{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.23758259707476428, "eval_steps": 500, "global_step": 1600, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "grad_norm": 1.75, "learning_rate": 2.9673590504451043e-07, "loss": 1.433, "step": 1 }, { "epoch": 0.0, "grad_norm": 1.578125, "learning_rate": 1.483679525222552e-06, "loss": 1.5006, "step": 5 }, { "epoch": 0.0, "grad_norm": 1.4921875, "learning_rate": 2.967359050445104e-06, "loss": 1.4919, "step": 10 }, { "epoch": 0.0, "grad_norm": 1.4921875, "learning_rate": 4.451038575667656e-06, "loss": 1.4531, "step": 15 }, { "epoch": 0.0, "grad_norm": 1.0234375, "learning_rate": 5.934718100890208e-06, "loss": 1.4242, "step": 20 }, { "epoch": 0.0, "grad_norm": 0.9453125, "learning_rate": 7.418397626112759e-06, "loss": 1.3932, "step": 25 }, { "epoch": 0.0, "grad_norm": 0.8359375, "learning_rate": 8.902077151335312e-06, "loss": 1.3661, "step": 30 }, { "epoch": 0.01, "grad_norm": 0.828125, "learning_rate": 1.0385756676557864e-05, "loss": 1.3677, "step": 35 }, { "epoch": 0.01, "grad_norm": 0.59375, "learning_rate": 1.1869436201780416e-05, "loss": 1.3497, "step": 40 }, { "epoch": 0.01, "grad_norm": 0.56640625, "learning_rate": 1.3353115727002968e-05, "loss": 1.3049, "step": 45 }, { "epoch": 0.01, "grad_norm": 0.439453125, "learning_rate": 1.4836795252225518e-05, "loss": 1.2836, "step": 50 }, { "epoch": 0.01, "grad_norm": 0.46484375, "learning_rate": 1.6320474777448072e-05, "loss": 1.3133, "step": 55 }, { "epoch": 0.01, "grad_norm": 0.40625, "learning_rate": 1.7804154302670624e-05, "loss": 1.2596, "step": 60 }, { "epoch": 0.01, "grad_norm": 0.408203125, "learning_rate": 1.9287833827893176e-05, "loss": 1.2612, "step": 65 }, { "epoch": 0.01, "grad_norm": 0.376953125, "learning_rate": 2.0771513353115728e-05, "loss": 1.2584, "step": 70 }, { "epoch": 0.01, "grad_norm": 0.40625, "learning_rate": 2.225519287833828e-05, "loss": 1.2386, "step": 75 }, { "epoch": 0.01, "grad_norm": 0.353515625, "learning_rate": 2.3738872403560832e-05, "loss": 1.1833, "step": 80 }, { "epoch": 0.01, "grad_norm": 0.349609375, "learning_rate": 2.5222551928783384e-05, "loss": 1.2411, "step": 85 }, { "epoch": 0.01, "grad_norm": 0.36328125, "learning_rate": 2.6706231454005936e-05, "loss": 1.1956, "step": 90 }, { "epoch": 0.01, "grad_norm": 0.369140625, "learning_rate": 2.818991097922849e-05, "loss": 1.2327, "step": 95 }, { "epoch": 0.01, "grad_norm": 0.353515625, "learning_rate": 2.9673590504451037e-05, "loss": 1.1896, "step": 100 }, { "epoch": 0.02, "grad_norm": 0.3984375, "learning_rate": 3.115727002967359e-05, "loss": 1.1979, "step": 105 }, { "epoch": 0.02, "grad_norm": 0.388671875, "learning_rate": 3.2640949554896144e-05, "loss": 1.2004, "step": 110 }, { "epoch": 0.02, "grad_norm": 0.392578125, "learning_rate": 3.41246290801187e-05, "loss": 1.1624, "step": 115 }, { "epoch": 0.02, "grad_norm": 0.400390625, "learning_rate": 3.560830860534125e-05, "loss": 1.1914, "step": 120 }, { "epoch": 0.02, "grad_norm": 0.41796875, "learning_rate": 3.70919881305638e-05, "loss": 1.1952, "step": 125 }, { "epoch": 0.02, "grad_norm": 0.4140625, "learning_rate": 3.857566765578635e-05, "loss": 1.1538, "step": 130 }, { "epoch": 0.02, "grad_norm": 0.4375, "learning_rate": 4.005934718100891e-05, "loss": 1.1702, "step": 135 }, { "epoch": 0.02, "grad_norm": 0.44140625, "learning_rate": 4.1543026706231456e-05, "loss": 1.1683, "step": 140 }, { "epoch": 0.02, "grad_norm": 0.443359375, "learning_rate": 4.3026706231454005e-05, "loss": 1.1822, "step": 145 }, { "epoch": 0.02, "grad_norm": 0.482421875, "learning_rate": 4.451038575667656e-05, "loss": 1.1649, "step": 150 }, { "epoch": 0.02, "grad_norm": 0.46484375, "learning_rate": 4.5994065281899116e-05, "loss": 1.1868, "step": 155 }, { "epoch": 0.02, "grad_norm": 0.5, "learning_rate": 4.7477744807121664e-05, "loss": 1.161, "step": 160 }, { "epoch": 0.02, "grad_norm": 0.451171875, "learning_rate": 4.896142433234421e-05, "loss": 1.1454, "step": 165 }, { "epoch": 0.03, "grad_norm": 0.45703125, "learning_rate": 5.044510385756677e-05, "loss": 1.1388, "step": 170 }, { "epoch": 0.03, "grad_norm": 0.490234375, "learning_rate": 5.1928783382789324e-05, "loss": 1.1253, "step": 175 }, { "epoch": 0.03, "grad_norm": 0.51171875, "learning_rate": 5.341246290801187e-05, "loss": 1.1527, "step": 180 }, { "epoch": 0.03, "grad_norm": 0.486328125, "learning_rate": 5.489614243323442e-05, "loss": 1.1234, "step": 185 }, { "epoch": 0.03, "grad_norm": 0.46875, "learning_rate": 5.637982195845698e-05, "loss": 1.113, "step": 190 }, { "epoch": 0.03, "grad_norm": 0.458984375, "learning_rate": 5.7863501483679525e-05, "loss": 1.1368, "step": 195 }, { "epoch": 0.03, "grad_norm": 0.484375, "learning_rate": 5.9347181008902074e-05, "loss": 1.137, "step": 200 }, { "epoch": 0.03, "grad_norm": 0.4765625, "learning_rate": 6.0830860534124636e-05, "loss": 1.1286, "step": 205 }, { "epoch": 0.03, "grad_norm": 0.484375, "learning_rate": 6.231454005934718e-05, "loss": 1.1277, "step": 210 }, { "epoch": 0.03, "grad_norm": 0.474609375, "learning_rate": 6.379821958456974e-05, "loss": 1.1268, "step": 215 }, { "epoch": 0.03, "grad_norm": 0.47265625, "learning_rate": 6.528189910979229e-05, "loss": 1.0993, "step": 220 }, { "epoch": 0.03, "grad_norm": 0.50390625, "learning_rate": 6.676557863501484e-05, "loss": 1.1017, "step": 225 }, { "epoch": 0.03, "grad_norm": 0.4609375, "learning_rate": 6.82492581602374e-05, "loss": 1.1345, "step": 230 }, { "epoch": 0.03, "grad_norm": 0.48828125, "learning_rate": 6.973293768545995e-05, "loss": 1.1086, "step": 235 }, { "epoch": 0.04, "grad_norm": 0.45703125, "learning_rate": 7.12166172106825e-05, "loss": 1.0791, "step": 240 }, { "epoch": 0.04, "grad_norm": 0.470703125, "learning_rate": 7.270029673590505e-05, "loss": 1.1158, "step": 245 }, { "epoch": 0.04, "grad_norm": 0.46875, "learning_rate": 7.41839762611276e-05, "loss": 1.132, "step": 250 }, { "epoch": 0.04, "grad_norm": 0.4609375, "learning_rate": 7.566765578635016e-05, "loss": 1.1438, "step": 255 }, { "epoch": 0.04, "grad_norm": 0.470703125, "learning_rate": 7.71513353115727e-05, "loss": 1.1405, "step": 260 }, { "epoch": 0.04, "grad_norm": 0.447265625, "learning_rate": 7.863501483679525e-05, "loss": 1.1124, "step": 265 }, { "epoch": 0.04, "grad_norm": 0.486328125, "learning_rate": 8.011869436201782e-05, "loss": 1.0813, "step": 270 }, { "epoch": 0.04, "grad_norm": 0.48046875, "learning_rate": 8.160237388724036e-05, "loss": 1.1194, "step": 275 }, { "epoch": 0.04, "grad_norm": 0.462890625, "learning_rate": 8.308605341246291e-05, "loss": 1.0927, "step": 280 }, { "epoch": 0.04, "grad_norm": 0.48046875, "learning_rate": 8.456973293768546e-05, "loss": 1.1277, "step": 285 }, { "epoch": 0.04, "grad_norm": 0.4453125, "learning_rate": 8.605341246290801e-05, "loss": 1.1271, "step": 290 }, { "epoch": 0.04, "grad_norm": 0.435546875, "learning_rate": 8.753709198813057e-05, "loss": 1.1162, "step": 295 }, { "epoch": 0.04, "grad_norm": 0.44921875, "learning_rate": 8.902077151335312e-05, "loss": 1.0857, "step": 300 }, { "epoch": 0.05, "grad_norm": 0.4375, "learning_rate": 9.050445103857568e-05, "loss": 1.0869, "step": 305 }, { "epoch": 0.05, "grad_norm": 0.4453125, "learning_rate": 9.198813056379823e-05, "loss": 1.0655, "step": 310 }, { "epoch": 0.05, "grad_norm": 0.4296875, "learning_rate": 9.347181008902077e-05, "loss": 1.0585, "step": 315 }, { "epoch": 0.05, "grad_norm": 0.41796875, "learning_rate": 9.495548961424333e-05, "loss": 1.1144, "step": 320 }, { "epoch": 0.05, "grad_norm": 0.43359375, "learning_rate": 9.643916913946588e-05, "loss": 1.0719, "step": 325 }, { "epoch": 0.05, "grad_norm": 0.416015625, "learning_rate": 9.792284866468843e-05, "loss": 1.0919, "step": 330 }, { "epoch": 0.05, "grad_norm": 0.423828125, "learning_rate": 9.940652818991099e-05, "loss": 1.1223, "step": 335 }, { "epoch": 0.05, "grad_norm": 0.431640625, "learning_rate": 0.00010089020771513354, "loss": 1.0565, "step": 340 }, { "epoch": 0.05, "grad_norm": 0.431640625, "learning_rate": 0.00010237388724035609, "loss": 1.0962, "step": 345 }, { "epoch": 0.05, "grad_norm": 0.44921875, "learning_rate": 0.00010385756676557865, "loss": 1.0959, "step": 350 }, { "epoch": 0.05, "grad_norm": 0.43359375, "learning_rate": 0.0001053412462908012, "loss": 1.0628, "step": 355 }, { "epoch": 0.05, "grad_norm": 0.431640625, "learning_rate": 0.00010682492581602374, "loss": 1.0975, "step": 360 }, { "epoch": 0.05, "grad_norm": 0.42578125, "learning_rate": 0.0001083086053412463, "loss": 1.0727, "step": 365 }, { "epoch": 0.05, "grad_norm": 0.416015625, "learning_rate": 0.00010979228486646884, "loss": 1.0649, "step": 370 }, { "epoch": 0.06, "grad_norm": 0.4296875, "learning_rate": 0.00011127596439169139, "loss": 1.0904, "step": 375 }, { "epoch": 0.06, "grad_norm": 0.3984375, "learning_rate": 0.00011275964391691397, "loss": 1.079, "step": 380 }, { "epoch": 0.06, "grad_norm": 0.40234375, "learning_rate": 0.0001142433234421365, "loss": 1.0522, "step": 385 }, { "epoch": 0.06, "grad_norm": 0.431640625, "learning_rate": 0.00011572700296735905, "loss": 1.0579, "step": 390 }, { "epoch": 0.06, "grad_norm": 0.396484375, "learning_rate": 0.0001172106824925816, "loss": 1.0871, "step": 395 }, { "epoch": 0.06, "grad_norm": 0.41015625, "learning_rate": 0.00011869436201780415, "loss": 1.0936, "step": 400 }, { "epoch": 0.06, "grad_norm": 0.412109375, "learning_rate": 0.00012017804154302672, "loss": 1.0734, "step": 405 }, { "epoch": 0.06, "grad_norm": 0.392578125, "learning_rate": 0.00012166172106824927, "loss": 1.0657, "step": 410 }, { "epoch": 0.06, "grad_norm": 0.4140625, "learning_rate": 0.00012314540059347182, "loss": 1.0884, "step": 415 }, { "epoch": 0.06, "grad_norm": 0.408203125, "learning_rate": 0.00012462908011869436, "loss": 1.0683, "step": 420 }, { "epoch": 0.06, "grad_norm": 0.3984375, "learning_rate": 0.00012611275964391692, "loss": 1.1073, "step": 425 }, { "epoch": 0.06, "grad_norm": 0.412109375, "learning_rate": 0.00012759643916913948, "loss": 1.0849, "step": 430 }, { "epoch": 0.06, "grad_norm": 0.40625, "learning_rate": 0.00012908011869436204, "loss": 1.0798, "step": 435 }, { "epoch": 0.07, "grad_norm": 0.404296875, "learning_rate": 0.00013056379821958458, "loss": 1.1029, "step": 440 }, { "epoch": 0.07, "grad_norm": 0.3828125, "learning_rate": 0.0001320474777448071, "loss": 1.0664, "step": 445 }, { "epoch": 0.07, "grad_norm": 0.38671875, "learning_rate": 0.00013353115727002967, "loss": 1.0998, "step": 450 }, { "epoch": 0.07, "grad_norm": 0.40234375, "learning_rate": 0.00013501483679525224, "loss": 1.0834, "step": 455 }, { "epoch": 0.07, "grad_norm": 0.400390625, "learning_rate": 0.0001364985163204748, "loss": 1.062, "step": 460 }, { "epoch": 0.07, "grad_norm": 0.3984375, "learning_rate": 0.00013798219584569733, "loss": 1.0825, "step": 465 }, { "epoch": 0.07, "grad_norm": 0.388671875, "learning_rate": 0.0001394658753709199, "loss": 1.0689, "step": 470 }, { "epoch": 0.07, "grad_norm": 0.392578125, "learning_rate": 0.00014094955489614243, "loss": 1.0557, "step": 475 }, { "epoch": 0.07, "grad_norm": 0.380859375, "learning_rate": 0.000142433234421365, "loss": 1.0582, "step": 480 }, { "epoch": 0.07, "grad_norm": 0.380859375, "learning_rate": 0.00014391691394658756, "loss": 1.0921, "step": 485 }, { "epoch": 0.07, "grad_norm": 0.384765625, "learning_rate": 0.0001454005934718101, "loss": 1.0544, "step": 490 }, { "epoch": 0.07, "grad_norm": 0.39453125, "learning_rate": 0.00014688427299703265, "loss": 1.0333, "step": 495 }, { "epoch": 0.07, "grad_norm": 0.384765625, "learning_rate": 0.0001483679525222552, "loss": 1.0454, "step": 500 }, { "epoch": 0.07, "grad_norm": 0.37890625, "learning_rate": 0.00014985163204747775, "loss": 1.0434, "step": 505 }, { "epoch": 0.08, "grad_norm": 0.390625, "learning_rate": 0.0001513353115727003, "loss": 1.0768, "step": 510 }, { "epoch": 0.08, "grad_norm": 0.376953125, "learning_rate": 0.00015281899109792285, "loss": 1.0774, "step": 515 }, { "epoch": 0.08, "grad_norm": 0.365234375, "learning_rate": 0.0001543026706231454, "loss": 1.0438, "step": 520 }, { "epoch": 0.08, "grad_norm": 0.373046875, "learning_rate": 0.00015578635014836794, "loss": 1.0573, "step": 525 }, { "epoch": 0.08, "grad_norm": 0.380859375, "learning_rate": 0.0001572700296735905, "loss": 1.0501, "step": 530 }, { "epoch": 0.08, "grad_norm": 0.392578125, "learning_rate": 0.00015875370919881307, "loss": 1.0413, "step": 535 }, { "epoch": 0.08, "grad_norm": 0.37109375, "learning_rate": 0.00016023738872403563, "loss": 1.0234, "step": 540 }, { "epoch": 0.08, "grad_norm": 0.3671875, "learning_rate": 0.00016172106824925817, "loss": 1.0538, "step": 545 }, { "epoch": 0.08, "grad_norm": 0.41015625, "learning_rate": 0.00016320474777448073, "loss": 1.0493, "step": 550 }, { "epoch": 0.08, "grad_norm": 0.35546875, "learning_rate": 0.00016468842729970326, "loss": 1.0686, "step": 555 }, { "epoch": 0.08, "grad_norm": 0.36328125, "learning_rate": 0.00016617210682492583, "loss": 1.0794, "step": 560 }, { "epoch": 0.08, "grad_norm": 0.380859375, "learning_rate": 0.0001676557863501484, "loss": 1.0575, "step": 565 }, { "epoch": 0.08, "grad_norm": 0.37109375, "learning_rate": 0.00016913946587537092, "loss": 1.0682, "step": 570 }, { "epoch": 0.09, "grad_norm": 0.361328125, "learning_rate": 0.00017062314540059348, "loss": 1.0237, "step": 575 }, { "epoch": 0.09, "grad_norm": 0.37109375, "learning_rate": 0.00017210682492581602, "loss": 1.0493, "step": 580 }, { "epoch": 0.09, "grad_norm": 0.361328125, "learning_rate": 0.00017359050445103858, "loss": 1.0613, "step": 585 }, { "epoch": 0.09, "grad_norm": 0.380859375, "learning_rate": 0.00017507418397626114, "loss": 1.0346, "step": 590 }, { "epoch": 0.09, "grad_norm": 0.375, "learning_rate": 0.00017655786350148368, "loss": 1.0749, "step": 595 }, { "epoch": 0.09, "grad_norm": 0.34765625, "learning_rate": 0.00017804154302670624, "loss": 1.0289, "step": 600 }, { "epoch": 0.09, "grad_norm": 0.353515625, "learning_rate": 0.00017952522255192878, "loss": 1.0606, "step": 605 }, { "epoch": 0.09, "grad_norm": 0.36328125, "learning_rate": 0.00018100890207715137, "loss": 1.0166, "step": 610 }, { "epoch": 0.09, "grad_norm": 0.375, "learning_rate": 0.0001824925816023739, "loss": 1.0347, "step": 615 }, { "epoch": 0.09, "grad_norm": 0.3671875, "learning_rate": 0.00018397626112759646, "loss": 1.0388, "step": 620 }, { "epoch": 0.09, "grad_norm": 0.365234375, "learning_rate": 0.000185459940652819, "loss": 1.0545, "step": 625 }, { "epoch": 0.09, "grad_norm": 0.361328125, "learning_rate": 0.00018694362017804153, "loss": 1.0281, "step": 630 }, { "epoch": 0.09, "grad_norm": 0.34765625, "learning_rate": 0.0001884272997032641, "loss": 1.0347, "step": 635 }, { "epoch": 0.1, "grad_norm": 0.353515625, "learning_rate": 0.00018991097922848666, "loss": 1.0347, "step": 640 }, { "epoch": 0.1, "grad_norm": 0.357421875, "learning_rate": 0.00019139465875370922, "loss": 1.0594, "step": 645 }, { "epoch": 0.1, "grad_norm": 0.353515625, "learning_rate": 0.00019287833827893175, "loss": 1.0201, "step": 650 }, { "epoch": 0.1, "grad_norm": 0.384765625, "learning_rate": 0.00019436201780415432, "loss": 1.0287, "step": 655 }, { "epoch": 0.1, "grad_norm": 0.359375, "learning_rate": 0.00019584569732937685, "loss": 1.0777, "step": 660 }, { "epoch": 0.1, "grad_norm": 0.349609375, "learning_rate": 0.00019732937685459941, "loss": 1.0157, "step": 665 }, { "epoch": 0.1, "grad_norm": 0.349609375, "learning_rate": 0.00019881305637982198, "loss": 1.0474, "step": 670 }, { "epoch": 0.1, "grad_norm": 0.353515625, "learning_rate": 0.0001999999865623139, "loss": 1.0452, "step": 675 }, { "epoch": 0.1, "grad_norm": 0.3828125, "learning_rate": 0.00019999951624367985, "loss": 1.02, "step": 680 }, { "epoch": 0.1, "grad_norm": 0.345703125, "learning_rate": 0.0001999983740443526, "loss": 1.041, "step": 685 }, { "epoch": 0.1, "grad_norm": 0.353515625, "learning_rate": 0.00019999655997200635, "loss": 1.0305, "step": 690 }, { "epoch": 0.1, "grad_norm": 0.359375, "learning_rate": 0.0001999940740388296, "loss": 1.032, "step": 695 }, { "epoch": 0.1, "grad_norm": 0.35546875, "learning_rate": 0.00019999091626152492, "loss": 1.0027, "step": 700 }, { "epoch": 0.1, "grad_norm": 0.349609375, "learning_rate": 0.00019998708666130893, "loss": 1.0311, "step": 705 }, { "epoch": 0.11, "grad_norm": 0.37109375, "learning_rate": 0.00019998258526391207, "loss": 1.0165, "step": 710 }, { "epoch": 0.11, "grad_norm": 0.361328125, "learning_rate": 0.00019997741209957853, "loss": 1.013, "step": 715 }, { "epoch": 0.11, "grad_norm": 0.357421875, "learning_rate": 0.00019997156720306597, "loss": 0.9992, "step": 720 }, { "epoch": 0.11, "grad_norm": 0.34765625, "learning_rate": 0.00019996505061364527, "loss": 1.044, "step": 725 }, { "epoch": 0.11, "grad_norm": 0.55859375, "learning_rate": 0.0001999578623751004, "loss": 0.998, "step": 730 }, { "epoch": 0.11, "grad_norm": 0.349609375, "learning_rate": 0.00019995000253572798, "loss": 1.0354, "step": 735 }, { "epoch": 0.11, "grad_norm": 0.353515625, "learning_rate": 0.00019994147114833698, "loss": 1.0083, "step": 740 }, { "epoch": 0.11, "grad_norm": 0.359375, "learning_rate": 0.00019993226827024843, "loss": 1.012, "step": 745 }, { "epoch": 0.11, "grad_norm": 0.359375, "learning_rate": 0.00019992239396329498, "loss": 0.9953, "step": 750 }, { "epoch": 0.11, "grad_norm": 0.3515625, "learning_rate": 0.00019991184829382057, "loss": 1.0095, "step": 755 }, { "epoch": 0.11, "grad_norm": 0.349609375, "learning_rate": 0.00019990063133267975, "loss": 1.0518, "step": 760 }, { "epoch": 0.11, "grad_norm": 0.369140625, "learning_rate": 0.0001998887431552376, "loss": 1.0357, "step": 765 }, { "epoch": 0.11, "grad_norm": 0.359375, "learning_rate": 0.00019987618384136879, "loss": 1.0542, "step": 770 }, { "epoch": 0.12, "grad_norm": 0.361328125, "learning_rate": 0.0001998629534754574, "loss": 1.0071, "step": 775 }, { "epoch": 0.12, "grad_norm": 0.34375, "learning_rate": 0.00019984905214639608, "loss": 1.0414, "step": 780 }, { "epoch": 0.12, "grad_norm": 0.376953125, "learning_rate": 0.00019983447994758563, "loss": 1.003, "step": 785 }, { "epoch": 0.12, "grad_norm": 0.3515625, "learning_rate": 0.00019981923697693437, "loss": 1.0482, "step": 790 }, { "epoch": 0.12, "grad_norm": 0.361328125, "learning_rate": 0.00019980332333685729, "loss": 1.0394, "step": 795 }, { "epoch": 0.12, "grad_norm": 0.353515625, "learning_rate": 0.00019978673913427568, "loss": 0.985, "step": 800 }, { "epoch": 0.12, "grad_norm": 0.33984375, "learning_rate": 0.00019976948448061603, "loss": 1.0149, "step": 805 }, { "epoch": 0.12, "grad_norm": 0.345703125, "learning_rate": 0.00019975155949180967, "loss": 1.0253, "step": 810 }, { "epoch": 0.12, "grad_norm": 0.34765625, "learning_rate": 0.00019973296428829168, "loss": 1.0067, "step": 815 }, { "epoch": 0.12, "grad_norm": 0.3515625, "learning_rate": 0.00019971369899500024, "loss": 1.013, "step": 820 }, { "epoch": 0.12, "grad_norm": 0.3515625, "learning_rate": 0.00019969376374137578, "loss": 1.0253, "step": 825 }, { "epoch": 0.12, "grad_norm": 0.345703125, "learning_rate": 0.0001996731586613601, "loss": 1.0505, "step": 830 }, { "epoch": 0.12, "grad_norm": 0.3671875, "learning_rate": 0.00019965188389339537, "loss": 1.0199, "step": 835 }, { "epoch": 0.12, "grad_norm": 0.345703125, "learning_rate": 0.00019962993958042336, "loss": 1.0066, "step": 840 }, { "epoch": 0.13, "grad_norm": 0.33203125, "learning_rate": 0.00019960732586988438, "loss": 1.0226, "step": 845 }, { "epoch": 0.13, "grad_norm": 0.345703125, "learning_rate": 0.00019958404291371635, "loss": 0.9902, "step": 850 }, { "epoch": 0.13, "grad_norm": 0.365234375, "learning_rate": 0.0001995600908683537, "loss": 1.0338, "step": 855 }, { "epoch": 0.13, "grad_norm": 0.349609375, "learning_rate": 0.00019953546989472633, "loss": 1.0103, "step": 860 }, { "epoch": 0.13, "grad_norm": 0.3515625, "learning_rate": 0.00019951018015825866, "loss": 1.0237, "step": 865 }, { "epoch": 0.13, "grad_norm": 0.349609375, "learning_rate": 0.00019948422182886833, "loss": 0.972, "step": 870 }, { "epoch": 0.13, "grad_norm": 0.357421875, "learning_rate": 0.00019945759508096527, "loss": 1.0472, "step": 875 }, { "epoch": 0.13, "grad_norm": 0.34765625, "learning_rate": 0.00019943030009345023, "loss": 1.0156, "step": 880 }, { "epoch": 0.13, "grad_norm": 0.3515625, "learning_rate": 0.00019940233704971388, "loss": 1.0365, "step": 885 }, { "epoch": 0.13, "grad_norm": 0.33984375, "learning_rate": 0.00019937370613763543, "loss": 1.0098, "step": 890 }, { "epoch": 0.13, "grad_norm": 0.33203125, "learning_rate": 0.00019934440754958136, "loss": 1.0042, "step": 895 }, { "epoch": 0.13, "grad_norm": 0.33984375, "learning_rate": 0.00019931444148240423, "loss": 1.0053, "step": 900 }, { "epoch": 0.13, "grad_norm": 0.353515625, "learning_rate": 0.00019928380813744119, "loss": 1.0311, "step": 905 }, { "epoch": 0.14, "grad_norm": 0.345703125, "learning_rate": 0.00019925250772051276, "loss": 0.9973, "step": 910 }, { "epoch": 0.14, "grad_norm": 0.353515625, "learning_rate": 0.00019922054044192145, "loss": 1.0023, "step": 915 }, { "epoch": 0.14, "grad_norm": 0.341796875, "learning_rate": 0.00019918790651645023, "loss": 1.0097, "step": 920 }, { "epoch": 0.14, "grad_norm": 0.34375, "learning_rate": 0.00019915460616336126, "loss": 1.013, "step": 925 }, { "epoch": 0.14, "grad_norm": 0.34765625, "learning_rate": 0.0001991206396063942, "loss": 0.9989, "step": 930 }, { "epoch": 0.14, "grad_norm": 0.34375, "learning_rate": 0.00019908600707376495, "loss": 0.9815, "step": 935 }, { "epoch": 0.14, "grad_norm": 0.3515625, "learning_rate": 0.0001990507087981639, "loss": 0.9753, "step": 940 }, { "epoch": 0.14, "grad_norm": 0.33984375, "learning_rate": 0.0001990147450167545, "loss": 1.0145, "step": 945 }, { "epoch": 0.14, "grad_norm": 0.35546875, "learning_rate": 0.00019897811597117168, "loss": 1.0359, "step": 950 }, { "epoch": 0.14, "grad_norm": 0.333984375, "learning_rate": 0.00019894082190751998, "loss": 0.9902, "step": 955 }, { "epoch": 0.14, "grad_norm": 0.349609375, "learning_rate": 0.00019890286307637237, "loss": 0.9997, "step": 960 }, { "epoch": 0.14, "grad_norm": 0.33984375, "learning_rate": 0.0001988642397327681, "loss": 1.0412, "step": 965 }, { "epoch": 0.14, "grad_norm": 0.33984375, "learning_rate": 0.00019882495213621116, "loss": 0.9552, "step": 970 }, { "epoch": 0.14, "grad_norm": 0.3515625, "learning_rate": 0.00019878500055066866, "loss": 0.9901, "step": 975 }, { "epoch": 0.15, "grad_norm": 0.337890625, "learning_rate": 0.00019874438524456888, "loss": 1.035, "step": 980 }, { "epoch": 0.15, "grad_norm": 0.34765625, "learning_rate": 0.0001987031064907995, "loss": 0.996, "step": 985 }, { "epoch": 0.15, "grad_norm": 0.333984375, "learning_rate": 0.0001986611645667059, "loss": 1.0248, "step": 990 }, { "epoch": 0.15, "grad_norm": 0.3515625, "learning_rate": 0.0001986185597540891, "loss": 0.9889, "step": 995 }, { "epoch": 0.15, "grad_norm": 0.341796875, "learning_rate": 0.00019857529233920397, "loss": 1.0151, "step": 1000 }, { "epoch": 0.15, "grad_norm": 0.3671875, "learning_rate": 0.00019853136261275737, "loss": 0.9882, "step": 1005 }, { "epoch": 0.15, "grad_norm": 0.33984375, "learning_rate": 0.00019848677086990605, "loss": 1.0051, "step": 1010 }, { "epoch": 0.15, "grad_norm": 0.34375, "learning_rate": 0.0001984415174102548, "loss": 0.9904, "step": 1015 }, { "epoch": 0.15, "grad_norm": 0.34765625, "learning_rate": 0.0001983956025378543, "loss": 0.9899, "step": 1020 }, { "epoch": 0.15, "grad_norm": 0.341796875, "learning_rate": 0.00019834902656119924, "loss": 0.999, "step": 1025 }, { "epoch": 0.15, "grad_norm": 0.3515625, "learning_rate": 0.00019830178979322614, "loss": 0.9798, "step": 1030 }, { "epoch": 0.15, "grad_norm": 0.349609375, "learning_rate": 0.00019825389255131125, "loss": 0.984, "step": 1035 }, { "epoch": 0.15, "grad_norm": 0.34765625, "learning_rate": 0.0001982053351572684, "loss": 1.0014, "step": 1040 }, { "epoch": 0.16, "grad_norm": 0.345703125, "learning_rate": 0.000198156117937347, "loss": 1.0148, "step": 1045 }, { "epoch": 0.16, "grad_norm": 0.330078125, "learning_rate": 0.0001981062412222296, "loss": 1.0271, "step": 1050 }, { "epoch": 0.16, "grad_norm": 0.341796875, "learning_rate": 0.00019805570534702987, "loss": 0.9902, "step": 1055 }, { "epoch": 0.16, "grad_norm": 0.330078125, "learning_rate": 0.00019800451065129018, "loss": 0.9699, "step": 1060 }, { "epoch": 0.16, "grad_norm": 0.337890625, "learning_rate": 0.00019795265747897956, "loss": 0.9911, "step": 1065 }, { "epoch": 0.16, "grad_norm": 0.345703125, "learning_rate": 0.00019790014617849106, "loss": 0.9923, "step": 1070 }, { "epoch": 0.16, "grad_norm": 0.341796875, "learning_rate": 0.00019784697710263974, "loss": 0.9976, "step": 1075 }, { "epoch": 0.16, "grad_norm": 0.3515625, "learning_rate": 0.00019779315060866007, "loss": 0.9647, "step": 1080 }, { "epoch": 0.16, "grad_norm": 0.353515625, "learning_rate": 0.00019773866705820363, "loss": 0.9589, "step": 1085 }, { "epoch": 0.16, "grad_norm": 0.330078125, "learning_rate": 0.00019768352681733662, "loss": 1.005, "step": 1090 }, { "epoch": 0.16, "grad_norm": 0.341796875, "learning_rate": 0.00019762773025653747, "loss": 1.0066, "step": 1095 }, { "epoch": 0.16, "grad_norm": 0.345703125, "learning_rate": 0.0001975712777506943, "loss": 1.0118, "step": 1100 }, { "epoch": 0.16, "grad_norm": 0.34375, "learning_rate": 0.00019751416967910248, "loss": 1.0445, "step": 1105 }, { "epoch": 0.16, "grad_norm": 0.34765625, "learning_rate": 0.00019745640642546196, "loss": 0.9766, "step": 1110 }, { "epoch": 0.17, "grad_norm": 0.353515625, "learning_rate": 0.0001973979883778747, "loss": 0.9897, "step": 1115 }, { "epoch": 0.17, "grad_norm": 0.3515625, "learning_rate": 0.00019733891592884227, "loss": 0.9826, "step": 1120 }, { "epoch": 0.17, "grad_norm": 0.34765625, "learning_rate": 0.00019727918947526292, "loss": 0.988, "step": 1125 }, { "epoch": 0.17, "grad_norm": 0.349609375, "learning_rate": 0.00019721880941842913, "loss": 1.0101, "step": 1130 }, { "epoch": 0.17, "grad_norm": 0.34375, "learning_rate": 0.00019715777616402479, "loss": 0.9744, "step": 1135 }, { "epoch": 0.17, "grad_norm": 0.33984375, "learning_rate": 0.0001970960901221225, "loss": 0.9896, "step": 1140 }, { "epoch": 0.17, "grad_norm": 0.357421875, "learning_rate": 0.00019703375170718093, "loss": 0.9786, "step": 1145 }, { "epoch": 0.17, "grad_norm": 0.34765625, "learning_rate": 0.00019697076133804185, "loss": 0.9958, "step": 1150 }, { "epoch": 0.17, "grad_norm": 0.341796875, "learning_rate": 0.0001969071194379275, "loss": 0.9803, "step": 1155 }, { "epoch": 0.17, "grad_norm": 0.3515625, "learning_rate": 0.00019684282643443748, "loss": 0.9894, "step": 1160 }, { "epoch": 0.17, "grad_norm": 0.357421875, "learning_rate": 0.00019677788275954624, "loss": 1.0073, "step": 1165 }, { "epoch": 0.17, "grad_norm": 0.33984375, "learning_rate": 0.00019671228884959987, "loss": 1.0145, "step": 1170 }, { "epoch": 0.17, "grad_norm": 0.357421875, "learning_rate": 0.00019664604514531332, "loss": 1.0035, "step": 1175 }, { "epoch": 0.18, "grad_norm": 0.353515625, "learning_rate": 0.0001965791520917674, "loss": 0.9856, "step": 1180 }, { "epoch": 0.18, "grad_norm": 0.33984375, "learning_rate": 0.00019651161013840583, "loss": 0.9573, "step": 1185 }, { "epoch": 0.18, "grad_norm": 0.328125, "learning_rate": 0.00019644341973903208, "loss": 0.9824, "step": 1190 }, { "epoch": 0.18, "grad_norm": 0.333984375, "learning_rate": 0.00019637458135180657, "loss": 0.9873, "step": 1195 }, { "epoch": 0.18, "grad_norm": 0.3515625, "learning_rate": 0.0001963050954392433, "loss": 0.9743, "step": 1200 }, { "epoch": 0.18, "grad_norm": 0.345703125, "learning_rate": 0.00019623496246820704, "loss": 0.9987, "step": 1205 }, { "epoch": 0.18, "grad_norm": 0.349609375, "learning_rate": 0.00019616418290990993, "loss": 1.0153, "step": 1210 }, { "epoch": 0.18, "grad_norm": 0.39453125, "learning_rate": 0.0001960927572399084, "loss": 0.9907, "step": 1215 }, { "epoch": 0.18, "grad_norm": 0.337890625, "learning_rate": 0.00019602068593810014, "loss": 1.0148, "step": 1220 }, { "epoch": 0.18, "grad_norm": 0.34375, "learning_rate": 0.0001959479694887206, "loss": 1.0024, "step": 1225 }, { "epoch": 0.18, "grad_norm": 0.34765625, "learning_rate": 0.00019587460838033996, "loss": 1.0298, "step": 1230 }, { "epoch": 0.18, "grad_norm": 0.357421875, "learning_rate": 0.00019580060310585973, "loss": 0.9946, "step": 1235 }, { "epoch": 0.18, "grad_norm": 0.349609375, "learning_rate": 0.00019572595416250942, "loss": 0.9961, "step": 1240 }, { "epoch": 0.18, "grad_norm": 0.333984375, "learning_rate": 0.00019565066205184332, "loss": 1.0016, "step": 1245 }, { "epoch": 0.19, "grad_norm": 0.353515625, "learning_rate": 0.00019557472727973707, "loss": 0.9829, "step": 1250 }, { "epoch": 0.19, "grad_norm": 0.34765625, "learning_rate": 0.00019549815035638414, "loss": 0.9781, "step": 1255 }, { "epoch": 0.19, "grad_norm": 0.353515625, "learning_rate": 0.00019542093179629268, "loss": 1.0089, "step": 1260 }, { "epoch": 0.19, "grad_norm": 0.34375, "learning_rate": 0.0001953430721182817, "loss": 0.9942, "step": 1265 }, { "epoch": 0.19, "grad_norm": 0.357421875, "learning_rate": 0.00019526457184547793, "loss": 0.9818, "step": 1270 }, { "epoch": 0.19, "grad_norm": 0.359375, "learning_rate": 0.00019518543150531207, "loss": 1.018, "step": 1275 }, { "epoch": 0.19, "grad_norm": 0.345703125, "learning_rate": 0.00019510565162951537, "loss": 0.9697, "step": 1280 }, { "epoch": 0.19, "grad_norm": 0.359375, "learning_rate": 0.00019502523275411599, "loss": 1.0331, "step": 1285 }, { "epoch": 0.19, "grad_norm": 0.34375, "learning_rate": 0.00019494417541943547, "loss": 1.0071, "step": 1290 }, { "epoch": 0.19, "grad_norm": 0.345703125, "learning_rate": 0.00019486248017008503, "loss": 0.9699, "step": 1295 }, { "epoch": 0.19, "grad_norm": 0.34765625, "learning_rate": 0.00019478014755496196, "loss": 1.0168, "step": 1300 }, { "epoch": 0.19, "grad_norm": 0.345703125, "learning_rate": 0.00019469717812724575, "loss": 0.9719, "step": 1305 }, { "epoch": 0.19, "grad_norm": 0.345703125, "learning_rate": 0.00019461357244439479, "loss": 0.9974, "step": 1310 }, { "epoch": 0.2, "grad_norm": 0.34375, "learning_rate": 0.00019452933106814223, "loss": 0.9897, "step": 1315 }, { "epoch": 0.2, "grad_norm": 0.345703125, "learning_rate": 0.0001944444545644923, "loss": 0.9809, "step": 1320 }, { "epoch": 0.2, "grad_norm": 0.341796875, "learning_rate": 0.0001943589435037166, "loss": 0.9601, "step": 1325 }, { "epoch": 0.2, "grad_norm": 0.345703125, "learning_rate": 0.00019427279846035025, "loss": 0.9615, "step": 1330 }, { "epoch": 0.2, "grad_norm": 0.337890625, "learning_rate": 0.00019418602001318797, "loss": 0.9888, "step": 1335 }, { "epoch": 0.2, "grad_norm": 0.341796875, "learning_rate": 0.00019409860874528017, "loss": 1.0099, "step": 1340 }, { "epoch": 0.2, "grad_norm": 0.333984375, "learning_rate": 0.00019401056524392916, "loss": 0.9911, "step": 1345 }, { "epoch": 0.2, "grad_norm": 0.3515625, "learning_rate": 0.00019392189010068508, "loss": 0.992, "step": 1350 }, { "epoch": 0.2, "grad_norm": 0.337890625, "learning_rate": 0.000193832583911342, "loss": 0.9827, "step": 1355 }, { "epoch": 0.2, "grad_norm": 0.3515625, "learning_rate": 0.0001937426472759338, "loss": 0.9473, "step": 1360 }, { "epoch": 0.2, "grad_norm": 0.3671875, "learning_rate": 0.00019365208079873036, "loss": 0.9863, "step": 1365 }, { "epoch": 0.2, "grad_norm": 0.349609375, "learning_rate": 0.0001935608850882333, "loss": 0.9662, "step": 1370 }, { "epoch": 0.2, "grad_norm": 0.3515625, "learning_rate": 0.0001934690607571719, "loss": 0.9864, "step": 1375 }, { "epoch": 0.2, "grad_norm": 0.345703125, "learning_rate": 0.00019337660842249914, "loss": 1.0151, "step": 1380 }, { "epoch": 0.21, "grad_norm": 0.337890625, "learning_rate": 0.0001932835287053874, "loss": 0.9638, "step": 1385 }, { "epoch": 0.21, "grad_norm": 0.341796875, "learning_rate": 0.00019318982223122437, "loss": 1.0149, "step": 1390 }, { "epoch": 0.21, "grad_norm": 0.341796875, "learning_rate": 0.00019309548962960876, "loss": 0.9827, "step": 1395 }, { "epoch": 0.21, "grad_norm": 0.349609375, "learning_rate": 0.00019300053153434622, "loss": 0.9726, "step": 1400 }, { "epoch": 0.21, "grad_norm": 0.33984375, "learning_rate": 0.00019290494858344493, "loss": 0.9742, "step": 1405 }, { "epoch": 0.21, "grad_norm": 0.345703125, "learning_rate": 0.00019280874141911137, "loss": 0.9987, "step": 1410 }, { "epoch": 0.21, "grad_norm": 0.349609375, "learning_rate": 0.00019271191068774606, "loss": 1.0067, "step": 1415 }, { "epoch": 0.21, "grad_norm": 0.33984375, "learning_rate": 0.00019261445703993912, "loss": 0.9833, "step": 1420 }, { "epoch": 0.21, "grad_norm": 0.35546875, "learning_rate": 0.00019251638113046597, "loss": 0.9599, "step": 1425 }, { "epoch": 0.21, "grad_norm": 0.3671875, "learning_rate": 0.0001924176836182829, "loss": 0.9793, "step": 1430 }, { "epoch": 0.21, "grad_norm": 0.337890625, "learning_rate": 0.00019231836516652261, "loss": 0.965, "step": 1435 }, { "epoch": 0.21, "grad_norm": 0.353515625, "learning_rate": 0.0001922184264424899, "loss": 0.9598, "step": 1440 }, { "epoch": 0.21, "grad_norm": 0.3359375, "learning_rate": 0.00019211786811765692, "loss": 0.9931, "step": 1445 }, { "epoch": 0.22, "grad_norm": 0.345703125, "learning_rate": 0.00019201669086765902, "loss": 0.9571, "step": 1450 }, { "epoch": 0.22, "grad_norm": 0.328125, "learning_rate": 0.0001919148953722898, "loss": 0.9448, "step": 1455 }, { "epoch": 0.22, "grad_norm": 0.328125, "learning_rate": 0.0001918124823154969, "loss": 0.9523, "step": 1460 }, { "epoch": 0.22, "grad_norm": 0.34765625, "learning_rate": 0.00019170945238537718, "loss": 0.9888, "step": 1465 }, { "epoch": 0.22, "grad_norm": 0.34765625, "learning_rate": 0.00019160580627417223, "loss": 0.9735, "step": 1470 }, { "epoch": 0.22, "grad_norm": 0.337890625, "learning_rate": 0.00019150154467826357, "loss": 0.9726, "step": 1475 }, { "epoch": 0.22, "grad_norm": 0.34765625, "learning_rate": 0.00019139666829816817, "loss": 0.9679, "step": 1480 }, { "epoch": 0.22, "grad_norm": 0.33203125, "learning_rate": 0.0001912911778385336, "loss": 0.9913, "step": 1485 }, { "epoch": 0.22, "grad_norm": 0.33203125, "learning_rate": 0.00019118507400813325, "loss": 0.9736, "step": 1490 }, { "epoch": 0.22, "grad_norm": 0.337890625, "learning_rate": 0.0001910783575198618, "loss": 0.9535, "step": 1495 }, { "epoch": 0.22, "grad_norm": 0.341796875, "learning_rate": 0.0001909710290907302, "loss": 0.993, "step": 1500 }, { "epoch": 0.22, "grad_norm": 0.3515625, "learning_rate": 0.00019086308944186084, "loss": 0.9645, "step": 1505 }, { "epoch": 0.22, "grad_norm": 0.33984375, "learning_rate": 0.000190754539298483, "loss": 0.9694, "step": 1510 }, { "epoch": 0.22, "grad_norm": 0.337890625, "learning_rate": 0.00019064537938992757, "loss": 0.9775, "step": 1515 }, { "epoch": 0.23, "grad_norm": 0.34375, "learning_rate": 0.0001905356104496225, "loss": 0.9388, "step": 1520 }, { "epoch": 0.23, "grad_norm": 0.357421875, "learning_rate": 0.00019042523321508768, "loss": 0.9755, "step": 1525 }, { "epoch": 0.23, "grad_norm": 0.359375, "learning_rate": 0.00019031424842793, "loss": 1.0159, "step": 1530 }, { "epoch": 0.23, "grad_norm": 0.3515625, "learning_rate": 0.00019020265683383842, "loss": 0.9835, "step": 1535 }, { "epoch": 0.23, "grad_norm": 0.349609375, "learning_rate": 0.000190090459182579, "loss": 0.9711, "step": 1540 }, { "epoch": 0.23, "grad_norm": 0.341796875, "learning_rate": 0.00018997765622798967, "loss": 0.9742, "step": 1545 }, { "epoch": 0.23, "grad_norm": 0.33984375, "learning_rate": 0.0001898642487279754, "loss": 0.9669, "step": 1550 }, { "epoch": 0.23, "grad_norm": 0.3359375, "learning_rate": 0.000189750237444503, "loss": 0.9516, "step": 1555 }, { "epoch": 0.23, "grad_norm": 0.349609375, "learning_rate": 0.00018963562314359595, "loss": 1.0002, "step": 1560 }, { "epoch": 0.23, "grad_norm": 0.341796875, "learning_rate": 0.00018952040659532936, "loss": 0.937, "step": 1565 }, { "epoch": 0.23, "grad_norm": 0.33984375, "learning_rate": 0.00018940458857382467, "loss": 0.9757, "step": 1570 }, { "epoch": 0.23, "grad_norm": 0.333984375, "learning_rate": 0.00018928816985724458, "loss": 0.9968, "step": 1575 }, { "epoch": 0.23, "grad_norm": 0.3359375, "learning_rate": 0.0001891711512277878, "loss": 0.974, "step": 1580 }, { "epoch": 0.24, "grad_norm": 0.34765625, "learning_rate": 0.00018905353347168366, "loss": 0.9641, "step": 1585 }, { "epoch": 0.24, "grad_norm": 0.33984375, "learning_rate": 0.00018893531737918702, "loss": 0.9799, "step": 1590 }, { "epoch": 0.24, "grad_norm": 0.3359375, "learning_rate": 0.0001888165037445728, "loss": 0.9385, "step": 1595 }, { "epoch": 0.24, "grad_norm": 0.357421875, "learning_rate": 0.00018869709336613073, "loss": 1.0199, "step": 1600 } ], "logging_steps": 5, "max_steps": 6734, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 100, "total_flos": 2.2500144814112113e+18, "train_batch_size": 2, "trial_name": null, "trial_params": null }