 lr 3.43e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.3h left [routing] mean entropy_norm 0.965 max_load(range) 0.048-0.228 step 39560/40589 loss 0.2612 acc 0.946 lb 1.960 z 0.332 lr 3.37e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.3h left step 39570/40589 loss 0.2525 acc 0.949 lb 1.963 z 0.399 lr 3.30e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.3h left [routing] mean entropy_norm 0.938 max_load(range) 0.058-0.130 step 39580/40589 loss 0.2665 acc 0.945 lb 1.961 z 0.347 lr 3.24e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.2h left step 39590/40589 loss 0.2634 acc 0.946 lb 1.961 z 0.334 lr 3.18e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.2h left step 39600/40589 loss 0.2601 acc 0.947 lb 1.958 z 0.336 lr 3.11e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.2h left [routing] mean entropy_norm 0.956 max_load(range) 0.056-0.146 step 39610/40589 loss 0.2465 acc 0.950 lb 1.960 z 0.393 lr 3.05e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.1h left step 39620/40589 loss 0.2648 acc 0.946 lb 1.963 z 0.328 lr 2.99e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.1h left [routing] mean entropy_norm 0.940 max_load(range) 0.059-0.150 step 39630/40589 loss 0.2623 acc 0.946 lb 1.960 z 0.322 lr 2.93e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.1h left step 39640/40589 loss 0.2615 acc 0.947 lb 1.957 z 0.342 lr 2.87e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.1h left step 39650/40589 loss 0.2487 acc 0.949 lb 1.959 z 0.410 lr 2.81e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.0h left [routing] mean entropy_norm 0.908 max_load(range) 0.063-0.299 step 39660/40589 loss 0.2662 acc 0.946 lb 1.962 z 0.338 lr 2.75e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.0h left step 39670/40589 loss 0.2639 acc 0.945 lb 1.958 z 0.313 lr 2.69e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 3.0h left [routing] mean entropy_norm 0.961 max_load(range) 0.048-0.198 step 39680/40589 loss 0.2610 acc 0.947 lb 1.959 z 0.325 lr 2.63e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.9h left step 39690/40589 loss 0.2496 acc 0.949 lb 1.961 z 0.400 lr 2.57e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.9h left step 39700/40589 loss 0.2650 acc 0.946 lb 1.963 z 0.349 lr 2.51e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.9h left [routing] mean entropy_norm 0.940 max_load(range) 0.054-0.236 step 39710/40589 loss 0.2677 acc 0.945 lb 1.961 z 0.321 lr 2.46e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.8h left step 39720/40589 loss 0.2610 acc 0.946 lb 1.958 z 0.329 lr 2.40e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.8h left [routing] mean entropy_norm 0.947 max_load(range) 0.055-0.364 step 39730/40589 loss 0.2463 acc 0.950 lb 1.963 z 0.414 lr 2.35e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.8h left step 39740/40589 loss 0.2664 acc 0.946 lb 1.964 z 0.350 lr 2.29e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.7h left step 39750/40589 loss 0.2620 acc 0.946 lb 1.958 z 0.318 lr 2.24e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.7h left [routing] mean entropy_norm 0.964 max_load(range) 0.047-0.217 step 39760/40589 loss 0.2608 acc 0.946 lb 1.956 z 0.335 lr 2.19e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.7h left step 39770/40589 loss 0.2491 acc 0.950 lb 1.963 z 0.402 lr 2.13e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.6h left [routing] mean entropy_norm 0.945 max_load(range) 0.058-0.156 step 39780/40589 loss 0.2700 acc 0.945 lb 1.962 z 0.354 lr 2.08e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.6h left step 39790/40589 loss 0.2657 acc 0.946 lb 1.959 z 0.338 lr 2.03e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.6h left step 39800/40589 loss 0.2623 acc 0.946 lb 1.957 z 0.335 lr 1.98e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.5h left [routing] mean entropy_norm 0.955 max_load(range) 0.062-0.169 step 39810/40589 loss 0.2484 acc 0.950 lb 1.964 z 0.412 lr 1.93e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.5h left step 39820/40589 loss 0.2681 acc 0.945 lb 1.966 z 0.338 lr 1.88e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.5h left [routing] mean entropy_norm 0.940 max_load(range) 0.058-0.164 step 39830/40589 loss 0.2631 acc 0.946 lb 1.961 z 0.337 lr 1.83e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.4h left step 39840/40589 loss 0.2635 acc 0.946 lb 1.955 z 0.331 lr 1.79e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.4h left step 39850/40589 loss 0.2456 acc 0.950 lb 1.964 z 0.416 lr 1.74e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.4h left [routing] mean entropy_norm 0.895 max_load(range) 0.067-0.322 step 39860/40589 loss 0.2689 acc 0.945 lb 1.962 z 0.345 lr 1.69e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.3h left step 39870/40589 loss 0.2663 acc 0.945 lb 1.960 z 0.329 lr 1.65e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.3h left [routing] mean entropy_norm 0.960 max_load(range) 0.053-0.197 step 39880/40589 loss 0.2613 acc 0.947 lb 1.958 z 0.341 lr 1.60e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.3h left step 39890/40589 loss 0.2483 acc 0.949 lb 1.964 z 0.386 lr 1.56e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.2h left step 39900/40589 loss 0.2654 acc 0.946 lb 1.963 z 0.353 lr 1.51e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.2h left [routing] mean entropy_norm 0.932 max_load(range) 0.056-0.251 step 39910/40589 loss 0.2655 acc 0.945 lb 1.959 z 0.311 lr 1.47e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.2h left step 39920/40589 loss 0.2610 acc 0.946 lb 1.956 z 0.332 lr 1.42e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.2h left [routing] mean entropy_norm 0.951 max_load(range) 0.054-0.365 step 39930/40589 loss 0.2495 acc 0.949 lb 1.960 z 0.397 lr 1.38e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.1h left step 39940/40589 loss 0.2613 acc 0.946 lb 1.962 z 0.329 lr 1.34e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.1h left step 39950/40589 loss 0.2651 acc 0.945 lb 1.960 z 0.320 lr 1.30e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.1h left [routing] mean entropy_norm 0.962 max_load(range) 0.049-0.207 step 39960/40589 loss 0.2644 acc 0.945 lb 1.957 z 0.333 lr 1.26e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.0h left step 39970/40589 loss 0.2468 acc 0.950 lb 1.961 z 0.370 lr 1.22e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.0h left [routing] mean entropy_norm 0.939 max_load(range) 0.058-0.138 step 39980/40589 loss 0.2687 acc 0.945 lb 1.963 z 0.341 lr 1.18e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2.0h left step 39990/40589 loss 0.2672 acc 0.945 lb 1.961 z 0.340 lr 1.14e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.9h left step 40000/40589 loss 0.2598 acc 0.947 lb 1.956 z 0.342 lr 1.10e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.9h left [routing] mean entropy_norm 0.952 max_load(range) 0.060-0.165 [save] runs/pretrain_flagship_h100_4x8/step40000 step 40010/40589 loss 0.2514 acc 0.949 lb 1.961 z 0.371 lr 1.07e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.9h left step 40020/40589 loss 0.2665 acc 0.946 lb 1.962 z 0.341 lr 1.03e-07 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.8h left [routing] mean entropy_norm 0.943 max_load(range) 0.060-0.156 step 40030/40589 loss 0.2652 acc 0.945 lb 1.958 z 0.320 lr 9.95e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.8h left step 40040/40589 loss 0.2621 acc 0.946 lb 1.958 z 0.326 lr 9.59e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.8h left step 40050/40589 loss 0.2476 acc 0.950 lb 1.961 z 0.408 lr 9.25e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.7h left [routing] mean entropy_norm 0.891 max_load(range) 0.072-0.332 step 40060/40589 loss 0.2616 acc 0.947 lb 1.962 z 0.317 lr 8.91e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.7h left step 40070/40589 loss 0.2672 acc 0.945 lb 1.960 z 0.336 lr 8.57e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.7h left [routing] mean entropy_norm 0.957 max_load(range) 0.050-0.216 step 40080/40589 loss 0.2619 acc 0.947 lb 1.957 z 0.327 lr 8.25e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.6h left step 40090/40589 loss 0.2484 acc 0.950 lb 1.962 z 0.409 lr 7.93e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.6h left step 40100/40589 loss 0.2635 acc 0.946 lb 1.963 z 0.340 lr 7.61e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.6h left [routing] mean entropy_norm 0.928 max_load(range) 0.053-0.247 step 40110/40589 loss 0.2680 acc 0.945 lb 1.959 z 0.328 lr 7.30e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.5h left step 40120/40589 loss 0.2593 acc 0.947 lb 1.958 z 0.332 lr 7.00e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.5h left [routing] mean entropy_norm 0.945 max_load(range) 0.055-0.364 step 40130/40589 loss 0.2476 acc 0.949 lb 1.961 z 0.408 lr 6.71e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.5h left step 40140/40589 loss 0.2641 acc 0.946 lb 1.963 z 0.357 lr 6.42e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.4h left step 40150/40589 loss 0.2689 acc 0.945 lb 1.961 z 0.343 lr 6.13e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.4h left [routing] mean entropy_norm 0.963 max_load(range) 0.044-0.228 step 40160/40589 loss 0.2617 acc 0.947 lb 1.958 z 0.338 lr 5.86e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.4h left step 40170/40589 loss 0.2507 acc 0.948 lb 1.962 z 0.408 lr 5.59e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.3h left [routing] mean entropy_norm 0.940 max_load(range) 0.058-0.135 step 40180/40589 loss 0.2641 acc 0.946 lb 1.961 z 0.339 lr 5.32e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.3h left step 40190/40589 loss 0.2675 acc 0.946 lb 1.963 z 0.344 lr 5.07e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.3h left step 40200/40589 loss 0.2612 acc 0.946 lb 1.957 z 0.330 lr 4.82e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.3h left [routing] mean entropy_norm 0.957 max_load(range) 0.060-0.168 step 40210/40589 loss 0.2502 acc 0.949 lb 1.962 z 0.423 lr 4.57e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.2h left step 40220/40589 loss 0.2686 acc 0.945 lb 1.966 z 0.334 lr 4.33e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.2h left [routing] mean entropy_norm 0.944 max_load(range) 0.056-0.172 step 40230/40589 loss 0.2681 acc 0.945 lb 1.959 z 0.336 lr 4.10e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.2h left step 40240/40589 loss 0.2605 acc 0.947 lb 1.955 z 0.319 lr 3.88e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.1h left step 40250/40589 loss 0.2506 acc 0.949 lb 1.961 z 0.408 lr 3.66e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.1h left [routing] mean entropy_norm 0.901 max_load(range) 0.066-0.313 step 40260/40589 loss 0.2698 acc 0.945 lb 1.962 z 0.336 lr 3.45e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.1h left step 40270/40589 loss 0.2664 acc 0.945 lb 1.959 z 0.334 lr 3.24e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 1.0h left [routing] mean entropy_norm 0.959 max_load(range) 0.047-0.207 step 40280/40589 loss 0.2602 acc 0.946 lb 1.954 z 0.323 lr 3.04e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 60m left step 40290/40589 loss 0.2479 acc 0.950 lb 1.962 z 0.408 lr 2.85e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 58m left step 40300/40589 loss 0.2647 acc 0.946 lb 1.964 z 0.348 lr 2.66e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 56m left [routing] mean entropy_norm 0.938 max_load(range) 0.055-0.233 step 40310/40589 loss 0.2644 acc 0.946 lb 1.959 z 0.328 lr 2.48e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 54m left step 40320/40589 loss 0.2644 acc 0.946 lb 1.962 z 0.320 lr 2.30e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 52m left [routing] mean entropy_norm 0.947 max_load(range) 0.050-0.354 step 40330/40589 loss 0.2513 acc 0.949 lb 1.963 z 0.407 lr 2.14e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 50m left step 40340/40589 loss 0.2612 acc 0.946 lb 1.962 z 0.317 lr 1.97e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 48m left step 40350/40589 loss 0.2697 acc 0.944 lb 1.959 z 0.335 lr 1.82e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 46m left [routing] mean entropy_norm 0.964 max_load(range) 0.049-0.228 step 40360/40589 loss 0.2607 acc 0.946 lb 1.958 z 0.332 lr 1.67e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 44m left step 40370/40589 loss 0.2481 acc 0.950 lb 1.961 z 0.403 lr 1.53e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 42m left [routing] mean entropy_norm 0.936 max_load(range) 0.061-0.129 step 40380/40589 loss 0.2641 acc 0.946 lb 1.961 z 0.341 lr 1.39e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 40m left step 40390/40589 loss 0.2664 acc 0.945 lb 1.960 z 0.326 lr 1.26e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 38m left step 40400/40589 loss 0.2618 acc 0.946 lb 1.956 z 0.321 lr 1.14e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 36m left [routing] mean entropy_norm 0.956 max_load(range) 0.060-0.165 step 40410/40589 loss 0.2502 acc 0.949 lb 1.962 z 0.413 lr 1.02e-08 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 35m left step 40420/40589 loss 0.2656 acc 0.946 lb 1.963 z 0.339 lr 9.09e-09 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 33m left [routing] mean entropy_norm 0.942 max_load(range) 0.061-0.143 step 40430/40589 loss 0.2668 acc 0.945 lb 1.958 z 0.314 lr 8.05e-09 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 31m left step 40440/40589 loss 0.2645 acc 0.946 lb 1.957 z 0.337 lr 7.07e-09 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 29m left step 40450/40589 loss 0.2518 acc 0.949 lb 1.962 z 0.401 lr 6.15e-09 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 27m left [routing] mean entropy_norm 0.894 max_load(range) 0.067-0.320 step 40460/40589 loss 0.2642 acc 0.946 lb 1.962 z 0.351 lr 5.30e-09 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 25m left step 40470/40589 loss 0.2660 acc 0.946 lb 1.961 z 0.323 lr 4.51e-09 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 23m left [routing] mean entropy_norm 0.957 max_load(range) 0.050-0.217 step 40480/40589 loss 0.2617 acc 0.946 lb 1.958 z 0.344 lr 3.78e-09 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 21m left step 40490/40589 loss 0.2473 acc 0.950 lb 1.963 z 0.399 lr 3.12e-09 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 19m left step 40500/40589 loss 0.2644 acc 0.946 lb 1.963 z 0.342 lr 2.52e-09 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 17m left [routing] mean entropy_norm 0.942 max_load(range) 0.053-0.230 step 40510/40589 loss 0.2647 acc 0.945 lb 1.959 z 0.347 lr 1.99e-09 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 15m left step 40520/40589 loss 0.2635 acc 0.946 lb 1.953 z 0.305 lr 1.52e-09 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 13m left [routing] mean entropy_norm 0.948 max_load(range) 0.054-0.350 step 40530/40589 loss 0.2521 acc 0.949 lb 1.961 z 0.382 lr 1.11e-09 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 11m left step 40540/40589 loss 0.2702 acc 0.944 lb 1.965 z 0.331 lr 7.64e-10 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 9m left step 40550/40589 loss 0.2625 acc 0.946 lb 1.957 z 0.326 lr 4.84e-10 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 8m left [routing] mean entropy_norm 0.964 max_load(range) 0.047-0.227 step 40560/40589 loss 0.2604 acc 0.947 lb 1.955 z 0.338 lr 2.68e-10 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 6m left step 40570/40589 loss 0.2507 acc 0.949 lb 1.963 z 0.388 lr 1.15e-10 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 4m left [routing] mean entropy_norm 0.940 max_load(range) 0.059-0.143 step 40580/40589 loss 0.2620 acc 0.946 lb 1.962 z 0.336 lr 2.58e-11 | 1415 pairs/s | mem 64.0/79 GiB (min 62.8) | 2m left [done] 40589 steps in 130.52 h -> runs/pretrain_flagship_h100_4x8 [done] 40589 steps in 130.52 h -> runs/pretrain_flagship_h100_4x8 [done] 40589 steps in 130.52 h -> runs/pretrain_flagship_h100_4x8 [save] runs/pretrain_flagship_h100_4x8/final [done] 40589 steps in 130.53 h -> runs/pretrain_flagship_h100_4x8 === flagship job exited ===