[13:59:56] device=cuda:0 bf16=True dev=NVIDIA GeForce RTX 4090 world=1 [13:59:56] dane: train=2,500,001,211B block=4096 batch=128 tok/step=524,288 [13:59:58] model GPT-ref: 183.6M param (L16 d896 h14) [13:59:59] optimizer=muon muon_lr=0.02 (mult=33.3x) [13:59:59] r11 pamiec: act_ckpt=True loss_chunk=1024 [13:59:59] r13: mixer_layout=AAAAAAAAAAAAAAAA gdn_head_dim=64 attn_bias=0 rope_style=interleaved [14:00:00] torch.compile enabled [14:00:00] START gpt-ref: steps=4768 (od 0) lr=0.0006->6e-05 [14:04:22] step 20/4768 loss 10.3912 lr 2.40e-05 gnorm 2.73 39,938 tok/s peakVRAM 13.7GB [14:07:35] step 40/4768 loss 9.5518 lr 4.80e-05 gnorm 1.36 54,379 tok/s peakVRAM 13.7GB [14:10:48] step 60/4768 loss 8.6782 lr 7.20e-05 gnorm 0.73 54,389 tok/s peakVRAM 13.7GB [14:14:00] step 80/4768 loss 8.0385 lr 9.60e-05 gnorm 1.30 54,384 tok/s peakVRAM 13.7GB [14:17:13] step 100/4768 loss 7.5974 lr 1.20e-04 gnorm 0.84 54,364 tok/s peakVRAM 13.7GB [14:20:26] step 120/4768 loss 7.1265 lr 1.44e-04 gnorm 0.78 54,354 tok/s peakVRAM 13.7GB [14:23:39] step 140/4768 loss 6.7082 lr 1.68e-04 gnorm 0.88 54,350 tok/s peakVRAM 13.7GB [14:26:52] step 160/4768 loss 6.3537 lr 1.92e-04 gnorm 0.78 54,345 tok/s peakVRAM 13.7GB [14:30:05] step 180/4768 loss 6.0894 lr 2.16e-04 gnorm 0.86 54,335 tok/s peakVRAM 13.7GB [14:33:18] step 200/4768 loss 5.8676 lr 2.40e-04 gnorm 0.76 54,330 tok/s peakVRAM 13.7GB [14:36:31] step 220/4768 loss 5.6845 lr 2.64e-04 gnorm 0.84 54,330 tok/s peakVRAM 13.7GB [14:39:44] step 240/4768 loss 5.5133 lr 2.88e-04 gnorm 0.76 54,330 tok/s peakVRAM 13.7GB [14:42:57] step 260/4768 loss 5.3498 lr 3.12e-04 gnorm 0.70 54,329 tok/s peakVRAM 13.7GB [14:46:10] step 280/4768 loss 5.1918 lr 3.36e-04 gnorm 0.66 54,330 tok/s peakVRAM 13.7GB [14:49:23] step 300/4768 loss 5.0466 lr 3.60e-04 gnorm 0.66 54,328 tok/s peakVRAM 13.7GB [14:52:36] step 320/4768 loss 4.9124 lr 3.84e-04 gnorm 0.72 54,329 tok/s peakVRAM 13.7GB [14:55:49] step 340/4768 loss 4.7663 lr 4.08e-04 gnorm 0.64 54,326 tok/s peakVRAM 13.7GB [14:59:02] step 360/4768 loss 4.6379 lr 4.32e-04 gnorm 0.60 54,329 tok/s peakVRAM 13.7GB [15:02:15] step 380/4768 loss 4.5207 lr 4.56e-04 gnorm 0.50 54,328 tok/s peakVRAM 13.7GB [15:05:28] step 400/4768 loss 4.4147 lr 4.80e-04 gnorm 0.44 54,326 tok/s peakVRAM 13.7GB [15:08:41] step 420/4768 loss 4.3344 lr 5.04e-04 gnorm 0.48 54,325 tok/s peakVRAM 13.7GB [15:11:54] step 440/4768 loss 4.2526 lr 5.28e-04 gnorm 0.36 54,324 tok/s peakVRAM 13.7GB [15:15:07] step 460/4768 loss 4.1822 lr 5.52e-04 gnorm 0.47 54,322 tok/s peakVRAM 13.7GB [15:18:20] step 480/4768 loss 4.1330 lr 5.76e-04 gnorm 0.38 54,320 tok/s peakVRAM 13.7GB [15:21:33] step 500/4768 loss 4.0766 lr 6.00e-04 gnorm 0.30 54,320 tok/s peakVRAM 13.7GB [15:24:46] step 520/4768 loss 4.0389 lr 6.00e-04 gnorm 0.35 54,316 tok/s peakVRAM 13.7GB [15:27:59] step 540/4768 loss 3.9892 lr 6.00e-04 gnorm 0.26 54,316 tok/s peakVRAM 13.7GB [15:31:13] step 560/4768 loss 3.9361 lr 6.00e-04 gnorm 0.24 54,314 tok/s peakVRAM 13.7GB [15:34:26] step 580/4768 loss 3.8941 lr 6.00e-04 gnorm 0.24 54,311 tok/s peakVRAM 13.7GB [15:37:39] step 600/4768 loss 3.8519 lr 6.00e-04 gnorm 0.22 54,312 tok/s peakVRAM 13.7GB [15:40:52] step 620/4768 loss 3.8298 lr 6.00e-04 gnorm 0.22 54,314 tok/s peakVRAM 13.7GB [15:44:05] step 640/4768 loss 3.8032 lr 6.00e-04 gnorm 0.19 54,308 tok/s peakVRAM 13.7GB [15:47:18] step 660/4768 loss 3.7710 lr 6.00e-04 gnorm 0.18 54,309 tok/s peakVRAM 13.7GB [15:50:31] step 680/4768 loss 3.7483 lr 6.00e-04 gnorm 0.18 54,309 tok/s peakVRAM 13.7GB [15:53:44] step 700/4768 loss 3.7176 lr 6.00e-04 gnorm 0.17 54,304 tok/s peakVRAM 13.7GB [15:56:57] step 720/4768 loss 3.6874 lr 6.00e-04 gnorm 0.18 54,296 tok/s peakVRAM 13.7GB [16:00:10] step 740/4768 loss 3.6829 lr 6.00e-04 gnorm 0.18 54,301 tok/s peakVRAM 13.7GB [16:03:23] step 760/4768 loss 3.6580 lr 6.00e-04 gnorm 0.15 54,298 tok/s peakVRAM 13.7GB [16:06:36] step 780/4768 loss 3.6315 lr 6.00e-04 gnorm 0.16 54,301 tok/s peakVRAM 13.7GB [16:09:50] step 800/4768 loss 3.6186 lr 6.00e-04 gnorm 0.17 54,303 tok/s peakVRAM 13.7GB [16:13:03] step 820/4768 loss 3.6022 lr 6.00e-04 gnorm 0.15 54,305 tok/s peakVRAM 13.7GB [16:16:16] step 840/4768 loss 3.5874 lr 6.00e-04 gnorm 0.17 54,300 tok/s peakVRAM 13.7GB [16:19:29] step 860/4768 loss 3.5672 lr 6.00e-04 gnorm 0.16 54,296 tok/s peakVRAM 13.7GB [16:22:42] step 880/4768 loss 3.5612 lr 6.00e-04 gnorm 0.13 54,301 tok/s peakVRAM 13.7GB [16:25:55] step 900/4768 loss 3.5428 lr 6.00e-04 gnorm 0.14 54,306 tok/s peakVRAM 13.7GB [16:29:08] step 920/4768 loss 3.5357 lr 6.00e-04 gnorm 0.14 54,304 tok/s peakVRAM 13.7GB [16:32:21] step 940/4768 loss 3.5316 lr 6.00e-04 gnorm 0.13 54,308 tok/s peakVRAM 13.7GB [16:35:34] step 960/4768 loss 3.5057 lr 6.00e-04 gnorm 0.13 54,309 tok/s peakVRAM 13.7GB [16:38:47] step 980/4768 loss 3.5031 lr 6.00e-04 gnorm 0.14 54,307 tok/s peakVRAM 13.7GB [16:42:01] step 1000/4768 loss 3.4927 lr 6.00e-04 gnorm 0.13 54,302 tok/s peakVRAM 13.7GB [16:45:14] step 1020/4768 loss 3.4872 lr 6.00e-04 gnorm 0.13 54,303 tok/s peakVRAM 13.7GB [16:48:27] step 1040/4768 loss 3.4604 lr 6.00e-04 gnorm 0.13 54,306 tok/s peakVRAM 13.7GB [16:51:40] step 1060/4768 loss 3.4721 lr 6.00e-04 gnorm 0.13 54,302 tok/s peakVRAM 13.7GB [16:54:53] step 1080/4768 loss 3.4612 lr 6.00e-04 gnorm 0.13 54,303 tok/s peakVRAM 13.7GB [16:58:06] step 1100/4768 loss 3.4462 lr 6.00e-04 gnorm 0.14 54,300 tok/s peakVRAM 13.7GB [17:01:19] step 1120/4768 loss 3.4286 lr 6.00e-04 gnorm 0.13 54,299 tok/s peakVRAM 13.7GB [17:04:32] step 1140/4768 loss 3.4225 lr 6.00e-04 gnorm 0.12 54,296 tok/s peakVRAM 13.7GB [17:07:45] step 1160/4768 loss 3.4164 lr 6.00e-04 gnorm 0.12 54,303 tok/s peakVRAM 13.7GB [17:10:58] step 1180/4768 loss 3.4192 lr 6.00e-04 gnorm 0.12 54,303 tok/s peakVRAM 13.7GB [17:14:12] step 1200/4768 loss 3.4006 lr 6.00e-04 gnorm 0.12 54,301 tok/s peakVRAM 13.7GB [17:17:25] step 1220/4768 loss 3.3984 lr 6.00e-04 gnorm 0.11 54,291 tok/s peakVRAM 13.7GB [17:20:38] step 1240/4768 loss 3.3901 lr 6.00e-04 gnorm 0.12 54,296 tok/s peakVRAM 13.7GB [17:23:51] step 1260/4768 loss 3.3763 lr 6.00e-04 gnorm 0.13 54,303 tok/s peakVRAM 13.7GB [17:27:04] step 1280/4768 loss 3.3792 lr 6.00e-04 gnorm 0.12 54,301 tok/s peakVRAM 13.7GB [17:30:17] step 1300/4768 loss 3.3721 lr 6.00e-04 gnorm 0.13 54,297 tok/s peakVRAM 13.7GB [17:33:30] step 1320/4768 loss 3.3646 lr 6.00e-04 gnorm 0.11 54,295 tok/s peakVRAM 13.7GB [17:36:43] step 1340/4768 loss 3.3652 lr 6.00e-04 gnorm 0.13 54,297 tok/s peakVRAM 13.7GB [17:39:57] step 1360/4768 loss 3.3561 lr 6.00e-04 gnorm 0.12 54,297 tok/s peakVRAM 13.7GB [17:43:10] step 1380/4768 loss 3.3482 lr 6.00e-04 gnorm 0.13 54,301 tok/s peakVRAM 13.7GB [17:46:23] step 1400/4768 loss 3.3375 lr 6.00e-04 gnorm 0.12 54,298 tok/s peakVRAM 13.7GB [17:49:36] step 1420/4768 loss 3.3256 lr 6.00e-04 gnorm 0.11 54,301 tok/s peakVRAM 13.7GB [17:52:49] step 1440/4768 loss 3.3326 lr 6.00e-04 gnorm 0.12 54,303 tok/s peakVRAM 13.7GB [17:56:02] step 1460/4768 loss 3.3301 lr 6.00e-04 gnorm 0.12 54,296 tok/s peakVRAM 13.7GB [17:59:15] step 1480/4768 loss 3.3341 lr 6.00e-04 gnorm 0.11 54,299 tok/s peakVRAM 13.7GB [18:02:28] step 1500/4768 loss 3.3077 lr 6.00e-04 gnorm 0.12 54,299 tok/s peakVRAM 13.7GB [18:05:41] step 1520/4768 loss 3.3145 lr 6.00e-04 gnorm 0.11 54,297 tok/s peakVRAM 13.7GB [18:08:55] step 1540/4768 loss 3.2987 lr 6.00e-04 gnorm 0.11 54,295 tok/s peakVRAM 13.7GB [18:12:08] step 1560/4768 loss 3.3075 lr 6.00e-04 gnorm 0.12 54,296 tok/s peakVRAM 13.7GB [18:15:21] step 1580/4768 loss 3.2953 lr 6.00e-04 gnorm 0.12 54,292 tok/s peakVRAM 13.7GB [18:18:34] step 1600/4768 loss 3.2850 lr 6.00e-04 gnorm 0.11 54,290 tok/s peakVRAM 13.7GB [18:21:47] step 1620/4768 loss 3.2859 lr 6.00e-04 gnorm 0.11 54,296 tok/s peakVRAM 13.7GB [18:25:00] step 1640/4768 loss 3.2823 lr 6.00e-04 gnorm 0.11 54,294 tok/s peakVRAM 13.7GB [18:28:13] step 1660/4768 loss 3.2832 lr 6.00e-04 gnorm 0.11 54,295 tok/s peakVRAM 13.7GB [18:31:26] step 1680/4768 loss 3.2724 lr 6.00e-04 gnorm 0.11 54,296 tok/s peakVRAM 13.7GB [18:34:40] step 1700/4768 loss 3.2652 lr 6.00e-04 gnorm 0.12 54,297 tok/s peakVRAM 13.7GB [18:37:53] step 1720/4768 loss 3.2673 lr 6.00e-04 gnorm 0.11 54,291 tok/s peakVRAM 13.7GB [18:41:06] step 1740/4768 loss 3.2613 lr 6.00e-04 gnorm 0.10 54,301 tok/s peakVRAM 13.7GB [18:44:19] step 1760/4768 loss 3.2605 lr 6.00e-04 gnorm 0.11 54,297 tok/s peakVRAM 13.7GB [18:47:32] step 1780/4768 loss 3.2474 lr 6.00e-04 gnorm 0.10 54,297 tok/s peakVRAM 13.7GB [18:50:45] step 1800/4768 loss 3.2512 lr 6.00e-04 gnorm 0.11 54,296 tok/s peakVRAM 13.7GB [18:53:58] step 1820/4768 loss 3.2471 lr 6.00e-04 gnorm 0.11 54,297 tok/s peakVRAM 13.7GB [18:57:11] step 1840/4768 loss 3.2405 lr 6.00e-04 gnorm 0.11 54,297 tok/s peakVRAM 13.7GB [19:00:25] step 1860/4768 loss 3.2397 lr 6.00e-04 gnorm 0.11 54,291 tok/s peakVRAM 13.7GB [19:03:38] step 1880/4768 loss 3.2355 lr 6.00e-04 gnorm 0.10 54,290 tok/s peakVRAM 13.7GB [19:06:51] step 1900/4768 loss 3.2295 lr 6.00e-04 gnorm 0.11 54,292 tok/s peakVRAM 13.7GB [19:10:04] step 1920/4768 loss 3.2302 lr 6.00e-04 gnorm 0.11 54,291 tok/s peakVRAM 13.7GB [19:13:17] step 1940/4768 loss 3.2224 lr 6.00e-04 gnorm 0.10 54,290 tok/s peakVRAM 13.7GB [19:16:30] step 1960/4768 loss 3.2266 lr 6.00e-04 gnorm 0.11 54,292 tok/s peakVRAM 13.7GB [19:19:43] step 1980/4768 loss 3.2208 lr 6.00e-04 gnorm 0.09 54,293 tok/s peakVRAM 13.7GB [19:22:56] step 2000/4768 loss 3.2143 lr 6.00e-04 gnorm 0.12 54,302 tok/s peakVRAM 13.7GB [19:26:10] step 2020/4768 loss 3.2143 lr 6.00e-04 gnorm 0.11 54,303 tok/s peakVRAM 13.7GB [19:29:23] step 2040/4768 loss 3.2074 lr 6.00e-04 gnorm 0.12 54,303 tok/s peakVRAM 13.7GB [19:32:36] step 2060/4768 loss 3.2075 lr 6.00e-04 gnorm 0.10 54,298 tok/s peakVRAM 13.7GB [19:35:49] step 2080/4768 loss 3.1964 lr 6.00e-04 gnorm 0.09 54,300 tok/s peakVRAM 13.7GB [19:39:02] step 2100/4768 loss 3.2047 lr 6.00e-04 gnorm 0.10 54,299 tok/s peakVRAM 13.7GB [19:42:15] step 2120/4768 loss 3.1944 lr 6.00e-04 gnorm 0.11 54,296 tok/s peakVRAM 13.7GB [19:45:28] step 2140/4768 loss 3.2072 lr 6.00e-04 gnorm 0.11 54,298 tok/s peakVRAM 13.7GB [19:48:41] step 2160/4768 loss 3.1865 lr 6.00e-04 gnorm 0.10 54,300 tok/s peakVRAM 13.7GB [19:51:54] step 2180/4768 loss 3.1878 lr 6.00e-04 gnorm 0.10 54,301 tok/s peakVRAM 13.7GB [19:55:08] step 2200/4768 loss 3.1888 lr 6.00e-04 gnorm 0.11 54,297 tok/s peakVRAM 13.7GB [19:58:21] step 2220/4768 loss 3.1810 lr 6.00e-04 gnorm 0.10 54,295 tok/s peakVRAM 13.7GB [20:01:34] step 2240/4768 loss 3.1782 lr 6.00e-04 gnorm 0.11 54,293 tok/s peakVRAM 13.7GB [20:04:47] step 2260/4768 loss 3.1746 lr 6.00e-04 gnorm 0.11 54,296 tok/s peakVRAM 13.7GB [20:08:00] step 2280/4768 loss 3.1712 lr 6.00e-04 gnorm 0.11 54,291 tok/s peakVRAM 13.7GB [20:11:13] step 2300/4768 loss 3.1743 lr 6.00e-04 gnorm 0.11 54,296 tok/s peakVRAM 13.7GB [20:14:26] step 2320/4768 loss 3.1707 lr 6.00e-04 gnorm 0.10 54,297 tok/s peakVRAM 13.7GB [20:17:39] step 2340/4768 loss 3.1656 lr 6.00e-04 gnorm 0.10 54,299 tok/s peakVRAM 13.7GB [20:20:53] step 2360/4768 loss 3.1564 lr 6.00e-04 gnorm 0.10 54,294 tok/s peakVRAM 13.7GB [20:24:06] step 2380/4768 loss 3.1555 lr 6.00e-04 gnorm 0.10 54,296 tok/s peakVRAM 13.7GB [20:27:19] step 2400/4768 loss 3.1551 lr 6.00e-04 gnorm 0.10 54,296 tok/s peakVRAM 13.7GB [20:30:32] step 2420/4768 loss 3.1583 lr 6.00e-04 gnorm 0.10 54,299 tok/s peakVRAM 13.7GB [20:33:45] step 2440/4768 loss 3.1520 lr 6.00e-04 gnorm 0.10 54,302 tok/s peakVRAM 13.7GB [20:36:58] step 2460/4768 loss 3.1510 lr 6.00e-04 gnorm 0.11 54,303 tok/s peakVRAM 13.7GB [20:40:11] step 2480/4768 loss 3.1482 lr 6.00e-04 gnorm 0.10 54,302 tok/s peakVRAM 13.7GB [20:43:24] step 2500/4768 loss 3.1426 lr 6.00e-04 gnorm 0.10 54,300 tok/s peakVRAM 13.7GB [20:46:37] step 2520/4768 loss 3.1444 lr 6.00e-04 gnorm 0.10 54,302 tok/s peakVRAM 13.7GB [20:49:51] step 2540/4768 loss 3.1448 lr 6.00e-04 gnorm 0.11 54,301 tok/s peakVRAM 13.7GB [20:53:04] step 2560/4768 loss 3.1416 lr 6.00e-04 gnorm 0.09 54,302 tok/s peakVRAM 13.7GB [20:56:17] step 2580/4768 loss 3.1369 lr 6.00e-04 gnorm 0.09 54,297 tok/s peakVRAM 13.7GB [20:59:30] step 2600/4768 loss 3.1276 lr 6.00e-04 gnorm 0.10 54,298 tok/s peakVRAM 13.7GB [21:02:43] step 2620/4768 loss 3.1362 lr 6.00e-04 gnorm 0.10 54,301 tok/s peakVRAM 13.7GB [21:05:56] step 2640/4768 loss 3.1374 lr 6.00e-04 gnorm 0.10 54,303 tok/s peakVRAM 13.7GB [21:09:09] step 2660/4768 loss 3.1362 lr 6.00e-04 gnorm 0.11 54,301 tok/s peakVRAM 13.7GB [21:12:22] step 2680/4768 loss 3.1310 lr 6.00e-04 gnorm 0.11 54,301 tok/s peakVRAM 13.7GB [21:15:35] step 2700/4768 loss 3.1338 lr 6.00e-04 gnorm 0.09 54,299 tok/s peakVRAM 13.7GB [21:18:48] step 2720/4768 loss 3.1165 lr 6.00e-04 gnorm 0.10 54,302 tok/s peakVRAM 13.7GB [21:22:02] step 2740/4768 loss 3.1183 lr 6.00e-04 gnorm 0.10 54,302 tok/s peakVRAM 13.7GB [21:25:15] step 2760/4768 loss 3.1214 lr 6.00e-04 gnorm 0.10 54,301 tok/s peakVRAM 13.7GB [21:28:28] step 2780/4768 loss 3.1122 lr 6.00e-04 gnorm 0.10 54,299 tok/s peakVRAM 13.7GB [21:31:41] step 2800/4768 loss 3.1071 lr 6.00e-04 gnorm 0.11 54,294 tok/s peakVRAM 13.7GB [21:34:54] step 2820/4768 loss 3.1044 lr 6.00e-04 gnorm 0.09 54,294 tok/s peakVRAM 13.7GB [21:38:07] step 2840/4768 loss 3.1124 lr 6.00e-04 gnorm 0.10 54,299 tok/s peakVRAM 13.7GB [21:41:20] step 2860/4768 loss 3.1040 lr 6.00e-04 gnorm 0.09 54,296 tok/s peakVRAM 13.7GB [21:44:33] step 2880/4768 loss 3.1058 lr 6.00e-04 gnorm 0.10 54,297 tok/s peakVRAM 13.7GB [21:47:47] step 2900/4768 loss 3.1077 lr 6.00e-04 gnorm 0.10 54,301 tok/s peakVRAM 13.7GB [21:51:00] step 2920/4768 loss 3.1145 lr 6.00e-04 gnorm 0.09 54,298 tok/s peakVRAM 13.7GB [21:54:13] step 2940/4768 loss 3.1047 lr 6.00e-04 gnorm 0.08 54,296 tok/s peakVRAM 13.7GB [21:57:26] step 2960/4768 loss 3.1066 lr 6.00e-04 gnorm 0.10 54,297 tok/s peakVRAM 13.7GB [22:00:39] step 2980/4768 loss 3.0944 lr 6.00e-04 gnorm 0.11 54,300 tok/s peakVRAM 13.7GB [22:03:52] step 3000/4768 loss 3.0945 lr 6.00e-04 gnorm 0.09 54,295 tok/s peakVRAM 13.7GB [22:07:05] step 3020/4768 loss 3.0901 lr 6.00e-04 gnorm 0.09 54,296 tok/s peakVRAM 13.7GB [22:10:18] step 3040/4768 loss 3.0874 lr 6.00e-04 gnorm 0.10 54,295 tok/s peakVRAM 13.7GB [22:13:31] step 3060/4768 loss 3.0911 lr 6.00e-04 gnorm 0.10 54,304 tok/s peakVRAM 13.7GB [22:16:45] step 3080/4768 loss 3.0834 lr 6.00e-04 gnorm 0.09 54,304 tok/s peakVRAM 13.7GB [22:19:58] step 3100/4768 loss 3.0775 lr 6.00e-04 gnorm 0.10 54,303 tok/s peakVRAM 13.7GB [22:23:11] step 3120/4768 loss 3.0931 lr 6.00e-04 gnorm 0.10 54,302 tok/s peakVRAM 13.7GB [22:26:24] step 3140/4768 loss 3.0785 lr 6.00e-04 gnorm 0.10 54,302 tok/s peakVRAM 13.7GB [22:29:37] step 3160/4768 loss 3.0747 lr 6.00e-04 gnorm 0.09 54,304 tok/s peakVRAM 13.7GB [22:32:50] step 3180/4768 loss 3.0803 lr 6.00e-04 gnorm 0.09 54,303 tok/s peakVRAM 13.7GB [22:36:03] step 3200/4768 loss 3.0820 lr 6.00e-04 gnorm 0.09 54,306 tok/s peakVRAM 13.7GB [22:39:16] step 3220/4768 loss 3.0916 lr 6.00e-04 gnorm 0.09 54,302 tok/s peakVRAM 13.7GB [22:42:29] step 3240/4768 loss 3.0735 lr 6.00e-04 gnorm 0.09 54,302 tok/s peakVRAM 13.7GB [22:45:42] step 3260/4768 loss 3.0701 lr 6.00e-04 gnorm 0.09 54,303 tok/s peakVRAM 13.7GB [22:48:56] step 3280/4768 loss 3.0658 lr 6.00e-04 gnorm 0.09 54,301 tok/s peakVRAM 13.7GB [22:52:09] step 3300/4768 loss 3.0763 lr 6.00e-04 gnorm 0.08 54,302 tok/s peakVRAM 13.7GB [22:55:22] step 3320/4768 loss 3.0704 lr 6.00e-04 gnorm 0.09 54,303 tok/s peakVRAM 13.7GB [22:58:35] step 3340/4768 loss 3.0707 lr 6.00e-04 gnorm 0.09 54,305 tok/s peakVRAM 13.7GB [23:01:48] step 3360/4768 loss 3.0617 lr 6.00e-04 gnorm 0.10 54,304 tok/s peakVRAM 13.7GB [23:05:01] step 3380/4768 loss 3.0715 lr 6.00e-04 gnorm 0.10 54,304 tok/s peakVRAM 13.7GB [23:08:14] step 3400/4768 loss 3.0564 lr 6.00e-04 gnorm 0.09 54,306 tok/s peakVRAM 13.7GB [23:11:27] step 3420/4768 loss 3.0647 lr 6.00e-04 gnorm 0.09 54,303 tok/s peakVRAM 13.7GB [23:14:40] step 3440/4768 loss 3.0560 lr 6.00e-04 gnorm 0.10 54,306 tok/s peakVRAM 13.7GB [23:17:53] step 3460/4768 loss 3.0657 lr 6.00e-04 gnorm 0.09 54,307 tok/s peakVRAM 13.7GB [23:21:06] step 3480/4768 loss 3.0652 lr 6.00e-04 gnorm 0.09 54,304 tok/s peakVRAM 13.7GB [23:24:20] step 3500/4768 loss 3.0513 lr 6.00e-04 gnorm 0.10 54,308 tok/s peakVRAM 13.7GB [23:27:33] step 3520/4768 loss 3.0545 lr 6.00e-04 gnorm 0.10 54,304 tok/s peakVRAM 13.7GB [23:30:46] step 3540/4768 loss 3.0561 lr 6.00e-04 gnorm 0.09 54,305 tok/s peakVRAM 13.7GB [23:33:59] step 3560/4768 loss 3.0606 lr 6.00e-04 gnorm 0.09 54,303 tok/s peakVRAM 13.7GB [23:37:12] step 3580/4768 loss 3.0544 lr 6.00e-04 gnorm 0.08 54,307 tok/s peakVRAM 13.7GB [23:40:25] step 3600/4768 loss 3.0568 lr 6.00e-04 gnorm 0.10 54,309 tok/s peakVRAM 13.7GB [23:43:38] step 3620/4768 loss 3.0479 lr 6.00e-04 gnorm 0.10 54,307 tok/s peakVRAM 13.7GB [23:46:51] step 3640/4768 loss 3.0562 lr 6.00e-04 gnorm 0.09 54,308 tok/s peakVRAM 13.7GB [23:50:04] step 3660/4768 loss 3.0452 lr 6.00e-04 gnorm 0.09 54,304 tok/s peakVRAM 13.7GB [23:53:17] step 3680/4768 loss 3.0520 lr 6.00e-04 gnorm 0.09 54,305 tok/s peakVRAM 13.7GB [23:56:30] step 3700/4768 loss 3.0491 lr 6.00e-04 gnorm 0.10 54,306 tok/s peakVRAM 13.7GB [23:59:44] step 3720/4768 loss 3.0422 lr 6.00e-04 gnorm 0.10 54,305 tok/s peakVRAM 13.7GB [00:02:57] step 3740/4768 loss 3.0386 lr 6.00e-04 gnorm 0.09 54,297 tok/s peakVRAM 13.7GB [00:06:10] step 3760/4768 loss 3.0374 lr 6.00e-04 gnorm 0.10 54,301 tok/s peakVRAM 13.7GB [00:09:23] step 3780/4768 loss 3.0376 lr 6.00e-04 gnorm 0.09 54,300 tok/s peakVRAM 13.7GB [00:12:36] step 3800/4768 loss 3.0388 lr 6.00e-04 gnorm 0.09 54,301 tok/s peakVRAM 13.7GB [00:15:49] step 3820/4768 loss 3.0327 lr 5.61e-04 gnorm 0.09 54,302 tok/s peakVRAM 13.7GB [00:19:02] step 3840/4768 loss 3.0306 lr 5.13e-04 gnorm 0.08 54,308 tok/s peakVRAM 13.7GB [00:22:15] step 3860/4768 loss 3.0235 lr 4.83e-04 gnorm 0.07 54,307 tok/s peakVRAM 13.7GB [00:25:28] step 3880/4768 loss 3.0207 lr 4.59e-04 gnorm 0.07 54,307 tok/s peakVRAM 13.7GB [00:28:41] step 3900/4768 loss 3.0140 lr 4.39e-04 gnorm 0.08 54,308 tok/s peakVRAM 13.7GB [00:31:54] step 3920/4768 loss 3.0089 lr 4.21e-04 gnorm 0.08 54,310 tok/s peakVRAM 13.7GB [00:35:08] step 3940/4768 loss 3.0008 lr 4.05e-04 gnorm 0.07 54,312 tok/s peakVRAM 13.7GB [00:38:21] step 3960/4768 loss 3.0012 lr 3.89e-04 gnorm 0.08 54,309 tok/s peakVRAM 13.7GB [00:41:34] step 3980/4768 loss 2.9935 lr 3.75e-04 gnorm 0.07 54,310 tok/s peakVRAM 13.7GB [00:44:47] step 4000/4768 loss 2.9921 lr 3.62e-04 gnorm 0.07 54,306 tok/s peakVRAM 13.7GB [00:48:00] step 4020/4768 loss 2.9892 lr 3.50e-04 gnorm 0.07 54,309 tok/s peakVRAM 13.7GB [00:51:13] step 4040/4768 loss 2.9804 lr 3.38e-04 gnorm 0.07 54,307 tok/s peakVRAM 13.7GB [00:54:26] step 4060/4768 loss 2.9800 lr 3.26e-04 gnorm 0.07 54,307 tok/s peakVRAM 13.7GB [00:57:39] step 4080/4768 loss 2.9711 lr 3.15e-04 gnorm 0.07 54,306 tok/s peakVRAM 13.7GB [01:00:52] step 4100/4768 loss 2.9771 lr 3.05e-04 gnorm 0.07 54,300 tok/s peakVRAM 13.7GB [01:04:05] step 4120/4768 loss 2.9652 lr 2.95e-04 gnorm 0.07 54,296 tok/s peakVRAM 13.7GB [01:07:18] step 4140/4768 loss 2.9576 lr 2.85e-04 gnorm 0.07 54,296 tok/s peakVRAM 13.7GB [01:10:32] step 4160/4768 loss 2.9754 lr 2.75e-04 gnorm 0.07 54,296 tok/s peakVRAM 13.7GB [01:13:45] step 4180/4768 loss 2.9682 lr 2.66e-04 gnorm 0.07 54,299 tok/s peakVRAM 13.7GB [01:16:58] step 4200/4768 loss 2.9618 lr 2.57e-04 gnorm 0.07 54,299 tok/s peakVRAM 13.7GB [01:20:11] step 4220/4768 loss 2.9553 lr 2.48e-04 gnorm 0.07 54,302 tok/s peakVRAM 13.7GB [01:23:24] step 4240/4768 loss 2.9614 lr 2.40e-04 gnorm 0.07 54,297 tok/s peakVRAM 13.7GB [01:26:37] step 4260/4768 loss 2.9540 lr 2.31e-04 gnorm 0.07 54,297 tok/s peakVRAM 13.7GB [01:29:50] step 4280/4768 loss 2.9529 lr 2.23e-04 gnorm 0.07 54,296 tok/s peakVRAM 13.7GB [01:33:03] step 4300/4768 loss 2.9506 lr 2.15e-04 gnorm 0.07 54,296 tok/s peakVRAM 13.7GB [01:36:16] step 4320/4768 loss 2.9459 lr 2.07e-04 gnorm 0.07 54,297 tok/s peakVRAM 13.7GB [01:39:30] step 4340/4768 loss 2.9443 lr 1.99e-04 gnorm 0.07 54,300 tok/s peakVRAM 13.7GB [01:42:43] step 4360/4768 loss 2.9391 lr 1.92e-04 gnorm 0.07 54,298 tok/s peakVRAM 13.7GB [01:45:56] step 4380/4768 loss 2.9453 lr 1.84e-04 gnorm 0.07 54,296 tok/s peakVRAM 13.7GB [01:49:09] step 4400/4768 loss 2.9431 lr 1.77e-04 gnorm 0.07 54,300 tok/s peakVRAM 13.7GB [01:52:22] step 4420/4768 loss 2.9360 lr 1.70e-04 gnorm 0.07 54,300 tok/s peakVRAM 13.7GB [01:55:35] step 4440/4768 loss 2.9422 lr 1.63e-04 gnorm 0.07 54,300 tok/s peakVRAM 13.7GB [01:58:48] step 4460/4768 loss 2.9381 lr 1.56e-04 gnorm 0.07 54,299 tok/s peakVRAM 13.7GB [02:02:01] step 4480/4768 loss 2.9331 lr 1.49e-04 gnorm 0.07 54,298 tok/s peakVRAM 13.7GB [02:05:15] step 4500/4768 loss 2.9303 lr 1.42e-04 gnorm 0.07 54,299 tok/s peakVRAM 13.7GB [02:08:28] step 4520/4768 loss 2.9317 lr 1.36e-04 gnorm 0.07 54,299 tok/s peakVRAM 13.7GB [02:11:41] step 4540/4768 loss 2.9233 lr 1.29e-04 gnorm 0.07 54,293 tok/s peakVRAM 13.7GB [02:14:54] step 4560/4768 loss 2.9264 lr 1.23e-04 gnorm 0.07 54,295 tok/s peakVRAM 13.7GB [02:18:07] step 4580/4768 loss 2.9300 lr 1.16e-04 gnorm 0.07 54,293 tok/s peakVRAM 13.7GB [02:21:20] step 4600/4768 loss 2.9284 lr 1.10e-04 gnorm 0.07 54,295 tok/s peakVRAM 13.7GB [02:24:33] step 4620/4768 loss 2.9155 lr 1.04e-04 gnorm 0.07 54,290 tok/s peakVRAM 13.7GB [02:27:46] step 4640/4768 loss 2.9244 lr 9.78e-05 gnorm 0.07 54,291 tok/s peakVRAM 13.7GB [02:31:00] step 4660/4768 loss 2.9173 lr 9.18e-05 gnorm 0.07 54,298 tok/s peakVRAM 13.7GB [02:34:13] step 4680/4768 loss 2.9151 lr 8.58e-05 gnorm 0.07 54,299 tok/s peakVRAM 13.7GB [02:37:26] step 4700/4768 loss 2.9109 lr 7.99e-05 gnorm 0.07 54,299 tok/s peakVRAM 13.7GB [02:40:39] step 4720/4768 loss 2.9122 lr 7.41e-05 gnorm 0.07 54,298 tok/s peakVRAM 13.7GB [02:43:52] step 4740/4768 loss 2.9100 lr 6.83e-05 gnorm 0.07 54,301 tok/s peakVRAM 13.7GB [02:47:05] step 4760/4768 loss 2.9194 lr 6.26e-05 gnorm 0.07 54,313 tok/s peakVRAM 13.7GB [02:48:24] ckpt @ 4768 [02:48:24] DONE