| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 1.835493862567397, |
| "eval_steps": 10, |
| "global_step": 16000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.001147183664104623, |
| "eval_loss": 1.873344898223877, |
| "eval_runtime": 12.7647, |
| "eval_samples_per_second": 515.797, |
| "eval_steps_per_second": 8.069, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.002294367328209246, |
| "eval_loss": 1.8726389408111572, |
| "eval_runtime": 12.8667, |
| "eval_samples_per_second": 511.709, |
| "eval_steps_per_second": 8.005, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.0034415509923138693, |
| "eval_loss": 1.8714078664779663, |
| "eval_runtime": 12.9103, |
| "eval_samples_per_second": 509.979, |
| "eval_steps_per_second": 7.978, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.004588734656418492, |
| "eval_loss": 1.8696790933609009, |
| "eval_runtime": 12.947, |
| "eval_samples_per_second": 508.534, |
| "eval_steps_per_second": 7.955, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.0057359183205231154, |
| "eval_loss": 1.8675329685211182, |
| "eval_runtime": 12.9458, |
| "eval_samples_per_second": 508.582, |
| "eval_steps_per_second": 7.956, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.006883101984627739, |
| "eval_loss": 1.8649154901504517, |
| "eval_runtime": 13.0432, |
| "eval_samples_per_second": 504.785, |
| "eval_steps_per_second": 7.897, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.008030285648732363, |
| "eval_loss": 1.8619294166564941, |
| "eval_runtime": 13.0638, |
| "eval_samples_per_second": 503.988, |
| "eval_steps_per_second": 7.884, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.009177469312836984, |
| "eval_loss": 1.8583979606628418, |
| "eval_runtime": 13.0482, |
| "eval_samples_per_second": 504.592, |
| "eval_steps_per_second": 7.894, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.010324652976941608, |
| "eval_loss": 1.85438871383667, |
| "eval_runtime": 13.0615, |
| "eval_samples_per_second": 504.075, |
| "eval_steps_per_second": 7.886, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.011471836641046231, |
| "grad_norm": 9.938580513000488, |
| "learning_rate": 3.8226299694189603e-07, |
| "loss": 3.1046, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.011471836641046231, |
| "eval_loss": 1.849947214126587, |
| "eval_runtime": 13.0663, |
| "eval_samples_per_second": 503.89, |
| "eval_steps_per_second": 7.883, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.012619020305150854, |
| "eval_loss": 1.8451412916183472, |
| "eval_runtime": 12.9771, |
| "eval_samples_per_second": 507.357, |
| "eval_steps_per_second": 7.937, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.013766203969255477, |
| "eval_loss": 1.8399487733840942, |
| "eval_runtime": 13.0209, |
| "eval_samples_per_second": 505.648, |
| "eval_steps_per_second": 7.91, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.0149133876333601, |
| "eval_loss": 1.8342881202697754, |
| "eval_runtime": 13.0369, |
| "eval_samples_per_second": 505.028, |
| "eval_steps_per_second": 7.901, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.016060571297464726, |
| "eval_loss": 1.8283486366271973, |
| "eval_runtime": 13.0149, |
| "eval_samples_per_second": 505.88, |
| "eval_steps_per_second": 7.914, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.017207754961569347, |
| "eval_loss": 1.822334885597229, |
| "eval_runtime": 13.0213, |
| "eval_samples_per_second": 505.632, |
| "eval_steps_per_second": 7.91, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.01835493862567397, |
| "eval_loss": 1.8158738613128662, |
| "eval_runtime": 13.0599, |
| "eval_samples_per_second": 504.14, |
| "eval_steps_per_second": 7.887, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.019502122289778594, |
| "eval_loss": 1.8090614080429077, |
| "eval_runtime": 13.034, |
| "eval_samples_per_second": 505.14, |
| "eval_steps_per_second": 7.902, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.020649305953883215, |
| "eval_loss": 1.8015782833099365, |
| "eval_runtime": 13.0665, |
| "eval_samples_per_second": 503.885, |
| "eval_steps_per_second": 7.883, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.02179648961798784, |
| "eval_loss": 1.793796420097351, |
| "eval_runtime": 13.0555, |
| "eval_samples_per_second": 504.31, |
| "eval_steps_per_second": 7.889, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.022943673282092462, |
| "grad_norm": 4.906337738037109, |
| "learning_rate": 7.645259938837921e-07, |
| "loss": 3.0303, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.022943673282092462, |
| "eval_loss": 1.785815715789795, |
| "eval_runtime": 12.9925, |
| "eval_samples_per_second": 506.754, |
| "eval_steps_per_second": 7.928, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.024090856946197087, |
| "eval_loss": 1.7775053977966309, |
| "eval_runtime": 13.0639, |
| "eval_samples_per_second": 503.986, |
| "eval_steps_per_second": 7.884, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.025238040610301708, |
| "eval_loss": 1.7692992687225342, |
| "eval_runtime": 13.0129, |
| "eval_samples_per_second": 505.96, |
| "eval_steps_per_second": 7.915, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.026385224274406333, |
| "eval_loss": 1.760453224182129, |
| "eval_runtime": 13.0078, |
| "eval_samples_per_second": 506.158, |
| "eval_steps_per_second": 7.918, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.027532407938510955, |
| "eval_loss": 1.751396656036377, |
| "eval_runtime": 12.9957, |
| "eval_samples_per_second": 506.628, |
| "eval_steps_per_second": 7.926, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.02867959160261558, |
| "eval_loss": 1.7417218685150146, |
| "eval_runtime": 12.9774, |
| "eval_samples_per_second": 507.344, |
| "eval_steps_per_second": 7.937, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.0298267752667202, |
| "eval_loss": 1.7319914102554321, |
| "eval_runtime": 13.0219, |
| "eval_samples_per_second": 505.611, |
| "eval_steps_per_second": 7.91, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.030973958930824826, |
| "eval_loss": 1.7227253913879395, |
| "eval_runtime": 13.0026, |
| "eval_samples_per_second": 506.361, |
| "eval_steps_per_second": 7.922, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.03212114259492945, |
| "eval_loss": 1.7133797407150269, |
| "eval_runtime": 12.9757, |
| "eval_samples_per_second": 507.409, |
| "eval_steps_per_second": 7.938, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.03326832625903407, |
| "eval_loss": 1.704041600227356, |
| "eval_runtime": 12.9845, |
| "eval_samples_per_second": 507.065, |
| "eval_steps_per_second": 7.933, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.034415509923138694, |
| "grad_norm": 4.665822505950928, |
| "learning_rate": 1.1467889908256882e-06, |
| "loss": 2.9459, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.034415509923138694, |
| "eval_loss": 1.6940686702728271, |
| "eval_runtime": 13.0019, |
| "eval_samples_per_second": 506.387, |
| "eval_steps_per_second": 7.922, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.035562693587243316, |
| "eval_loss": 1.683342695236206, |
| "eval_runtime": 13.0065, |
| "eval_samples_per_second": 506.209, |
| "eval_steps_per_second": 7.919, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.03670987725134794, |
| "eval_loss": 1.6724653244018555, |
| "eval_runtime": 13.0129, |
| "eval_samples_per_second": 505.96, |
| "eval_steps_per_second": 7.915, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.037857060915452566, |
| "eval_loss": 1.6614341735839844, |
| "eval_runtime": 12.9921, |
| "eval_samples_per_second": 506.769, |
| "eval_steps_per_second": 7.928, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.03900424457955719, |
| "eval_loss": 1.6510112285614014, |
| "eval_runtime": 13.0242, |
| "eval_samples_per_second": 505.52, |
| "eval_steps_per_second": 7.908, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.04015142824366181, |
| "eval_loss": 1.6401513814926147, |
| "eval_runtime": 12.9214, |
| "eval_samples_per_second": 509.542, |
| "eval_steps_per_second": 7.971, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.04129861190776643, |
| "eval_loss": 1.6295816898345947, |
| "eval_runtime": 12.9563, |
| "eval_samples_per_second": 508.171, |
| "eval_steps_per_second": 7.95, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.04244579557187106, |
| "eval_loss": 1.6187150478363037, |
| "eval_runtime": 12.9758, |
| "eval_samples_per_second": 507.405, |
| "eval_steps_per_second": 7.938, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.04359297923597568, |
| "eval_loss": 1.607272982597351, |
| "eval_runtime": 12.9876, |
| "eval_samples_per_second": 506.947, |
| "eval_steps_per_second": 7.931, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.0447401629000803, |
| "eval_loss": 1.5961676836013794, |
| "eval_runtime": 12.9782, |
| "eval_samples_per_second": 507.313, |
| "eval_steps_per_second": 7.936, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.045887346564184923, |
| "grad_norm": 4.870114326477051, |
| "learning_rate": 1.5290519877675841e-06, |
| "loss": 2.7813, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.045887346564184923, |
| "eval_loss": 1.5848218202590942, |
| "eval_runtime": 12.9783, |
| "eval_samples_per_second": 507.309, |
| "eval_steps_per_second": 7.936, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.04703453022828955, |
| "eval_loss": 1.5734797716140747, |
| "eval_runtime": 12.9739, |
| "eval_samples_per_second": 507.482, |
| "eval_steps_per_second": 7.939, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.04818171389239417, |
| "eval_loss": 1.562021255493164, |
| "eval_runtime": 12.9388, |
| "eval_samples_per_second": 508.855, |
| "eval_steps_per_second": 7.961, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.049328897556498795, |
| "eval_loss": 1.5495364665985107, |
| "eval_runtime": 12.9412, |
| "eval_samples_per_second": 508.764, |
| "eval_steps_per_second": 7.959, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.050476081220603417, |
| "eval_loss": 1.5375314950942993, |
| "eval_runtime": 12.9686, |
| "eval_samples_per_second": 507.687, |
| "eval_steps_per_second": 7.942, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.051623264884708045, |
| "eval_loss": 1.525598168373108, |
| "eval_runtime": 12.9695, |
| "eval_samples_per_second": 507.651, |
| "eval_steps_per_second": 7.942, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.052770448548812667, |
| "eval_loss": 1.5132672786712646, |
| "eval_runtime": 12.8961, |
| "eval_samples_per_second": 510.543, |
| "eval_steps_per_second": 7.987, |
| "step": 460 |
| }, |
| { |
| "epoch": 0.05391763221291729, |
| "eval_loss": 1.5012215375900269, |
| "eval_runtime": 12.9428, |
| "eval_samples_per_second": 508.7, |
| "eval_steps_per_second": 7.958, |
| "step": 470 |
| }, |
| { |
| "epoch": 0.05506481587702191, |
| "eval_loss": 1.4892219305038452, |
| "eval_runtime": 12.9208, |
| "eval_samples_per_second": 509.567, |
| "eval_steps_per_second": 7.972, |
| "step": 480 |
| }, |
| { |
| "epoch": 0.05621199954112653, |
| "eval_loss": 1.4768636226654053, |
| "eval_runtime": 12.9423, |
| "eval_samples_per_second": 508.721, |
| "eval_steps_per_second": 7.958, |
| "step": 490 |
| }, |
| { |
| "epoch": 0.05735918320523116, |
| "grad_norm": 4.155641555786133, |
| "learning_rate": 1.9113149847094803e-06, |
| "loss": 2.6308, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.05735918320523116, |
| "eval_loss": 1.4640088081359863, |
| "eval_runtime": 12.8729, |
| "eval_samples_per_second": 511.462, |
| "eval_steps_per_second": 8.001, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.05850636686933578, |
| "eval_loss": 1.4513096809387207, |
| "eval_runtime": 12.9653, |
| "eval_samples_per_second": 507.817, |
| "eval_steps_per_second": 7.944, |
| "step": 510 |
| }, |
| { |
| "epoch": 0.0596535505334404, |
| "eval_loss": 1.439149260520935, |
| "eval_runtime": 12.9443, |
| "eval_samples_per_second": 508.639, |
| "eval_steps_per_second": 7.957, |
| "step": 520 |
| }, |
| { |
| "epoch": 0.060800734197545024, |
| "eval_loss": 1.426237940788269, |
| "eval_runtime": 12.9496, |
| "eval_samples_per_second": 508.433, |
| "eval_steps_per_second": 7.954, |
| "step": 530 |
| }, |
| { |
| "epoch": 0.06194791786164965, |
| "eval_loss": 1.4129557609558105, |
| "eval_runtime": 12.9822, |
| "eval_samples_per_second": 507.157, |
| "eval_steps_per_second": 7.934, |
| "step": 540 |
| }, |
| { |
| "epoch": 0.06309510152575427, |
| "eval_loss": 1.3997886180877686, |
| "eval_runtime": 12.9979, |
| "eval_samples_per_second": 506.542, |
| "eval_steps_per_second": 7.924, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.0642422851898589, |
| "eval_loss": 1.3873906135559082, |
| "eval_runtime": 12.9378, |
| "eval_samples_per_second": 508.895, |
| "eval_steps_per_second": 7.961, |
| "step": 560 |
| }, |
| { |
| "epoch": 0.06538946885396352, |
| "eval_loss": 1.3751789331436157, |
| "eval_runtime": 12.9624, |
| "eval_samples_per_second": 507.932, |
| "eval_steps_per_second": 7.946, |
| "step": 570 |
| }, |
| { |
| "epoch": 0.06653665251806815, |
| "eval_loss": 1.3620370626449585, |
| "eval_runtime": 12.9498, |
| "eval_samples_per_second": 508.426, |
| "eval_steps_per_second": 7.954, |
| "step": 580 |
| }, |
| { |
| "epoch": 0.06768383618217276, |
| "eval_loss": 1.3485124111175537, |
| "eval_runtime": 12.9759, |
| "eval_samples_per_second": 507.403, |
| "eval_steps_per_second": 7.938, |
| "step": 590 |
| }, |
| { |
| "epoch": 0.06883101984627739, |
| "grad_norm": 5.262124061584473, |
| "learning_rate": 2.2935779816513764e-06, |
| "loss": 2.4452, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.06883101984627739, |
| "eval_loss": 1.3349775075912476, |
| "eval_runtime": 12.9631, |
| "eval_samples_per_second": 507.902, |
| "eval_steps_per_second": 7.946, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.06997820351038202, |
| "eval_loss": 1.3213400840759277, |
| "eval_runtime": 12.9619, |
| "eval_samples_per_second": 507.951, |
| "eval_steps_per_second": 7.946, |
| "step": 610 |
| }, |
| { |
| "epoch": 0.07112538717448663, |
| "eval_loss": 1.308822512626648, |
| "eval_runtime": 12.9652, |
| "eval_samples_per_second": 507.82, |
| "eval_steps_per_second": 7.944, |
| "step": 620 |
| }, |
| { |
| "epoch": 0.07227257083859126, |
| "eval_loss": 1.296485185623169, |
| "eval_runtime": 13.0441, |
| "eval_samples_per_second": 504.75, |
| "eval_steps_per_second": 7.896, |
| "step": 630 |
| }, |
| { |
| "epoch": 0.07341975450269587, |
| "eval_loss": 1.283867597579956, |
| "eval_runtime": 12.9822, |
| "eval_samples_per_second": 507.154, |
| "eval_steps_per_second": 7.934, |
| "step": 640 |
| }, |
| { |
| "epoch": 0.0745669381668005, |
| "eval_loss": 1.2713148593902588, |
| "eval_runtime": 12.9775, |
| "eval_samples_per_second": 507.338, |
| "eval_steps_per_second": 7.937, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.07571412183090513, |
| "eval_loss": 1.2591922283172607, |
| "eval_runtime": 13.0227, |
| "eval_samples_per_second": 505.578, |
| "eval_steps_per_second": 7.909, |
| "step": 660 |
| }, |
| { |
| "epoch": 0.07686130549500975, |
| "eval_loss": 1.246610164642334, |
| "eval_runtime": 13.014, |
| "eval_samples_per_second": 505.917, |
| "eval_steps_per_second": 7.915, |
| "step": 670 |
| }, |
| { |
| "epoch": 0.07800848915911437, |
| "eval_loss": 1.2331972122192383, |
| "eval_runtime": 12.9634, |
| "eval_samples_per_second": 507.891, |
| "eval_steps_per_second": 7.945, |
| "step": 680 |
| }, |
| { |
| "epoch": 0.079155672823219, |
| "eval_loss": 1.2203081846237183, |
| "eval_runtime": 12.9664, |
| "eval_samples_per_second": 507.775, |
| "eval_steps_per_second": 7.944, |
| "step": 690 |
| }, |
| { |
| "epoch": 0.08030285648732362, |
| "grad_norm": 3.824066400527954, |
| "learning_rate": 2.6758409785932725e-06, |
| "loss": 2.2626, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.08030285648732362, |
| "eval_loss": 1.207729697227478, |
| "eval_runtime": 13.037, |
| "eval_samples_per_second": 505.025, |
| "eval_steps_per_second": 7.901, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.08145004015142825, |
| "eval_loss": 1.195885419845581, |
| "eval_runtime": 12.9625, |
| "eval_samples_per_second": 507.928, |
| "eval_steps_per_second": 7.946, |
| "step": 710 |
| }, |
| { |
| "epoch": 0.08259722381553286, |
| "eval_loss": 1.1840639114379883, |
| "eval_runtime": 13.0124, |
| "eval_samples_per_second": 505.98, |
| "eval_steps_per_second": 7.916, |
| "step": 720 |
| }, |
| { |
| "epoch": 0.08374440747963749, |
| "eval_loss": 1.1725258827209473, |
| "eval_runtime": 12.9851, |
| "eval_samples_per_second": 507.043, |
| "eval_steps_per_second": 7.932, |
| "step": 730 |
| }, |
| { |
| "epoch": 0.08489159114374212, |
| "eval_loss": 1.1619255542755127, |
| "eval_runtime": 13.1041, |
| "eval_samples_per_second": 502.438, |
| "eval_steps_per_second": 7.86, |
| "step": 740 |
| }, |
| { |
| "epoch": 0.08603877480784673, |
| "eval_loss": 1.1515777111053467, |
| "eval_runtime": 12.9648, |
| "eval_samples_per_second": 507.837, |
| "eval_steps_per_second": 7.945, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.08718595847195136, |
| "eval_loss": 1.1416434049606323, |
| "eval_runtime": 13.0349, |
| "eval_samples_per_second": 505.104, |
| "eval_steps_per_second": 7.902, |
| "step": 760 |
| }, |
| { |
| "epoch": 0.08833314213605599, |
| "eval_loss": 1.1320044994354248, |
| "eval_runtime": 13.0667, |
| "eval_samples_per_second": 503.875, |
| "eval_steps_per_second": 7.883, |
| "step": 770 |
| }, |
| { |
| "epoch": 0.0894803258001606, |
| "eval_loss": 1.1226693391799927, |
| "eval_runtime": 13.3034, |
| "eval_samples_per_second": 494.912, |
| "eval_steps_per_second": 7.742, |
| "step": 780 |
| }, |
| { |
| "epoch": 0.09062750946426523, |
| "eval_loss": 1.113792061805725, |
| "eval_runtime": 12.9793, |
| "eval_samples_per_second": 507.271, |
| "eval_steps_per_second": 7.936, |
| "step": 790 |
| }, |
| { |
| "epoch": 0.09177469312836985, |
| "grad_norm": 3.358140707015991, |
| "learning_rate": 3.0581039755351682e-06, |
| "loss": 2.0044, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.09177469312836985, |
| "eval_loss": 1.1052675247192383, |
| "eval_runtime": 13.0193, |
| "eval_samples_per_second": 505.709, |
| "eval_steps_per_second": 7.911, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.09292187679247448, |
| "eval_loss": 1.0964922904968262, |
| "eval_runtime": 13.1289, |
| "eval_samples_per_second": 501.489, |
| "eval_steps_per_second": 7.845, |
| "step": 810 |
| }, |
| { |
| "epoch": 0.0940690604565791, |
| "eval_loss": 1.0879426002502441, |
| "eval_runtime": 12.9727, |
| "eval_samples_per_second": 507.526, |
| "eval_steps_per_second": 7.94, |
| "step": 820 |
| }, |
| { |
| "epoch": 0.09521624412068372, |
| "eval_loss": 1.0795575380325317, |
| "eval_runtime": 13.1138, |
| "eval_samples_per_second": 502.068, |
| "eval_steps_per_second": 7.854, |
| "step": 830 |
| }, |
| { |
| "epoch": 0.09636342778478835, |
| "eval_loss": 1.0718028545379639, |
| "eval_runtime": 12.9813, |
| "eval_samples_per_second": 507.191, |
| "eval_steps_per_second": 7.934, |
| "step": 840 |
| }, |
| { |
| "epoch": 0.09751061144889296, |
| "eval_loss": 1.0643888711929321, |
| "eval_runtime": 13.0171, |
| "eval_samples_per_second": 505.795, |
| "eval_steps_per_second": 7.913, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.09865779511299759, |
| "eval_loss": 1.0564391613006592, |
| "eval_runtime": 13.0169, |
| "eval_samples_per_second": 505.806, |
| "eval_steps_per_second": 7.913, |
| "step": 860 |
| }, |
| { |
| "epoch": 0.09980497877710222, |
| "eval_loss": 1.0490267276763916, |
| "eval_runtime": 13.0703, |
| "eval_samples_per_second": 503.738, |
| "eval_steps_per_second": 7.88, |
| "step": 870 |
| }, |
| { |
| "epoch": 0.10095216244120683, |
| "eval_loss": 1.0417358875274658, |
| "eval_runtime": 13.0303, |
| "eval_samples_per_second": 505.284, |
| "eval_steps_per_second": 7.905, |
| "step": 880 |
| }, |
| { |
| "epoch": 0.10209934610531146, |
| "eval_loss": 1.0353840589523315, |
| "eval_runtime": 12.9765, |
| "eval_samples_per_second": 507.381, |
| "eval_steps_per_second": 7.937, |
| "step": 890 |
| }, |
| { |
| "epoch": 0.10324652976941609, |
| "grad_norm": 3.705247402191162, |
| "learning_rate": 3.4403669724770644e-06, |
| "loss": 1.8763, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.10324652976941609, |
| "eval_loss": 1.0295778512954712, |
| "eval_runtime": 13.0396, |
| "eval_samples_per_second": 504.924, |
| "eval_steps_per_second": 7.899, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.1043937134335207, |
| "eval_loss": 1.0238651037216187, |
| "eval_runtime": 12.973, |
| "eval_samples_per_second": 507.515, |
| "eval_steps_per_second": 7.94, |
| "step": 910 |
| }, |
| { |
| "epoch": 0.10554089709762533, |
| "eval_loss": 1.0180109739303589, |
| "eval_runtime": 13.0196, |
| "eval_samples_per_second": 505.7, |
| "eval_steps_per_second": 7.911, |
| "step": 920 |
| }, |
| { |
| "epoch": 0.10668808076172995, |
| "eval_loss": 1.0122654438018799, |
| "eval_runtime": 12.9877, |
| "eval_samples_per_second": 506.942, |
| "eval_steps_per_second": 7.931, |
| "step": 930 |
| }, |
| { |
| "epoch": 0.10783526442583458, |
| "eval_loss": 1.0065423250198364, |
| "eval_runtime": 13.0039, |
| "eval_samples_per_second": 506.312, |
| "eval_steps_per_second": 7.921, |
| "step": 940 |
| }, |
| { |
| "epoch": 0.1089824480899392, |
| "eval_loss": 1.0008338689804077, |
| "eval_runtime": 13.1129, |
| "eval_samples_per_second": 502.101, |
| "eval_steps_per_second": 7.855, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.11012963175404382, |
| "eval_loss": 0.9950375556945801, |
| "eval_runtime": 13.0191, |
| "eval_samples_per_second": 505.718, |
| "eval_steps_per_second": 7.911, |
| "step": 960 |
| }, |
| { |
| "epoch": 0.11127681541814845, |
| "eval_loss": 0.9893631935119629, |
| "eval_runtime": 12.9794, |
| "eval_samples_per_second": 507.267, |
| "eval_steps_per_second": 7.936, |
| "step": 970 |
| }, |
| { |
| "epoch": 0.11242399908225306, |
| "eval_loss": 0.9840025305747986, |
| "eval_runtime": 12.9799, |
| "eval_samples_per_second": 507.247, |
| "eval_steps_per_second": 7.935, |
| "step": 980 |
| }, |
| { |
| "epoch": 0.11357118274635769, |
| "eval_loss": 0.9792994856834412, |
| "eval_runtime": 12.9866, |
| "eval_samples_per_second": 506.983, |
| "eval_steps_per_second": 7.931, |
| "step": 990 |
| }, |
| { |
| "epoch": 0.11471836641046232, |
| "grad_norm": 4.240755558013916, |
| "learning_rate": 3.8226299694189605e-06, |
| "loss": 1.7287, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.11471836641046232, |
| "eval_loss": 0.9751574993133545, |
| "eval_runtime": 13.0083, |
| "eval_samples_per_second": 506.136, |
| "eval_steps_per_second": 7.918, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.11586555007456693, |
| "eval_loss": 0.9705988764762878, |
| "eval_runtime": 12.9835, |
| "eval_samples_per_second": 507.107, |
| "eval_steps_per_second": 7.933, |
| "step": 1010 |
| }, |
| { |
| "epoch": 0.11701273373867156, |
| "eval_loss": 0.9658572673797607, |
| "eval_runtime": 13.0627, |
| "eval_samples_per_second": 504.031, |
| "eval_steps_per_second": 7.885, |
| "step": 1020 |
| }, |
| { |
| "epoch": 0.11815991740277619, |
| "eval_loss": 0.9614543318748474, |
| "eval_runtime": 13.0169, |
| "eval_samples_per_second": 505.803, |
| "eval_steps_per_second": 7.913, |
| "step": 1030 |
| }, |
| { |
| "epoch": 0.1193071010668808, |
| "eval_loss": 0.9571945667266846, |
| "eval_runtime": 12.9762, |
| "eval_samples_per_second": 507.389, |
| "eval_steps_per_second": 7.938, |
| "step": 1040 |
| }, |
| { |
| "epoch": 0.12045428473098543, |
| "eval_loss": 0.9531480073928833, |
| "eval_runtime": 13.0036, |
| "eval_samples_per_second": 506.323, |
| "eval_steps_per_second": 7.921, |
| "step": 1050 |
| }, |
| { |
| "epoch": 0.12160146839509005, |
| "eval_loss": 0.9493557810783386, |
| "eval_runtime": 13.0046, |
| "eval_samples_per_second": 506.282, |
| "eval_steps_per_second": 7.92, |
| "step": 1060 |
| }, |
| { |
| "epoch": 0.12274865205919468, |
| "eval_loss": 0.9455849528312683, |
| "eval_runtime": 13.0676, |
| "eval_samples_per_second": 503.843, |
| "eval_steps_per_second": 7.882, |
| "step": 1070 |
| }, |
| { |
| "epoch": 0.1238958357232993, |
| "eval_loss": 0.9415374398231506, |
| "eval_runtime": 13.0698, |
| "eval_samples_per_second": 503.755, |
| "eval_steps_per_second": 7.881, |
| "step": 1080 |
| }, |
| { |
| "epoch": 0.12504301938740392, |
| "eval_loss": 0.9376588463783264, |
| "eval_runtime": 12.9843, |
| "eval_samples_per_second": 507.072, |
| "eval_steps_per_second": 7.933, |
| "step": 1090 |
| }, |
| { |
| "epoch": 0.12619020305150855, |
| "grad_norm": 3.3081679344177246, |
| "learning_rate": 4.204892966360857e-06, |
| "loss": 1.6312, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.12619020305150855, |
| "eval_loss": 0.9338813424110413, |
| "eval_runtime": 13.0365, |
| "eval_samples_per_second": 505.043, |
| "eval_steps_per_second": 7.901, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.12733738671561318, |
| "eval_loss": 0.9302825927734375, |
| "eval_runtime": 13.0692, |
| "eval_samples_per_second": 503.779, |
| "eval_steps_per_second": 7.881, |
| "step": 1110 |
| }, |
| { |
| "epoch": 0.1284845703797178, |
| "eval_loss": 0.9267016053199768, |
| "eval_runtime": 13.0455, |
| "eval_samples_per_second": 504.694, |
| "eval_steps_per_second": 7.895, |
| "step": 1120 |
| }, |
| { |
| "epoch": 0.1296317540438224, |
| "eval_loss": 0.9232119917869568, |
| "eval_runtime": 13.0051, |
| "eval_samples_per_second": 506.262, |
| "eval_steps_per_second": 7.92, |
| "step": 1130 |
| }, |
| { |
| "epoch": 0.13077893770792703, |
| "eval_loss": 0.9197220206260681, |
| "eval_runtime": 13.0069, |
| "eval_samples_per_second": 506.192, |
| "eval_steps_per_second": 7.919, |
| "step": 1140 |
| }, |
| { |
| "epoch": 0.13192612137203166, |
| "eval_loss": 0.9162309765815735, |
| "eval_runtime": 12.9851, |
| "eval_samples_per_second": 507.044, |
| "eval_steps_per_second": 7.932, |
| "step": 1150 |
| }, |
| { |
| "epoch": 0.1330733050361363, |
| "eval_loss": 0.9127652049064636, |
| "eval_runtime": 12.9768, |
| "eval_samples_per_second": 507.366, |
| "eval_steps_per_second": 7.937, |
| "step": 1160 |
| }, |
| { |
| "epoch": 0.13422048870024092, |
| "eval_loss": 0.9096914529800415, |
| "eval_runtime": 12.9953, |
| "eval_samples_per_second": 506.645, |
| "eval_steps_per_second": 7.926, |
| "step": 1170 |
| }, |
| { |
| "epoch": 0.13536767236434552, |
| "eval_loss": 0.9069137573242188, |
| "eval_runtime": 13.0109, |
| "eval_samples_per_second": 506.038, |
| "eval_steps_per_second": 7.916, |
| "step": 1180 |
| }, |
| { |
| "epoch": 0.13651485602845015, |
| "eval_loss": 0.9039744734764099, |
| "eval_runtime": 12.9772, |
| "eval_samples_per_second": 507.352, |
| "eval_steps_per_second": 7.937, |
| "step": 1190 |
| }, |
| { |
| "epoch": 0.13766203969255478, |
| "grad_norm": 3.420077085494995, |
| "learning_rate": 4.587155963302753e-06, |
| "loss": 1.5316, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.13766203969255478, |
| "eval_loss": 0.9009912610054016, |
| "eval_runtime": 12.9848, |
| "eval_samples_per_second": 507.056, |
| "eval_steps_per_second": 7.932, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.1388092233566594, |
| "eval_loss": 0.8979319930076599, |
| "eval_runtime": 13.0295, |
| "eval_samples_per_second": 505.313, |
| "eval_steps_per_second": 7.905, |
| "step": 1210 |
| }, |
| { |
| "epoch": 0.13995640702076403, |
| "eval_loss": 0.8946565985679626, |
| "eval_runtime": 13.0119, |
| "eval_samples_per_second": 505.997, |
| "eval_steps_per_second": 7.916, |
| "step": 1220 |
| }, |
| { |
| "epoch": 0.14110359068486863, |
| "eval_loss": 0.8915460705757141, |
| "eval_runtime": 12.9979, |
| "eval_samples_per_second": 506.545, |
| "eval_steps_per_second": 7.924, |
| "step": 1230 |
| }, |
| { |
| "epoch": 0.14225077434897326, |
| "eval_loss": 0.8887993097305298, |
| "eval_runtime": 12.9633, |
| "eval_samples_per_second": 507.895, |
| "eval_steps_per_second": 7.945, |
| "step": 1240 |
| }, |
| { |
| "epoch": 0.1433979580130779, |
| "eval_loss": 0.886054277420044, |
| "eval_runtime": 12.9996, |
| "eval_samples_per_second": 506.477, |
| "eval_steps_per_second": 7.923, |
| "step": 1250 |
| }, |
| { |
| "epoch": 0.14454514167718252, |
| "eval_loss": 0.8833284974098206, |
| "eval_runtime": 13.1133, |
| "eval_samples_per_second": 502.086, |
| "eval_steps_per_second": 7.855, |
| "step": 1260 |
| }, |
| { |
| "epoch": 0.14569232534128715, |
| "eval_loss": 0.8806452751159668, |
| "eval_runtime": 13.0208, |
| "eval_samples_per_second": 505.654, |
| "eval_steps_per_second": 7.91, |
| "step": 1270 |
| }, |
| { |
| "epoch": 0.14683950900539175, |
| "eval_loss": 0.8778645992279053, |
| "eval_runtime": 13.1071, |
| "eval_samples_per_second": 502.322, |
| "eval_steps_per_second": 7.858, |
| "step": 1280 |
| }, |
| { |
| "epoch": 0.14798669266949638, |
| "eval_loss": 0.8747658133506775, |
| "eval_runtime": 13.0362, |
| "eval_samples_per_second": 505.055, |
| "eval_steps_per_second": 7.901, |
| "step": 1290 |
| }, |
| { |
| "epoch": 0.149133876333601, |
| "grad_norm": 3.3882789611816406, |
| "learning_rate": 4.969418960244649e-06, |
| "loss": 1.4961, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.149133876333601, |
| "eval_loss": 0.8717615008354187, |
| "eval_runtime": 13.0104, |
| "eval_samples_per_second": 506.058, |
| "eval_steps_per_second": 7.917, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.15028105999770563, |
| "eval_loss": 0.8690453767776489, |
| "eval_runtime": 13.0189, |
| "eval_samples_per_second": 505.727, |
| "eval_steps_per_second": 7.912, |
| "step": 1310 |
| }, |
| { |
| "epoch": 0.15142824366181026, |
| "eval_loss": 0.8664180040359497, |
| "eval_runtime": 12.988, |
| "eval_samples_per_second": 506.931, |
| "eval_steps_per_second": 7.93, |
| "step": 1320 |
| }, |
| { |
| "epoch": 0.1525754273259149, |
| "eval_loss": 0.8634527921676636, |
| "eval_runtime": 12.9897, |
| "eval_samples_per_second": 506.861, |
| "eval_steps_per_second": 7.929, |
| "step": 1330 |
| }, |
| { |
| "epoch": 0.1537226109900195, |
| "eval_loss": 0.8603318929672241, |
| "eval_runtime": 13.0269, |
| "eval_samples_per_second": 505.417, |
| "eval_steps_per_second": 7.907, |
| "step": 1340 |
| }, |
| { |
| "epoch": 0.15486979465412412, |
| "eval_loss": 0.8573653697967529, |
| "eval_runtime": 12.9945, |
| "eval_samples_per_second": 506.677, |
| "eval_steps_per_second": 7.926, |
| "step": 1350 |
| }, |
| { |
| "epoch": 0.15601697831822875, |
| "eval_loss": 0.8545491695404053, |
| "eval_runtime": 12.9981, |
| "eval_samples_per_second": 506.536, |
| "eval_steps_per_second": 7.924, |
| "step": 1360 |
| }, |
| { |
| "epoch": 0.15716416198233338, |
| "eval_loss": 0.85209721326828, |
| "eval_runtime": 13.0301, |
| "eval_samples_per_second": 505.293, |
| "eval_steps_per_second": 7.905, |
| "step": 1370 |
| }, |
| { |
| "epoch": 0.158311345646438, |
| "eval_loss": 0.8497452139854431, |
| "eval_runtime": 13.1229, |
| "eval_samples_per_second": 501.717, |
| "eval_steps_per_second": 7.849, |
| "step": 1380 |
| }, |
| { |
| "epoch": 0.1594585293105426, |
| "eval_loss": 0.8474416136741638, |
| "eval_runtime": 13.0624, |
| "eval_samples_per_second": 504.043, |
| "eval_steps_per_second": 7.885, |
| "step": 1390 |
| }, |
| { |
| "epoch": 0.16060571297464724, |
| "grad_norm": 3.7449283599853516, |
| "learning_rate": 5.351681957186545e-06, |
| "loss": 1.451, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.16060571297464724, |
| "eval_loss": 0.8452581763267517, |
| "eval_runtime": 13.0191, |
| "eval_samples_per_second": 505.72, |
| "eval_steps_per_second": 7.911, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.16175289663875186, |
| "eval_loss": 0.8428558111190796, |
| "eval_runtime": 13.0038, |
| "eval_samples_per_second": 506.314, |
| "eval_steps_per_second": 7.921, |
| "step": 1410 |
| }, |
| { |
| "epoch": 0.1629000803028565, |
| "eval_loss": 0.8404139876365662, |
| "eval_runtime": 12.9904, |
| "eval_samples_per_second": 506.837, |
| "eval_steps_per_second": 7.929, |
| "step": 1420 |
| }, |
| { |
| "epoch": 0.16404726396696112, |
| "eval_loss": 0.8380302786827087, |
| "eval_runtime": 12.9812, |
| "eval_samples_per_second": 507.195, |
| "eval_steps_per_second": 7.935, |
| "step": 1430 |
| }, |
| { |
| "epoch": 0.16519444763106572, |
| "eval_loss": 0.8357470631599426, |
| "eval_runtime": 13.0241, |
| "eval_samples_per_second": 505.526, |
| "eval_steps_per_second": 7.908, |
| "step": 1440 |
| }, |
| { |
| "epoch": 0.16634163129517035, |
| "eval_loss": 0.8336036205291748, |
| "eval_runtime": 13.142, |
| "eval_samples_per_second": 500.99, |
| "eval_steps_per_second": 7.837, |
| "step": 1450 |
| }, |
| { |
| "epoch": 0.16748881495927498, |
| "eval_loss": 0.8312162756919861, |
| "eval_runtime": 12.9935, |
| "eval_samples_per_second": 506.714, |
| "eval_steps_per_second": 7.927, |
| "step": 1460 |
| }, |
| { |
| "epoch": 0.1686359986233796, |
| "eval_loss": 0.8288681507110596, |
| "eval_runtime": 13.1018, |
| "eval_samples_per_second": 502.527, |
| "eval_steps_per_second": 7.862, |
| "step": 1470 |
| }, |
| { |
| "epoch": 0.16978318228748424, |
| "eval_loss": 0.8261700868606567, |
| "eval_runtime": 13.0327, |
| "eval_samples_per_second": 505.19, |
| "eval_steps_per_second": 7.903, |
| "step": 1480 |
| }, |
| { |
| "epoch": 0.17093036595158884, |
| "eval_loss": 0.8235771656036377, |
| "eval_runtime": 12.9895, |
| "eval_samples_per_second": 506.873, |
| "eval_steps_per_second": 7.93, |
| "step": 1490 |
| }, |
| { |
| "epoch": 0.17207754961569346, |
| "grad_norm": 2.5695531368255615, |
| "learning_rate": 5.733944954128441e-06, |
| "loss": 1.4177, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.17207754961569346, |
| "eval_loss": 0.8213275074958801, |
| "eval_runtime": 13.0468, |
| "eval_samples_per_second": 504.647, |
| "eval_steps_per_second": 7.895, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.1732247332797981, |
| "eval_loss": 0.8189388513565063, |
| "eval_runtime": 12.9877, |
| "eval_samples_per_second": 506.943, |
| "eval_steps_per_second": 7.931, |
| "step": 1510 |
| }, |
| { |
| "epoch": 0.17437191694390272, |
| "eval_loss": 0.8168380856513977, |
| "eval_runtime": 12.9915, |
| "eval_samples_per_second": 506.793, |
| "eval_steps_per_second": 7.928, |
| "step": 1520 |
| }, |
| { |
| "epoch": 0.17551910060800735, |
| "eval_loss": 0.814733624458313, |
| "eval_runtime": 13.1726, |
| "eval_samples_per_second": 499.824, |
| "eval_steps_per_second": 7.819, |
| "step": 1530 |
| }, |
| { |
| "epoch": 0.17666628427211198, |
| "eval_loss": 0.8127268552780151, |
| "eval_runtime": 13.0207, |
| "eval_samples_per_second": 505.657, |
| "eval_steps_per_second": 7.91, |
| "step": 1540 |
| }, |
| { |
| "epoch": 0.17781346793621658, |
| "eval_loss": 0.8106749057769775, |
| "eval_runtime": 13.0973, |
| "eval_samples_per_second": 502.697, |
| "eval_steps_per_second": 7.864, |
| "step": 1550 |
| }, |
| { |
| "epoch": 0.1789606516003212, |
| "eval_loss": 0.808172345161438, |
| "eval_runtime": 13.2116, |
| "eval_samples_per_second": 498.349, |
| "eval_steps_per_second": 7.796, |
| "step": 1560 |
| }, |
| { |
| "epoch": 0.18010783526442584, |
| "eval_loss": 0.8059112429618835, |
| "eval_runtime": 12.9871, |
| "eval_samples_per_second": 506.963, |
| "eval_steps_per_second": 7.931, |
| "step": 1570 |
| }, |
| { |
| "epoch": 0.18125501892853046, |
| "eval_loss": 0.8035888671875, |
| "eval_runtime": 13.0963, |
| "eval_samples_per_second": 502.737, |
| "eval_steps_per_second": 7.865, |
| "step": 1580 |
| }, |
| { |
| "epoch": 0.1824022025926351, |
| "eval_loss": 0.8014644384384155, |
| "eval_runtime": 13.0028, |
| "eval_samples_per_second": 506.351, |
| "eval_steps_per_second": 7.921, |
| "step": 1590 |
| }, |
| { |
| "epoch": 0.1835493862567397, |
| "grad_norm": 2.681454658508301, |
| "learning_rate": 6.1162079510703365e-06, |
| "loss": 1.3734, |
| "step": 1600 |
| }, |
| { |
| "epoch": 0.1835493862567397, |
| "eval_loss": 0.7993160486221313, |
| "eval_runtime": 12.9893, |
| "eval_samples_per_second": 506.879, |
| "eval_steps_per_second": 7.93, |
| "step": 1600 |
| }, |
| { |
| "epoch": 0.18469656992084432, |
| "eval_loss": 0.7969831228256226, |
| "eval_runtime": 13.0137, |
| "eval_samples_per_second": 505.929, |
| "eval_steps_per_second": 7.915, |
| "step": 1610 |
| }, |
| { |
| "epoch": 0.18584375358494895, |
| "eval_loss": 0.7947937250137329, |
| "eval_runtime": 13.0467, |
| "eval_samples_per_second": 504.648, |
| "eval_steps_per_second": 7.895, |
| "step": 1620 |
| }, |
| { |
| "epoch": 0.18699093724905358, |
| "eval_loss": 0.7922271490097046, |
| "eval_runtime": 13.1133, |
| "eval_samples_per_second": 502.086, |
| "eval_steps_per_second": 7.855, |
| "step": 1630 |
| }, |
| { |
| "epoch": 0.1881381209131582, |
| "eval_loss": 0.790046751499176, |
| "eval_runtime": 13.0687, |
| "eval_samples_per_second": 503.801, |
| "eval_steps_per_second": 7.881, |
| "step": 1640 |
| }, |
| { |
| "epoch": 0.1892853045772628, |
| "eval_loss": 0.7877430319786072, |
| "eval_runtime": 13.1712, |
| "eval_samples_per_second": 499.877, |
| "eval_steps_per_second": 7.82, |
| "step": 1650 |
| }, |
| { |
| "epoch": 0.19043248824136744, |
| "eval_loss": 0.7852274179458618, |
| "eval_runtime": 13.0721, |
| "eval_samples_per_second": 503.669, |
| "eval_steps_per_second": 7.879, |
| "step": 1660 |
| }, |
| { |
| "epoch": 0.19157967190547207, |
| "eval_loss": 0.782863438129425, |
| "eval_runtime": 12.9741, |
| "eval_samples_per_second": 507.471, |
| "eval_steps_per_second": 7.939, |
| "step": 1670 |
| }, |
| { |
| "epoch": 0.1927268555695767, |
| "eval_loss": 0.7804363965988159, |
| "eval_runtime": 12.9826, |
| "eval_samples_per_second": 507.142, |
| "eval_steps_per_second": 7.934, |
| "step": 1680 |
| }, |
| { |
| "epoch": 0.19387403923368132, |
| "eval_loss": 0.7779432535171509, |
| "eval_runtime": 12.9731, |
| "eval_samples_per_second": 507.514, |
| "eval_steps_per_second": 7.94, |
| "step": 1690 |
| }, |
| { |
| "epoch": 0.19502122289778592, |
| "grad_norm": 3.1423346996307373, |
| "learning_rate": 6.4984709480122335e-06, |
| "loss": 1.3327, |
| "step": 1700 |
| }, |
| { |
| "epoch": 0.19502122289778592, |
| "eval_loss": 0.7756665945053101, |
| "eval_runtime": 12.9767, |
| "eval_samples_per_second": 507.371, |
| "eval_steps_per_second": 7.937, |
| "step": 1700 |
| }, |
| { |
| "epoch": 0.19616840656189055, |
| "eval_loss": 0.7737661004066467, |
| "eval_runtime": 12.9633, |
| "eval_samples_per_second": 507.896, |
| "eval_steps_per_second": 7.946, |
| "step": 1710 |
| }, |
| { |
| "epoch": 0.19731559022599518, |
| "eval_loss": 0.7719039916992188, |
| "eval_runtime": 12.9798, |
| "eval_samples_per_second": 507.248, |
| "eval_steps_per_second": 7.935, |
| "step": 1720 |
| }, |
| { |
| "epoch": 0.1984627738900998, |
| "eval_loss": 0.7699739933013916, |
| "eval_runtime": 12.9817, |
| "eval_samples_per_second": 507.177, |
| "eval_steps_per_second": 7.934, |
| "step": 1730 |
| }, |
| { |
| "epoch": 0.19960995755420444, |
| "eval_loss": 0.7679038643836975, |
| "eval_runtime": 13.0593, |
| "eval_samples_per_second": 504.164, |
| "eval_steps_per_second": 7.887, |
| "step": 1740 |
| }, |
| { |
| "epoch": 0.20075714121830904, |
| "eval_loss": 0.765801191329956, |
| "eval_runtime": 12.9827, |
| "eval_samples_per_second": 507.136, |
| "eval_steps_per_second": 7.934, |
| "step": 1750 |
| }, |
| { |
| "epoch": 0.20190432488241367, |
| "eval_loss": 0.7640795707702637, |
| "eval_runtime": 12.9885, |
| "eval_samples_per_second": 506.908, |
| "eval_steps_per_second": 7.93, |
| "step": 1760 |
| }, |
| { |
| "epoch": 0.2030515085465183, |
| "eval_loss": 0.7621497511863708, |
| "eval_runtime": 12.9872, |
| "eval_samples_per_second": 506.96, |
| "eval_steps_per_second": 7.931, |
| "step": 1770 |
| }, |
| { |
| "epoch": 0.20419869221062292, |
| "eval_loss": 0.7601312398910522, |
| "eval_runtime": 12.9805, |
| "eval_samples_per_second": 507.223, |
| "eval_steps_per_second": 7.935, |
| "step": 1780 |
| }, |
| { |
| "epoch": 0.20534587587472755, |
| "eval_loss": 0.7580214738845825, |
| "eval_runtime": 12.9849, |
| "eval_samples_per_second": 507.05, |
| "eval_steps_per_second": 7.932, |
| "step": 1790 |
| }, |
| { |
| "epoch": 0.20649305953883218, |
| "grad_norm": 3.4953625202178955, |
| "learning_rate": 6.880733944954129e-06, |
| "loss": 1.2804, |
| "step": 1800 |
| }, |
| { |
| "epoch": 0.20649305953883218, |
| "eval_loss": 0.7558028697967529, |
| "eval_runtime": 12.989, |
| "eval_samples_per_second": 506.889, |
| "eval_steps_per_second": 7.93, |
| "step": 1800 |
| }, |
| { |
| "epoch": 0.20764024320293678, |
| "eval_loss": 0.7535884380340576, |
| "eval_runtime": 12.9795, |
| "eval_samples_per_second": 507.262, |
| "eval_steps_per_second": 7.936, |
| "step": 1810 |
| }, |
| { |
| "epoch": 0.2087874268670414, |
| "eval_loss": 0.7514472603797913, |
| "eval_runtime": 12.9901, |
| "eval_samples_per_second": 506.846, |
| "eval_steps_per_second": 7.929, |
| "step": 1820 |
| }, |
| { |
| "epoch": 0.20993461053114604, |
| "eval_loss": 0.7493338584899902, |
| "eval_runtime": 13.0011, |
| "eval_samples_per_second": 506.418, |
| "eval_steps_per_second": 7.922, |
| "step": 1830 |
| }, |
| { |
| "epoch": 0.21108179419525067, |
| "eval_loss": 0.7473368644714355, |
| "eval_runtime": 13.0424, |
| "eval_samples_per_second": 504.815, |
| "eval_steps_per_second": 7.897, |
| "step": 1840 |
| }, |
| { |
| "epoch": 0.2122289778593553, |
| "eval_loss": 0.7450571060180664, |
| "eval_runtime": 13.1033, |
| "eval_samples_per_second": 502.468, |
| "eval_steps_per_second": 7.861, |
| "step": 1850 |
| }, |
| { |
| "epoch": 0.2133761615234599, |
| "eval_loss": 0.7429091930389404, |
| "eval_runtime": 12.9903, |
| "eval_samples_per_second": 506.841, |
| "eval_steps_per_second": 7.929, |
| "step": 1860 |
| }, |
| { |
| "epoch": 0.21452334518756452, |
| "eval_loss": 0.7408153414726257, |
| "eval_runtime": 12.983, |
| "eval_samples_per_second": 507.126, |
| "eval_steps_per_second": 7.933, |
| "step": 1870 |
| }, |
| { |
| "epoch": 0.21567052885166915, |
| "eval_loss": 0.7389461994171143, |
| "eval_runtime": 12.9922, |
| "eval_samples_per_second": 506.767, |
| "eval_steps_per_second": 7.928, |
| "step": 1880 |
| }, |
| { |
| "epoch": 0.21681771251577378, |
| "eval_loss": 0.7368388772010803, |
| "eval_runtime": 13.0283, |
| "eval_samples_per_second": 505.36, |
| "eval_steps_per_second": 7.906, |
| "step": 1890 |
| }, |
| { |
| "epoch": 0.2179648961798784, |
| "grad_norm": 2.869717597961426, |
| "learning_rate": 7.262996941896026e-06, |
| "loss": 1.2255, |
| "step": 1900 |
| }, |
| { |
| "epoch": 0.2179648961798784, |
| "eval_loss": 0.73488450050354, |
| "eval_runtime": 12.9924, |
| "eval_samples_per_second": 506.759, |
| "eval_steps_per_second": 7.928, |
| "step": 1900 |
| }, |
| { |
| "epoch": 0.219112079843983, |
| "eval_loss": 0.7328305244445801, |
| "eval_runtime": 12.9892, |
| "eval_samples_per_second": 506.881, |
| "eval_steps_per_second": 7.93, |
| "step": 1910 |
| }, |
| { |
| "epoch": 0.22025926350808764, |
| "eval_loss": 0.7309767007827759, |
| "eval_runtime": 13.1016, |
| "eval_samples_per_second": 502.534, |
| "eval_steps_per_second": 7.862, |
| "step": 1920 |
| }, |
| { |
| "epoch": 0.22140644717219227, |
| "eval_loss": 0.7293325662612915, |
| "eval_runtime": 12.9943, |
| "eval_samples_per_second": 506.684, |
| "eval_steps_per_second": 7.927, |
| "step": 1930 |
| }, |
| { |
| "epoch": 0.2225536308362969, |
| "eval_loss": 0.7277292013168335, |
| "eval_runtime": 13.0122, |
| "eval_samples_per_second": 505.986, |
| "eval_steps_per_second": 7.916, |
| "step": 1940 |
| }, |
| { |
| "epoch": 0.22370081450040152, |
| "eval_loss": 0.7258683443069458, |
| "eval_runtime": 13.0042, |
| "eval_samples_per_second": 506.298, |
| "eval_steps_per_second": 7.921, |
| "step": 1950 |
| }, |
| { |
| "epoch": 0.22484799816450612, |
| "eval_loss": 0.7239750027656555, |
| "eval_runtime": 13.0619, |
| "eval_samples_per_second": 504.061, |
| "eval_steps_per_second": 7.886, |
| "step": 1960 |
| }, |
| { |
| "epoch": 0.22599518182861075, |
| "eval_loss": 0.7221319079399109, |
| "eval_runtime": 12.9944, |
| "eval_samples_per_second": 506.679, |
| "eval_steps_per_second": 7.926, |
| "step": 1970 |
| }, |
| { |
| "epoch": 0.22714236549271538, |
| "eval_loss": 0.720324695110321, |
| "eval_runtime": 12.982, |
| "eval_samples_per_second": 507.165, |
| "eval_steps_per_second": 7.934, |
| "step": 1980 |
| }, |
| { |
| "epoch": 0.22828954915682, |
| "eval_loss": 0.7184233069419861, |
| "eval_runtime": 12.9988, |
| "eval_samples_per_second": 506.507, |
| "eval_steps_per_second": 7.924, |
| "step": 1990 |
| }, |
| { |
| "epoch": 0.22943673282092464, |
| "grad_norm": 9.275369644165039, |
| "learning_rate": 7.645259938837921e-06, |
| "loss": 1.2635, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.22943673282092464, |
| "eval_loss": 0.7165008187294006, |
| "eval_runtime": 12.9966, |
| "eval_samples_per_second": 506.594, |
| "eval_steps_per_second": 7.925, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.23058391648502927, |
| "eval_loss": 0.7150009870529175, |
| "eval_runtime": 12.9901, |
| "eval_samples_per_second": 506.849, |
| "eval_steps_per_second": 7.929, |
| "step": 2010 |
| }, |
| { |
| "epoch": 0.23173110014913387, |
| "eval_loss": 0.7134894728660583, |
| "eval_runtime": 13.0392, |
| "eval_samples_per_second": 504.938, |
| "eval_steps_per_second": 7.899, |
| "step": 2020 |
| }, |
| { |
| "epoch": 0.2328782838132385, |
| "eval_loss": 0.7117303013801575, |
| "eval_runtime": 13.0384, |
| "eval_samples_per_second": 504.971, |
| "eval_steps_per_second": 7.9, |
| "step": 2030 |
| }, |
| { |
| "epoch": 0.23402546747734312, |
| "eval_loss": 0.7099109292030334, |
| "eval_runtime": 12.9879, |
| "eval_samples_per_second": 506.932, |
| "eval_steps_per_second": 7.93, |
| "step": 2040 |
| }, |
| { |
| "epoch": 0.23517265114144775, |
| "eval_loss": 0.7084246873855591, |
| "eval_runtime": 12.9995, |
| "eval_samples_per_second": 506.481, |
| "eval_steps_per_second": 7.923, |
| "step": 2050 |
| }, |
| { |
| "epoch": 0.23631983480555238, |
| "eval_loss": 0.7067718505859375, |
| "eval_runtime": 12.9881, |
| "eval_samples_per_second": 506.925, |
| "eval_steps_per_second": 7.93, |
| "step": 2060 |
| }, |
| { |
| "epoch": 0.23746701846965698, |
| "eval_loss": 0.7053564786911011, |
| "eval_runtime": 13.1243, |
| "eval_samples_per_second": 501.665, |
| "eval_steps_per_second": 7.848, |
| "step": 2070 |
| }, |
| { |
| "epoch": 0.2386142021337616, |
| "eval_loss": 0.7037129402160645, |
| "eval_runtime": 12.9868, |
| "eval_samples_per_second": 506.975, |
| "eval_steps_per_second": 7.931, |
| "step": 2080 |
| }, |
| { |
| "epoch": 0.23976138579786624, |
| "eval_loss": 0.7023361921310425, |
| "eval_runtime": 12.9909, |
| "eval_samples_per_second": 506.818, |
| "eval_steps_per_second": 7.929, |
| "step": 2090 |
| }, |
| { |
| "epoch": 0.24090856946197087, |
| "grad_norm": 2.9726195335388184, |
| "learning_rate": 8.027522935779817e-06, |
| "loss": 1.1912, |
| "step": 2100 |
| }, |
| { |
| "epoch": 0.24090856946197087, |
| "eval_loss": 0.7009308338165283, |
| "eval_runtime": 13.1789, |
| "eval_samples_per_second": 499.587, |
| "eval_steps_per_second": 7.816, |
| "step": 2100 |
| }, |
| { |
| "epoch": 0.2420557531260755, |
| "eval_loss": 0.6990575790405273, |
| "eval_runtime": 13.0982, |
| "eval_samples_per_second": 502.665, |
| "eval_steps_per_second": 7.864, |
| "step": 2110 |
| }, |
| { |
| "epoch": 0.2432029367901801, |
| "eval_loss": 0.6973636746406555, |
| "eval_runtime": 13.0024, |
| "eval_samples_per_second": 506.369, |
| "eval_steps_per_second": 7.922, |
| "step": 2120 |
| }, |
| { |
| "epoch": 0.24435012045428472, |
| "eval_loss": 0.6962077617645264, |
| "eval_runtime": 12.9266, |
| "eval_samples_per_second": 509.337, |
| "eval_steps_per_second": 7.968, |
| "step": 2130 |
| }, |
| { |
| "epoch": 0.24549730411838935, |
| "eval_loss": 0.6949887275695801, |
| "eval_runtime": 12.9495, |
| "eval_samples_per_second": 508.438, |
| "eval_steps_per_second": 7.954, |
| "step": 2140 |
| }, |
| { |
| "epoch": 0.24664448778249398, |
| "eval_loss": 0.6937941312789917, |
| "eval_runtime": 12.9625, |
| "eval_samples_per_second": 507.926, |
| "eval_steps_per_second": 7.946, |
| "step": 2150 |
| }, |
| { |
| "epoch": 0.2477916714465986, |
| "eval_loss": 0.6922134160995483, |
| "eval_runtime": 13.0384, |
| "eval_samples_per_second": 504.971, |
| "eval_steps_per_second": 7.9, |
| "step": 2160 |
| }, |
| { |
| "epoch": 0.2489388551107032, |
| "eval_loss": 0.6908959150314331, |
| "eval_runtime": 13.0162, |
| "eval_samples_per_second": 505.831, |
| "eval_steps_per_second": 7.913, |
| "step": 2170 |
| }, |
| { |
| "epoch": 0.25008603877480784, |
| "eval_loss": 0.6896905899047852, |
| "eval_runtime": 13.0281, |
| "eval_samples_per_second": 505.368, |
| "eval_steps_per_second": 7.906, |
| "step": 2180 |
| }, |
| { |
| "epoch": 0.2512332224389125, |
| "eval_loss": 0.6883853673934937, |
| "eval_runtime": 13.059, |
| "eval_samples_per_second": 504.173, |
| "eval_steps_per_second": 7.887, |
| "step": 2190 |
| }, |
| { |
| "epoch": 0.2523804061030171, |
| "grad_norm": 9.038525581359863, |
| "learning_rate": 8.409785932721713e-06, |
| "loss": 1.2144, |
| "step": 2200 |
| }, |
| { |
| "epoch": 0.2523804061030171, |
| "eval_loss": 0.68684321641922, |
| "eval_runtime": 12.9342, |
| "eval_samples_per_second": 509.039, |
| "eval_steps_per_second": 7.963, |
| "step": 2200 |
| }, |
| { |
| "epoch": 0.2535275897671217, |
| "eval_loss": 0.6855539679527283, |
| "eval_runtime": 12.947, |
| "eval_samples_per_second": 508.536, |
| "eval_steps_per_second": 7.956, |
| "step": 2210 |
| }, |
| { |
| "epoch": 0.25467477343122635, |
| "eval_loss": 0.684305727481842, |
| "eval_runtime": 13.0482, |
| "eval_samples_per_second": 504.59, |
| "eval_steps_per_second": 7.894, |
| "step": 2220 |
| }, |
| { |
| "epoch": 0.25582195709533095, |
| "eval_loss": 0.6828807592391968, |
| "eval_runtime": 13.0317, |
| "eval_samples_per_second": 505.23, |
| "eval_steps_per_second": 7.904, |
| "step": 2230 |
| }, |
| { |
| "epoch": 0.2569691407594356, |
| "eval_loss": 0.6817071437835693, |
| "eval_runtime": 13.0489, |
| "eval_samples_per_second": 504.564, |
| "eval_steps_per_second": 7.893, |
| "step": 2240 |
| }, |
| { |
| "epoch": 0.2581163244235402, |
| "eval_loss": 0.6803813576698303, |
| "eval_runtime": 13.0166, |
| "eval_samples_per_second": 505.817, |
| "eval_steps_per_second": 7.913, |
| "step": 2250 |
| }, |
| { |
| "epoch": 0.2592635080876448, |
| "eval_loss": 0.6788561344146729, |
| "eval_runtime": 13.018, |
| "eval_samples_per_second": 505.76, |
| "eval_steps_per_second": 7.912, |
| "step": 2260 |
| }, |
| { |
| "epoch": 0.26041069175174947, |
| "eval_loss": 0.6775069832801819, |
| "eval_runtime": 12.9958, |
| "eval_samples_per_second": 506.623, |
| "eval_steps_per_second": 7.926, |
| "step": 2270 |
| }, |
| { |
| "epoch": 0.26155787541585407, |
| "eval_loss": 0.6762964725494385, |
| "eval_runtime": 12.9456, |
| "eval_samples_per_second": 508.59, |
| "eval_steps_per_second": 7.956, |
| "step": 2280 |
| }, |
| { |
| "epoch": 0.2627050590799587, |
| "eval_loss": 0.6750586032867432, |
| "eval_runtime": 12.9511, |
| "eval_samples_per_second": 508.372, |
| "eval_steps_per_second": 7.953, |
| "step": 2290 |
| }, |
| { |
| "epoch": 0.2638522427440633, |
| "grad_norm": 3.329761266708374, |
| "learning_rate": 8.79204892966361e-06, |
| "loss": 1.1498, |
| "step": 2300 |
| }, |
| { |
| "epoch": 0.2638522427440633, |
| "eval_loss": 0.6738935112953186, |
| "eval_runtime": 13.0495, |
| "eval_samples_per_second": 504.542, |
| "eval_steps_per_second": 7.893, |
| "step": 2300 |
| }, |
| { |
| "epoch": 0.2649994264081679, |
| "eval_loss": 0.6725099682807922, |
| "eval_runtime": 13.0328, |
| "eval_samples_per_second": 505.188, |
| "eval_steps_per_second": 7.903, |
| "step": 2310 |
| }, |
| { |
| "epoch": 0.2661466100722726, |
| "eval_loss": 0.6710599660873413, |
| "eval_runtime": 13.0505, |
| "eval_samples_per_second": 504.502, |
| "eval_steps_per_second": 7.892, |
| "step": 2320 |
| }, |
| { |
| "epoch": 0.2672937937363772, |
| "eval_loss": 0.6697894930839539, |
| "eval_runtime": 13.0616, |
| "eval_samples_per_second": 504.074, |
| "eval_steps_per_second": 7.886, |
| "step": 2330 |
| }, |
| { |
| "epoch": 0.26844097740048184, |
| "eval_loss": 0.668381929397583, |
| "eval_runtime": 12.9197, |
| "eval_samples_per_second": 509.608, |
| "eval_steps_per_second": 7.972, |
| "step": 2340 |
| }, |
| { |
| "epoch": 0.26958816106458644, |
| "eval_loss": 0.6666426062583923, |
| "eval_runtime": 13.0254, |
| "eval_samples_per_second": 505.475, |
| "eval_steps_per_second": 7.908, |
| "step": 2350 |
| }, |
| { |
| "epoch": 0.27073534472869104, |
| "eval_loss": 0.6652824878692627, |
| "eval_runtime": 13.0301, |
| "eval_samples_per_second": 505.293, |
| "eval_steps_per_second": 7.905, |
| "step": 2360 |
| }, |
| { |
| "epoch": 0.2718825283927957, |
| "eval_loss": 0.6638170480728149, |
| "eval_runtime": 12.917, |
| "eval_samples_per_second": 509.715, |
| "eval_steps_per_second": 7.974, |
| "step": 2370 |
| }, |
| { |
| "epoch": 0.2730297120569003, |
| "eval_loss": 0.6620729565620422, |
| "eval_runtime": 13.0292, |
| "eval_samples_per_second": 505.327, |
| "eval_steps_per_second": 7.905, |
| "step": 2380 |
| }, |
| { |
| "epoch": 0.27417689572100495, |
| "eval_loss": 0.660891056060791, |
| "eval_runtime": 13.022, |
| "eval_samples_per_second": 505.604, |
| "eval_steps_per_second": 7.91, |
| "step": 2390 |
| }, |
| { |
| "epoch": 0.27532407938510955, |
| "grad_norm": 2.91483211517334, |
| "learning_rate": 9.174311926605506e-06, |
| "loss": 1.1446, |
| "step": 2400 |
| }, |
| { |
| "epoch": 0.27532407938510955, |
| "eval_loss": 0.6595732569694519, |
| "eval_runtime": 13.036, |
| "eval_samples_per_second": 505.063, |
| "eval_steps_per_second": 7.901, |
| "step": 2400 |
| }, |
| { |
| "epoch": 0.27647126304921416, |
| "eval_loss": 0.6582459807395935, |
| "eval_runtime": 12.9309, |
| "eval_samples_per_second": 509.168, |
| "eval_steps_per_second": 7.965, |
| "step": 2410 |
| }, |
| { |
| "epoch": 0.2776184467133188, |
| "eval_loss": 0.6568032503128052, |
| "eval_runtime": 13.0039, |
| "eval_samples_per_second": 506.309, |
| "eval_steps_per_second": 7.921, |
| "step": 2420 |
| }, |
| { |
| "epoch": 0.2787656303774234, |
| "eval_loss": 0.6553359031677246, |
| "eval_runtime": 12.9157, |
| "eval_samples_per_second": 509.766, |
| "eval_steps_per_second": 7.975, |
| "step": 2430 |
| }, |
| { |
| "epoch": 0.27991281404152807, |
| "eval_loss": 0.6541372537612915, |
| "eval_runtime": 12.9901, |
| "eval_samples_per_second": 506.846, |
| "eval_steps_per_second": 7.929, |
| "step": 2440 |
| }, |
| { |
| "epoch": 0.28105999770563267, |
| "eval_loss": 0.6527132391929626, |
| "eval_runtime": 12.9209, |
| "eval_samples_per_second": 509.562, |
| "eval_steps_per_second": 7.972, |
| "step": 2450 |
| }, |
| { |
| "epoch": 0.28220718136973727, |
| "eval_loss": 0.6512923836708069, |
| "eval_runtime": 12.9637, |
| "eval_samples_per_second": 507.879, |
| "eval_steps_per_second": 7.945, |
| "step": 2460 |
| }, |
| { |
| "epoch": 0.2833543650338419, |
| "eval_loss": 0.6496260762214661, |
| "eval_runtime": 13.0604, |
| "eval_samples_per_second": 504.119, |
| "eval_steps_per_second": 7.886, |
| "step": 2470 |
| }, |
| { |
| "epoch": 0.2845015486979465, |
| "eval_loss": 0.6483332514762878, |
| "eval_runtime": 13.0233, |
| "eval_samples_per_second": 505.557, |
| "eval_steps_per_second": 7.909, |
| "step": 2480 |
| }, |
| { |
| "epoch": 0.2856487323620512, |
| "eval_loss": 0.647476077079773, |
| "eval_runtime": 13.0344, |
| "eval_samples_per_second": 505.124, |
| "eval_steps_per_second": 7.902, |
| "step": 2490 |
| }, |
| { |
| "epoch": 0.2867959160261558, |
| "grad_norm": 3.6412861347198486, |
| "learning_rate": 9.556574923547402e-06, |
| "loss": 1.1309, |
| "step": 2500 |
| }, |
| { |
| "epoch": 0.2867959160261558, |
| "eval_loss": 0.6465412378311157, |
| "eval_runtime": 12.9143, |
| "eval_samples_per_second": 509.824, |
| "eval_steps_per_second": 7.976, |
| "step": 2500 |
| }, |
| { |
| "epoch": 0.2879430996902604, |
| "eval_loss": 0.6455466747283936, |
| "eval_runtime": 13.0269, |
| "eval_samples_per_second": 505.417, |
| "eval_steps_per_second": 7.907, |
| "step": 2510 |
| }, |
| { |
| "epoch": 0.28909028335436504, |
| "eval_loss": 0.6446804404258728, |
| "eval_runtime": 13.0271, |
| "eval_samples_per_second": 505.407, |
| "eval_steps_per_second": 7.907, |
| "step": 2520 |
| }, |
| { |
| "epoch": 0.29023746701846964, |
| "eval_loss": 0.6436929702758789, |
| "eval_runtime": 13.0149, |
| "eval_samples_per_second": 505.882, |
| "eval_steps_per_second": 7.914, |
| "step": 2530 |
| }, |
| { |
| "epoch": 0.2913846506825743, |
| "eval_loss": 0.6427727341651917, |
| "eval_runtime": 13.0187, |
| "eval_samples_per_second": 505.734, |
| "eval_steps_per_second": 7.912, |
| "step": 2540 |
| }, |
| { |
| "epoch": 0.2925318343466789, |
| "eval_loss": 0.6415053009986877, |
| "eval_runtime": 13.0192, |
| "eval_samples_per_second": 505.716, |
| "eval_steps_per_second": 7.911, |
| "step": 2550 |
| }, |
| { |
| "epoch": 0.2936790180107835, |
| "eval_loss": 0.6403249502182007, |
| "eval_runtime": 13.0143, |
| "eval_samples_per_second": 505.905, |
| "eval_steps_per_second": 7.914, |
| "step": 2560 |
| }, |
| { |
| "epoch": 0.29482620167488816, |
| "eval_loss": 0.639183759689331, |
| "eval_runtime": 13.0127, |
| "eval_samples_per_second": 505.966, |
| "eval_steps_per_second": 7.915, |
| "step": 2570 |
| }, |
| { |
| "epoch": 0.29597338533899276, |
| "eval_loss": 0.6380952000617981, |
| "eval_runtime": 13.0253, |
| "eval_samples_per_second": 505.478, |
| "eval_steps_per_second": 7.908, |
| "step": 2580 |
| }, |
| { |
| "epoch": 0.2971205690030974, |
| "eval_loss": 0.6371238231658936, |
| "eval_runtime": 12.9266, |
| "eval_samples_per_second": 509.339, |
| "eval_steps_per_second": 7.968, |
| "step": 2590 |
| }, |
| { |
| "epoch": 0.298267752667202, |
| "grad_norm": 3.0183558464050293, |
| "learning_rate": 9.938837920489298e-06, |
| "loss": 1.1006, |
| "step": 2600 |
| }, |
| { |
| "epoch": 0.298267752667202, |
| "eval_loss": 0.6357853412628174, |
| "eval_runtime": 12.9136, |
| "eval_samples_per_second": 509.848, |
| "eval_steps_per_second": 7.976, |
| "step": 2600 |
| }, |
| { |
| "epoch": 0.29941493633130667, |
| "eval_loss": 0.6347528696060181, |
| "eval_runtime": 13.031, |
| "eval_samples_per_second": 505.258, |
| "eval_steps_per_second": 7.904, |
| "step": 2610 |
| }, |
| { |
| "epoch": 0.30056211999541127, |
| "eval_loss": 0.634018063545227, |
| "eval_runtime": 13.0415, |
| "eval_samples_per_second": 504.85, |
| "eval_steps_per_second": 7.898, |
| "step": 2620 |
| }, |
| { |
| "epoch": 0.30170930365951587, |
| "eval_loss": 0.6330114006996155, |
| "eval_runtime": 12.9387, |
| "eval_samples_per_second": 508.863, |
| "eval_steps_per_second": 7.961, |
| "step": 2630 |
| }, |
| { |
| "epoch": 0.3028564873236205, |
| "eval_loss": 0.6319145560264587, |
| "eval_runtime": 12.919, |
| "eval_samples_per_second": 509.636, |
| "eval_steps_per_second": 7.973, |
| "step": 2640 |
| }, |
| { |
| "epoch": 0.3040036709877251, |
| "eval_loss": 0.6307594180107117, |
| "eval_runtime": 12.9268, |
| "eval_samples_per_second": 509.328, |
| "eval_steps_per_second": 7.968, |
| "step": 2650 |
| }, |
| { |
| "epoch": 0.3051508546518298, |
| "eval_loss": 0.6300274133682251, |
| "eval_runtime": 13.0, |
| "eval_samples_per_second": 506.461, |
| "eval_steps_per_second": 7.923, |
| "step": 2660 |
| }, |
| { |
| "epoch": 0.3062980383159344, |
| "eval_loss": 0.6291391849517822, |
| "eval_runtime": 13.023, |
| "eval_samples_per_second": 505.566, |
| "eval_steps_per_second": 7.909, |
| "step": 2670 |
| }, |
| { |
| "epoch": 0.307445221980039, |
| "eval_loss": 0.6279686093330383, |
| "eval_runtime": 12.9099, |
| "eval_samples_per_second": 509.995, |
| "eval_steps_per_second": 7.978, |
| "step": 2680 |
| }, |
| { |
| "epoch": 0.30859240564414364, |
| "eval_loss": 0.6267827153205872, |
| "eval_runtime": 12.9942, |
| "eval_samples_per_second": 506.687, |
| "eval_steps_per_second": 7.927, |
| "step": 2690 |
| }, |
| { |
| "epoch": 0.30973958930824824, |
| "grad_norm": 2.923506021499634, |
| "learning_rate": 9.964308476736776e-06, |
| "loss": 1.0772, |
| "step": 2700 |
| }, |
| { |
| "epoch": 0.30973958930824824, |
| "eval_loss": 0.6254075765609741, |
| "eval_runtime": 13.0366, |
| "eval_samples_per_second": 505.041, |
| "eval_steps_per_second": 7.901, |
| "step": 2700 |
| }, |
| { |
| "epoch": 0.3108867729723529, |
| "eval_loss": 0.6242946982383728, |
| "eval_runtime": 13.0323, |
| "eval_samples_per_second": 505.207, |
| "eval_steps_per_second": 7.903, |
| "step": 2710 |
| }, |
| { |
| "epoch": 0.3120339566364575, |
| "eval_loss": 0.6231787204742432, |
| "eval_runtime": 12.9273, |
| "eval_samples_per_second": 509.309, |
| "eval_steps_per_second": 7.968, |
| "step": 2720 |
| }, |
| { |
| "epoch": 0.3131811403005621, |
| "eval_loss": 0.62239009141922, |
| "eval_runtime": 12.9153, |
| "eval_samples_per_second": 509.781, |
| "eval_steps_per_second": 7.975, |
| "step": 2730 |
| }, |
| { |
| "epoch": 0.31432832396466676, |
| "eval_loss": 0.6214902400970459, |
| "eval_runtime": 12.9178, |
| "eval_samples_per_second": 509.684, |
| "eval_steps_per_second": 7.973, |
| "step": 2740 |
| }, |
| { |
| "epoch": 0.31547550762877136, |
| "eval_loss": 0.6204728484153748, |
| "eval_runtime": 12.9141, |
| "eval_samples_per_second": 509.831, |
| "eval_steps_per_second": 7.976, |
| "step": 2750 |
| }, |
| { |
| "epoch": 0.316622691292876, |
| "eval_loss": 0.6194381713867188, |
| "eval_runtime": 12.9151, |
| "eval_samples_per_second": 509.791, |
| "eval_steps_per_second": 7.975, |
| "step": 2760 |
| }, |
| { |
| "epoch": 0.3177698749569806, |
| "eval_loss": 0.6183201670646667, |
| "eval_runtime": 12.9127, |
| "eval_samples_per_second": 509.887, |
| "eval_steps_per_second": 7.977, |
| "step": 2770 |
| }, |
| { |
| "epoch": 0.3189170586210852, |
| "eval_loss": 0.6171374917030334, |
| "eval_runtime": 12.9112, |
| "eval_samples_per_second": 509.944, |
| "eval_steps_per_second": 7.978, |
| "step": 2780 |
| }, |
| { |
| "epoch": 0.32006424228518987, |
| "eval_loss": 0.6160247921943665, |
| "eval_runtime": 13.0361, |
| "eval_samples_per_second": 505.06, |
| "eval_steps_per_second": 7.901, |
| "step": 2790 |
| }, |
| { |
| "epoch": 0.32121142594929447, |
| "grad_norm": 2.8031253814697266, |
| "learning_rate": 9.92181856809008e-06, |
| "loss": 1.0648, |
| "step": 2800 |
| }, |
| { |
| "epoch": 0.32121142594929447, |
| "eval_loss": 0.6153244376182556, |
| "eval_runtime": 13.0215, |
| "eval_samples_per_second": 505.624, |
| "eval_steps_per_second": 7.91, |
| "step": 2800 |
| }, |
| { |
| "epoch": 0.3223586096133991, |
| "eval_loss": 0.6140704154968262, |
| "eval_runtime": 13.0201, |
| "eval_samples_per_second": 505.679, |
| "eval_steps_per_second": 7.911, |
| "step": 2810 |
| }, |
| { |
| "epoch": 0.32350579327750373, |
| "eval_loss": 0.6129491329193115, |
| "eval_runtime": 13.0287, |
| "eval_samples_per_second": 505.347, |
| "eval_steps_per_second": 7.906, |
| "step": 2820 |
| }, |
| { |
| "epoch": 0.32465297694160833, |
| "eval_loss": 0.6119452118873596, |
| "eval_runtime": 13.045, |
| "eval_samples_per_second": 504.716, |
| "eval_steps_per_second": 7.896, |
| "step": 2830 |
| }, |
| { |
| "epoch": 0.325800160605713, |
| "eval_loss": 0.6109396815299988, |
| "eval_runtime": 13.0301, |
| "eval_samples_per_second": 505.29, |
| "eval_steps_per_second": 7.905, |
| "step": 2840 |
| }, |
| { |
| "epoch": 0.3269473442698176, |
| "eval_loss": 0.6099222898483276, |
| "eval_runtime": 13.0261, |
| "eval_samples_per_second": 505.446, |
| "eval_steps_per_second": 7.907, |
| "step": 2850 |
| }, |
| { |
| "epoch": 0.32809452793392224, |
| "eval_loss": 0.6087589859962463, |
| "eval_runtime": 12.9193, |
| "eval_samples_per_second": 509.625, |
| "eval_steps_per_second": 7.973, |
| "step": 2860 |
| }, |
| { |
| "epoch": 0.32924171159802684, |
| "eval_loss": 0.6078950762748718, |
| "eval_runtime": 12.9854, |
| "eval_samples_per_second": 507.031, |
| "eval_steps_per_second": 7.932, |
| "step": 2870 |
| }, |
| { |
| "epoch": 0.33038889526213144, |
| "eval_loss": 0.60728919506073, |
| "eval_runtime": 13.0282, |
| "eval_samples_per_second": 505.366, |
| "eval_steps_per_second": 7.906, |
| "step": 2880 |
| }, |
| { |
| "epoch": 0.3315360789262361, |
| "eval_loss": 0.606285810470581, |
| "eval_runtime": 13.0385, |
| "eval_samples_per_second": 504.964, |
| "eval_steps_per_second": 7.9, |
| "step": 2890 |
| }, |
| { |
| "epoch": 0.3326832625903407, |
| "grad_norm": 3.0763423442840576, |
| "learning_rate": 9.879328659443383e-06, |
| "loss": 1.0398, |
| "step": 2900 |
| }, |
| { |
| "epoch": 0.3326832625903407, |
| "eval_loss": 0.6053714752197266, |
| "eval_runtime": 13.0412, |
| "eval_samples_per_second": 504.863, |
| "eval_steps_per_second": 7.898, |
| "step": 2900 |
| }, |
| { |
| "epoch": 0.33383044625444536, |
| "eval_loss": 0.6043825149536133, |
| "eval_runtime": 13.0358, |
| "eval_samples_per_second": 505.072, |
| "eval_steps_per_second": 7.901, |
| "step": 2910 |
| }, |
| { |
| "epoch": 0.33497762991854996, |
| "eval_loss": 0.6032926440238953, |
| "eval_runtime": 13.0011, |
| "eval_samples_per_second": 506.418, |
| "eval_steps_per_second": 7.922, |
| "step": 2920 |
| }, |
| { |
| "epoch": 0.33612481358265456, |
| "eval_loss": 0.6022117137908936, |
| "eval_runtime": 12.922, |
| "eval_samples_per_second": 509.517, |
| "eval_steps_per_second": 7.971, |
| "step": 2930 |
| }, |
| { |
| "epoch": 0.3372719972467592, |
| "eval_loss": 0.6011677980422974, |
| "eval_runtime": 12.9117, |
| "eval_samples_per_second": 509.924, |
| "eval_steps_per_second": 7.977, |
| "step": 2940 |
| }, |
| { |
| "epoch": 0.3384191809108638, |
| "eval_loss": 0.6003371477127075, |
| "eval_runtime": 13.0239, |
| "eval_samples_per_second": 505.531, |
| "eval_steps_per_second": 7.909, |
| "step": 2950 |
| }, |
| { |
| "epoch": 0.33956636457496847, |
| "eval_loss": 0.5993051528930664, |
| "eval_runtime": 12.9359, |
| "eval_samples_per_second": 508.971, |
| "eval_steps_per_second": 7.962, |
| "step": 2960 |
| }, |
| { |
| "epoch": 0.34071354823907307, |
| "eval_loss": 0.5986045598983765, |
| "eval_runtime": 12.9502, |
| "eval_samples_per_second": 508.409, |
| "eval_steps_per_second": 7.954, |
| "step": 2970 |
| }, |
| { |
| "epoch": 0.3418607319031777, |
| "eval_loss": 0.5977791547775269, |
| "eval_runtime": 13.0367, |
| "eval_samples_per_second": 505.037, |
| "eval_steps_per_second": 7.901, |
| "step": 2980 |
| }, |
| { |
| "epoch": 0.34300791556728233, |
| "eval_loss": 0.596732497215271, |
| "eval_runtime": 13.0228, |
| "eval_samples_per_second": 505.576, |
| "eval_steps_per_second": 7.909, |
| "step": 2990 |
| }, |
| { |
| "epoch": 0.34415509923138693, |
| "grad_norm": 2.941866636276245, |
| "learning_rate": 9.836838750796687e-06, |
| "loss": 1.0256, |
| "step": 3000 |
| }, |
| { |
| "epoch": 0.34415509923138693, |
| "eval_loss": 0.5959278345108032, |
| "eval_runtime": 12.9239, |
| "eval_samples_per_second": 509.443, |
| "eval_steps_per_second": 7.97, |
| "step": 3000 |
| }, |
| { |
| "epoch": 0.3453022828954916, |
| "eval_loss": 0.5946770906448364, |
| "eval_runtime": 12.9123, |
| "eval_samples_per_second": 509.903, |
| "eval_steps_per_second": 7.977, |
| "step": 3010 |
| }, |
| { |
| "epoch": 0.3464494665595962, |
| "eval_loss": 0.593720555305481, |
| "eval_runtime": 12.92, |
| "eval_samples_per_second": 509.596, |
| "eval_steps_per_second": 7.972, |
| "step": 3020 |
| }, |
| { |
| "epoch": 0.3475966502237008, |
| "eval_loss": 0.5929400324821472, |
| "eval_runtime": 12.9184, |
| "eval_samples_per_second": 509.661, |
| "eval_steps_per_second": 7.973, |
| "step": 3030 |
| }, |
| { |
| "epoch": 0.34874383388780544, |
| "eval_loss": 0.5920357704162598, |
| "eval_runtime": 12.9232, |
| "eval_samples_per_second": 509.47, |
| "eval_steps_per_second": 7.97, |
| "step": 3040 |
| }, |
| { |
| "epoch": 0.34989101755191004, |
| "eval_loss": 0.5908279418945312, |
| "eval_runtime": 12.9197, |
| "eval_samples_per_second": 509.609, |
| "eval_steps_per_second": 7.972, |
| "step": 3050 |
| }, |
| { |
| "epoch": 0.3510382012160147, |
| "eval_loss": 0.5896555185317993, |
| "eval_runtime": 13.0254, |
| "eval_samples_per_second": 505.473, |
| "eval_steps_per_second": 7.908, |
| "step": 3060 |
| }, |
| { |
| "epoch": 0.3521853848801193, |
| "eval_loss": 0.5887726545333862, |
| "eval_runtime": 13.0242, |
| "eval_samples_per_second": 505.522, |
| "eval_steps_per_second": 7.908, |
| "step": 3070 |
| }, |
| { |
| "epoch": 0.35333256854422396, |
| "eval_loss": 0.5881541967391968, |
| "eval_runtime": 13.0358, |
| "eval_samples_per_second": 505.069, |
| "eval_steps_per_second": 7.901, |
| "step": 3080 |
| }, |
| { |
| "epoch": 0.35447975220832856, |
| "eval_loss": 0.5873861312866211, |
| "eval_runtime": 12.9582, |
| "eval_samples_per_second": 508.095, |
| "eval_steps_per_second": 7.949, |
| "step": 3090 |
| }, |
| { |
| "epoch": 0.35562693587243316, |
| "grad_norm": 3.1805124282836914, |
| "learning_rate": 9.79434884214999e-06, |
| "loss": 1.0489, |
| "step": 3100 |
| }, |
| { |
| "epoch": 0.35562693587243316, |
| "eval_loss": 0.5868309736251831, |
| "eval_runtime": 13.0375, |
| "eval_samples_per_second": 505.003, |
| "eval_steps_per_second": 7.9, |
| "step": 3100 |
| }, |
| { |
| "epoch": 0.3567741195365378, |
| "eval_loss": 0.5860410928726196, |
| "eval_runtime": 12.9224, |
| "eval_samples_per_second": 509.501, |
| "eval_steps_per_second": 7.971, |
| "step": 3110 |
| }, |
| { |
| "epoch": 0.3579213032006424, |
| "eval_loss": 0.5853646993637085, |
| "eval_runtime": 12.9171, |
| "eval_samples_per_second": 509.711, |
| "eval_steps_per_second": 7.974, |
| "step": 3120 |
| }, |
| { |
| "epoch": 0.35906848686474707, |
| "eval_loss": 0.5839424729347229, |
| "eval_runtime": 12.9136, |
| "eval_samples_per_second": 509.848, |
| "eval_steps_per_second": 7.976, |
| "step": 3130 |
| }, |
| { |
| "epoch": 0.3602156705288517, |
| "eval_loss": 0.5830027461051941, |
| "eval_runtime": 12.9177, |
| "eval_samples_per_second": 509.686, |
| "eval_steps_per_second": 7.974, |
| "step": 3140 |
| }, |
| { |
| "epoch": 0.3613628541929563, |
| "eval_loss": 0.5822171568870544, |
| "eval_runtime": 13.0378, |
| "eval_samples_per_second": 504.995, |
| "eval_steps_per_second": 7.9, |
| "step": 3150 |
| }, |
| { |
| "epoch": 0.36251003785706093, |
| "eval_loss": 0.5814335942268372, |
| "eval_runtime": 12.9196, |
| "eval_samples_per_second": 509.612, |
| "eval_steps_per_second": 7.972, |
| "step": 3160 |
| }, |
| { |
| "epoch": 0.36365722152116553, |
| "eval_loss": 0.5807969570159912, |
| "eval_runtime": 12.9184, |
| "eval_samples_per_second": 509.659, |
| "eval_steps_per_second": 7.973, |
| "step": 3170 |
| }, |
| { |
| "epoch": 0.3648044051852702, |
| "eval_loss": 0.5801600813865662, |
| "eval_runtime": 12.9016, |
| "eval_samples_per_second": 510.323, |
| "eval_steps_per_second": 7.983, |
| "step": 3180 |
| }, |
| { |
| "epoch": 0.3659515888493748, |
| "eval_loss": 0.5793916583061218, |
| "eval_runtime": 12.9241, |
| "eval_samples_per_second": 509.434, |
| "eval_steps_per_second": 7.97, |
| "step": 3190 |
| }, |
| { |
| "epoch": 0.3670987725134794, |
| "grad_norm": 3.2195188999176025, |
| "learning_rate": 9.751858933503294e-06, |
| "loss": 1.038, |
| "step": 3200 |
| }, |
| { |
| "epoch": 0.3670987725134794, |
| "eval_loss": 0.578770637512207, |
| "eval_runtime": 13.0163, |
| "eval_samples_per_second": 505.828, |
| "eval_steps_per_second": 7.913, |
| "step": 3200 |
| }, |
| { |
| "epoch": 0.36824595617758404, |
| "eval_loss": 0.5778489112854004, |
| "eval_runtime": 13.0447, |
| "eval_samples_per_second": 504.726, |
| "eval_steps_per_second": 7.896, |
| "step": 3210 |
| }, |
| { |
| "epoch": 0.36939313984168864, |
| "eval_loss": 0.577012300491333, |
| "eval_runtime": 13.038, |
| "eval_samples_per_second": 504.984, |
| "eval_steps_per_second": 7.9, |
| "step": 3220 |
| }, |
| { |
| "epoch": 0.3705403235057933, |
| "eval_loss": 0.5763078331947327, |
| "eval_runtime": 12.9775, |
| "eval_samples_per_second": 507.339, |
| "eval_steps_per_second": 7.937, |
| "step": 3230 |
| }, |
| { |
| "epoch": 0.3716875071698979, |
| "eval_loss": 0.5751996040344238, |
| "eval_runtime": 12.9227, |
| "eval_samples_per_second": 509.492, |
| "eval_steps_per_second": 7.97, |
| "step": 3240 |
| }, |
| { |
| "epoch": 0.3728346908340025, |
| "eval_loss": 0.5744589567184448, |
| "eval_runtime": 12.9251, |
| "eval_samples_per_second": 509.398, |
| "eval_steps_per_second": 7.969, |
| "step": 3250 |
| }, |
| { |
| "epoch": 0.37398187449810716, |
| "eval_loss": 0.5737271308898926, |
| "eval_runtime": 12.9481, |
| "eval_samples_per_second": 508.493, |
| "eval_steps_per_second": 7.955, |
| "step": 3260 |
| }, |
| { |
| "epoch": 0.37512905816221176, |
| "eval_loss": 0.5727923512458801, |
| "eval_runtime": 13.0412, |
| "eval_samples_per_second": 504.861, |
| "eval_steps_per_second": 7.898, |
| "step": 3270 |
| }, |
| { |
| "epoch": 0.3762762418263164, |
| "eval_loss": 0.5719892382621765, |
| "eval_runtime": 12.9946, |
| "eval_samples_per_second": 506.674, |
| "eval_steps_per_second": 7.926, |
| "step": 3280 |
| }, |
| { |
| "epoch": 0.377423425490421, |
| "eval_loss": 0.5713133811950684, |
| "eval_runtime": 12.9204, |
| "eval_samples_per_second": 509.58, |
| "eval_steps_per_second": 7.972, |
| "step": 3290 |
| }, |
| { |
| "epoch": 0.3785706091545256, |
| "grad_norm": 2.951904058456421, |
| "learning_rate": 9.709369024856598e-06, |
| "loss": 1.0058, |
| "step": 3300 |
| }, |
| { |
| "epoch": 0.3785706091545256, |
| "eval_loss": 0.5707195401191711, |
| "eval_runtime": 12.9678, |
| "eval_samples_per_second": 507.718, |
| "eval_steps_per_second": 7.943, |
| "step": 3300 |
| }, |
| { |
| "epoch": 0.3797177928186303, |
| "eval_loss": 0.5699546933174133, |
| "eval_runtime": 12.9635, |
| "eval_samples_per_second": 507.889, |
| "eval_steps_per_second": 7.945, |
| "step": 3310 |
| }, |
| { |
| "epoch": 0.3808649764827349, |
| "eval_loss": 0.5689657926559448, |
| "eval_runtime": 12.9065, |
| "eval_samples_per_second": 510.131, |
| "eval_steps_per_second": 7.98, |
| "step": 3320 |
| }, |
| { |
| "epoch": 0.38201216014683953, |
| "eval_loss": 0.5680745244026184, |
| "eval_runtime": 12.9206, |
| "eval_samples_per_second": 509.574, |
| "eval_steps_per_second": 7.972, |
| "step": 3330 |
| }, |
| { |
| "epoch": 0.38315934381094413, |
| "eval_loss": 0.5672716498374939, |
| "eval_runtime": 12.9202, |
| "eval_samples_per_second": 509.591, |
| "eval_steps_per_second": 7.972, |
| "step": 3340 |
| }, |
| { |
| "epoch": 0.38430652747504873, |
| "eval_loss": 0.5669021606445312, |
| "eval_runtime": 12.9282, |
| "eval_samples_per_second": 509.275, |
| "eval_steps_per_second": 7.967, |
| "step": 3350 |
| }, |
| { |
| "epoch": 0.3854537111391534, |
| "eval_loss": 0.5667062997817993, |
| "eval_runtime": 12.9165, |
| "eval_samples_per_second": 509.737, |
| "eval_steps_per_second": 7.974, |
| "step": 3360 |
| }, |
| { |
| "epoch": 0.386600894803258, |
| "eval_loss": 0.5664732456207275, |
| "eval_runtime": 12.9163, |
| "eval_samples_per_second": 509.742, |
| "eval_steps_per_second": 7.974, |
| "step": 3370 |
| }, |
| { |
| "epoch": 0.38774807846736264, |
| "eval_loss": 0.5658523440361023, |
| "eval_runtime": 12.9214, |
| "eval_samples_per_second": 509.542, |
| "eval_steps_per_second": 7.971, |
| "step": 3380 |
| }, |
| { |
| "epoch": 0.38889526213146725, |
| "eval_loss": 0.5650349855422974, |
| "eval_runtime": 12.9209, |
| "eval_samples_per_second": 509.562, |
| "eval_steps_per_second": 7.972, |
| "step": 3390 |
| }, |
| { |
| "epoch": 0.39004244579557185, |
| "grad_norm": 2.8982646465301514, |
| "learning_rate": 9.666879116209901e-06, |
| "loss": 1.0413, |
| "step": 3400 |
| }, |
| { |
| "epoch": 0.39004244579557185, |
| "eval_loss": 0.5644938349723816, |
| "eval_runtime": 12.9147, |
| "eval_samples_per_second": 509.808, |
| "eval_steps_per_second": 7.975, |
| "step": 3400 |
| }, |
| { |
| "epoch": 0.3911896294596765, |
| "eval_loss": 0.5641273260116577, |
| "eval_runtime": 12.9416, |
| "eval_samples_per_second": 508.745, |
| "eval_steps_per_second": 7.959, |
| "step": 3410 |
| }, |
| { |
| "epoch": 0.3923368131237811, |
| "eval_loss": 0.5635139346122742, |
| "eval_runtime": 12.9481, |
| "eval_samples_per_second": 508.491, |
| "eval_steps_per_second": 7.955, |
| "step": 3420 |
| }, |
| { |
| "epoch": 0.39348399678788576, |
| "eval_loss": 0.5628678202629089, |
| "eval_runtime": 12.9965, |
| "eval_samples_per_second": 506.596, |
| "eval_steps_per_second": 7.925, |
| "step": 3430 |
| }, |
| { |
| "epoch": 0.39463118045199036, |
| "eval_loss": 0.5622407793998718, |
| "eval_runtime": 12.9926, |
| "eval_samples_per_second": 506.748, |
| "eval_steps_per_second": 7.928, |
| "step": 3440 |
| }, |
| { |
| "epoch": 0.39577836411609496, |
| "eval_loss": 0.5617032051086426, |
| "eval_runtime": 12.92, |
| "eval_samples_per_second": 509.598, |
| "eval_steps_per_second": 7.972, |
| "step": 3450 |
| }, |
| { |
| "epoch": 0.3969255477801996, |
| "eval_loss": 0.5613731741905212, |
| "eval_runtime": 12.9152, |
| "eval_samples_per_second": 509.786, |
| "eval_steps_per_second": 7.975, |
| "step": 3460 |
| }, |
| { |
| "epoch": 0.3980727314443042, |
| "eval_loss": 0.5606963038444519, |
| "eval_runtime": 12.9194, |
| "eval_samples_per_second": 509.62, |
| "eval_steps_per_second": 7.972, |
| "step": 3470 |
| }, |
| { |
| "epoch": 0.3992199151084089, |
| "eval_loss": 0.5602635145187378, |
| "eval_runtime": 12.9242, |
| "eval_samples_per_second": 509.432, |
| "eval_steps_per_second": 7.97, |
| "step": 3480 |
| }, |
| { |
| "epoch": 0.4003670987725135, |
| "eval_loss": 0.5598347783088684, |
| "eval_runtime": 12.9579, |
| "eval_samples_per_second": 508.107, |
| "eval_steps_per_second": 7.949, |
| "step": 3490 |
| }, |
| { |
| "epoch": 0.4015142824366181, |
| "grad_norm": 3.217618465423584, |
| "learning_rate": 9.624389207563205e-06, |
| "loss": 0.938, |
| "step": 3500 |
| }, |
| { |
| "epoch": 0.4015142824366181, |
| "eval_loss": 0.5596053600311279, |
| "eval_runtime": 13.0466, |
| "eval_samples_per_second": 504.652, |
| "eval_steps_per_second": 7.895, |
| "step": 3500 |
| }, |
| { |
| "epoch": 0.40266146610072273, |
| "eval_loss": 0.5589408874511719, |
| "eval_runtime": 13.0715, |
| "eval_samples_per_second": 503.69, |
| "eval_steps_per_second": 7.88, |
| "step": 3510 |
| }, |
| { |
| "epoch": 0.40380864976482733, |
| "eval_loss": 0.558090329170227, |
| "eval_runtime": 12.9472, |
| "eval_samples_per_second": 508.528, |
| "eval_steps_per_second": 7.955, |
| "step": 3520 |
| }, |
| { |
| "epoch": 0.404955833428932, |
| "eval_loss": 0.5570879578590393, |
| "eval_runtime": 12.9169, |
| "eval_samples_per_second": 509.72, |
| "eval_steps_per_second": 7.974, |
| "step": 3530 |
| }, |
| { |
| "epoch": 0.4061030170930366, |
| "eval_loss": 0.5562835931777954, |
| "eval_runtime": 12.9892, |
| "eval_samples_per_second": 506.881, |
| "eval_steps_per_second": 7.93, |
| "step": 3540 |
| }, |
| { |
| "epoch": 0.40725020075714125, |
| "eval_loss": 0.5557334423065186, |
| "eval_runtime": 12.9142, |
| "eval_samples_per_second": 509.826, |
| "eval_steps_per_second": 7.976, |
| "step": 3550 |
| }, |
| { |
| "epoch": 0.40839738442124585, |
| "eval_loss": 0.5551453828811646, |
| "eval_runtime": 12.9967, |
| "eval_samples_per_second": 506.59, |
| "eval_steps_per_second": 7.925, |
| "step": 3560 |
| }, |
| { |
| "epoch": 0.40954456808535045, |
| "eval_loss": 0.5546151995658875, |
| "eval_runtime": 13.0283, |
| "eval_samples_per_second": 505.362, |
| "eval_steps_per_second": 7.906, |
| "step": 3570 |
| }, |
| { |
| "epoch": 0.4106917517494551, |
| "eval_loss": 0.5540978312492371, |
| "eval_runtime": 13.0306, |
| "eval_samples_per_second": 505.273, |
| "eval_steps_per_second": 7.904, |
| "step": 3580 |
| }, |
| { |
| "epoch": 0.4118389354135597, |
| "eval_loss": 0.5535085797309875, |
| "eval_runtime": 13.0453, |
| "eval_samples_per_second": 504.704, |
| "eval_steps_per_second": 7.896, |
| "step": 3590 |
| }, |
| { |
| "epoch": 0.41298611907766436, |
| "grad_norm": 3.073883056640625, |
| "learning_rate": 9.581899298916509e-06, |
| "loss": 0.955, |
| "step": 3600 |
| }, |
| { |
| "epoch": 0.41298611907766436, |
| "eval_loss": 0.5527840256690979, |
| "eval_runtime": 13.0322, |
| "eval_samples_per_second": 505.21, |
| "eval_steps_per_second": 7.903, |
| "step": 3600 |
| }, |
| { |
| "epoch": 0.41413330274176896, |
| "eval_loss": 0.5521849989891052, |
| "eval_runtime": 13.0324, |
| "eval_samples_per_second": 505.202, |
| "eval_steps_per_second": 7.903, |
| "step": 3610 |
| }, |
| { |
| "epoch": 0.41528048640587356, |
| "eval_loss": 0.5515934824943542, |
| "eval_runtime": 13.0428, |
| "eval_samples_per_second": 504.799, |
| "eval_steps_per_second": 7.897, |
| "step": 3620 |
| }, |
| { |
| "epoch": 0.4164276700699782, |
| "eval_loss": 0.5508650541305542, |
| "eval_runtime": 13.0451, |
| "eval_samples_per_second": 504.709, |
| "eval_steps_per_second": 7.896, |
| "step": 3630 |
| }, |
| { |
| "epoch": 0.4175748537340828, |
| "eval_loss": 0.5502671003341675, |
| "eval_runtime": 13.0251, |
| "eval_samples_per_second": 505.484, |
| "eval_steps_per_second": 7.908, |
| "step": 3640 |
| }, |
| { |
| "epoch": 0.4187220373981875, |
| "eval_loss": 0.5495434403419495, |
| "eval_runtime": 13.0242, |
| "eval_samples_per_second": 505.519, |
| "eval_steps_per_second": 7.908, |
| "step": 3650 |
| }, |
| { |
| "epoch": 0.4198692210622921, |
| "eval_loss": 0.5489790439605713, |
| "eval_runtime": 13.0389, |
| "eval_samples_per_second": 504.951, |
| "eval_steps_per_second": 7.899, |
| "step": 3660 |
| }, |
| { |
| "epoch": 0.4210164047263967, |
| "eval_loss": 0.5481202006340027, |
| "eval_runtime": 12.9193, |
| "eval_samples_per_second": 509.627, |
| "eval_steps_per_second": 7.973, |
| "step": 3670 |
| }, |
| { |
| "epoch": 0.42216358839050133, |
| "eval_loss": 0.5474987030029297, |
| "eval_runtime": 13.0455, |
| "eval_samples_per_second": 504.697, |
| "eval_steps_per_second": 7.895, |
| "step": 3680 |
| }, |
| { |
| "epoch": 0.42331077205460593, |
| "eval_loss": 0.5467217564582825, |
| "eval_runtime": 12.9516, |
| "eval_samples_per_second": 508.356, |
| "eval_steps_per_second": 7.953, |
| "step": 3690 |
| }, |
| { |
| "epoch": 0.4244579557187106, |
| "grad_norm": 4.54811954498291, |
| "learning_rate": 9.539409390269812e-06, |
| "loss": 0.9387, |
| "step": 3700 |
| }, |
| { |
| "epoch": 0.4244579557187106, |
| "eval_loss": 0.5463354587554932, |
| "eval_runtime": 12.949, |
| "eval_samples_per_second": 508.456, |
| "eval_steps_per_second": 7.954, |
| "step": 3700 |
| }, |
| { |
| "epoch": 0.4256051393828152, |
| "eval_loss": 0.5458938479423523, |
| "eval_runtime": 12.9107, |
| "eval_samples_per_second": 509.966, |
| "eval_steps_per_second": 7.978, |
| "step": 3710 |
| }, |
| { |
| "epoch": 0.4267523230469198, |
| "eval_loss": 0.5451832413673401, |
| "eval_runtime": 13.0127, |
| "eval_samples_per_second": 505.968, |
| "eval_steps_per_second": 7.915, |
| "step": 3720 |
| }, |
| { |
| "epoch": 0.42789950671102445, |
| "eval_loss": 0.5448161363601685, |
| "eval_runtime": 13.033, |
| "eval_samples_per_second": 505.18, |
| "eval_steps_per_second": 7.903, |
| "step": 3730 |
| }, |
| { |
| "epoch": 0.42904669037512905, |
| "eval_loss": 0.5443439483642578, |
| "eval_runtime": 13.0146, |
| "eval_samples_per_second": 505.895, |
| "eval_steps_per_second": 7.914, |
| "step": 3740 |
| }, |
| { |
| "epoch": 0.4301938740392337, |
| "eval_loss": 0.544032633304596, |
| "eval_runtime": 12.9589, |
| "eval_samples_per_second": 508.069, |
| "eval_steps_per_second": 7.948, |
| "step": 3750 |
| }, |
| { |
| "epoch": 0.4313410577033383, |
| "eval_loss": 0.5435395836830139, |
| "eval_runtime": 12.9098, |
| "eval_samples_per_second": 510.001, |
| "eval_steps_per_second": 7.978, |
| "step": 3760 |
| }, |
| { |
| "epoch": 0.4324882413674429, |
| "eval_loss": 0.5429771542549133, |
| "eval_runtime": 12.9366, |
| "eval_samples_per_second": 508.944, |
| "eval_steps_per_second": 7.962, |
| "step": 3770 |
| }, |
| { |
| "epoch": 0.43363542503154756, |
| "eval_loss": 0.5423203706741333, |
| "eval_runtime": 13.1085, |
| "eval_samples_per_second": 502.268, |
| "eval_steps_per_second": 7.857, |
| "step": 3780 |
| }, |
| { |
| "epoch": 0.43478260869565216, |
| "eval_loss": 0.5417954921722412, |
| "eval_runtime": 13.0184, |
| "eval_samples_per_second": 505.747, |
| "eval_steps_per_second": 7.912, |
| "step": 3790 |
| }, |
| { |
| "epoch": 0.4359297923597568, |
| "grad_norm": 6.273746967315674, |
| "learning_rate": 9.496919481623116e-06, |
| "loss": 0.9672, |
| "step": 3800 |
| }, |
| { |
| "epoch": 0.4359297923597568, |
| "eval_loss": 0.5414963364601135, |
| "eval_runtime": 13.0358, |
| "eval_samples_per_second": 505.07, |
| "eval_steps_per_second": 7.901, |
| "step": 3800 |
| }, |
| { |
| "epoch": 0.4370769760238614, |
| "eval_loss": 0.541257381439209, |
| "eval_runtime": 12.9726, |
| "eval_samples_per_second": 507.53, |
| "eval_steps_per_second": 7.94, |
| "step": 3810 |
| }, |
| { |
| "epoch": 0.438224159687966, |
| "eval_loss": 0.5409945845603943, |
| "eval_runtime": 13.0274, |
| "eval_samples_per_second": 505.398, |
| "eval_steps_per_second": 7.906, |
| "step": 3820 |
| }, |
| { |
| "epoch": 0.4393713433520707, |
| "eval_loss": 0.5406415462493896, |
| "eval_runtime": 13.0472, |
| "eval_samples_per_second": 504.629, |
| "eval_steps_per_second": 7.894, |
| "step": 3830 |
| }, |
| { |
| "epoch": 0.4405185270161753, |
| "eval_loss": 0.5403361916542053, |
| "eval_runtime": 12.9682, |
| "eval_samples_per_second": 507.704, |
| "eval_steps_per_second": 7.943, |
| "step": 3840 |
| }, |
| { |
| "epoch": 0.44166571068027993, |
| "eval_loss": 0.5397402048110962, |
| "eval_runtime": 13.0503, |
| "eval_samples_per_second": 504.509, |
| "eval_steps_per_second": 7.893, |
| "step": 3850 |
| }, |
| { |
| "epoch": 0.44281289434438453, |
| "eval_loss": 0.5394212007522583, |
| "eval_runtime": 12.9962, |
| "eval_samples_per_second": 506.611, |
| "eval_steps_per_second": 7.925, |
| "step": 3860 |
| }, |
| { |
| "epoch": 0.44396007800848913, |
| "eval_loss": 0.538590669631958, |
| "eval_runtime": 13.0433, |
| "eval_samples_per_second": 504.78, |
| "eval_steps_per_second": 7.897, |
| "step": 3870 |
| }, |
| { |
| "epoch": 0.4451072616725938, |
| "eval_loss": 0.537807822227478, |
| "eval_runtime": 12.9414, |
| "eval_samples_per_second": 508.756, |
| "eval_steps_per_second": 7.959, |
| "step": 3880 |
| }, |
| { |
| "epoch": 0.4462544453366984, |
| "eval_loss": 0.536992073059082, |
| "eval_runtime": 12.9902, |
| "eval_samples_per_second": 506.844, |
| "eval_steps_per_second": 7.929, |
| "step": 3890 |
| }, |
| { |
| "epoch": 0.44740162900080305, |
| "grad_norm": 2.6543989181518555, |
| "learning_rate": 9.45442957297642e-06, |
| "loss": 0.926, |
| "step": 3900 |
| }, |
| { |
| "epoch": 0.44740162900080305, |
| "eval_loss": 0.5359557867050171, |
| "eval_runtime": 13.0495, |
| "eval_samples_per_second": 504.542, |
| "eval_steps_per_second": 7.893, |
| "step": 3900 |
| }, |
| { |
| "epoch": 0.44854881266490765, |
| "eval_loss": 0.5351366400718689, |
| "eval_runtime": 12.923, |
| "eval_samples_per_second": 509.48, |
| "eval_steps_per_second": 7.97, |
| "step": 3910 |
| }, |
| { |
| "epoch": 0.44969599632901225, |
| "eval_loss": 0.5345980525016785, |
| "eval_runtime": 12.9402, |
| "eval_samples_per_second": 508.803, |
| "eval_steps_per_second": 7.96, |
| "step": 3920 |
| }, |
| { |
| "epoch": 0.4508431799931169, |
| "eval_loss": 0.5343027710914612, |
| "eval_runtime": 13.0184, |
| "eval_samples_per_second": 505.747, |
| "eval_steps_per_second": 7.912, |
| "step": 3930 |
| }, |
| { |
| "epoch": 0.4519903636572215, |
| "eval_loss": 0.5339150428771973, |
| "eval_runtime": 13.0383, |
| "eval_samples_per_second": 504.974, |
| "eval_steps_per_second": 7.9, |
| "step": 3940 |
| }, |
| { |
| "epoch": 0.45313754732132616, |
| "eval_loss": 0.533658504486084, |
| "eval_runtime": 13.0155, |
| "eval_samples_per_second": 505.857, |
| "eval_steps_per_second": 7.914, |
| "step": 3950 |
| }, |
| { |
| "epoch": 0.45428473098543076, |
| "eval_loss": 0.5334100127220154, |
| "eval_runtime": 12.9238, |
| "eval_samples_per_second": 509.45, |
| "eval_steps_per_second": 7.97, |
| "step": 3960 |
| }, |
| { |
| "epoch": 0.45543191464953536, |
| "eval_loss": 0.5330451726913452, |
| "eval_runtime": 12.9194, |
| "eval_samples_per_second": 509.619, |
| "eval_steps_per_second": 7.972, |
| "step": 3970 |
| }, |
| { |
| "epoch": 0.45657909831364, |
| "eval_loss": 0.5327152609825134, |
| "eval_runtime": 13.015, |
| "eval_samples_per_second": 505.879, |
| "eval_steps_per_second": 7.914, |
| "step": 3980 |
| }, |
| { |
| "epoch": 0.4577262819777446, |
| "eval_loss": 0.5323871374130249, |
| "eval_runtime": 13.021, |
| "eval_samples_per_second": 505.646, |
| "eval_steps_per_second": 7.91, |
| "step": 3990 |
| }, |
| { |
| "epoch": 0.4588734656418493, |
| "grad_norm": 2.330146074295044, |
| "learning_rate": 9.411939664329721e-06, |
| "loss": 0.867, |
| "step": 4000 |
| }, |
| { |
| "epoch": 0.4588734656418493, |
| "eval_loss": 0.5319190621376038, |
| "eval_runtime": 13.0222, |
| "eval_samples_per_second": 505.6, |
| "eval_steps_per_second": 7.91, |
| "step": 4000 |
| }, |
| { |
| "epoch": 0.4600206493059539, |
| "eval_loss": 0.5313332676887512, |
| "eval_runtime": 13.0478, |
| "eval_samples_per_second": 504.605, |
| "eval_steps_per_second": 7.894, |
| "step": 4010 |
| }, |
| { |
| "epoch": 0.46116783297005853, |
| "eval_loss": 0.5308486223220825, |
| "eval_runtime": 12.964, |
| "eval_samples_per_second": 507.868, |
| "eval_steps_per_second": 7.945, |
| "step": 4020 |
| }, |
| { |
| "epoch": 0.46231501663416313, |
| "eval_loss": 0.5300149917602539, |
| "eval_runtime": 12.914, |
| "eval_samples_per_second": 509.835, |
| "eval_steps_per_second": 7.976, |
| "step": 4030 |
| }, |
| { |
| "epoch": 0.46346220029826773, |
| "eval_loss": 0.5292573571205139, |
| "eval_runtime": 12.9769, |
| "eval_samples_per_second": 507.361, |
| "eval_steps_per_second": 7.937, |
| "step": 4040 |
| }, |
| { |
| "epoch": 0.4646093839623724, |
| "eval_loss": 0.5286610722541809, |
| "eval_runtime": 13.0368, |
| "eval_samples_per_second": 505.031, |
| "eval_steps_per_second": 7.901, |
| "step": 4050 |
| }, |
| { |
| "epoch": 0.465756567626477, |
| "eval_loss": 0.5283547043800354, |
| "eval_runtime": 13.0356, |
| "eval_samples_per_second": 505.077, |
| "eval_steps_per_second": 7.901, |
| "step": 4060 |
| }, |
| { |
| "epoch": 0.46690375129058165, |
| "eval_loss": 0.5280731916427612, |
| "eval_runtime": 13.0048, |
| "eval_samples_per_second": 506.275, |
| "eval_steps_per_second": 7.92, |
| "step": 4070 |
| }, |
| { |
| "epoch": 0.46805093495468625, |
| "eval_loss": 0.5276577472686768, |
| "eval_runtime": 12.9174, |
| "eval_samples_per_second": 509.699, |
| "eval_steps_per_second": 7.974, |
| "step": 4080 |
| }, |
| { |
| "epoch": 0.46919811861879085, |
| "eval_loss": 0.5272203087806702, |
| "eval_runtime": 12.9134, |
| "eval_samples_per_second": 509.857, |
| "eval_steps_per_second": 7.976, |
| "step": 4090 |
| }, |
| { |
| "epoch": 0.4703453022828955, |
| "grad_norm": 2.8037149906158447, |
| "learning_rate": 9.369449755683025e-06, |
| "loss": 0.916, |
| "step": 4100 |
| }, |
| { |
| "epoch": 0.4703453022828955, |
| "eval_loss": 0.5267237424850464, |
| "eval_runtime": 12.9151, |
| "eval_samples_per_second": 509.79, |
| "eval_steps_per_second": 7.975, |
| "step": 4100 |
| }, |
| { |
| "epoch": 0.4714924859470001, |
| "eval_loss": 0.5260052680969238, |
| "eval_runtime": 12.9148, |
| "eval_samples_per_second": 509.801, |
| "eval_steps_per_second": 7.975, |
| "step": 4110 |
| }, |
| { |
| "epoch": 0.47263966961110476, |
| "eval_loss": 0.5252425074577332, |
| "eval_runtime": 13.014, |
| "eval_samples_per_second": 505.915, |
| "eval_steps_per_second": 7.915, |
| "step": 4120 |
| }, |
| { |
| "epoch": 0.47378685327520936, |
| "eval_loss": 0.5246235132217407, |
| "eval_runtime": 13.0291, |
| "eval_samples_per_second": 505.332, |
| "eval_steps_per_second": 7.905, |
| "step": 4130 |
| }, |
| { |
| "epoch": 0.47493403693931396, |
| "eval_loss": 0.5238820314407349, |
| "eval_runtime": 12.9264, |
| "eval_samples_per_second": 509.345, |
| "eval_steps_per_second": 7.968, |
| "step": 4140 |
| }, |
| { |
| "epoch": 0.4760812206034186, |
| "eval_loss": 0.5231978297233582, |
| "eval_runtime": 12.9421, |
| "eval_samples_per_second": 508.729, |
| "eval_steps_per_second": 7.959, |
| "step": 4150 |
| }, |
| { |
| "epoch": 0.4772284042675232, |
| "eval_loss": 0.5225175023078918, |
| "eval_runtime": 13.0292, |
| "eval_samples_per_second": 505.328, |
| "eval_steps_per_second": 7.905, |
| "step": 4160 |
| }, |
| { |
| "epoch": 0.4783755879316279, |
| "eval_loss": 0.522089958190918, |
| "eval_runtime": 13.0462, |
| "eval_samples_per_second": 504.666, |
| "eval_steps_per_second": 7.895, |
| "step": 4170 |
| }, |
| { |
| "epoch": 0.4795227715957325, |
| "eval_loss": 0.5216270685195923, |
| "eval_runtime": 13.0261, |
| "eval_samples_per_second": 505.448, |
| "eval_steps_per_second": 7.907, |
| "step": 4180 |
| }, |
| { |
| "epoch": 0.4806699552598371, |
| "eval_loss": 0.5211179256439209, |
| "eval_runtime": 13.0229, |
| "eval_samples_per_second": 505.569, |
| "eval_steps_per_second": 7.909, |
| "step": 4190 |
| }, |
| { |
| "epoch": 0.48181713892394173, |
| "grad_norm": 4.981523036956787, |
| "learning_rate": 9.32695984703633e-06, |
| "loss": 0.9667, |
| "step": 4200 |
| }, |
| { |
| "epoch": 0.48181713892394173, |
| "eval_loss": 0.5206459760665894, |
| "eval_runtime": 13.0189, |
| "eval_samples_per_second": 505.728, |
| "eval_steps_per_second": 7.912, |
| "step": 4200 |
| }, |
| { |
| "epoch": 0.48296432258804634, |
| "eval_loss": 0.5203832983970642, |
| "eval_runtime": 13.0372, |
| "eval_samples_per_second": 505.016, |
| "eval_steps_per_second": 7.9, |
| "step": 4210 |
| }, |
| { |
| "epoch": 0.484111506252151, |
| "eval_loss": 0.5200340151786804, |
| "eval_runtime": 12.921, |
| "eval_samples_per_second": 509.558, |
| "eval_steps_per_second": 7.972, |
| "step": 4220 |
| }, |
| { |
| "epoch": 0.4852586899162556, |
| "eval_loss": 0.5191519856452942, |
| "eval_runtime": 12.9412, |
| "eval_samples_per_second": 508.761, |
| "eval_steps_per_second": 7.959, |
| "step": 4230 |
| }, |
| { |
| "epoch": 0.4864058735803602, |
| "eval_loss": 0.5187414884567261, |
| "eval_runtime": 13.0244, |
| "eval_samples_per_second": 505.513, |
| "eval_steps_per_second": 7.908, |
| "step": 4240 |
| }, |
| { |
| "epoch": 0.48755305724446485, |
| "eval_loss": 0.5184528827667236, |
| "eval_runtime": 13.0462, |
| "eval_samples_per_second": 504.669, |
| "eval_steps_per_second": 7.895, |
| "step": 4250 |
| }, |
| { |
| "epoch": 0.48870024090856945, |
| "eval_loss": 0.5179461240768433, |
| "eval_runtime": 12.9404, |
| "eval_samples_per_second": 508.793, |
| "eval_steps_per_second": 7.96, |
| "step": 4260 |
| }, |
| { |
| "epoch": 0.4898474245726741, |
| "eval_loss": 0.5173225402832031, |
| "eval_runtime": 12.9886, |
| "eval_samples_per_second": 506.904, |
| "eval_steps_per_second": 7.93, |
| "step": 4270 |
| }, |
| { |
| "epoch": 0.4909946082367787, |
| "eval_loss": 0.5169667601585388, |
| "eval_runtime": 12.918, |
| "eval_samples_per_second": 509.678, |
| "eval_steps_per_second": 7.973, |
| "step": 4280 |
| }, |
| { |
| "epoch": 0.4921417919008833, |
| "eval_loss": 0.5165331363677979, |
| "eval_runtime": 12.9162, |
| "eval_samples_per_second": 509.747, |
| "eval_steps_per_second": 7.974, |
| "step": 4290 |
| }, |
| { |
| "epoch": 0.49328897556498796, |
| "grad_norm": 2.352576732635498, |
| "learning_rate": 9.284469938389632e-06, |
| "loss": 0.9276, |
| "step": 4300 |
| }, |
| { |
| "epoch": 0.49328897556498796, |
| "eval_loss": 0.5160226821899414, |
| "eval_runtime": 12.9928, |
| "eval_samples_per_second": 506.741, |
| "eval_steps_per_second": 7.927, |
| "step": 4300 |
| }, |
| { |
| "epoch": 0.49443615922909256, |
| "eval_loss": 0.5154353976249695, |
| "eval_runtime": 13.0317, |
| "eval_samples_per_second": 505.228, |
| "eval_steps_per_second": 7.904, |
| "step": 4310 |
| }, |
| { |
| "epoch": 0.4955833428931972, |
| "eval_loss": 0.5149873495101929, |
| "eval_runtime": 12.9748, |
| "eval_samples_per_second": 507.445, |
| "eval_steps_per_second": 7.938, |
| "step": 4320 |
| }, |
| { |
| "epoch": 0.4967305265573018, |
| "eval_loss": 0.514439582824707, |
| "eval_runtime": 12.9649, |
| "eval_samples_per_second": 507.833, |
| "eval_steps_per_second": 7.945, |
| "step": 4330 |
| }, |
| { |
| "epoch": 0.4978777102214064, |
| "eval_loss": 0.5140998363494873, |
| "eval_runtime": 13.0221, |
| "eval_samples_per_second": 505.603, |
| "eval_steps_per_second": 7.91, |
| "step": 4340 |
| }, |
| { |
| "epoch": 0.4990248938855111, |
| "eval_loss": 0.5139394402503967, |
| "eval_runtime": 13.0098, |
| "eval_samples_per_second": 506.081, |
| "eval_steps_per_second": 7.917, |
| "step": 4350 |
| }, |
| { |
| "epoch": 0.5001720775496157, |
| "eval_loss": 0.5137957334518433, |
| "eval_runtime": 13.0381, |
| "eval_samples_per_second": 504.983, |
| "eval_steps_per_second": 7.9, |
| "step": 4360 |
| }, |
| { |
| "epoch": 0.5013192612137203, |
| "eval_loss": 0.5136282444000244, |
| "eval_runtime": 13.0808, |
| "eval_samples_per_second": 503.332, |
| "eval_steps_per_second": 7.874, |
| "step": 4370 |
| }, |
| { |
| "epoch": 0.502466444877825, |
| "eval_loss": 0.5132753849029541, |
| "eval_runtime": 13.0402, |
| "eval_samples_per_second": 504.902, |
| "eval_steps_per_second": 7.899, |
| "step": 4380 |
| }, |
| { |
| "epoch": 0.5036136285419296, |
| "eval_loss": 0.5129442811012268, |
| "eval_runtime": 12.9179, |
| "eval_samples_per_second": 509.682, |
| "eval_steps_per_second": 7.973, |
| "step": 4390 |
| }, |
| { |
| "epoch": 0.5047608122060342, |
| "grad_norm": 2.744755983352661, |
| "learning_rate": 9.241980029742936e-06, |
| "loss": 0.9331, |
| "step": 4400 |
| }, |
| { |
| "epoch": 0.5047608122060342, |
| "eval_loss": 0.5125741362571716, |
| "eval_runtime": 12.9316, |
| "eval_samples_per_second": 509.142, |
| "eval_steps_per_second": 7.965, |
| "step": 4400 |
| }, |
| { |
| "epoch": 0.5059079958701388, |
| "eval_loss": 0.5123194456100464, |
| "eval_runtime": 13.0104, |
| "eval_samples_per_second": 506.056, |
| "eval_steps_per_second": 7.917, |
| "step": 4410 |
| }, |
| { |
| "epoch": 0.5070551795342434, |
| "eval_loss": 0.5116878747940063, |
| "eval_runtime": 13.0458, |
| "eval_samples_per_second": 504.684, |
| "eval_steps_per_second": 7.895, |
| "step": 4420 |
| }, |
| { |
| "epoch": 0.5082023631983481, |
| "eval_loss": 0.5112791061401367, |
| "eval_runtime": 13.0399, |
| "eval_samples_per_second": 504.911, |
| "eval_steps_per_second": 7.899, |
| "step": 4430 |
| }, |
| { |
| "epoch": 0.5093495468624527, |
| "eval_loss": 0.5108391642570496, |
| "eval_runtime": 13.0444, |
| "eval_samples_per_second": 504.738, |
| "eval_steps_per_second": 7.896, |
| "step": 4440 |
| }, |
| { |
| "epoch": 0.5104967305265573, |
| "eval_loss": 0.5106413960456848, |
| "eval_runtime": 13.0495, |
| "eval_samples_per_second": 504.542, |
| "eval_steps_per_second": 7.893, |
| "step": 4450 |
| }, |
| { |
| "epoch": 0.5116439141906619, |
| "eval_loss": 0.5105773210525513, |
| "eval_runtime": 13.026, |
| "eval_samples_per_second": 505.453, |
| "eval_steps_per_second": 7.907, |
| "step": 4460 |
| }, |
| { |
| "epoch": 0.5127910978547665, |
| "eval_loss": 0.5106124877929688, |
| "eval_runtime": 13.0887, |
| "eval_samples_per_second": 503.028, |
| "eval_steps_per_second": 7.869, |
| "step": 4470 |
| }, |
| { |
| "epoch": 0.5139382815188712, |
| "eval_loss": 0.5104395151138306, |
| "eval_runtime": 13.0488, |
| "eval_samples_per_second": 504.567, |
| "eval_steps_per_second": 7.893, |
| "step": 4480 |
| }, |
| { |
| "epoch": 0.5150854651829758, |
| "eval_loss": 0.510217547416687, |
| "eval_runtime": 13.0309, |
| "eval_samples_per_second": 505.261, |
| "eval_steps_per_second": 7.904, |
| "step": 4490 |
| }, |
| { |
| "epoch": 0.5162326488470804, |
| "grad_norm": 3.2935314178466797, |
| "learning_rate": 9.199490121096241e-06, |
| "loss": 0.907, |
| "step": 4500 |
| }, |
| { |
| "epoch": 0.5162326488470804, |
| "eval_loss": 0.5096931457519531, |
| "eval_runtime": 13.0419, |
| "eval_samples_per_second": 504.834, |
| "eval_steps_per_second": 7.898, |
| "step": 4500 |
| }, |
| { |
| "epoch": 0.517379832511185, |
| "eval_loss": 0.5092429518699646, |
| "eval_runtime": 13.0377, |
| "eval_samples_per_second": 504.999, |
| "eval_steps_per_second": 7.9, |
| "step": 4510 |
| }, |
| { |
| "epoch": 0.5185270161752896, |
| "eval_loss": 0.5086391568183899, |
| "eval_runtime": 12.9418, |
| "eval_samples_per_second": 508.739, |
| "eval_steps_per_second": 7.959, |
| "step": 4520 |
| }, |
| { |
| "epoch": 0.5196741998393943, |
| "eval_loss": 0.508176326751709, |
| "eval_runtime": 13.154, |
| "eval_samples_per_second": 500.533, |
| "eval_steps_per_second": 7.83, |
| "step": 4530 |
| }, |
| { |
| "epoch": 0.5208213835034989, |
| "eval_loss": 0.5079353451728821, |
| "eval_runtime": 13.0269, |
| "eval_samples_per_second": 505.417, |
| "eval_steps_per_second": 7.907, |
| "step": 4540 |
| }, |
| { |
| "epoch": 0.5219685671676035, |
| "eval_loss": 0.5075249671936035, |
| "eval_runtime": 13.0283, |
| "eval_samples_per_second": 505.36, |
| "eval_steps_per_second": 7.906, |
| "step": 4550 |
| }, |
| { |
| "epoch": 0.5231157508317081, |
| "eval_loss": 0.5070711970329285, |
| "eval_runtime": 13.0867, |
| "eval_samples_per_second": 503.106, |
| "eval_steps_per_second": 7.871, |
| "step": 4560 |
| }, |
| { |
| "epoch": 0.5242629344958127, |
| "eval_loss": 0.5066924095153809, |
| "eval_runtime": 13.0158, |
| "eval_samples_per_second": 505.847, |
| "eval_steps_per_second": 7.913, |
| "step": 4570 |
| }, |
| { |
| "epoch": 0.5254101181599174, |
| "eval_loss": 0.5066002607345581, |
| "eval_runtime": 13.0247, |
| "eval_samples_per_second": 505.499, |
| "eval_steps_per_second": 7.908, |
| "step": 4580 |
| }, |
| { |
| "epoch": 0.526557301824022, |
| "eval_loss": 0.5061988830566406, |
| "eval_runtime": 12.9452, |
| "eval_samples_per_second": 508.604, |
| "eval_steps_per_second": 7.957, |
| "step": 4590 |
| }, |
| { |
| "epoch": 0.5277044854881267, |
| "grad_norm": 2.8888604640960693, |
| "learning_rate": 9.157000212449543e-06, |
| "loss": 0.913, |
| "step": 4600 |
| }, |
| { |
| "epoch": 0.5277044854881267, |
| "eval_loss": 0.5059130787849426, |
| "eval_runtime": 13.0522, |
| "eval_samples_per_second": 504.435, |
| "eval_steps_per_second": 7.891, |
| "step": 4600 |
| }, |
| { |
| "epoch": 0.5288516691522313, |
| "eval_loss": 0.5056325793266296, |
| "eval_runtime": 13.0474, |
| "eval_samples_per_second": 504.62, |
| "eval_steps_per_second": 7.894, |
| "step": 4610 |
| }, |
| { |
| "epoch": 0.5299988528163359, |
| "eval_loss": 0.5051873922348022, |
| "eval_runtime": 13.0297, |
| "eval_samples_per_second": 505.306, |
| "eval_steps_per_second": 7.905, |
| "step": 4620 |
| }, |
| { |
| "epoch": 0.5311460364804406, |
| "eval_loss": 0.5046107769012451, |
| "eval_runtime": 13.0544, |
| "eval_samples_per_second": 504.351, |
| "eval_steps_per_second": 7.89, |
| "step": 4630 |
| }, |
| { |
| "epoch": 0.5322932201445452, |
| "eval_loss": 0.5039039254188538, |
| "eval_runtime": 13.0434, |
| "eval_samples_per_second": 504.778, |
| "eval_steps_per_second": 7.897, |
| "step": 4640 |
| }, |
| { |
| "epoch": 0.5334404038086498, |
| "eval_loss": 0.5032686591148376, |
| "eval_runtime": 13.0225, |
| "eval_samples_per_second": 505.586, |
| "eval_steps_per_second": 7.909, |
| "step": 4650 |
| }, |
| { |
| "epoch": 0.5345875874727544, |
| "eval_loss": 0.5030205249786377, |
| "eval_runtime": 13.044, |
| "eval_samples_per_second": 504.753, |
| "eval_steps_per_second": 7.896, |
| "step": 4660 |
| }, |
| { |
| "epoch": 0.535734771136859, |
| "eval_loss": 0.5027660727500916, |
| "eval_runtime": 13.0891, |
| "eval_samples_per_second": 503.013, |
| "eval_steps_per_second": 7.869, |
| "step": 4670 |
| }, |
| { |
| "epoch": 0.5368819548009637, |
| "eval_loss": 0.5026541352272034, |
| "eval_runtime": 13.0208, |
| "eval_samples_per_second": 505.653, |
| "eval_steps_per_second": 7.91, |
| "step": 4680 |
| }, |
| { |
| "epoch": 0.5380291384650683, |
| "eval_loss": 0.5023172497749329, |
| "eval_runtime": 13.0369, |
| "eval_samples_per_second": 505.028, |
| "eval_steps_per_second": 7.901, |
| "step": 4690 |
| }, |
| { |
| "epoch": 0.5391763221291729, |
| "grad_norm": 2.457648992538452, |
| "learning_rate": 9.114510303802847e-06, |
| "loss": 0.9047, |
| "step": 4700 |
| }, |
| { |
| "epoch": 0.5391763221291729, |
| "eval_loss": 0.5020408034324646, |
| "eval_runtime": 13.0428, |
| "eval_samples_per_second": 504.8, |
| "eval_steps_per_second": 7.897, |
| "step": 4700 |
| }, |
| { |
| "epoch": 0.5403235057932775, |
| "eval_loss": 0.5017600059509277, |
| "eval_runtime": 13.0177, |
| "eval_samples_per_second": 505.772, |
| "eval_steps_per_second": 7.912, |
| "step": 4710 |
| }, |
| { |
| "epoch": 0.5414706894573821, |
| "eval_loss": 0.5015026926994324, |
| "eval_runtime": 13.0186, |
| "eval_samples_per_second": 505.738, |
| "eval_steps_per_second": 7.912, |
| "step": 4720 |
| }, |
| { |
| "epoch": 0.5426178731214868, |
| "eval_loss": 0.5008870959281921, |
| "eval_runtime": 12.9903, |
| "eval_samples_per_second": 506.841, |
| "eval_steps_per_second": 7.929, |
| "step": 4730 |
| }, |
| { |
| "epoch": 0.5437650567855914, |
| "eval_loss": 0.500299334526062, |
| "eval_runtime": 13.032, |
| "eval_samples_per_second": 505.218, |
| "eval_steps_per_second": 7.904, |
| "step": 4740 |
| }, |
| { |
| "epoch": 0.544912240449696, |
| "eval_loss": 0.49969592690467834, |
| "eval_runtime": 13.0199, |
| "eval_samples_per_second": 505.686, |
| "eval_steps_per_second": 7.911, |
| "step": 4750 |
| }, |
| { |
| "epoch": 0.5460594241138006, |
| "eval_loss": 0.4990648329257965, |
| "eval_runtime": 13.0229, |
| "eval_samples_per_second": 505.573, |
| "eval_steps_per_second": 7.909, |
| "step": 4760 |
| }, |
| { |
| "epoch": 0.5472066077779052, |
| "eval_loss": 0.4983595013618469, |
| "eval_runtime": 13.0558, |
| "eval_samples_per_second": 504.298, |
| "eval_steps_per_second": 7.889, |
| "step": 4770 |
| }, |
| { |
| "epoch": 0.5483537914420099, |
| "eval_loss": 0.4980465769767761, |
| "eval_runtime": 13.0075, |
| "eval_samples_per_second": 506.171, |
| "eval_steps_per_second": 7.919, |
| "step": 4780 |
| }, |
| { |
| "epoch": 0.5495009751061145, |
| "eval_loss": 0.4980180859565735, |
| "eval_runtime": 13.0189, |
| "eval_samples_per_second": 505.727, |
| "eval_steps_per_second": 7.912, |
| "step": 4790 |
| }, |
| { |
| "epoch": 0.5506481587702191, |
| "grad_norm": 2.9509494304656982, |
| "learning_rate": 9.07202039515615e-06, |
| "loss": 0.887, |
| "step": 4800 |
| }, |
| { |
| "epoch": 0.5506481587702191, |
| "eval_loss": 0.4978945553302765, |
| "eval_runtime": 13.073, |
| "eval_samples_per_second": 503.632, |
| "eval_steps_per_second": 7.879, |
| "step": 4800 |
| }, |
| { |
| "epoch": 0.5517953424343237, |
| "eval_loss": 0.49748191237449646, |
| "eval_runtime": 13.0456, |
| "eval_samples_per_second": 504.69, |
| "eval_steps_per_second": 7.895, |
| "step": 4810 |
| }, |
| { |
| "epoch": 0.5529425260984283, |
| "eval_loss": 0.49734070897102356, |
| "eval_runtime": 12.917, |
| "eval_samples_per_second": 509.716, |
| "eval_steps_per_second": 7.974, |
| "step": 4820 |
| }, |
| { |
| "epoch": 0.554089709762533, |
| "eval_loss": 0.4968598783016205, |
| "eval_runtime": 12.9242, |
| "eval_samples_per_second": 509.434, |
| "eval_steps_per_second": 7.97, |
| "step": 4830 |
| }, |
| { |
| "epoch": 0.5552368934266376, |
| "eval_loss": 0.49663135409355164, |
| "eval_runtime": 12.917, |
| "eval_samples_per_second": 509.714, |
| "eval_steps_per_second": 7.974, |
| "step": 4840 |
| }, |
| { |
| "epoch": 0.5563840770907422, |
| "eval_loss": 0.4964074492454529, |
| "eval_runtime": 12.9307, |
| "eval_samples_per_second": 509.176, |
| "eval_steps_per_second": 7.966, |
| "step": 4850 |
| }, |
| { |
| "epoch": 0.5575312607548468, |
| "eval_loss": 0.49635863304138184, |
| "eval_runtime": 13.0256, |
| "eval_samples_per_second": 505.465, |
| "eval_steps_per_second": 7.907, |
| "step": 4860 |
| }, |
| { |
| "epoch": 0.5586784444189514, |
| "eval_loss": 0.4959893226623535, |
| "eval_runtime": 13.019, |
| "eval_samples_per_second": 505.724, |
| "eval_steps_per_second": 7.912, |
| "step": 4870 |
| }, |
| { |
| "epoch": 0.5598256280830561, |
| "eval_loss": 0.495743066072464, |
| "eval_runtime": 12.9529, |
| "eval_samples_per_second": 508.302, |
| "eval_steps_per_second": 7.952, |
| "step": 4880 |
| }, |
| { |
| "epoch": 0.5609728117471607, |
| "eval_loss": 0.49548840522766113, |
| "eval_runtime": 12.9165, |
| "eval_samples_per_second": 509.734, |
| "eval_steps_per_second": 7.974, |
| "step": 4890 |
| }, |
| { |
| "epoch": 0.5621199954112653, |
| "grad_norm": 2.523698329925537, |
| "learning_rate": 9.029530486509454e-06, |
| "loss": 0.8645, |
| "step": 4900 |
| }, |
| { |
| "epoch": 0.5621199954112653, |
| "eval_loss": 0.4952070116996765, |
| "eval_runtime": 12.9252, |
| "eval_samples_per_second": 509.394, |
| "eval_steps_per_second": 7.969, |
| "step": 4900 |
| }, |
| { |
| "epoch": 0.5632671790753699, |
| "eval_loss": 0.49501940608024597, |
| "eval_runtime": 12.9174, |
| "eval_samples_per_second": 509.699, |
| "eval_steps_per_second": 7.974, |
| "step": 4910 |
| }, |
| { |
| "epoch": 0.5644143627394745, |
| "eval_loss": 0.49515894055366516, |
| "eval_runtime": 12.9248, |
| "eval_samples_per_second": 509.409, |
| "eval_steps_per_second": 7.969, |
| "step": 4920 |
| }, |
| { |
| "epoch": 0.5655615464035793, |
| "eval_loss": 0.4948519468307495, |
| "eval_runtime": 13.0296, |
| "eval_samples_per_second": 505.311, |
| "eval_steps_per_second": 7.905, |
| "step": 4930 |
| }, |
| { |
| "epoch": 0.5667087300676839, |
| "eval_loss": 0.4942854344844818, |
| "eval_runtime": 13.0211, |
| "eval_samples_per_second": 505.639, |
| "eval_steps_per_second": 7.91, |
| "step": 4940 |
| }, |
| { |
| "epoch": 0.5678559137317885, |
| "eval_loss": 0.49377119541168213, |
| "eval_runtime": 13.0383, |
| "eval_samples_per_second": 504.973, |
| "eval_steps_per_second": 7.9, |
| "step": 4950 |
| }, |
| { |
| "epoch": 0.569003097395893, |
| "eval_loss": 0.4936191439628601, |
| "eval_runtime": 13.02, |
| "eval_samples_per_second": 505.683, |
| "eval_steps_per_second": 7.911, |
| "step": 4960 |
| }, |
| { |
| "epoch": 0.5701502810599977, |
| "eval_loss": 0.49330177903175354, |
| "eval_runtime": 13.023, |
| "eval_samples_per_second": 505.566, |
| "eval_steps_per_second": 7.909, |
| "step": 4970 |
| }, |
| { |
| "epoch": 0.5712974647241024, |
| "eval_loss": 0.4930713176727295, |
| "eval_runtime": 13.0236, |
| "eval_samples_per_second": 505.545, |
| "eval_steps_per_second": 7.909, |
| "step": 4980 |
| }, |
| { |
| "epoch": 0.572444648388207, |
| "eval_loss": 0.49287787079811096, |
| "eval_runtime": 12.9359, |
| "eval_samples_per_second": 508.972, |
| "eval_steps_per_second": 7.962, |
| "step": 4990 |
| }, |
| { |
| "epoch": 0.5735918320523116, |
| "grad_norm": 8.273838996887207, |
| "learning_rate": 8.987040577862758e-06, |
| "loss": 0.8348, |
| "step": 5000 |
| }, |
| { |
| "epoch": 0.5735918320523116, |
| "eval_loss": 0.49241214990615845, |
| "eval_runtime": 12.9315, |
| "eval_samples_per_second": 509.146, |
| "eval_steps_per_second": 7.965, |
| "step": 5000 |
| }, |
| { |
| "epoch": 0.5747390157164162, |
| "eval_loss": 0.49207600951194763, |
| "eval_runtime": 13.022, |
| "eval_samples_per_second": 505.606, |
| "eval_steps_per_second": 7.91, |
| "step": 5010 |
| }, |
| { |
| "epoch": 0.5758861993805208, |
| "eval_loss": 0.4914530813694, |
| "eval_runtime": 13.0318, |
| "eval_samples_per_second": 505.225, |
| "eval_steps_per_second": 7.904, |
| "step": 5020 |
| }, |
| { |
| "epoch": 0.5770333830446255, |
| "eval_loss": 0.49109673500061035, |
| "eval_runtime": 12.9148, |
| "eval_samples_per_second": 509.805, |
| "eval_steps_per_second": 7.975, |
| "step": 5030 |
| }, |
| { |
| "epoch": 0.5781805667087301, |
| "eval_loss": 0.49086394906044006, |
| "eval_runtime": 12.9787, |
| "eval_samples_per_second": 507.293, |
| "eval_steps_per_second": 7.936, |
| "step": 5040 |
| }, |
| { |
| "epoch": 0.5793277503728347, |
| "eval_loss": 0.4904956519603729, |
| "eval_runtime": 12.9257, |
| "eval_samples_per_second": 509.373, |
| "eval_steps_per_second": 7.969, |
| "step": 5050 |
| }, |
| { |
| "epoch": 0.5804749340369393, |
| "eval_loss": 0.4899919331073761, |
| "eval_runtime": 12.9192, |
| "eval_samples_per_second": 509.627, |
| "eval_steps_per_second": 7.973, |
| "step": 5060 |
| }, |
| { |
| "epoch": 0.5816221177010439, |
| "eval_loss": 0.48917409777641296, |
| "eval_runtime": 13.0398, |
| "eval_samples_per_second": 504.917, |
| "eval_steps_per_second": 7.899, |
| "step": 5070 |
| }, |
| { |
| "epoch": 0.5827693013651486, |
| "eval_loss": 0.4885985851287842, |
| "eval_runtime": 13.0538, |
| "eval_samples_per_second": 504.375, |
| "eval_steps_per_second": 7.89, |
| "step": 5080 |
| }, |
| { |
| "epoch": 0.5839164850292532, |
| "eval_loss": 0.48832204937934875, |
| "eval_runtime": 13.0334, |
| "eval_samples_per_second": 505.163, |
| "eval_steps_per_second": 7.903, |
| "step": 5090 |
| }, |
| { |
| "epoch": 0.5850636686933578, |
| "grad_norm": 4.226832389831543, |
| "learning_rate": 8.944550669216062e-06, |
| "loss": 0.871, |
| "step": 5100 |
| }, |
| { |
| "epoch": 0.5850636686933578, |
| "eval_loss": 0.4879455268383026, |
| "eval_runtime": 13.0288, |
| "eval_samples_per_second": 505.341, |
| "eval_steps_per_second": 7.906, |
| "step": 5100 |
| }, |
| { |
| "epoch": 0.5862108523574624, |
| "eval_loss": 0.4876773953437805, |
| "eval_runtime": 13.0376, |
| "eval_samples_per_second": 505.002, |
| "eval_steps_per_second": 7.9, |
| "step": 5110 |
| }, |
| { |
| "epoch": 0.587358036021567, |
| "eval_loss": 0.4874019920825958, |
| "eval_runtime": 13.0366, |
| "eval_samples_per_second": 505.039, |
| "eval_steps_per_second": 7.901, |
| "step": 5120 |
| }, |
| { |
| "epoch": 0.5885052196856717, |
| "eval_loss": 0.4869537651538849, |
| "eval_runtime": 12.9317, |
| "eval_samples_per_second": 509.137, |
| "eval_steps_per_second": 7.965, |
| "step": 5130 |
| }, |
| { |
| "epoch": 0.5896524033497763, |
| "eval_loss": 0.4866671860218048, |
| "eval_runtime": 12.9189, |
| "eval_samples_per_second": 509.641, |
| "eval_steps_per_second": 7.973, |
| "step": 5140 |
| }, |
| { |
| "epoch": 0.5907995870138809, |
| "eval_loss": 0.48636749386787415, |
| "eval_runtime": 12.9954, |
| "eval_samples_per_second": 506.64, |
| "eval_steps_per_second": 7.926, |
| "step": 5150 |
| }, |
| { |
| "epoch": 0.5919467706779855, |
| "eval_loss": 0.4861582815647125, |
| "eval_runtime": 13.0674, |
| "eval_samples_per_second": 503.849, |
| "eval_steps_per_second": 7.882, |
| "step": 5160 |
| }, |
| { |
| "epoch": 0.5930939543420902, |
| "eval_loss": 0.4859904944896698, |
| "eval_runtime": 13.0241, |
| "eval_samples_per_second": 505.523, |
| "eval_steps_per_second": 7.908, |
| "step": 5170 |
| }, |
| { |
| "epoch": 0.5942411380061948, |
| "eval_loss": 0.4856555759906769, |
| "eval_runtime": 13.0743, |
| "eval_samples_per_second": 503.584, |
| "eval_steps_per_second": 7.878, |
| "step": 5180 |
| }, |
| { |
| "epoch": 0.5953883216702994, |
| "eval_loss": 0.4855053722858429, |
| "eval_runtime": 13.0348, |
| "eval_samples_per_second": 505.11, |
| "eval_steps_per_second": 7.902, |
| "step": 5190 |
| }, |
| { |
| "epoch": 0.596535505334404, |
| "grad_norm": 2.4225528240203857, |
| "learning_rate": 8.902060760569365e-06, |
| "loss": 0.8522, |
| "step": 5200 |
| }, |
| { |
| "epoch": 0.596535505334404, |
| "eval_loss": 0.48495832085609436, |
| "eval_runtime": 13.0482, |
| "eval_samples_per_second": 504.589, |
| "eval_steps_per_second": 7.894, |
| "step": 5200 |
| }, |
| { |
| "epoch": 0.5976826889985086, |
| "eval_loss": 0.4846104681491852, |
| "eval_runtime": 13.0504, |
| "eval_samples_per_second": 504.506, |
| "eval_steps_per_second": 7.892, |
| "step": 5210 |
| }, |
| { |
| "epoch": 0.5988298726626133, |
| "eval_loss": 0.484415203332901, |
| "eval_runtime": 13.0394, |
| "eval_samples_per_second": 504.931, |
| "eval_steps_per_second": 7.899, |
| "step": 5220 |
| }, |
| { |
| "epoch": 0.5999770563267179, |
| "eval_loss": 0.48415929079055786, |
| "eval_runtime": 12.9158, |
| "eval_samples_per_second": 509.763, |
| "eval_steps_per_second": 7.975, |
| "step": 5230 |
| }, |
| { |
| "epoch": 0.6011242399908225, |
| "eval_loss": 0.48370271921157837, |
| "eval_runtime": 12.9157, |
| "eval_samples_per_second": 509.767, |
| "eval_steps_per_second": 7.975, |
| "step": 5240 |
| }, |
| { |
| "epoch": 0.6022714236549271, |
| "eval_loss": 0.4835106432437897, |
| "eval_runtime": 13.0501, |
| "eval_samples_per_second": 504.517, |
| "eval_steps_per_second": 7.893, |
| "step": 5250 |
| }, |
| { |
| "epoch": 0.6034186073190317, |
| "eval_loss": 0.4830651879310608, |
| "eval_runtime": 13.0173, |
| "eval_samples_per_second": 505.788, |
| "eval_steps_per_second": 7.913, |
| "step": 5260 |
| }, |
| { |
| "epoch": 0.6045657909831365, |
| "eval_loss": 0.48263636231422424, |
| "eval_runtime": 13.0256, |
| "eval_samples_per_second": 505.468, |
| "eval_steps_per_second": 7.908, |
| "step": 5270 |
| }, |
| { |
| "epoch": 0.605712974647241, |
| "eval_loss": 0.48216575384140015, |
| "eval_runtime": 12.9588, |
| "eval_samples_per_second": 508.073, |
| "eval_steps_per_second": 7.948, |
| "step": 5280 |
| }, |
| { |
| "epoch": 0.6068601583113457, |
| "eval_loss": 0.48219650983810425, |
| "eval_runtime": 13.0536, |
| "eval_samples_per_second": 504.381, |
| "eval_steps_per_second": 7.891, |
| "step": 5290 |
| }, |
| { |
| "epoch": 0.6080073419754503, |
| "grad_norm": 3.061457395553589, |
| "learning_rate": 8.859570851922669e-06, |
| "loss": 0.869, |
| "step": 5300 |
| }, |
| { |
| "epoch": 0.6080073419754503, |
| "eval_loss": 0.4820204973220825, |
| "eval_runtime": 13.0472, |
| "eval_samples_per_second": 504.628, |
| "eval_steps_per_second": 7.894, |
| "step": 5300 |
| }, |
| { |
| "epoch": 0.6091545256395549, |
| "eval_loss": 0.48179325461387634, |
| "eval_runtime": 13.0288, |
| "eval_samples_per_second": 505.341, |
| "eval_steps_per_second": 7.906, |
| "step": 5310 |
| }, |
| { |
| "epoch": 0.6103017093036596, |
| "eval_loss": 0.48185059428215027, |
| "eval_runtime": 13.0315, |
| "eval_samples_per_second": 505.236, |
| "eval_steps_per_second": 7.904, |
| "step": 5320 |
| }, |
| { |
| "epoch": 0.6114488929677642, |
| "eval_loss": 0.4818604290485382, |
| "eval_runtime": 13.0386, |
| "eval_samples_per_second": 504.961, |
| "eval_steps_per_second": 7.9, |
| "step": 5330 |
| }, |
| { |
| "epoch": 0.6125960766318688, |
| "eval_loss": 0.4814794957637787, |
| "eval_runtime": 13.0315, |
| "eval_samples_per_second": 505.236, |
| "eval_steps_per_second": 7.904, |
| "step": 5340 |
| }, |
| { |
| "epoch": 0.6137432602959734, |
| "eval_loss": 0.48126035928726196, |
| "eval_runtime": 12.9876, |
| "eval_samples_per_second": 506.943, |
| "eval_steps_per_second": 7.931, |
| "step": 5350 |
| }, |
| { |
| "epoch": 0.614890443960078, |
| "eval_loss": 0.48123446106910706, |
| "eval_runtime": 12.9132, |
| "eval_samples_per_second": 509.864, |
| "eval_steps_per_second": 7.976, |
| "step": 5360 |
| }, |
| { |
| "epoch": 0.6160376276241827, |
| "eval_loss": 0.48100870847702026, |
| "eval_runtime": 12.951, |
| "eval_samples_per_second": 508.378, |
| "eval_steps_per_second": 7.953, |
| "step": 5370 |
| }, |
| { |
| "epoch": 0.6171848112882873, |
| "eval_loss": 0.48086845874786377, |
| "eval_runtime": 13.0228, |
| "eval_samples_per_second": 505.575, |
| "eval_steps_per_second": 7.909, |
| "step": 5380 |
| }, |
| { |
| "epoch": 0.6183319949523919, |
| "eval_loss": 0.48063814640045166, |
| "eval_runtime": 13.0057, |
| "eval_samples_per_second": 506.238, |
| "eval_steps_per_second": 7.92, |
| "step": 5390 |
| }, |
| { |
| "epoch": 0.6194791786164965, |
| "grad_norm": 3.033811092376709, |
| "learning_rate": 8.817080943275972e-06, |
| "loss": 0.8548, |
| "step": 5400 |
| }, |
| { |
| "epoch": 0.6194791786164965, |
| "eval_loss": 0.4804742634296417, |
| "eval_runtime": 13.067, |
| "eval_samples_per_second": 503.865, |
| "eval_steps_per_second": 7.882, |
| "step": 5400 |
| }, |
| { |
| "epoch": 0.6206263622806011, |
| "eval_loss": 0.47996437549591064, |
| "eval_runtime": 13.0358, |
| "eval_samples_per_second": 505.07, |
| "eval_steps_per_second": 7.901, |
| "step": 5410 |
| }, |
| { |
| "epoch": 0.6217735459447058, |
| "eval_loss": 0.479777067899704, |
| "eval_runtime": 13.0527, |
| "eval_samples_per_second": 504.416, |
| "eval_steps_per_second": 7.891, |
| "step": 5420 |
| }, |
| { |
| "epoch": 0.6229207296088104, |
| "eval_loss": 0.4794636368751526, |
| "eval_runtime": 13.0443, |
| "eval_samples_per_second": 504.743, |
| "eval_steps_per_second": 7.896, |
| "step": 5430 |
| }, |
| { |
| "epoch": 0.624067913272915, |
| "eval_loss": 0.47919943928718567, |
| "eval_runtime": 13.029, |
| "eval_samples_per_second": 505.333, |
| "eval_steps_per_second": 7.905, |
| "step": 5440 |
| }, |
| { |
| "epoch": 0.6252150969370196, |
| "eval_loss": 0.47902804613113403, |
| "eval_runtime": 13.0497, |
| "eval_samples_per_second": 504.533, |
| "eval_steps_per_second": 7.893, |
| "step": 5450 |
| }, |
| { |
| "epoch": 0.6263622806011242, |
| "eval_loss": 0.47903186082839966, |
| "eval_runtime": 13.0455, |
| "eval_samples_per_second": 504.695, |
| "eval_steps_per_second": 7.895, |
| "step": 5460 |
| }, |
| { |
| "epoch": 0.6275094642652289, |
| "eval_loss": 0.47908058762550354, |
| "eval_runtime": 12.925, |
| "eval_samples_per_second": 509.401, |
| "eval_steps_per_second": 7.969, |
| "step": 5470 |
| }, |
| { |
| "epoch": 0.6286566479293335, |
| "eval_loss": 0.47938427329063416, |
| "eval_runtime": 12.9195, |
| "eval_samples_per_second": 509.618, |
| "eval_steps_per_second": 7.972, |
| "step": 5480 |
| }, |
| { |
| "epoch": 0.6298038315934381, |
| "eval_loss": 0.4791744649410248, |
| "eval_runtime": 13.0292, |
| "eval_samples_per_second": 505.326, |
| "eval_steps_per_second": 7.905, |
| "step": 5490 |
| }, |
| { |
| "epoch": 0.6309510152575427, |
| "grad_norm": 3.4439074993133545, |
| "learning_rate": 8.774591034629276e-06, |
| "loss": 0.8366, |
| "step": 5500 |
| }, |
| { |
| "epoch": 0.6309510152575427, |
| "eval_loss": 0.47903913259506226, |
| "eval_runtime": 13.0858, |
| "eval_samples_per_second": 503.143, |
| "eval_steps_per_second": 7.871, |
| "step": 5500 |
| }, |
| { |
| "epoch": 0.6320981989216473, |
| "eval_loss": 0.47857555747032166, |
| "eval_runtime": 13.0402, |
| "eval_samples_per_second": 504.9, |
| "eval_steps_per_second": 7.899, |
| "step": 5510 |
| }, |
| { |
| "epoch": 0.633245382585752, |
| "eval_loss": 0.47797891497612, |
| "eval_runtime": 13.0407, |
| "eval_samples_per_second": 504.881, |
| "eval_steps_per_second": 7.898, |
| "step": 5520 |
| }, |
| { |
| "epoch": 0.6343925662498566, |
| "eval_loss": 0.4772583246231079, |
| "eval_runtime": 13.0406, |
| "eval_samples_per_second": 504.887, |
| "eval_steps_per_second": 7.898, |
| "step": 5530 |
| }, |
| { |
| "epoch": 0.6355397499139612, |
| "eval_loss": 0.4768436551094055, |
| "eval_runtime": 13.0417, |
| "eval_samples_per_second": 504.841, |
| "eval_steps_per_second": 7.898, |
| "step": 5540 |
| }, |
| { |
| "epoch": 0.6366869335780658, |
| "eval_loss": 0.4766625463962555, |
| "eval_runtime": 13.0285, |
| "eval_samples_per_second": 505.353, |
| "eval_steps_per_second": 7.906, |
| "step": 5550 |
| }, |
| { |
| "epoch": 0.6378341172421704, |
| "eval_loss": 0.4764641523361206, |
| "eval_runtime": 13.0165, |
| "eval_samples_per_second": 505.82, |
| "eval_steps_per_second": 7.913, |
| "step": 5560 |
| }, |
| { |
| "epoch": 0.6389813009062751, |
| "eval_loss": 0.4764559268951416, |
| "eval_runtime": 13.0425, |
| "eval_samples_per_second": 504.811, |
| "eval_steps_per_second": 7.897, |
| "step": 5570 |
| }, |
| { |
| "epoch": 0.6401284845703797, |
| "eval_loss": 0.4763183891773224, |
| "eval_runtime": 12.9185, |
| "eval_samples_per_second": 509.656, |
| "eval_steps_per_second": 7.973, |
| "step": 5580 |
| }, |
| { |
| "epoch": 0.6412756682344843, |
| "eval_loss": 0.4759766161441803, |
| "eval_runtime": 12.9253, |
| "eval_samples_per_second": 509.389, |
| "eval_steps_per_second": 7.969, |
| "step": 5590 |
| }, |
| { |
| "epoch": 0.6424228518985889, |
| "grad_norm": 3.5839452743530273, |
| "learning_rate": 8.73210112598258e-06, |
| "loss": 0.8696, |
| "step": 5600 |
| }, |
| { |
| "epoch": 0.6424228518985889, |
| "eval_loss": 0.4756835401058197, |
| "eval_runtime": 12.9294, |
| "eval_samples_per_second": 509.228, |
| "eval_steps_per_second": 7.966, |
| "step": 5600 |
| }, |
| { |
| "epoch": 0.6435700355626935, |
| "eval_loss": 0.47543641924858093, |
| "eval_runtime": 12.9219, |
| "eval_samples_per_second": 509.524, |
| "eval_steps_per_second": 7.971, |
| "step": 5610 |
| }, |
| { |
| "epoch": 0.6447172192267983, |
| "eval_loss": 0.475233793258667, |
| "eval_runtime": 12.9273, |
| "eval_samples_per_second": 509.312, |
| "eval_steps_per_second": 7.968, |
| "step": 5620 |
| }, |
| { |
| "epoch": 0.6458644028909029, |
| "eval_loss": 0.4750543534755707, |
| "eval_runtime": 13.0516, |
| "eval_samples_per_second": 504.461, |
| "eval_steps_per_second": 7.892, |
| "step": 5630 |
| }, |
| { |
| "epoch": 0.6470115865550075, |
| "eval_loss": 0.47466254234313965, |
| "eval_runtime": 12.9649, |
| "eval_samples_per_second": 507.834, |
| "eval_steps_per_second": 7.945, |
| "step": 5640 |
| }, |
| { |
| "epoch": 0.6481587702191121, |
| "eval_loss": 0.47465524077415466, |
| "eval_runtime": 13.0683, |
| "eval_samples_per_second": 503.815, |
| "eval_steps_per_second": 7.882, |
| "step": 5650 |
| }, |
| { |
| "epoch": 0.6493059538832167, |
| "eval_loss": 0.4742332994937897, |
| "eval_runtime": 13.0318, |
| "eval_samples_per_second": 505.226, |
| "eval_steps_per_second": 7.904, |
| "step": 5660 |
| }, |
| { |
| "epoch": 0.6504531375473214, |
| "eval_loss": 0.4739505648612976, |
| "eval_runtime": 13.0398, |
| "eval_samples_per_second": 504.914, |
| "eval_steps_per_second": 7.899, |
| "step": 5670 |
| }, |
| { |
| "epoch": 0.651600321211426, |
| "eval_loss": 0.47357243299484253, |
| "eval_runtime": 12.9824, |
| "eval_samples_per_second": 507.146, |
| "eval_steps_per_second": 7.934, |
| "step": 5680 |
| }, |
| { |
| "epoch": 0.6527475048755306, |
| "eval_loss": 0.4729899764060974, |
| "eval_runtime": 12.9723, |
| "eval_samples_per_second": 507.544, |
| "eval_steps_per_second": 7.94, |
| "step": 5690 |
| }, |
| { |
| "epoch": 0.6538946885396352, |
| "grad_norm": 2.302046060562134, |
| "learning_rate": 8.689611217335883e-06, |
| "loss": 0.8302, |
| "step": 5700 |
| }, |
| { |
| "epoch": 0.6538946885396352, |
| "eval_loss": 0.4725137948989868, |
| "eval_runtime": 13.084, |
| "eval_samples_per_second": 503.21, |
| "eval_steps_per_second": 7.872, |
| "step": 5700 |
| }, |
| { |
| "epoch": 0.6550418722037398, |
| "eval_loss": 0.4722561538219452, |
| "eval_runtime": 13.1585, |
| "eval_samples_per_second": 500.362, |
| "eval_steps_per_second": 7.828, |
| "step": 5710 |
| }, |
| { |
| "epoch": 0.6561890558678445, |
| "eval_loss": 0.4719793498516083, |
| "eval_runtime": 13.054, |
| "eval_samples_per_second": 504.367, |
| "eval_steps_per_second": 7.89, |
| "step": 5720 |
| }, |
| { |
| "epoch": 0.6573362395319491, |
| "eval_loss": 0.47180819511413574, |
| "eval_runtime": 13.0838, |
| "eval_samples_per_second": 503.216, |
| "eval_steps_per_second": 7.872, |
| "step": 5730 |
| }, |
| { |
| "epoch": 0.6584834231960537, |
| "eval_loss": 0.4715099334716797, |
| "eval_runtime": 13.02, |
| "eval_samples_per_second": 505.682, |
| "eval_steps_per_second": 7.911, |
| "step": 5740 |
| }, |
| { |
| "epoch": 0.6596306068601583, |
| "eval_loss": 0.471357524394989, |
| "eval_runtime": 13.0013, |
| "eval_samples_per_second": 506.411, |
| "eval_steps_per_second": 7.922, |
| "step": 5750 |
| }, |
| { |
| "epoch": 0.6607777905242629, |
| "eval_loss": 0.47113123536109924, |
| "eval_runtime": 13.0177, |
| "eval_samples_per_second": 505.774, |
| "eval_steps_per_second": 7.912, |
| "step": 5760 |
| }, |
| { |
| "epoch": 0.6619249741883676, |
| "eval_loss": 0.4707399010658264, |
| "eval_runtime": 12.9866, |
| "eval_samples_per_second": 506.984, |
| "eval_steps_per_second": 7.931, |
| "step": 5770 |
| }, |
| { |
| "epoch": 0.6630721578524722, |
| "eval_loss": 0.47066983580589294, |
| "eval_runtime": 12.9825, |
| "eval_samples_per_second": 507.146, |
| "eval_steps_per_second": 7.934, |
| "step": 5780 |
| }, |
| { |
| "epoch": 0.6642193415165768, |
| "eval_loss": 0.4702962636947632, |
| "eval_runtime": 12.9831, |
| "eval_samples_per_second": 507.12, |
| "eval_steps_per_second": 7.933, |
| "step": 5790 |
| }, |
| { |
| "epoch": 0.6653665251806814, |
| "grad_norm": 2.9902710914611816, |
| "learning_rate": 8.647121308689187e-06, |
| "loss": 0.8128, |
| "step": 5800 |
| }, |
| { |
| "epoch": 0.6653665251806814, |
| "eval_loss": 0.47025352716445923, |
| "eval_runtime": 12.9869, |
| "eval_samples_per_second": 506.973, |
| "eval_steps_per_second": 7.931, |
| "step": 5800 |
| }, |
| { |
| "epoch": 0.666513708844786, |
| "eval_loss": 0.47005340456962585, |
| "eval_runtime": 13.176, |
| "eval_samples_per_second": 499.695, |
| "eval_steps_per_second": 7.817, |
| "step": 5810 |
| }, |
| { |
| "epoch": 0.6676608925088907, |
| "eval_loss": 0.469906210899353, |
| "eval_runtime": 13.0218, |
| "eval_samples_per_second": 505.613, |
| "eval_steps_per_second": 7.91, |
| "step": 5820 |
| }, |
| { |
| "epoch": 0.6688080761729953, |
| "eval_loss": 0.4697432816028595, |
| "eval_runtime": 12.9919, |
| "eval_samples_per_second": 506.779, |
| "eval_steps_per_second": 7.928, |
| "step": 5830 |
| }, |
| { |
| "epoch": 0.6699552598370999, |
| "eval_loss": 0.4697941243648529, |
| "eval_runtime": 13.0225, |
| "eval_samples_per_second": 505.586, |
| "eval_steps_per_second": 7.909, |
| "step": 5840 |
| }, |
| { |
| "epoch": 0.6711024435012045, |
| "eval_loss": 0.4695260226726532, |
| "eval_runtime": 13.1086, |
| "eval_samples_per_second": 502.264, |
| "eval_steps_per_second": 7.857, |
| "step": 5850 |
| }, |
| { |
| "epoch": 0.6722496271653091, |
| "eval_loss": 0.46907415986061096, |
| "eval_runtime": 13.1209, |
| "eval_samples_per_second": 501.796, |
| "eval_steps_per_second": 7.85, |
| "step": 5860 |
| }, |
| { |
| "epoch": 0.6733968108294138, |
| "eval_loss": 0.46890172362327576, |
| "eval_runtime": 13.0057, |
| "eval_samples_per_second": 506.24, |
| "eval_steps_per_second": 7.92, |
| "step": 5870 |
| }, |
| { |
| "epoch": 0.6745439944935184, |
| "eval_loss": 0.4689192771911621, |
| "eval_runtime": 13.0073, |
| "eval_samples_per_second": 506.176, |
| "eval_steps_per_second": 7.919, |
| "step": 5880 |
| }, |
| { |
| "epoch": 0.675691178157623, |
| "eval_loss": 0.46877485513687134, |
| "eval_runtime": 13.0048, |
| "eval_samples_per_second": 506.273, |
| "eval_steps_per_second": 7.92, |
| "step": 5890 |
| }, |
| { |
| "epoch": 0.6768383618217276, |
| "grad_norm": 3.034991979598999, |
| "learning_rate": 8.60463140004249e-06, |
| "loss": 0.8437, |
| "step": 5900 |
| }, |
| { |
| "epoch": 0.6768383618217276, |
| "eval_loss": 0.4682934582233429, |
| "eval_runtime": 13.0727, |
| "eval_samples_per_second": 503.647, |
| "eval_steps_per_second": 7.879, |
| "step": 5900 |
| }, |
| { |
| "epoch": 0.6779855454858322, |
| "eval_loss": 0.46829932928085327, |
| "eval_runtime": 12.9784, |
| "eval_samples_per_second": 507.306, |
| "eval_steps_per_second": 7.936, |
| "step": 5910 |
| }, |
| { |
| "epoch": 0.6791327291499369, |
| "eval_loss": 0.46807605028152466, |
| "eval_runtime": 13.0164, |
| "eval_samples_per_second": 505.823, |
| "eval_steps_per_second": 7.913, |
| "step": 5920 |
| }, |
| { |
| "epoch": 0.6802799128140415, |
| "eval_loss": 0.4678213596343994, |
| "eval_runtime": 13.0054, |
| "eval_samples_per_second": 506.252, |
| "eval_steps_per_second": 7.92, |
| "step": 5930 |
| }, |
| { |
| "epoch": 0.6814270964781461, |
| "eval_loss": 0.4677067697048187, |
| "eval_runtime": 13.1833, |
| "eval_samples_per_second": 499.422, |
| "eval_steps_per_second": 7.813, |
| "step": 5940 |
| }, |
| { |
| "epoch": 0.6825742801422507, |
| "eval_loss": 0.46759942173957825, |
| "eval_runtime": 13.0106, |
| "eval_samples_per_second": 506.05, |
| "eval_steps_per_second": 7.917, |
| "step": 5950 |
| }, |
| { |
| "epoch": 0.6837214638063553, |
| "eval_loss": 0.4672621786594391, |
| "eval_runtime": 13.0065, |
| "eval_samples_per_second": 506.21, |
| "eval_steps_per_second": 7.919, |
| "step": 5960 |
| }, |
| { |
| "epoch": 0.6848686474704601, |
| "eval_loss": 0.4668444097042084, |
| "eval_runtime": 12.9914, |
| "eval_samples_per_second": 506.798, |
| "eval_steps_per_second": 7.928, |
| "step": 5970 |
| }, |
| { |
| "epoch": 0.6860158311345647, |
| "eval_loss": 0.4666861891746521, |
| "eval_runtime": 13.0444, |
| "eval_samples_per_second": 504.739, |
| "eval_steps_per_second": 7.896, |
| "step": 5980 |
| }, |
| { |
| "epoch": 0.6871630147986693, |
| "eval_loss": 0.4661363661289215, |
| "eval_runtime": 13.0273, |
| "eval_samples_per_second": 505.401, |
| "eval_steps_per_second": 7.906, |
| "step": 5990 |
| }, |
| { |
| "epoch": 0.6883101984627739, |
| "grad_norm": 2.8303074836730957, |
| "learning_rate": 8.562141491395794e-06, |
| "loss": 0.7774, |
| "step": 6000 |
| }, |
| { |
| "epoch": 0.6883101984627739, |
| "eval_loss": 0.4656630754470825, |
| "eval_runtime": 13.1208, |
| "eval_samples_per_second": 501.8, |
| "eval_steps_per_second": 7.85, |
| "step": 6000 |
| }, |
| { |
| "epoch": 0.6894573821268785, |
| "eval_loss": 0.46537360548973083, |
| "eval_runtime": 13.2041, |
| "eval_samples_per_second": 498.635, |
| "eval_steps_per_second": 7.801, |
| "step": 6010 |
| }, |
| { |
| "epoch": 0.6906045657909832, |
| "eval_loss": 0.46501070261001587, |
| "eval_runtime": 13.0241, |
| "eval_samples_per_second": 505.525, |
| "eval_steps_per_second": 7.908, |
| "step": 6020 |
| }, |
| { |
| "epoch": 0.6917517494550878, |
| "eval_loss": 0.4648328423500061, |
| "eval_runtime": 13.1998, |
| "eval_samples_per_second": 498.796, |
| "eval_steps_per_second": 7.803, |
| "step": 6030 |
| }, |
| { |
| "epoch": 0.6928989331191924, |
| "eval_loss": 0.46458834409713745, |
| "eval_runtime": 12.9951, |
| "eval_samples_per_second": 506.654, |
| "eval_steps_per_second": 7.926, |
| "step": 6040 |
| }, |
| { |
| "epoch": 0.694046116783297, |
| "eval_loss": 0.46444424986839294, |
| "eval_runtime": 13.0891, |
| "eval_samples_per_second": 503.014, |
| "eval_steps_per_second": 7.869, |
| "step": 6050 |
| }, |
| { |
| "epoch": 0.6951933004474016, |
| "eval_loss": 0.4643203914165497, |
| "eval_runtime": 13.1478, |
| "eval_samples_per_second": 500.768, |
| "eval_steps_per_second": 7.834, |
| "step": 6060 |
| }, |
| { |
| "epoch": 0.6963404841115063, |
| "eval_loss": 0.46412837505340576, |
| "eval_runtime": 13.2014, |
| "eval_samples_per_second": 498.736, |
| "eval_steps_per_second": 7.802, |
| "step": 6070 |
| }, |
| { |
| "epoch": 0.6974876677756109, |
| "eval_loss": 0.46399426460266113, |
| "eval_runtime": 12.9915, |
| "eval_samples_per_second": 506.793, |
| "eval_steps_per_second": 7.928, |
| "step": 6080 |
| }, |
| { |
| "epoch": 0.6986348514397155, |
| "eval_loss": 0.4638455808162689, |
| "eval_runtime": 13.0404, |
| "eval_samples_per_second": 504.893, |
| "eval_steps_per_second": 7.899, |
| "step": 6090 |
| }, |
| { |
| "epoch": 0.6997820351038201, |
| "grad_norm": 2.879986524581909, |
| "learning_rate": 8.519651582749098e-06, |
| "loss": 0.834, |
| "step": 6100 |
| }, |
| { |
| "epoch": 0.6997820351038201, |
| "eval_loss": 0.4636705815792084, |
| "eval_runtime": 13.1806, |
| "eval_samples_per_second": 499.521, |
| "eval_steps_per_second": 7.814, |
| "step": 6100 |
| }, |
| { |
| "epoch": 0.7009292187679248, |
| "eval_loss": 0.46326521039009094, |
| "eval_runtime": 13.0092, |
| "eval_samples_per_second": 506.105, |
| "eval_steps_per_second": 7.918, |
| "step": 6110 |
| }, |
| { |
| "epoch": 0.7020764024320294, |
| "eval_loss": 0.4632194936275482, |
| "eval_runtime": 13.0162, |
| "eval_samples_per_second": 505.832, |
| "eval_steps_per_second": 7.913, |
| "step": 6120 |
| }, |
| { |
| "epoch": 0.703223586096134, |
| "eval_loss": 0.46308234333992004, |
| "eval_runtime": 13.0142, |
| "eval_samples_per_second": 505.91, |
| "eval_steps_per_second": 7.914, |
| "step": 6130 |
| }, |
| { |
| "epoch": 0.7043707697602386, |
| "eval_loss": 0.4628349244594574, |
| "eval_runtime": 13.1584, |
| "eval_samples_per_second": 500.363, |
| "eval_steps_per_second": 7.828, |
| "step": 6140 |
| }, |
| { |
| "epoch": 0.7055179534243432, |
| "eval_loss": 0.4627319872379303, |
| "eval_runtime": 13.0258, |
| "eval_samples_per_second": 505.458, |
| "eval_steps_per_second": 7.907, |
| "step": 6150 |
| }, |
| { |
| "epoch": 0.7066651370884479, |
| "eval_loss": 0.46227478981018066, |
| "eval_runtime": 13.132, |
| "eval_samples_per_second": 501.371, |
| "eval_steps_per_second": 7.843, |
| "step": 6160 |
| }, |
| { |
| "epoch": 0.7078123207525525, |
| "eval_loss": 0.461668998003006, |
| "eval_runtime": 13.0493, |
| "eval_samples_per_second": 504.548, |
| "eval_steps_per_second": 7.893, |
| "step": 6170 |
| }, |
| { |
| "epoch": 0.7089595044166571, |
| "eval_loss": 0.4614720642566681, |
| "eval_runtime": 13.0212, |
| "eval_samples_per_second": 505.639, |
| "eval_steps_per_second": 7.91, |
| "step": 6180 |
| }, |
| { |
| "epoch": 0.7101066880807617, |
| "eval_loss": 0.4614366888999939, |
| "eval_runtime": 13.2061, |
| "eval_samples_per_second": 498.559, |
| "eval_steps_per_second": 7.799, |
| "step": 6190 |
| }, |
| { |
| "epoch": 0.7112538717448663, |
| "grad_norm": 4.115511894226074, |
| "learning_rate": 8.477161674102402e-06, |
| "loss": 0.8118, |
| "step": 6200 |
| }, |
| { |
| "epoch": 0.7112538717448663, |
| "eval_loss": 0.46122443675994873, |
| "eval_runtime": 13.0866, |
| "eval_samples_per_second": 503.111, |
| "eval_steps_per_second": 7.871, |
| "step": 6200 |
| }, |
| { |
| "epoch": 0.712401055408971, |
| "eval_loss": 0.4612271785736084, |
| "eval_runtime": 12.9924, |
| "eval_samples_per_second": 506.757, |
| "eval_steps_per_second": 7.928, |
| "step": 6210 |
| }, |
| { |
| "epoch": 0.7135482390730756, |
| "eval_loss": 0.46115055680274963, |
| "eval_runtime": 13.477, |
| "eval_samples_per_second": 488.536, |
| "eval_steps_per_second": 7.643, |
| "step": 6220 |
| }, |
| { |
| "epoch": 0.7146954227371802, |
| "eval_loss": 0.46099165081977844, |
| "eval_runtime": 13.0008, |
| "eval_samples_per_second": 506.43, |
| "eval_steps_per_second": 7.923, |
| "step": 6230 |
| }, |
| { |
| "epoch": 0.7158426064012848, |
| "eval_loss": 0.46086400747299194, |
| "eval_runtime": 13.0372, |
| "eval_samples_per_second": 505.016, |
| "eval_steps_per_second": 7.9, |
| "step": 6240 |
| }, |
| { |
| "epoch": 0.7169897900653894, |
| "eval_loss": 0.4609854817390442, |
| "eval_runtime": 13.0338, |
| "eval_samples_per_second": 505.147, |
| "eval_steps_per_second": 7.903, |
| "step": 6250 |
| }, |
| { |
| "epoch": 0.7181369737294941, |
| "eval_loss": 0.46107491850852966, |
| "eval_runtime": 13.014, |
| "eval_samples_per_second": 505.918, |
| "eval_steps_per_second": 7.915, |
| "step": 6260 |
| }, |
| { |
| "epoch": 0.7192841573935987, |
| "eval_loss": 0.4607292115688324, |
| "eval_runtime": 13.0185, |
| "eval_samples_per_second": 505.743, |
| "eval_steps_per_second": 7.912, |
| "step": 6270 |
| }, |
| { |
| "epoch": 0.7204313410577033, |
| "eval_loss": 0.4598737359046936, |
| "eval_runtime": 12.9916, |
| "eval_samples_per_second": 506.79, |
| "eval_steps_per_second": 7.928, |
| "step": 6280 |
| }, |
| { |
| "epoch": 0.721578524721808, |
| "eval_loss": 0.4598146975040436, |
| "eval_runtime": 13.0078, |
| "eval_samples_per_second": 506.157, |
| "eval_steps_per_second": 7.918, |
| "step": 6290 |
| }, |
| { |
| "epoch": 0.7227257083859125, |
| "grad_norm": 2.5154004096984863, |
| "learning_rate": 8.434671765455705e-06, |
| "loss": 0.7884, |
| "step": 6300 |
| }, |
| { |
| "epoch": 0.7227257083859125, |
| "eval_loss": 0.4600335359573364, |
| "eval_runtime": 13.1719, |
| "eval_samples_per_second": 499.852, |
| "eval_steps_per_second": 7.82, |
| "step": 6300 |
| }, |
| { |
| "epoch": 0.7238728920500173, |
| "eval_loss": 0.459884911775589, |
| "eval_runtime": 13.0046, |
| "eval_samples_per_second": 506.282, |
| "eval_steps_per_second": 7.92, |
| "step": 6310 |
| }, |
| { |
| "epoch": 0.7250200757141219, |
| "eval_loss": 0.4599865972995758, |
| "eval_runtime": 13.0329, |
| "eval_samples_per_second": 505.183, |
| "eval_steps_per_second": 7.903, |
| "step": 6320 |
| }, |
| { |
| "epoch": 0.7261672593782265, |
| "eval_loss": 0.460053950548172, |
| "eval_runtime": 13.1305, |
| "eval_samples_per_second": 501.429, |
| "eval_steps_per_second": 7.844, |
| "step": 6330 |
| }, |
| { |
| "epoch": 0.7273144430423311, |
| "eval_loss": 0.4602521061897278, |
| "eval_runtime": 12.9902, |
| "eval_samples_per_second": 506.844, |
| "eval_steps_per_second": 7.929, |
| "step": 6340 |
| }, |
| { |
| "epoch": 0.7284616267064357, |
| "eval_loss": 0.46030986309051514, |
| "eval_runtime": 12.9894, |
| "eval_samples_per_second": 506.875, |
| "eval_steps_per_second": 7.93, |
| "step": 6350 |
| }, |
| { |
| "epoch": 0.7296088103705404, |
| "eval_loss": 0.459776908159256, |
| "eval_runtime": 12.9821, |
| "eval_samples_per_second": 507.161, |
| "eval_steps_per_second": 7.934, |
| "step": 6360 |
| }, |
| { |
| "epoch": 0.730755994034645, |
| "eval_loss": 0.4597093462944031, |
| "eval_runtime": 12.9882, |
| "eval_samples_per_second": 506.922, |
| "eval_steps_per_second": 7.93, |
| "step": 6370 |
| }, |
| { |
| "epoch": 0.7319031776987496, |
| "eval_loss": 0.4596352279186249, |
| "eval_runtime": 13.0647, |
| "eval_samples_per_second": 503.953, |
| "eval_steps_per_second": 7.884, |
| "step": 6380 |
| }, |
| { |
| "epoch": 0.7330503613628542, |
| "eval_loss": 0.4594053030014038, |
| "eval_runtime": 13.021, |
| "eval_samples_per_second": 505.645, |
| "eval_steps_per_second": 7.91, |
| "step": 6390 |
| }, |
| { |
| "epoch": 0.7341975450269588, |
| "grad_norm": 3.104555606842041, |
| "learning_rate": 8.392181856809009e-06, |
| "loss": 0.8092, |
| "step": 6400 |
| }, |
| { |
| "epoch": 0.7341975450269588, |
| "eval_loss": 0.4590100347995758, |
| "eval_runtime": 13.2643, |
| "eval_samples_per_second": 496.37, |
| "eval_steps_per_second": 7.765, |
| "step": 6400 |
| }, |
| { |
| "epoch": 0.7353447286910635, |
| "eval_loss": 0.4587569236755371, |
| "eval_runtime": 13.2732, |
| "eval_samples_per_second": 496.038, |
| "eval_steps_per_second": 7.76, |
| "step": 6410 |
| }, |
| { |
| "epoch": 0.7364919123551681, |
| "eval_loss": 0.4585030674934387, |
| "eval_runtime": 13.2129, |
| "eval_samples_per_second": 498.299, |
| "eval_steps_per_second": 7.795, |
| "step": 6420 |
| }, |
| { |
| "epoch": 0.7376390960192727, |
| "eval_loss": 0.4584300220012665, |
| "eval_runtime": 12.9947, |
| "eval_samples_per_second": 506.669, |
| "eval_steps_per_second": 7.926, |
| "step": 6430 |
| }, |
| { |
| "epoch": 0.7387862796833773, |
| "eval_loss": 0.458046019077301, |
| "eval_runtime": 12.987, |
| "eval_samples_per_second": 506.968, |
| "eval_steps_per_second": 7.931, |
| "step": 6440 |
| }, |
| { |
| "epoch": 0.7399334633474819, |
| "eval_loss": 0.4573623836040497, |
| "eval_runtime": 12.986, |
| "eval_samples_per_second": 507.008, |
| "eval_steps_per_second": 7.932, |
| "step": 6450 |
| }, |
| { |
| "epoch": 0.7410806470115866, |
| "eval_loss": 0.45698782801628113, |
| "eval_runtime": 12.9844, |
| "eval_samples_per_second": 507.07, |
| "eval_steps_per_second": 7.933, |
| "step": 6460 |
| }, |
| { |
| "epoch": 0.7422278306756912, |
| "eval_loss": 0.4565979540348053, |
| "eval_runtime": 13.0011, |
| "eval_samples_per_second": 506.417, |
| "eval_steps_per_second": 7.922, |
| "step": 6470 |
| }, |
| { |
| "epoch": 0.7433750143397958, |
| "eval_loss": 0.4562567174434662, |
| "eval_runtime": 13.0884, |
| "eval_samples_per_second": 503.041, |
| "eval_steps_per_second": 7.87, |
| "step": 6480 |
| }, |
| { |
| "epoch": 0.7445221980039004, |
| "eval_loss": 0.4559851884841919, |
| "eval_runtime": 13.0011, |
| "eval_samples_per_second": 506.419, |
| "eval_steps_per_second": 7.922, |
| "step": 6490 |
| }, |
| { |
| "epoch": 0.745669381668005, |
| "grad_norm": 3.0330851078033447, |
| "learning_rate": 8.349691948162313e-06, |
| "loss": 0.8195, |
| "step": 6500 |
| }, |
| { |
| "epoch": 0.745669381668005, |
| "eval_loss": 0.4557217061519623, |
| "eval_runtime": 13.1027, |
| "eval_samples_per_second": 502.491, |
| "eval_steps_per_second": 7.861, |
| "step": 6500 |
| }, |
| { |
| "epoch": 0.7468165653321097, |
| "eval_loss": 0.4556296765804291, |
| "eval_runtime": 12.9912, |
| "eval_samples_per_second": 506.805, |
| "eval_steps_per_second": 7.928, |
| "step": 6510 |
| }, |
| { |
| "epoch": 0.7479637489962143, |
| "eval_loss": 0.4553574323654175, |
| "eval_runtime": 12.9979, |
| "eval_samples_per_second": 506.541, |
| "eval_steps_per_second": 7.924, |
| "step": 6520 |
| }, |
| { |
| "epoch": 0.7491109326603189, |
| "eval_loss": 0.45506760478019714, |
| "eval_runtime": 12.9893, |
| "eval_samples_per_second": 506.878, |
| "eval_steps_per_second": 7.93, |
| "step": 6530 |
| }, |
| { |
| "epoch": 0.7502581163244235, |
| "eval_loss": 0.4548060894012451, |
| "eval_runtime": 13.0037, |
| "eval_samples_per_second": 506.316, |
| "eval_steps_per_second": 7.921, |
| "step": 6540 |
| }, |
| { |
| "epoch": 0.7514052999885281, |
| "eval_loss": 0.45448189973831177, |
| "eval_runtime": 13.0658, |
| "eval_samples_per_second": 503.911, |
| "eval_steps_per_second": 7.883, |
| "step": 6550 |
| }, |
| { |
| "epoch": 0.7525524836526328, |
| "eval_loss": 0.4542520344257355, |
| "eval_runtime": 12.9994, |
| "eval_samples_per_second": 506.485, |
| "eval_steps_per_second": 7.923, |
| "step": 6560 |
| }, |
| { |
| "epoch": 0.7536996673167374, |
| "eval_loss": 0.45408859848976135, |
| "eval_runtime": 12.9946, |
| "eval_samples_per_second": 506.67, |
| "eval_steps_per_second": 7.926, |
| "step": 6570 |
| }, |
| { |
| "epoch": 0.754846850980842, |
| "eval_loss": 0.45401427149772644, |
| "eval_runtime": 13.093, |
| "eval_samples_per_second": 502.866, |
| "eval_steps_per_second": 7.867, |
| "step": 6580 |
| }, |
| { |
| "epoch": 0.7559940346449466, |
| "eval_loss": 0.4538465142250061, |
| "eval_runtime": 13.0205, |
| "eval_samples_per_second": 505.664, |
| "eval_steps_per_second": 7.911, |
| "step": 6590 |
| }, |
| { |
| "epoch": 0.7571412183090512, |
| "grad_norm": 3.7183449268341064, |
| "learning_rate": 8.307202039515616e-06, |
| "loss": 0.8163, |
| "step": 6600 |
| }, |
| { |
| "epoch": 0.7571412183090512, |
| "eval_loss": 0.45350751280784607, |
| "eval_runtime": 12.9867, |
| "eval_samples_per_second": 506.979, |
| "eval_steps_per_second": 7.931, |
| "step": 6600 |
| }, |
| { |
| "epoch": 0.758288401973156, |
| "eval_loss": 0.4532826542854309, |
| "eval_runtime": 13.0195, |
| "eval_samples_per_second": 505.704, |
| "eval_steps_per_second": 7.911, |
| "step": 6610 |
| }, |
| { |
| "epoch": 0.7594355856372605, |
| "eval_loss": 0.4535670280456543, |
| "eval_runtime": 12.9905, |
| "eval_samples_per_second": 506.833, |
| "eval_steps_per_second": 7.929, |
| "step": 6620 |
| }, |
| { |
| "epoch": 0.7605827693013651, |
| "eval_loss": 0.4535196125507355, |
| "eval_runtime": 12.9785, |
| "eval_samples_per_second": 507.301, |
| "eval_steps_per_second": 7.936, |
| "step": 6630 |
| }, |
| { |
| "epoch": 0.7617299529654697, |
| "eval_loss": 0.4532821476459503, |
| "eval_runtime": 12.9651, |
| "eval_samples_per_second": 507.826, |
| "eval_steps_per_second": 7.944, |
| "step": 6640 |
| }, |
| { |
| "epoch": 0.7628771366295743, |
| "eval_loss": 0.4532470405101776, |
| "eval_runtime": 13.0126, |
| "eval_samples_per_second": 505.971, |
| "eval_steps_per_second": 7.915, |
| "step": 6650 |
| }, |
| { |
| "epoch": 0.7640243202936791, |
| "eval_loss": 0.4531117379665375, |
| "eval_runtime": 13.0241, |
| "eval_samples_per_second": 505.526, |
| "eval_steps_per_second": 7.908, |
| "step": 6660 |
| }, |
| { |
| "epoch": 0.7651715039577837, |
| "eval_loss": 0.4530920088291168, |
| "eval_runtime": 13.0159, |
| "eval_samples_per_second": 505.842, |
| "eval_steps_per_second": 7.913, |
| "step": 6670 |
| }, |
| { |
| "epoch": 0.7663186876218883, |
| "eval_loss": 0.4529833197593689, |
| "eval_runtime": 13.103, |
| "eval_samples_per_second": 502.481, |
| "eval_steps_per_second": 7.861, |
| "step": 6680 |
| }, |
| { |
| "epoch": 0.7674658712859929, |
| "eval_loss": 0.45283904671669006, |
| "eval_runtime": 13.1618, |
| "eval_samples_per_second": 500.235, |
| "eval_steps_per_second": 7.826, |
| "step": 6690 |
| }, |
| { |
| "epoch": 0.7686130549500975, |
| "grad_norm": 6.719768047332764, |
| "learning_rate": 8.26471213086892e-06, |
| "loss": 0.8091, |
| "step": 6700 |
| }, |
| { |
| "epoch": 0.7686130549500975, |
| "eval_loss": 0.4526599049568176, |
| "eval_runtime": 13.0347, |
| "eval_samples_per_second": 505.112, |
| "eval_steps_per_second": 7.902, |
| "step": 6700 |
| }, |
| { |
| "epoch": 0.7697602386142022, |
| "eval_loss": 0.45269957184791565, |
| "eval_runtime": 13.1449, |
| "eval_samples_per_second": 500.88, |
| "eval_steps_per_second": 7.836, |
| "step": 6710 |
| }, |
| { |
| "epoch": 0.7709074222783068, |
| "eval_loss": 0.4525602161884308, |
| "eval_runtime": 13.1978, |
| "eval_samples_per_second": 498.872, |
| "eval_steps_per_second": 7.804, |
| "step": 6720 |
| }, |
| { |
| "epoch": 0.7720546059424114, |
| "eval_loss": 0.4524688124656677, |
| "eval_runtime": 13.0602, |
| "eval_samples_per_second": 504.126, |
| "eval_steps_per_second": 7.887, |
| "step": 6730 |
| }, |
| { |
| "epoch": 0.773201789606516, |
| "eval_loss": 0.45239222049713135, |
| "eval_runtime": 12.9858, |
| "eval_samples_per_second": 507.017, |
| "eval_steps_per_second": 7.932, |
| "step": 6740 |
| }, |
| { |
| "epoch": 0.7743489732706206, |
| "eval_loss": 0.4521077275276184, |
| "eval_runtime": 12.9829, |
| "eval_samples_per_second": 507.13, |
| "eval_steps_per_second": 7.934, |
| "step": 6750 |
| }, |
| { |
| "epoch": 0.7754961569347253, |
| "eval_loss": 0.451655775308609, |
| "eval_runtime": 13.0057, |
| "eval_samples_per_second": 506.239, |
| "eval_steps_per_second": 7.92, |
| "step": 6760 |
| }, |
| { |
| "epoch": 0.7766433405988299, |
| "eval_loss": 0.45140165090560913, |
| "eval_runtime": 12.9876, |
| "eval_samples_per_second": 506.945, |
| "eval_steps_per_second": 7.931, |
| "step": 6770 |
| }, |
| { |
| "epoch": 0.7777905242629345, |
| "eval_loss": 0.45122718811035156, |
| "eval_runtime": 12.984, |
| "eval_samples_per_second": 507.084, |
| "eval_steps_per_second": 7.933, |
| "step": 6780 |
| }, |
| { |
| "epoch": 0.7789377079270391, |
| "eval_loss": 0.45139777660369873, |
| "eval_runtime": 13.1503, |
| "eval_samples_per_second": 500.674, |
| "eval_steps_per_second": 7.833, |
| "step": 6790 |
| }, |
| { |
| "epoch": 0.7800848915911437, |
| "grad_norm": 2.830284595489502, |
| "learning_rate": 8.222222222222222e-06, |
| "loss": 0.8098, |
| "step": 6800 |
| }, |
| { |
| "epoch": 0.7800848915911437, |
| "eval_loss": 0.45153865218162537, |
| "eval_runtime": 13.0741, |
| "eval_samples_per_second": 503.591, |
| "eval_steps_per_second": 7.878, |
| "step": 6800 |
| }, |
| { |
| "epoch": 0.7812320752552484, |
| "eval_loss": 0.45143216848373413, |
| "eval_runtime": 13.0429, |
| "eval_samples_per_second": 504.797, |
| "eval_steps_per_second": 7.897, |
| "step": 6810 |
| }, |
| { |
| "epoch": 0.782379258919353, |
| "eval_loss": 0.45105916261672974, |
| "eval_runtime": 12.9831, |
| "eval_samples_per_second": 507.122, |
| "eval_steps_per_second": 7.933, |
| "step": 6820 |
| }, |
| { |
| "epoch": 0.7835264425834576, |
| "eval_loss": 0.45074450969696045, |
| "eval_runtime": 13.1325, |
| "eval_samples_per_second": 501.351, |
| "eval_steps_per_second": 7.843, |
| "step": 6830 |
| }, |
| { |
| "epoch": 0.7846736262475622, |
| "eval_loss": 0.45049625635147095, |
| "eval_runtime": 12.997, |
| "eval_samples_per_second": 506.579, |
| "eval_steps_per_second": 7.925, |
| "step": 6840 |
| }, |
| { |
| "epoch": 0.7858208099116668, |
| "eval_loss": 0.4503604769706726, |
| "eval_runtime": 12.9733, |
| "eval_samples_per_second": 507.502, |
| "eval_steps_per_second": 7.939, |
| "step": 6850 |
| }, |
| { |
| "epoch": 0.7869679935757715, |
| "eval_loss": 0.4502841532230377, |
| "eval_runtime": 12.9771, |
| "eval_samples_per_second": 507.355, |
| "eval_steps_per_second": 7.937, |
| "step": 6860 |
| }, |
| { |
| "epoch": 0.7881151772398761, |
| "eval_loss": 0.44999396800994873, |
| "eval_runtime": 13.087, |
| "eval_samples_per_second": 503.096, |
| "eval_steps_per_second": 7.87, |
| "step": 6870 |
| }, |
| { |
| "epoch": 0.7892623609039807, |
| "eval_loss": 0.4498303532600403, |
| "eval_runtime": 12.9848, |
| "eval_samples_per_second": 507.053, |
| "eval_steps_per_second": 7.932, |
| "step": 6880 |
| }, |
| { |
| "epoch": 0.7904095445680853, |
| "eval_loss": 0.4495447278022766, |
| "eval_runtime": 13.0431, |
| "eval_samples_per_second": 504.787, |
| "eval_steps_per_second": 7.897, |
| "step": 6890 |
| }, |
| { |
| "epoch": 0.7915567282321899, |
| "grad_norm": 2.750669240951538, |
| "learning_rate": 8.179732313575527e-06, |
| "loss": 0.7857, |
| "step": 6900 |
| }, |
| { |
| "epoch": 0.7915567282321899, |
| "eval_loss": 0.44914257526397705, |
| "eval_runtime": 13.0753, |
| "eval_samples_per_second": 503.546, |
| "eval_steps_per_second": 7.877, |
| "step": 6900 |
| }, |
| { |
| "epoch": 0.7927039118962946, |
| "eval_loss": 0.4489689767360687, |
| "eval_runtime": 12.989, |
| "eval_samples_per_second": 506.892, |
| "eval_steps_per_second": 7.93, |
| "step": 6910 |
| }, |
| { |
| "epoch": 0.7938510955603992, |
| "eval_loss": 0.44879305362701416, |
| "eval_runtime": 12.9912, |
| "eval_samples_per_second": 506.806, |
| "eval_steps_per_second": 7.928, |
| "step": 6920 |
| }, |
| { |
| "epoch": 0.7949982792245038, |
| "eval_loss": 0.44882142543792725, |
| "eval_runtime": 12.9886, |
| "eval_samples_per_second": 506.907, |
| "eval_steps_per_second": 7.93, |
| "step": 6930 |
| }, |
| { |
| "epoch": 0.7961454628886084, |
| "eval_loss": 0.4487648010253906, |
| "eval_runtime": 12.9888, |
| "eval_samples_per_second": 506.9, |
| "eval_steps_per_second": 7.93, |
| "step": 6940 |
| }, |
| { |
| "epoch": 0.797292646552713, |
| "eval_loss": 0.4487059712409973, |
| "eval_runtime": 13.026, |
| "eval_samples_per_second": 505.452, |
| "eval_steps_per_second": 7.907, |
| "step": 6950 |
| }, |
| { |
| "epoch": 0.7984398302168177, |
| "eval_loss": 0.4484008252620697, |
| "eval_runtime": 12.9783, |
| "eval_samples_per_second": 507.31, |
| "eval_steps_per_second": 7.936, |
| "step": 6960 |
| }, |
| { |
| "epoch": 0.7995870138809223, |
| "eval_loss": 0.44824883341789246, |
| "eval_runtime": 13.1902, |
| "eval_samples_per_second": 499.158, |
| "eval_steps_per_second": 7.809, |
| "step": 6970 |
| }, |
| { |
| "epoch": 0.800734197545027, |
| "eval_loss": 0.44833362102508545, |
| "eval_runtime": 13.1729, |
| "eval_samples_per_second": 499.816, |
| "eval_steps_per_second": 7.819, |
| "step": 6980 |
| }, |
| { |
| "epoch": 0.8018813812091315, |
| "eval_loss": 0.4480605721473694, |
| "eval_runtime": 13.1992, |
| "eval_samples_per_second": 498.816, |
| "eval_steps_per_second": 7.803, |
| "step": 6990 |
| }, |
| { |
| "epoch": 0.8030285648732362, |
| "grad_norm": 2.8977108001708984, |
| "learning_rate": 8.13724240492883e-06, |
| "loss": 0.7817, |
| "step": 7000 |
| }, |
| { |
| "epoch": 0.8030285648732362, |
| "eval_loss": 0.44771045446395874, |
| "eval_runtime": 13.1683, |
| "eval_samples_per_second": 499.988, |
| "eval_steps_per_second": 7.822, |
| "step": 7000 |
| }, |
| { |
| "epoch": 0.8041757485373409, |
| "eval_loss": 0.44761350750923157, |
| "eval_runtime": 13.0621, |
| "eval_samples_per_second": 504.054, |
| "eval_steps_per_second": 7.885, |
| "step": 7010 |
| }, |
| { |
| "epoch": 0.8053229322014455, |
| "eval_loss": 0.44709402322769165, |
| "eval_runtime": 13.0048, |
| "eval_samples_per_second": 506.276, |
| "eval_steps_per_second": 7.92, |
| "step": 7020 |
| }, |
| { |
| "epoch": 0.8064701158655501, |
| "eval_loss": 0.4468764364719391, |
| "eval_runtime": 13.0715, |
| "eval_samples_per_second": 503.693, |
| "eval_steps_per_second": 7.88, |
| "step": 7030 |
| }, |
| { |
| "epoch": 0.8076172995296547, |
| "eval_loss": 0.4468289315700531, |
| "eval_runtime": 13.1458, |
| "eval_samples_per_second": 500.843, |
| "eval_steps_per_second": 7.835, |
| "step": 7040 |
| }, |
| { |
| "epoch": 0.8087644831937594, |
| "eval_loss": 0.44648417830467224, |
| "eval_runtime": 12.9814, |
| "eval_samples_per_second": 507.186, |
| "eval_steps_per_second": 7.934, |
| "step": 7050 |
| }, |
| { |
| "epoch": 0.809911666857864, |
| "eval_loss": 0.44603875279426575, |
| "eval_runtime": 12.9715, |
| "eval_samples_per_second": 507.575, |
| "eval_steps_per_second": 7.941, |
| "step": 7060 |
| }, |
| { |
| "epoch": 0.8110588505219686, |
| "eval_loss": 0.4458308517932892, |
| "eval_runtime": 12.9963, |
| "eval_samples_per_second": 506.604, |
| "eval_steps_per_second": 7.925, |
| "step": 7070 |
| }, |
| { |
| "epoch": 0.8122060341860732, |
| "eval_loss": 0.44575193524360657, |
| "eval_runtime": 12.9902, |
| "eval_samples_per_second": 506.842, |
| "eval_steps_per_second": 7.929, |
| "step": 7080 |
| }, |
| { |
| "epoch": 0.8133532178501778, |
| "eval_loss": 0.4453906714916229, |
| "eval_runtime": 13.1119, |
| "eval_samples_per_second": 502.14, |
| "eval_steps_per_second": 7.855, |
| "step": 7090 |
| }, |
| { |
| "epoch": 0.8145004015142825, |
| "grad_norm": 4.991448879241943, |
| "learning_rate": 8.094752496282133e-06, |
| "loss": 0.779, |
| "step": 7100 |
| }, |
| { |
| "epoch": 0.8145004015142825, |
| "eval_loss": 0.44515758752822876, |
| "eval_runtime": 13.1581, |
| "eval_samples_per_second": 500.375, |
| "eval_steps_per_second": 7.828, |
| "step": 7100 |
| }, |
| { |
| "epoch": 0.8156475851783871, |
| "eval_loss": 0.4449334442615509, |
| "eval_runtime": 13.1519, |
| "eval_samples_per_second": 500.611, |
| "eval_steps_per_second": 7.832, |
| "step": 7110 |
| }, |
| { |
| "epoch": 0.8167947688424917, |
| "eval_loss": 0.44477564096450806, |
| "eval_runtime": 13.0668, |
| "eval_samples_per_second": 503.872, |
| "eval_steps_per_second": 7.883, |
| "step": 7120 |
| }, |
| { |
| "epoch": 0.8179419525065963, |
| "eval_loss": 0.4445902407169342, |
| "eval_runtime": 13.0492, |
| "eval_samples_per_second": 504.553, |
| "eval_steps_per_second": 7.893, |
| "step": 7130 |
| }, |
| { |
| "epoch": 0.8190891361707009, |
| "eval_loss": 0.4442489445209503, |
| "eval_runtime": 13.063, |
| "eval_samples_per_second": 504.021, |
| "eval_steps_per_second": 7.885, |
| "step": 7140 |
| }, |
| { |
| "epoch": 0.8202363198348056, |
| "eval_loss": 0.4442029893398285, |
| "eval_runtime": 12.9882, |
| "eval_samples_per_second": 506.923, |
| "eval_steps_per_second": 7.93, |
| "step": 7150 |
| }, |
| { |
| "epoch": 0.8213835034989102, |
| "eval_loss": 0.4440893530845642, |
| "eval_runtime": 12.9936, |
| "eval_samples_per_second": 506.712, |
| "eval_steps_per_second": 7.927, |
| "step": 7160 |
| }, |
| { |
| "epoch": 0.8225306871630148, |
| "eval_loss": 0.4440220892429352, |
| "eval_runtime": 12.9882, |
| "eval_samples_per_second": 506.92, |
| "eval_steps_per_second": 7.93, |
| "step": 7170 |
| }, |
| { |
| "epoch": 0.8236778708271194, |
| "eval_loss": 0.4436882734298706, |
| "eval_runtime": 12.9892, |
| "eval_samples_per_second": 506.882, |
| "eval_steps_per_second": 7.93, |
| "step": 7180 |
| }, |
| { |
| "epoch": 0.824825054491224, |
| "eval_loss": 0.44342365860939026, |
| "eval_runtime": 12.9866, |
| "eval_samples_per_second": 506.986, |
| "eval_steps_per_second": 7.931, |
| "step": 7190 |
| }, |
| { |
| "epoch": 0.8259722381553287, |
| "grad_norm": 3.3440160751342773, |
| "learning_rate": 8.052262587635438e-06, |
| "loss": 0.7807, |
| "step": 7200 |
| }, |
| { |
| "epoch": 0.8259722381553287, |
| "eval_loss": 0.44338205456733704, |
| "eval_runtime": 13.001, |
| "eval_samples_per_second": 506.424, |
| "eval_steps_per_second": 7.922, |
| "step": 7200 |
| }, |
| { |
| "epoch": 0.8271194218194333, |
| "eval_loss": 0.443464457988739, |
| "eval_runtime": 13.0209, |
| "eval_samples_per_second": 505.647, |
| "eval_steps_per_second": 7.91, |
| "step": 7210 |
| }, |
| { |
| "epoch": 0.8282666054835379, |
| "eval_loss": 0.4433028995990753, |
| "eval_runtime": 13.0102, |
| "eval_samples_per_second": 506.066, |
| "eval_steps_per_second": 7.917, |
| "step": 7220 |
| }, |
| { |
| "epoch": 0.8294137891476425, |
| "eval_loss": 0.44313445687294006, |
| "eval_runtime": 13.0757, |
| "eval_samples_per_second": 503.529, |
| "eval_steps_per_second": 7.877, |
| "step": 7230 |
| }, |
| { |
| "epoch": 0.8305609728117471, |
| "eval_loss": 0.4429507553577423, |
| "eval_runtime": 13.0812, |
| "eval_samples_per_second": 503.318, |
| "eval_steps_per_second": 7.874, |
| "step": 7240 |
| }, |
| { |
| "epoch": 0.8317081564758518, |
| "eval_loss": 0.44282954931259155, |
| "eval_runtime": 13.1417, |
| "eval_samples_per_second": 501.002, |
| "eval_steps_per_second": 7.838, |
| "step": 7250 |
| }, |
| { |
| "epoch": 0.8328553401399564, |
| "eval_loss": 0.44259387254714966, |
| "eval_runtime": 13.0693, |
| "eval_samples_per_second": 503.778, |
| "eval_steps_per_second": 7.881, |
| "step": 7260 |
| }, |
| { |
| "epoch": 0.834002523804061, |
| "eval_loss": 0.4424245059490204, |
| "eval_runtime": 13.0404, |
| "eval_samples_per_second": 504.893, |
| "eval_steps_per_second": 7.899, |
| "step": 7270 |
| }, |
| { |
| "epoch": 0.8351497074681656, |
| "eval_loss": 0.4427710473537445, |
| "eval_runtime": 12.9895, |
| "eval_samples_per_second": 506.872, |
| "eval_steps_per_second": 7.929, |
| "step": 7280 |
| }, |
| { |
| "epoch": 0.8362968911322702, |
| "eval_loss": 0.44257354736328125, |
| "eval_runtime": 13.1032, |
| "eval_samples_per_second": 502.474, |
| "eval_steps_per_second": 7.861, |
| "step": 7290 |
| }, |
| { |
| "epoch": 0.837444074796375, |
| "grad_norm": 2.637993097305298, |
| "learning_rate": 8.009772678988742e-06, |
| "loss": 0.7724, |
| "step": 7300 |
| }, |
| { |
| "epoch": 0.837444074796375, |
| "eval_loss": 0.4422559440135956, |
| "eval_runtime": 12.9798, |
| "eval_samples_per_second": 507.251, |
| "eval_steps_per_second": 7.935, |
| "step": 7300 |
| }, |
| { |
| "epoch": 0.8385912584604795, |
| "eval_loss": 0.44188517332077026, |
| "eval_runtime": 13.1857, |
| "eval_samples_per_second": 499.329, |
| "eval_steps_per_second": 7.811, |
| "step": 7310 |
| }, |
| { |
| "epoch": 0.8397384421245842, |
| "eval_loss": 0.4417719542980194, |
| "eval_runtime": 13.0086, |
| "eval_samples_per_second": 506.127, |
| "eval_steps_per_second": 7.918, |
| "step": 7320 |
| }, |
| { |
| "epoch": 0.8408856257886888, |
| "eval_loss": 0.44172385334968567, |
| "eval_runtime": 13.0294, |
| "eval_samples_per_second": 505.32, |
| "eval_steps_per_second": 7.905, |
| "step": 7330 |
| }, |
| { |
| "epoch": 0.8420328094527934, |
| "eval_loss": 0.44147878885269165, |
| "eval_runtime": 13.0966, |
| "eval_samples_per_second": 502.726, |
| "eval_steps_per_second": 7.865, |
| "step": 7340 |
| }, |
| { |
| "epoch": 0.8431799931168981, |
| "eval_loss": 0.4412890374660492, |
| "eval_runtime": 13.097, |
| "eval_samples_per_second": 502.71, |
| "eval_steps_per_second": 7.864, |
| "step": 7350 |
| }, |
| { |
| "epoch": 0.8443271767810027, |
| "eval_loss": 0.440915048122406, |
| "eval_runtime": 13.0035, |
| "eval_samples_per_second": 506.325, |
| "eval_steps_per_second": 7.921, |
| "step": 7360 |
| }, |
| { |
| "epoch": 0.8454743604451073, |
| "eval_loss": 0.44064828753471375, |
| "eval_runtime": 13.0917, |
| "eval_samples_per_second": 502.913, |
| "eval_steps_per_second": 7.868, |
| "step": 7370 |
| }, |
| { |
| "epoch": 0.8466215441092119, |
| "eval_loss": 0.44048747420310974, |
| "eval_runtime": 13.1371, |
| "eval_samples_per_second": 501.176, |
| "eval_steps_per_second": 7.84, |
| "step": 7380 |
| }, |
| { |
| "epoch": 0.8477687277733165, |
| "eval_loss": 0.4400079548358917, |
| "eval_runtime": 12.995, |
| "eval_samples_per_second": 506.658, |
| "eval_steps_per_second": 7.926, |
| "step": 7390 |
| }, |
| { |
| "epoch": 0.8489159114374212, |
| "grad_norm": 2.776257276535034, |
| "learning_rate": 7.967282770342044e-06, |
| "loss": 0.7898, |
| "step": 7400 |
| }, |
| { |
| "epoch": 0.8489159114374212, |
| "eval_loss": 0.43930843472480774, |
| "eval_runtime": 13.0797, |
| "eval_samples_per_second": 503.376, |
| "eval_steps_per_second": 7.875, |
| "step": 7400 |
| }, |
| { |
| "epoch": 0.8500630951015258, |
| "eval_loss": 0.43894800543785095, |
| "eval_runtime": 13.2518, |
| "eval_samples_per_second": 496.838, |
| "eval_steps_per_second": 7.773, |
| "step": 7410 |
| }, |
| { |
| "epoch": 0.8512102787656304, |
| "eval_loss": 0.4384170174598694, |
| "eval_runtime": 12.9959, |
| "eval_samples_per_second": 506.621, |
| "eval_steps_per_second": 7.926, |
| "step": 7420 |
| }, |
| { |
| "epoch": 0.852357462429735, |
| "eval_loss": 0.43813180923461914, |
| "eval_runtime": 13.0185, |
| "eval_samples_per_second": 505.744, |
| "eval_steps_per_second": 7.912, |
| "step": 7430 |
| }, |
| { |
| "epoch": 0.8535046460938396, |
| "eval_loss": 0.43801015615463257, |
| "eval_runtime": 12.9767, |
| "eval_samples_per_second": 507.37, |
| "eval_steps_per_second": 7.937, |
| "step": 7440 |
| }, |
| { |
| "epoch": 0.8546518297579443, |
| "eval_loss": 0.43804019689559937, |
| "eval_runtime": 13.017, |
| "eval_samples_per_second": 505.798, |
| "eval_steps_per_second": 7.913, |
| "step": 7450 |
| }, |
| { |
| "epoch": 0.8557990134220489, |
| "eval_loss": 0.4378819465637207, |
| "eval_runtime": 13.0288, |
| "eval_samples_per_second": 505.34, |
| "eval_steps_per_second": 7.906, |
| "step": 7460 |
| }, |
| { |
| "epoch": 0.8569461970861535, |
| "eval_loss": 0.4376521110534668, |
| "eval_runtime": 13.042, |
| "eval_samples_per_second": 504.83, |
| "eval_steps_per_second": 7.898, |
| "step": 7470 |
| }, |
| { |
| "epoch": 0.8580933807502581, |
| "eval_loss": 0.4377218782901764, |
| "eval_runtime": 13.0815, |
| "eval_samples_per_second": 503.307, |
| "eval_steps_per_second": 7.874, |
| "step": 7480 |
| }, |
| { |
| "epoch": 0.8592405644143627, |
| "eval_loss": 0.4375830590724945, |
| "eval_runtime": 13.0085, |
| "eval_samples_per_second": 506.13, |
| "eval_steps_per_second": 7.918, |
| "step": 7490 |
| }, |
| { |
| "epoch": 0.8603877480784674, |
| "grad_norm": 2.578803300857544, |
| "learning_rate": 7.924792861695349e-06, |
| "loss": 0.8009, |
| "step": 7500 |
| }, |
| { |
| "epoch": 0.8603877480784674, |
| "eval_loss": 0.4374997019767761, |
| "eval_runtime": 12.964, |
| "eval_samples_per_second": 507.87, |
| "eval_steps_per_second": 7.945, |
| "step": 7500 |
| }, |
| { |
| "epoch": 0.861534931742572, |
| "eval_loss": 0.4370783865451813, |
| "eval_runtime": 13.1536, |
| "eval_samples_per_second": 500.548, |
| "eval_steps_per_second": 7.831, |
| "step": 7510 |
| }, |
| { |
| "epoch": 0.8626821154066766, |
| "eval_loss": 0.4368523359298706, |
| "eval_runtime": 13.1596, |
| "eval_samples_per_second": 500.318, |
| "eval_steps_per_second": 7.827, |
| "step": 7520 |
| }, |
| { |
| "epoch": 0.8638292990707812, |
| "eval_loss": 0.4364851415157318, |
| "eval_runtime": 13.0167, |
| "eval_samples_per_second": 505.812, |
| "eval_steps_per_second": 7.913, |
| "step": 7530 |
| }, |
| { |
| "epoch": 0.8649764827348858, |
| "eval_loss": 0.43616530299186707, |
| "eval_runtime": 13.0258, |
| "eval_samples_per_second": 505.457, |
| "eval_steps_per_second": 7.907, |
| "step": 7540 |
| }, |
| { |
| "epoch": 0.8661236663989905, |
| "eval_loss": 0.43588289618492126, |
| "eval_runtime": 12.9544, |
| "eval_samples_per_second": 508.244, |
| "eval_steps_per_second": 7.951, |
| "step": 7550 |
| }, |
| { |
| "epoch": 0.8672708500630951, |
| "eval_loss": 0.4356750249862671, |
| "eval_runtime": 13.0073, |
| "eval_samples_per_second": 506.177, |
| "eval_steps_per_second": 7.919, |
| "step": 7560 |
| }, |
| { |
| "epoch": 0.8684180337271997, |
| "eval_loss": 0.4355123043060303, |
| "eval_runtime": 13.1717, |
| "eval_samples_per_second": 499.86, |
| "eval_steps_per_second": 7.82, |
| "step": 7570 |
| }, |
| { |
| "epoch": 0.8695652173913043, |
| "eval_loss": 0.4350906312465668, |
| "eval_runtime": 13.1392, |
| "eval_samples_per_second": 501.097, |
| "eval_steps_per_second": 7.839, |
| "step": 7580 |
| }, |
| { |
| "epoch": 0.8707124010554089, |
| "eval_loss": 0.43471574783325195, |
| "eval_runtime": 12.9597, |
| "eval_samples_per_second": 508.036, |
| "eval_steps_per_second": 7.948, |
| "step": 7590 |
| }, |
| { |
| "epoch": 0.8718595847195136, |
| "grad_norm": 2.563943386077881, |
| "learning_rate": 7.882302953048651e-06, |
| "loss": 0.7847, |
| "step": 7600 |
| }, |
| { |
| "epoch": 0.8718595847195136, |
| "eval_loss": 0.434588223695755, |
| "eval_runtime": 13.0423, |
| "eval_samples_per_second": 504.819, |
| "eval_steps_per_second": 7.897, |
| "step": 7600 |
| }, |
| { |
| "epoch": 0.8730067683836182, |
| "eval_loss": 0.4345826208591461, |
| "eval_runtime": 13.0101, |
| "eval_samples_per_second": 506.068, |
| "eval_steps_per_second": 7.917, |
| "step": 7610 |
| }, |
| { |
| "epoch": 0.8741539520477228, |
| "eval_loss": 0.4344346821308136, |
| "eval_runtime": 13.0885, |
| "eval_samples_per_second": 503.036, |
| "eval_steps_per_second": 7.869, |
| "step": 7620 |
| }, |
| { |
| "epoch": 0.8753011357118274, |
| "eval_loss": 0.43427374958992004, |
| "eval_runtime": 12.9682, |
| "eval_samples_per_second": 507.702, |
| "eval_steps_per_second": 7.942, |
| "step": 7630 |
| }, |
| { |
| "epoch": 0.876448319375932, |
| "eval_loss": 0.43383604288101196, |
| "eval_runtime": 13.0749, |
| "eval_samples_per_second": 503.561, |
| "eval_steps_per_second": 7.878, |
| "step": 7640 |
| }, |
| { |
| "epoch": 0.8775955030400368, |
| "eval_loss": 0.43361151218414307, |
| "eval_runtime": 12.9663, |
| "eval_samples_per_second": 507.778, |
| "eval_steps_per_second": 7.944, |
| "step": 7650 |
| }, |
| { |
| "epoch": 0.8787426867041414, |
| "eval_loss": 0.4332073926925659, |
| "eval_runtime": 12.9541, |
| "eval_samples_per_second": 508.255, |
| "eval_steps_per_second": 7.951, |
| "step": 7660 |
| }, |
| { |
| "epoch": 0.879889870368246, |
| "eval_loss": 0.43305349349975586, |
| "eval_runtime": 12.9935, |
| "eval_samples_per_second": 506.714, |
| "eval_steps_per_second": 7.927, |
| "step": 7670 |
| }, |
| { |
| "epoch": 0.8810370540323506, |
| "eval_loss": 0.43292590975761414, |
| "eval_runtime": 13.1467, |
| "eval_samples_per_second": 500.812, |
| "eval_steps_per_second": 7.835, |
| "step": 7680 |
| }, |
| { |
| "epoch": 0.8821842376964552, |
| "eval_loss": 0.43263253569602966, |
| "eval_runtime": 13.0897, |
| "eval_samples_per_second": 502.992, |
| "eval_steps_per_second": 7.869, |
| "step": 7690 |
| }, |
| { |
| "epoch": 0.8833314213605599, |
| "grad_norm": 1.9172425270080566, |
| "learning_rate": 7.839813044401955e-06, |
| "loss": 0.7668, |
| "step": 7700 |
| }, |
| { |
| "epoch": 0.8833314213605599, |
| "eval_loss": 0.43244412541389465, |
| "eval_runtime": 12.9892, |
| "eval_samples_per_second": 506.881, |
| "eval_steps_per_second": 7.93, |
| "step": 7700 |
| }, |
| { |
| "epoch": 0.8844786050246645, |
| "eval_loss": 0.4325471520423889, |
| "eval_runtime": 13.0449, |
| "eval_samples_per_second": 504.72, |
| "eval_steps_per_second": 7.896, |
| "step": 7710 |
| }, |
| { |
| "epoch": 0.8856257886887691, |
| "eval_loss": 0.4327198266983032, |
| "eval_runtime": 13.0782, |
| "eval_samples_per_second": 503.432, |
| "eval_steps_per_second": 7.876, |
| "step": 7720 |
| }, |
| { |
| "epoch": 0.8867729723528737, |
| "eval_loss": 0.4329088628292084, |
| "eval_runtime": 13.0079, |
| "eval_samples_per_second": 506.153, |
| "eval_steps_per_second": 7.918, |
| "step": 7730 |
| }, |
| { |
| "epoch": 0.8879201560169783, |
| "eval_loss": 0.4328489899635315, |
| "eval_runtime": 12.9762, |
| "eval_samples_per_second": 507.39, |
| "eval_steps_per_second": 7.938, |
| "step": 7740 |
| }, |
| { |
| "epoch": 0.889067339681083, |
| "eval_loss": 0.43251752853393555, |
| "eval_runtime": 13.0108, |
| "eval_samples_per_second": 506.043, |
| "eval_steps_per_second": 7.917, |
| "step": 7750 |
| }, |
| { |
| "epoch": 0.8902145233451876, |
| "eval_loss": 0.4324817657470703, |
| "eval_runtime": 13.1917, |
| "eval_samples_per_second": 499.102, |
| "eval_steps_per_second": 7.808, |
| "step": 7760 |
| }, |
| { |
| "epoch": 0.8913617070092922, |
| "eval_loss": 0.432619571685791, |
| "eval_runtime": 12.9588, |
| "eval_samples_per_second": 508.07, |
| "eval_steps_per_second": 7.948, |
| "step": 7770 |
| }, |
| { |
| "epoch": 0.8925088906733968, |
| "eval_loss": 0.43235525488853455, |
| "eval_runtime": 13.1105, |
| "eval_samples_per_second": 502.193, |
| "eval_steps_per_second": 7.856, |
| "step": 7780 |
| }, |
| { |
| "epoch": 0.8936560743375014, |
| "eval_loss": 0.43218794465065, |
| "eval_runtime": 13.023, |
| "eval_samples_per_second": 505.566, |
| "eval_steps_per_second": 7.909, |
| "step": 7790 |
| }, |
| { |
| "epoch": 0.8948032580016061, |
| "grad_norm": 2.1403651237487793, |
| "learning_rate": 7.79732313575526e-06, |
| "loss": 0.7987, |
| "step": 7800 |
| }, |
| { |
| "epoch": 0.8948032580016061, |
| "eval_loss": 0.4319910705089569, |
| "eval_runtime": 12.9746, |
| "eval_samples_per_second": 507.452, |
| "eval_steps_per_second": 7.939, |
| "step": 7800 |
| }, |
| { |
| "epoch": 0.8959504416657107, |
| "eval_loss": 0.4318791329860687, |
| "eval_runtime": 13.0072, |
| "eval_samples_per_second": 506.18, |
| "eval_steps_per_second": 7.919, |
| "step": 7810 |
| }, |
| { |
| "epoch": 0.8970976253298153, |
| "eval_loss": 0.4317561089992523, |
| "eval_runtime": 12.9579, |
| "eval_samples_per_second": 508.108, |
| "eval_steps_per_second": 7.949, |
| "step": 7820 |
| }, |
| { |
| "epoch": 0.8982448089939199, |
| "eval_loss": 0.43148013949394226, |
| "eval_runtime": 12.9637, |
| "eval_samples_per_second": 507.879, |
| "eval_steps_per_second": 7.945, |
| "step": 7830 |
| }, |
| { |
| "epoch": 0.8993919926580245, |
| "eval_loss": 0.43115872144699097, |
| "eval_runtime": 12.9823, |
| "eval_samples_per_second": 507.152, |
| "eval_steps_per_second": 7.934, |
| "step": 7840 |
| }, |
| { |
| "epoch": 0.9005391763221292, |
| "eval_loss": 0.4308260977268219, |
| "eval_runtime": 12.9885, |
| "eval_samples_per_second": 506.908, |
| "eval_steps_per_second": 7.93, |
| "step": 7850 |
| }, |
| { |
| "epoch": 0.9016863599862338, |
| "eval_loss": 0.4307934045791626, |
| "eval_runtime": 13.0462, |
| "eval_samples_per_second": 504.668, |
| "eval_steps_per_second": 7.895, |
| "step": 7860 |
| }, |
| { |
| "epoch": 0.9028335436503384, |
| "eval_loss": 0.4309324622154236, |
| "eval_runtime": 12.9629, |
| "eval_samples_per_second": 507.91, |
| "eval_steps_per_second": 7.946, |
| "step": 7870 |
| }, |
| { |
| "epoch": 0.903980727314443, |
| "eval_loss": 0.4305669069290161, |
| "eval_runtime": 12.9661, |
| "eval_samples_per_second": 507.784, |
| "eval_steps_per_second": 7.944, |
| "step": 7880 |
| }, |
| { |
| "epoch": 0.9051279109785476, |
| "eval_loss": 0.43051281571388245, |
| "eval_runtime": 12.9389, |
| "eval_samples_per_second": 508.852, |
| "eval_steps_per_second": 7.96, |
| "step": 7890 |
| }, |
| { |
| "epoch": 0.9062750946426523, |
| "grad_norm": 2.909410238265991, |
| "learning_rate": 7.754833227108562e-06, |
| "loss": 0.7691, |
| "step": 7900 |
| }, |
| { |
| "epoch": 0.9062750946426523, |
| "eval_loss": 0.4305475056171417, |
| "eval_runtime": 12.9608, |
| "eval_samples_per_second": 507.992, |
| "eval_steps_per_second": 7.947, |
| "step": 7900 |
| }, |
| { |
| "epoch": 0.9074222783067569, |
| "eval_loss": 0.43047958612442017, |
| "eval_runtime": 12.9552, |
| "eval_samples_per_second": 508.214, |
| "eval_steps_per_second": 7.95, |
| "step": 7910 |
| }, |
| { |
| "epoch": 0.9085694619708615, |
| "eval_loss": 0.43082302808761597, |
| "eval_runtime": 12.9608, |
| "eval_samples_per_second": 507.994, |
| "eval_steps_per_second": 7.947, |
| "step": 7920 |
| }, |
| { |
| "epoch": 0.9097166456349661, |
| "eval_loss": 0.4308614730834961, |
| "eval_runtime": 12.9456, |
| "eval_samples_per_second": 508.588, |
| "eval_steps_per_second": 7.956, |
| "step": 7930 |
| }, |
| { |
| "epoch": 0.9108638292990707, |
| "eval_loss": 0.43091505765914917, |
| "eval_runtime": 12.945, |
| "eval_samples_per_second": 508.615, |
| "eval_steps_per_second": 7.957, |
| "step": 7940 |
| }, |
| { |
| "epoch": 0.9120110129631754, |
| "eval_loss": 0.4305214583873749, |
| "eval_runtime": 12.9756, |
| "eval_samples_per_second": 507.413, |
| "eval_steps_per_second": 7.938, |
| "step": 7950 |
| }, |
| { |
| "epoch": 0.91315819662728, |
| "eval_loss": 0.42969560623168945, |
| "eval_runtime": 12.9424, |
| "eval_samples_per_second": 508.716, |
| "eval_steps_per_second": 7.958, |
| "step": 7960 |
| }, |
| { |
| "epoch": 0.9143053802913846, |
| "eval_loss": 0.42939293384552, |
| "eval_runtime": 12.9916, |
| "eval_samples_per_second": 506.787, |
| "eval_steps_per_second": 7.928, |
| "step": 7970 |
| }, |
| { |
| "epoch": 0.9154525639554892, |
| "eval_loss": 0.4292427599430084, |
| "eval_runtime": 13.1323, |
| "eval_samples_per_second": 501.359, |
| "eval_steps_per_second": 7.843, |
| "step": 7980 |
| }, |
| { |
| "epoch": 0.916599747619594, |
| "eval_loss": 0.4292071461677551, |
| "eval_runtime": 13.1059, |
| "eval_samples_per_second": 502.367, |
| "eval_steps_per_second": 7.859, |
| "step": 7990 |
| }, |
| { |
| "epoch": 0.9177469312836986, |
| "grad_norm": 4.258418560028076, |
| "learning_rate": 7.712343318461866e-06, |
| "loss": 0.7828, |
| "step": 8000 |
| }, |
| { |
| "epoch": 0.9177469312836986, |
| "eval_loss": 0.4289082884788513, |
| "eval_runtime": 13.0768, |
| "eval_samples_per_second": 503.486, |
| "eval_steps_per_second": 7.877, |
| "step": 8000 |
| }, |
| { |
| "epoch": 0.9188941149478032, |
| "eval_loss": 0.4288167655467987, |
| "eval_runtime": 12.9488, |
| "eval_samples_per_second": 508.466, |
| "eval_steps_per_second": 7.954, |
| "step": 8010 |
| }, |
| { |
| "epoch": 0.9200412986119078, |
| "eval_loss": 0.42887839674949646, |
| "eval_runtime": 12.9558, |
| "eval_samples_per_second": 508.188, |
| "eval_steps_per_second": 7.95, |
| "step": 8020 |
| }, |
| { |
| "epoch": 0.9211884822760124, |
| "eval_loss": 0.4285453259944916, |
| "eval_runtime": 13.0794, |
| "eval_samples_per_second": 503.387, |
| "eval_steps_per_second": 7.875, |
| "step": 8030 |
| }, |
| { |
| "epoch": 0.9223356659401171, |
| "eval_loss": 0.42856818437576294, |
| "eval_runtime": 12.9353, |
| "eval_samples_per_second": 508.996, |
| "eval_steps_per_second": 7.963, |
| "step": 8040 |
| }, |
| { |
| "epoch": 0.9234828496042217, |
| "eval_loss": 0.42885395884513855, |
| "eval_runtime": 12.981, |
| "eval_samples_per_second": 507.202, |
| "eval_steps_per_second": 7.935, |
| "step": 8050 |
| }, |
| { |
| "epoch": 0.9246300332683263, |
| "eval_loss": 0.4287983775138855, |
| "eval_runtime": 12.9634, |
| "eval_samples_per_second": 507.892, |
| "eval_steps_per_second": 7.945, |
| "step": 8060 |
| }, |
| { |
| "epoch": 0.9257772169324309, |
| "eval_loss": 0.42901915311813354, |
| "eval_runtime": 12.9515, |
| "eval_samples_per_second": 508.36, |
| "eval_steps_per_second": 7.953, |
| "step": 8070 |
| }, |
| { |
| "epoch": 0.9269244005965355, |
| "eval_loss": 0.42891281843185425, |
| "eval_runtime": 13.0716, |
| "eval_samples_per_second": 503.687, |
| "eval_steps_per_second": 7.88, |
| "step": 8080 |
| }, |
| { |
| "epoch": 0.9280715842606402, |
| "eval_loss": 0.42868056893348694, |
| "eval_runtime": 13.0589, |
| "eval_samples_per_second": 504.177, |
| "eval_steps_per_second": 7.887, |
| "step": 8090 |
| }, |
| { |
| "epoch": 0.9292187679247448, |
| "grad_norm": 3.712153196334839, |
| "learning_rate": 7.66985340981517e-06, |
| "loss": 0.7544, |
| "step": 8100 |
| }, |
| { |
| "epoch": 0.9292187679247448, |
| "eval_loss": 0.42875462770462036, |
| "eval_runtime": 13.0317, |
| "eval_samples_per_second": 505.228, |
| "eval_steps_per_second": 7.904, |
| "step": 8100 |
| }, |
| { |
| "epoch": 0.9303659515888494, |
| "eval_loss": 0.42842909693717957, |
| "eval_runtime": 13.1365, |
| "eval_samples_per_second": 501.199, |
| "eval_steps_per_second": 7.841, |
| "step": 8110 |
| }, |
| { |
| "epoch": 0.931513135252954, |
| "eval_loss": 0.4286580979824066, |
| "eval_runtime": 12.9604, |
| "eval_samples_per_second": 508.01, |
| "eval_steps_per_second": 7.947, |
| "step": 8120 |
| }, |
| { |
| "epoch": 0.9326603189170586, |
| "eval_loss": 0.4289449155330658, |
| "eval_runtime": 12.9456, |
| "eval_samples_per_second": 508.59, |
| "eval_steps_per_second": 7.956, |
| "step": 8130 |
| }, |
| { |
| "epoch": 0.9338075025811633, |
| "eval_loss": 0.4292765259742737, |
| "eval_runtime": 13.2041, |
| "eval_samples_per_second": 498.633, |
| "eval_steps_per_second": 7.801, |
| "step": 8140 |
| }, |
| { |
| "epoch": 0.9349546862452679, |
| "eval_loss": 0.429167777299881, |
| "eval_runtime": 13.0311, |
| "eval_samples_per_second": 505.251, |
| "eval_steps_per_second": 7.904, |
| "step": 8150 |
| }, |
| { |
| "epoch": 0.9361018699093725, |
| "eval_loss": 0.4289201498031616, |
| "eval_runtime": 13.1313, |
| "eval_samples_per_second": 501.396, |
| "eval_steps_per_second": 7.844, |
| "step": 8160 |
| }, |
| { |
| "epoch": 0.9372490535734771, |
| "eval_loss": 0.42864990234375, |
| "eval_runtime": 12.985, |
| "eval_samples_per_second": 507.046, |
| "eval_steps_per_second": 7.932, |
| "step": 8170 |
| }, |
| { |
| "epoch": 0.9383962372375817, |
| "eval_loss": 0.4280158579349518, |
| "eval_runtime": 13.0779, |
| "eval_samples_per_second": 503.444, |
| "eval_steps_per_second": 7.876, |
| "step": 8180 |
| }, |
| { |
| "epoch": 0.9395434209016864, |
| "eval_loss": 0.4280916750431061, |
| "eval_runtime": 13.0317, |
| "eval_samples_per_second": 505.23, |
| "eval_steps_per_second": 7.904, |
| "step": 8190 |
| }, |
| { |
| "epoch": 0.940690604565791, |
| "grad_norm": 3.268425226211548, |
| "learning_rate": 7.627363501168474e-06, |
| "loss": 0.7502, |
| "step": 8200 |
| }, |
| { |
| "epoch": 0.940690604565791, |
| "eval_loss": 0.4280862808227539, |
| "eval_runtime": 13.1138, |
| "eval_samples_per_second": 502.065, |
| "eval_steps_per_second": 7.854, |
| "step": 8200 |
| }, |
| { |
| "epoch": 0.9418377882298956, |
| "eval_loss": 0.4278132915496826, |
| "eval_runtime": 12.9877, |
| "eval_samples_per_second": 506.941, |
| "eval_steps_per_second": 7.931, |
| "step": 8210 |
| }, |
| { |
| "epoch": 0.9429849718940002, |
| "eval_loss": 0.42760562896728516, |
| "eval_runtime": 12.954, |
| "eval_samples_per_second": 508.261, |
| "eval_steps_per_second": 7.951, |
| "step": 8220 |
| }, |
| { |
| "epoch": 0.9441321555581048, |
| "eval_loss": 0.427383154630661, |
| "eval_runtime": 12.9654, |
| "eval_samples_per_second": 507.814, |
| "eval_steps_per_second": 7.944, |
| "step": 8230 |
| }, |
| { |
| "epoch": 0.9452793392222095, |
| "eval_loss": 0.42702731490135193, |
| "eval_runtime": 12.9517, |
| "eval_samples_per_second": 508.351, |
| "eval_steps_per_second": 7.953, |
| "step": 8240 |
| }, |
| { |
| "epoch": 0.9464265228863141, |
| "eval_loss": 0.42669162154197693, |
| "eval_runtime": 13.1023, |
| "eval_samples_per_second": 502.508, |
| "eval_steps_per_second": 7.861, |
| "step": 8250 |
| }, |
| { |
| "epoch": 0.9475737065504187, |
| "eval_loss": 0.4262804388999939, |
| "eval_runtime": 12.9472, |
| "eval_samples_per_second": 508.525, |
| "eval_steps_per_second": 7.955, |
| "step": 8260 |
| }, |
| { |
| "epoch": 0.9487208902145233, |
| "eval_loss": 0.4261133372783661, |
| "eval_runtime": 12.9861, |
| "eval_samples_per_second": 507.004, |
| "eval_steps_per_second": 7.932, |
| "step": 8270 |
| }, |
| { |
| "epoch": 0.9498680738786279, |
| "eval_loss": 0.42570507526397705, |
| "eval_runtime": 12.9759, |
| "eval_samples_per_second": 507.403, |
| "eval_steps_per_second": 7.938, |
| "step": 8280 |
| }, |
| { |
| "epoch": 0.9510152575427326, |
| "eval_loss": 0.42537105083465576, |
| "eval_runtime": 13.1175, |
| "eval_samples_per_second": 501.924, |
| "eval_steps_per_second": 7.852, |
| "step": 8290 |
| }, |
| { |
| "epoch": 0.9521624412068372, |
| "grad_norm": 3.7640209197998047, |
| "learning_rate": 7.5848735925217765e-06, |
| "loss": 0.7818, |
| "step": 8300 |
| }, |
| { |
| "epoch": 0.9521624412068372, |
| "eval_loss": 0.42547184228897095, |
| "eval_runtime": 13.1141, |
| "eval_samples_per_second": 502.057, |
| "eval_steps_per_second": 7.854, |
| "step": 8300 |
| }, |
| { |
| "epoch": 0.9533096248709418, |
| "eval_loss": 0.4254818856716156, |
| "eval_runtime": 12.9518, |
| "eval_samples_per_second": 508.346, |
| "eval_steps_per_second": 7.953, |
| "step": 8310 |
| }, |
| { |
| "epoch": 0.9544568085350464, |
| "eval_loss": 0.4253588914871216, |
| "eval_runtime": 12.9518, |
| "eval_samples_per_second": 508.348, |
| "eval_steps_per_second": 7.953, |
| "step": 8320 |
| }, |
| { |
| "epoch": 0.955603992199151, |
| "eval_loss": 0.42518895864486694, |
| "eval_runtime": 12.9463, |
| "eval_samples_per_second": 508.563, |
| "eval_steps_per_second": 7.956, |
| "step": 8330 |
| }, |
| { |
| "epoch": 0.9567511758632558, |
| "eval_loss": 0.42494314908981323, |
| "eval_runtime": 13.144, |
| "eval_samples_per_second": 500.912, |
| "eval_steps_per_second": 7.836, |
| "step": 8340 |
| }, |
| { |
| "epoch": 0.9578983595273604, |
| "eval_loss": 0.42492446303367615, |
| "eval_runtime": 13.15, |
| "eval_samples_per_second": 500.686, |
| "eval_steps_per_second": 7.833, |
| "step": 8350 |
| }, |
| { |
| "epoch": 0.959045543191465, |
| "eval_loss": 0.42482849955558777, |
| "eval_runtime": 13.1567, |
| "eval_samples_per_second": 500.428, |
| "eval_steps_per_second": 7.829, |
| "step": 8360 |
| }, |
| { |
| "epoch": 0.9601927268555696, |
| "eval_loss": 0.4248596429824829, |
| "eval_runtime": 13.0146, |
| "eval_samples_per_second": 505.893, |
| "eval_steps_per_second": 7.914, |
| "step": 8370 |
| }, |
| { |
| "epoch": 0.9613399105196742, |
| "eval_loss": 0.4247839152812958, |
| "eval_runtime": 13.1006, |
| "eval_samples_per_second": 502.571, |
| "eval_steps_per_second": 7.862, |
| "step": 8380 |
| }, |
| { |
| "epoch": 0.9624870941837789, |
| "eval_loss": 0.42458516359329224, |
| "eval_runtime": 12.9446, |
| "eval_samples_per_second": 508.629, |
| "eval_steps_per_second": 7.957, |
| "step": 8390 |
| }, |
| { |
| "epoch": 0.9636342778478835, |
| "grad_norm": 3.5270276069641113, |
| "learning_rate": 7.542383683875081e-06, |
| "loss": 0.7606, |
| "step": 8400 |
| }, |
| { |
| "epoch": 0.9636342778478835, |
| "eval_loss": 0.4243062138557434, |
| "eval_runtime": 12.9351, |
| "eval_samples_per_second": 509.001, |
| "eval_steps_per_second": 7.963, |
| "step": 8400 |
| }, |
| { |
| "epoch": 0.9647814615119881, |
| "eval_loss": 0.42416658997535706, |
| "eval_runtime": 12.9581, |
| "eval_samples_per_second": 508.1, |
| "eval_steps_per_second": 7.949, |
| "step": 8410 |
| }, |
| { |
| "epoch": 0.9659286451760927, |
| "eval_loss": 0.42395856976509094, |
| "eval_runtime": 12.9512, |
| "eval_samples_per_second": 508.371, |
| "eval_steps_per_second": 7.953, |
| "step": 8420 |
| }, |
| { |
| "epoch": 0.9670758288401973, |
| "eval_loss": 0.42387980222702026, |
| "eval_runtime": 12.9381, |
| "eval_samples_per_second": 508.883, |
| "eval_steps_per_second": 7.961, |
| "step": 8430 |
| }, |
| { |
| "epoch": 0.968223012504302, |
| "eval_loss": 0.4238240420818329, |
| "eval_runtime": 12.9615, |
| "eval_samples_per_second": 507.967, |
| "eval_steps_per_second": 7.947, |
| "step": 8440 |
| }, |
| { |
| "epoch": 0.9693701961684066, |
| "eval_loss": 0.42380833625793457, |
| "eval_runtime": 12.9435, |
| "eval_samples_per_second": 508.671, |
| "eval_steps_per_second": 7.958, |
| "step": 8450 |
| }, |
| { |
| "epoch": 0.9705173798325112, |
| "eval_loss": 0.42374566197395325, |
| "eval_runtime": 13.0818, |
| "eval_samples_per_second": 503.295, |
| "eval_steps_per_second": 7.874, |
| "step": 8460 |
| }, |
| { |
| "epoch": 0.9716645634966158, |
| "eval_loss": 0.4235632121562958, |
| "eval_runtime": 13.0108, |
| "eval_samples_per_second": 506.043, |
| "eval_steps_per_second": 7.917, |
| "step": 8470 |
| }, |
| { |
| "epoch": 0.9728117471607204, |
| "eval_loss": 0.42322155833244324, |
| "eval_runtime": 12.9524, |
| "eval_samples_per_second": 508.325, |
| "eval_steps_per_second": 7.952, |
| "step": 8480 |
| }, |
| { |
| "epoch": 0.9739589308248251, |
| "eval_loss": 0.422890305519104, |
| "eval_runtime": 12.9575, |
| "eval_samples_per_second": 508.123, |
| "eval_steps_per_second": 7.949, |
| "step": 8490 |
| }, |
| { |
| "epoch": 0.9751061144889297, |
| "grad_norm": 3.317497730255127, |
| "learning_rate": 7.499893775228384e-06, |
| "loss": 0.7416, |
| "step": 8500 |
| }, |
| { |
| "epoch": 0.9751061144889297, |
| "eval_loss": 0.42266547679901123, |
| "eval_runtime": 12.9539, |
| "eval_samples_per_second": 508.265, |
| "eval_steps_per_second": 7.951, |
| "step": 8500 |
| }, |
| { |
| "epoch": 0.9762532981530343, |
| "eval_loss": 0.4226071238517761, |
| "eval_runtime": 12.9562, |
| "eval_samples_per_second": 508.175, |
| "eval_steps_per_second": 7.95, |
| "step": 8510 |
| }, |
| { |
| "epoch": 0.9774004818171389, |
| "eval_loss": 0.4220356047153473, |
| "eval_runtime": 12.967, |
| "eval_samples_per_second": 507.751, |
| "eval_steps_per_second": 7.943, |
| "step": 8520 |
| }, |
| { |
| "epoch": 0.9785476654812435, |
| "eval_loss": 0.4217708706855774, |
| "eval_runtime": 12.9761, |
| "eval_samples_per_second": 507.396, |
| "eval_steps_per_second": 7.938, |
| "step": 8530 |
| }, |
| { |
| "epoch": 0.9796948491453482, |
| "eval_loss": 0.4216693639755249, |
| "eval_runtime": 12.9585, |
| "eval_samples_per_second": 508.085, |
| "eval_steps_per_second": 7.948, |
| "step": 8540 |
| }, |
| { |
| "epoch": 0.9808420328094528, |
| "eval_loss": 0.4217049777507782, |
| "eval_runtime": 13.0231, |
| "eval_samples_per_second": 505.562, |
| "eval_steps_per_second": 7.909, |
| "step": 8550 |
| }, |
| { |
| "epoch": 0.9819892164735574, |
| "eval_loss": 0.42149025201797485, |
| "eval_runtime": 12.9442, |
| "eval_samples_per_second": 508.645, |
| "eval_steps_per_second": 7.957, |
| "step": 8560 |
| }, |
| { |
| "epoch": 0.983136400137662, |
| "eval_loss": 0.4216158092021942, |
| "eval_runtime": 12.9619, |
| "eval_samples_per_second": 507.949, |
| "eval_steps_per_second": 7.946, |
| "step": 8570 |
| }, |
| { |
| "epoch": 0.9842835838017666, |
| "eval_loss": 0.4216848611831665, |
| "eval_runtime": 12.9488, |
| "eval_samples_per_second": 508.465, |
| "eval_steps_per_second": 7.954, |
| "step": 8580 |
| }, |
| { |
| "epoch": 0.9854307674658713, |
| "eval_loss": 0.4216417968273163, |
| "eval_runtime": 13.0159, |
| "eval_samples_per_second": 505.843, |
| "eval_steps_per_second": 7.913, |
| "step": 8590 |
| }, |
| { |
| "epoch": 0.9865779511299759, |
| "grad_norm": 2.9382412433624268, |
| "learning_rate": 7.457403866581687e-06, |
| "loss": 0.748, |
| "step": 8600 |
| }, |
| { |
| "epoch": 0.9865779511299759, |
| "eval_loss": 0.42172494530677795, |
| "eval_runtime": 13.1039, |
| "eval_samples_per_second": 502.447, |
| "eval_steps_per_second": 7.86, |
| "step": 8600 |
| }, |
| { |
| "epoch": 0.9877251347940805, |
| "eval_loss": 0.4215153455734253, |
| "eval_runtime": 13.0819, |
| "eval_samples_per_second": 503.289, |
| "eval_steps_per_second": 7.873, |
| "step": 8610 |
| }, |
| { |
| "epoch": 0.9888723184581851, |
| "eval_loss": 0.42156848311424255, |
| "eval_runtime": 13.0123, |
| "eval_samples_per_second": 505.983, |
| "eval_steps_per_second": 7.916, |
| "step": 8620 |
| }, |
| { |
| "epoch": 0.9900195021222897, |
| "eval_loss": 0.4217861294746399, |
| "eval_runtime": 13.0506, |
| "eval_samples_per_second": 504.497, |
| "eval_steps_per_second": 7.892, |
| "step": 8630 |
| }, |
| { |
| "epoch": 0.9911666857863944, |
| "eval_loss": 0.42182856798171997, |
| "eval_runtime": 12.988, |
| "eval_samples_per_second": 506.931, |
| "eval_steps_per_second": 7.93, |
| "step": 8640 |
| }, |
| { |
| "epoch": 0.992313869450499, |
| "eval_loss": 0.4219215512275696, |
| "eval_runtime": 13.1011, |
| "eval_samples_per_second": 502.552, |
| "eval_steps_per_second": 7.862, |
| "step": 8650 |
| }, |
| { |
| "epoch": 0.9934610531146036, |
| "eval_loss": 0.42174986004829407, |
| "eval_runtime": 12.9979, |
| "eval_samples_per_second": 506.542, |
| "eval_steps_per_second": 7.924, |
| "step": 8660 |
| }, |
| { |
| "epoch": 0.9946082367787082, |
| "eval_loss": 0.4216502904891968, |
| "eval_runtime": 12.9467, |
| "eval_samples_per_second": 508.547, |
| "eval_steps_per_second": 7.956, |
| "step": 8670 |
| }, |
| { |
| "epoch": 0.9957554204428128, |
| "eval_loss": 0.4214436411857605, |
| "eval_runtime": 13.0134, |
| "eval_samples_per_second": 505.941, |
| "eval_steps_per_second": 7.915, |
| "step": 8680 |
| }, |
| { |
| "epoch": 0.9969026041069176, |
| "eval_loss": 0.4210452735424042, |
| "eval_runtime": 13.0878, |
| "eval_samples_per_second": 503.063, |
| "eval_steps_per_second": 7.87, |
| "step": 8690 |
| }, |
| { |
| "epoch": 0.9980497877710222, |
| "grad_norm": 2.704181432723999, |
| "learning_rate": 7.414913957934992e-06, |
| "loss": 0.7553, |
| "step": 8700 |
| }, |
| { |
| "epoch": 0.9980497877710222, |
| "eval_loss": 0.4204506576061249, |
| "eval_runtime": 13.0753, |
| "eval_samples_per_second": 503.544, |
| "eval_steps_per_second": 7.877, |
| "step": 8700 |
| }, |
| { |
| "epoch": 0.9991969714351268, |
| "eval_loss": 0.4200124144554138, |
| "eval_runtime": 13.0728, |
| "eval_samples_per_second": 503.641, |
| "eval_steps_per_second": 7.879, |
| "step": 8710 |
| }, |
| { |
| "epoch": 1.0003441550992314, |
| "eval_loss": 0.41992324590682983, |
| "eval_runtime": 12.9932, |
| "eval_samples_per_second": 506.726, |
| "eval_steps_per_second": 7.927, |
| "step": 8720 |
| }, |
| { |
| "epoch": 1.001491338763336, |
| "eval_loss": 0.41985148191452026, |
| "eval_runtime": 12.9442, |
| "eval_samples_per_second": 508.644, |
| "eval_steps_per_second": 7.957, |
| "step": 8730 |
| }, |
| { |
| "epoch": 1.0026385224274406, |
| "eval_loss": 0.41989246010780334, |
| "eval_runtime": 13.0701, |
| "eval_samples_per_second": 503.745, |
| "eval_steps_per_second": 7.881, |
| "step": 8740 |
| }, |
| { |
| "epoch": 1.0037857060915452, |
| "eval_loss": 0.41982072591781616, |
| "eval_runtime": 13.0432, |
| "eval_samples_per_second": 504.784, |
| "eval_steps_per_second": 7.897, |
| "step": 8750 |
| }, |
| { |
| "epoch": 1.00493288975565, |
| "eval_loss": 0.4199511706829071, |
| "eval_runtime": 13.028, |
| "eval_samples_per_second": 505.372, |
| "eval_steps_per_second": 7.906, |
| "step": 8760 |
| }, |
| { |
| "epoch": 1.0060800734197546, |
| "eval_loss": 0.41983067989349365, |
| "eval_runtime": 12.9453, |
| "eval_samples_per_second": 508.603, |
| "eval_steps_per_second": 7.957, |
| "step": 8770 |
| }, |
| { |
| "epoch": 1.0072272570838592, |
| "eval_loss": 0.4195476472377777, |
| "eval_runtime": 12.9443, |
| "eval_samples_per_second": 508.64, |
| "eval_steps_per_second": 7.957, |
| "step": 8780 |
| }, |
| { |
| "epoch": 1.0083744407479638, |
| "eval_loss": 0.41939350962638855, |
| "eval_runtime": 13.0833, |
| "eval_samples_per_second": 503.237, |
| "eval_steps_per_second": 7.873, |
| "step": 8790 |
| }, |
| { |
| "epoch": 1.0095216244120684, |
| "grad_norm": 2.3688294887542725, |
| "learning_rate": 7.372424049288295e-06, |
| "loss": 0.7202, |
| "step": 8800 |
| }, |
| { |
| "epoch": 1.0095216244120684, |
| "eval_loss": 0.4190540313720703, |
| "eval_runtime": 13.0885, |
| "eval_samples_per_second": 503.035, |
| "eval_steps_per_second": 7.869, |
| "step": 8800 |
| }, |
| { |
| "epoch": 1.010668808076173, |
| "eval_loss": 0.4189584255218506, |
| "eval_runtime": 13.0581, |
| "eval_samples_per_second": 504.207, |
| "eval_steps_per_second": 7.888, |
| "step": 8810 |
| }, |
| { |
| "epoch": 1.0118159917402776, |
| "eval_loss": 0.41879957914352417, |
| "eval_runtime": 13.0579, |
| "eval_samples_per_second": 504.215, |
| "eval_steps_per_second": 7.888, |
| "step": 8820 |
| }, |
| { |
| "epoch": 1.0129631754043822, |
| "eval_loss": 0.4187750518321991, |
| "eval_runtime": 13.0625, |
| "eval_samples_per_second": 504.036, |
| "eval_steps_per_second": 7.885, |
| "step": 8830 |
| }, |
| { |
| "epoch": 1.0141103590684868, |
| "eval_loss": 0.4191732108592987, |
| "eval_runtime": 13.0795, |
| "eval_samples_per_second": 503.383, |
| "eval_steps_per_second": 7.875, |
| "step": 8840 |
| }, |
| { |
| "epoch": 1.0152575427325914, |
| "eval_loss": 0.41903460025787354, |
| "eval_runtime": 13.0189, |
| "eval_samples_per_second": 505.727, |
| "eval_steps_per_second": 7.912, |
| "step": 8850 |
| }, |
| { |
| "epoch": 1.0164047263966962, |
| "eval_loss": 0.4190785884857178, |
| "eval_runtime": 13.0061, |
| "eval_samples_per_second": 506.225, |
| "eval_steps_per_second": 7.919, |
| "step": 8860 |
| }, |
| { |
| "epoch": 1.0175519100608008, |
| "eval_loss": 0.4190203547477722, |
| "eval_runtime": 13.0592, |
| "eval_samples_per_second": 504.165, |
| "eval_steps_per_second": 7.887, |
| "step": 8870 |
| }, |
| { |
| "epoch": 1.0186990937249054, |
| "eval_loss": 0.4191504716873169, |
| "eval_runtime": 13.0751, |
| "eval_samples_per_second": 503.551, |
| "eval_steps_per_second": 7.878, |
| "step": 8880 |
| }, |
| { |
| "epoch": 1.01984627738901, |
| "eval_loss": 0.41900479793548584, |
| "eval_runtime": 13.0713, |
| "eval_samples_per_second": 503.699, |
| "eval_steps_per_second": 7.88, |
| "step": 8890 |
| }, |
| { |
| "epoch": 1.0209934610531146, |
| "grad_norm": 2.6150166988372803, |
| "learning_rate": 7.329934140641598e-06, |
| "loss": 0.7567, |
| "step": 8900 |
| }, |
| { |
| "epoch": 1.0209934610531146, |
| "eval_loss": 0.41888001561164856, |
| "eval_runtime": 13.0327, |
| "eval_samples_per_second": 505.193, |
| "eval_steps_per_second": 7.903, |
| "step": 8900 |
| }, |
| { |
| "epoch": 1.0221406447172192, |
| "eval_loss": 0.41882747411727905, |
| "eval_runtime": 13.014, |
| "eval_samples_per_second": 505.917, |
| "eval_steps_per_second": 7.915, |
| "step": 8910 |
| }, |
| { |
| "epoch": 1.0232878283813238, |
| "eval_loss": 0.4189399778842926, |
| "eval_runtime": 13.0901, |
| "eval_samples_per_second": 502.976, |
| "eval_steps_per_second": 7.869, |
| "step": 8920 |
| }, |
| { |
| "epoch": 1.0244350120454284, |
| "eval_loss": 0.41882216930389404, |
| "eval_runtime": 13.0107, |
| "eval_samples_per_second": 506.046, |
| "eval_steps_per_second": 7.917, |
| "step": 8930 |
| }, |
| { |
| "epoch": 1.025582195709533, |
| "eval_loss": 0.4186832904815674, |
| "eval_runtime": 13.003, |
| "eval_samples_per_second": 506.345, |
| "eval_steps_per_second": 7.921, |
| "step": 8940 |
| }, |
| { |
| "epoch": 1.0267293793736376, |
| "eval_loss": 0.4183092415332794, |
| "eval_runtime": 13.0416, |
| "eval_samples_per_second": 504.847, |
| "eval_steps_per_second": 7.898, |
| "step": 8950 |
| }, |
| { |
| "epoch": 1.0278765630377424, |
| "eval_loss": 0.41818562150001526, |
| "eval_runtime": 13.0962, |
| "eval_samples_per_second": 502.74, |
| "eval_steps_per_second": 7.865, |
| "step": 8960 |
| }, |
| { |
| "epoch": 1.029023746701847, |
| "eval_loss": 0.41821199655532837, |
| "eval_runtime": 12.9531, |
| "eval_samples_per_second": 508.296, |
| "eval_steps_per_second": 7.952, |
| "step": 8970 |
| }, |
| { |
| "epoch": 1.0301709303659516, |
| "eval_loss": 0.41837337613105774, |
| "eval_runtime": 12.9361, |
| "eval_samples_per_second": 508.965, |
| "eval_steps_per_second": 7.962, |
| "step": 8980 |
| }, |
| { |
| "epoch": 1.0313181140300562, |
| "eval_loss": 0.4180542528629303, |
| "eval_runtime": 12.9379, |
| "eval_samples_per_second": 508.894, |
| "eval_steps_per_second": 7.961, |
| "step": 8990 |
| }, |
| { |
| "epoch": 1.0324652976941608, |
| "grad_norm": 2.393902540206909, |
| "learning_rate": 7.287444231994901e-06, |
| "loss": 0.7345, |
| "step": 9000 |
| }, |
| { |
| "epoch": 1.0324652976941608, |
| "eval_loss": 0.4176781475543976, |
| "eval_runtime": 13.0632, |
| "eval_samples_per_second": 504.011, |
| "eval_steps_per_second": 7.885, |
| "step": 9000 |
| }, |
| { |
| "epoch": 1.0336124813582654, |
| "eval_loss": 0.41729438304901123, |
| "eval_runtime": 13.0481, |
| "eval_samples_per_second": 504.593, |
| "eval_steps_per_second": 7.894, |
| "step": 9010 |
| }, |
| { |
| "epoch": 1.03475966502237, |
| "eval_loss": 0.4170686900615692, |
| "eval_runtime": 12.9512, |
| "eval_samples_per_second": 508.368, |
| "eval_steps_per_second": 7.953, |
| "step": 9020 |
| }, |
| { |
| "epoch": 1.0359068486864746, |
| "eval_loss": 0.4171694815158844, |
| "eval_runtime": 12.9493, |
| "eval_samples_per_second": 508.444, |
| "eval_steps_per_second": 7.954, |
| "step": 9030 |
| }, |
| { |
| "epoch": 1.0370540323505792, |
| "eval_loss": 0.41708752512931824, |
| "eval_runtime": 13.1159, |
| "eval_samples_per_second": 501.988, |
| "eval_steps_per_second": 7.853, |
| "step": 9040 |
| }, |
| { |
| "epoch": 1.0382012160146838, |
| "eval_loss": 0.41715097427368164, |
| "eval_runtime": 13.1392, |
| "eval_samples_per_second": 501.095, |
| "eval_steps_per_second": 7.839, |
| "step": 9050 |
| }, |
| { |
| "epoch": 1.0393483996787887, |
| "eval_loss": 0.41723066568374634, |
| "eval_runtime": 13.1566, |
| "eval_samples_per_second": 500.434, |
| "eval_steps_per_second": 7.829, |
| "step": 9060 |
| }, |
| { |
| "epoch": 1.0404955833428933, |
| "eval_loss": 0.41699478030204773, |
| "eval_runtime": 13.1258, |
| "eval_samples_per_second": 501.606, |
| "eval_steps_per_second": 7.847, |
| "step": 9070 |
| }, |
| { |
| "epoch": 1.0416427670069979, |
| "eval_loss": 0.4164867401123047, |
| "eval_runtime": 12.9417, |
| "eval_samples_per_second": 508.742, |
| "eval_steps_per_second": 7.959, |
| "step": 9080 |
| }, |
| { |
| "epoch": 1.0427899506711025, |
| "eval_loss": 0.41616734862327576, |
| "eval_runtime": 13.0739, |
| "eval_samples_per_second": 503.598, |
| "eval_steps_per_second": 7.878, |
| "step": 9090 |
| }, |
| { |
| "epoch": 1.043937134335207, |
| "grad_norm": 3.1782264709472656, |
| "learning_rate": 7.244954323348206e-06, |
| "loss": 0.7344, |
| "step": 9100 |
| }, |
| { |
| "epoch": 1.043937134335207, |
| "eval_loss": 0.4161781370639801, |
| "eval_runtime": 12.9607, |
| "eval_samples_per_second": 507.995, |
| "eval_steps_per_second": 7.947, |
| "step": 9100 |
| }, |
| { |
| "epoch": 1.0450843179993117, |
| "eval_loss": 0.4159933030605316, |
| "eval_runtime": 12.9522, |
| "eval_samples_per_second": 508.331, |
| "eval_steps_per_second": 7.952, |
| "step": 9110 |
| }, |
| { |
| "epoch": 1.0462315016634163, |
| "eval_loss": 0.4158400893211365, |
| "eval_runtime": 12.9833, |
| "eval_samples_per_second": 507.114, |
| "eval_steps_per_second": 7.933, |
| "step": 9120 |
| }, |
| { |
| "epoch": 1.0473786853275209, |
| "eval_loss": 0.41568663716316223, |
| "eval_runtime": 12.9792, |
| "eval_samples_per_second": 507.272, |
| "eval_steps_per_second": 7.936, |
| "step": 9130 |
| }, |
| { |
| "epoch": 1.0485258689916255, |
| "eval_loss": 0.4156523048877716, |
| "eval_runtime": 12.9674, |
| "eval_samples_per_second": 507.733, |
| "eval_steps_per_second": 7.943, |
| "step": 9140 |
| }, |
| { |
| "epoch": 1.04967305265573, |
| "eval_loss": 0.41555264592170715, |
| "eval_runtime": 12.9669, |
| "eval_samples_per_second": 507.753, |
| "eval_steps_per_second": 7.943, |
| "step": 9150 |
| }, |
| { |
| "epoch": 1.050820236319835, |
| "eval_loss": 0.41529908776283264, |
| "eval_runtime": 12.9744, |
| "eval_samples_per_second": 507.461, |
| "eval_steps_per_second": 7.939, |
| "step": 9160 |
| }, |
| { |
| "epoch": 1.0519674199839395, |
| "eval_loss": 0.41533347964286804, |
| "eval_runtime": 12.9449, |
| "eval_samples_per_second": 508.618, |
| "eval_steps_per_second": 7.957, |
| "step": 9170 |
| }, |
| { |
| "epoch": 1.053114603648044, |
| "eval_loss": 0.41539207100868225, |
| "eval_runtime": 13.0198, |
| "eval_samples_per_second": 505.69, |
| "eval_steps_per_second": 7.911, |
| "step": 9180 |
| }, |
| { |
| "epoch": 1.0542617873121487, |
| "eval_loss": 0.415426641702652, |
| "eval_runtime": 12.9514, |
| "eval_samples_per_second": 508.361, |
| "eval_steps_per_second": 7.953, |
| "step": 9190 |
| }, |
| { |
| "epoch": 1.0554089709762533, |
| "grad_norm": 1.90788733959198, |
| "learning_rate": 7.202464414701508e-06, |
| "loss": 0.7233, |
| "step": 9200 |
| }, |
| { |
| "epoch": 1.0554089709762533, |
| "eval_loss": 0.41567009687423706, |
| "eval_runtime": 13.0408, |
| "eval_samples_per_second": 504.878, |
| "eval_steps_per_second": 7.898, |
| "step": 9200 |
| }, |
| { |
| "epoch": 1.056556154640358, |
| "eval_loss": 0.4156700670719147, |
| "eval_runtime": 12.9664, |
| "eval_samples_per_second": 507.773, |
| "eval_steps_per_second": 7.944, |
| "step": 9210 |
| }, |
| { |
| "epoch": 1.0577033383044625, |
| "eval_loss": 0.41562220454216003, |
| "eval_runtime": 12.9666, |
| "eval_samples_per_second": 507.767, |
| "eval_steps_per_second": 7.943, |
| "step": 9220 |
| }, |
| { |
| "epoch": 1.058850521968567, |
| "eval_loss": 0.41545170545578003, |
| "eval_runtime": 13.0742, |
| "eval_samples_per_second": 503.587, |
| "eval_steps_per_second": 7.878, |
| "step": 9230 |
| }, |
| { |
| "epoch": 1.0599977056326717, |
| "eval_loss": 0.4153362214565277, |
| "eval_runtime": 12.9665, |
| "eval_samples_per_second": 507.77, |
| "eval_steps_per_second": 7.944, |
| "step": 9240 |
| }, |
| { |
| "epoch": 1.0611448892967763, |
| "eval_loss": 0.4154272675514221, |
| "eval_runtime": 12.9639, |
| "eval_samples_per_second": 507.874, |
| "eval_steps_per_second": 7.945, |
| "step": 9250 |
| }, |
| { |
| "epoch": 1.0622920729608811, |
| "eval_loss": 0.4155464768409729, |
| "eval_runtime": 12.968, |
| "eval_samples_per_second": 507.713, |
| "eval_steps_per_second": 7.943, |
| "step": 9260 |
| }, |
| { |
| "epoch": 1.0634392566249857, |
| "eval_loss": 0.41540005803108215, |
| "eval_runtime": 13.131, |
| "eval_samples_per_second": 501.408, |
| "eval_steps_per_second": 7.844, |
| "step": 9270 |
| }, |
| { |
| "epoch": 1.0645864402890903, |
| "eval_loss": 0.41506633162498474, |
| "eval_runtime": 12.9514, |
| "eval_samples_per_second": 508.363, |
| "eval_steps_per_second": 7.953, |
| "step": 9280 |
| }, |
| { |
| "epoch": 1.065733623953195, |
| "eval_loss": 0.4148799479007721, |
| "eval_runtime": 12.9705, |
| "eval_samples_per_second": 507.613, |
| "eval_steps_per_second": 7.941, |
| "step": 9290 |
| }, |
| { |
| "epoch": 1.0668808076172995, |
| "grad_norm": 2.8713057041168213, |
| "learning_rate": 7.159974506054812e-06, |
| "loss": 0.7442, |
| "step": 9300 |
| }, |
| { |
| "epoch": 1.0668808076172995, |
| "eval_loss": 0.4148331880569458, |
| "eval_runtime": 12.9674, |
| "eval_samples_per_second": 507.733, |
| "eval_steps_per_second": 7.943, |
| "step": 9300 |
| }, |
| { |
| "epoch": 1.0680279912814041, |
| "eval_loss": 0.4144473969936371, |
| "eval_runtime": 12.9684, |
| "eval_samples_per_second": 507.695, |
| "eval_steps_per_second": 7.942, |
| "step": 9310 |
| }, |
| { |
| "epoch": 1.0691751749455087, |
| "eval_loss": 0.41430482268333435, |
| "eval_runtime": 12.9747, |
| "eval_samples_per_second": 507.45, |
| "eval_steps_per_second": 7.939, |
| "step": 9320 |
| }, |
| { |
| "epoch": 1.0703223586096133, |
| "eval_loss": 0.4141288697719574, |
| "eval_runtime": 12.9959, |
| "eval_samples_per_second": 506.623, |
| "eval_steps_per_second": 7.926, |
| "step": 9330 |
| }, |
| { |
| "epoch": 1.071469542273718, |
| "eval_loss": 0.41395846009254456, |
| "eval_runtime": 12.987, |
| "eval_samples_per_second": 506.97, |
| "eval_steps_per_second": 7.931, |
| "step": 9340 |
| }, |
| { |
| "epoch": 1.0726167259378228, |
| "eval_loss": 0.413784921169281, |
| "eval_runtime": 12.9981, |
| "eval_samples_per_second": 506.537, |
| "eval_steps_per_second": 7.924, |
| "step": 9350 |
| }, |
| { |
| "epoch": 1.0737639096019274, |
| "eval_loss": 0.4135656952857971, |
| "eval_runtime": 12.9767, |
| "eval_samples_per_second": 507.373, |
| "eval_steps_per_second": 7.937, |
| "step": 9360 |
| }, |
| { |
| "epoch": 1.074911093266032, |
| "eval_loss": 0.41334104537963867, |
| "eval_runtime": 12.993, |
| "eval_samples_per_second": 506.734, |
| "eval_steps_per_second": 7.927, |
| "step": 9370 |
| }, |
| { |
| "epoch": 1.0760582769301366, |
| "eval_loss": 0.41317835450172424, |
| "eval_runtime": 12.9698, |
| "eval_samples_per_second": 507.643, |
| "eval_steps_per_second": 7.942, |
| "step": 9380 |
| }, |
| { |
| "epoch": 1.0772054605942412, |
| "eval_loss": 0.41296446323394775, |
| "eval_runtime": 12.9667, |
| "eval_samples_per_second": 507.761, |
| "eval_steps_per_second": 7.943, |
| "step": 9390 |
| }, |
| { |
| "epoch": 1.0783526442583458, |
| "grad_norm": 2.884150743484497, |
| "learning_rate": 7.1174845974081165e-06, |
| "loss": 0.722, |
| "step": 9400 |
| }, |
| { |
| "epoch": 1.0783526442583458, |
| "eval_loss": 0.41291168332099915, |
| "eval_runtime": 12.9936, |
| "eval_samples_per_second": 506.711, |
| "eval_steps_per_second": 7.927, |
| "step": 9400 |
| }, |
| { |
| "epoch": 1.0794998279224504, |
| "eval_loss": 0.4131143391132355, |
| "eval_runtime": 12.9612, |
| "eval_samples_per_second": 507.977, |
| "eval_steps_per_second": 7.947, |
| "step": 9410 |
| }, |
| { |
| "epoch": 1.080647011586555, |
| "eval_loss": 0.4132421314716339, |
| "eval_runtime": 12.9747, |
| "eval_samples_per_second": 507.449, |
| "eval_steps_per_second": 7.939, |
| "step": 9420 |
| }, |
| { |
| "epoch": 1.0817941952506596, |
| "eval_loss": 0.4132674038410187, |
| "eval_runtime": 13.0285, |
| "eval_samples_per_second": 505.352, |
| "eval_steps_per_second": 7.906, |
| "step": 9430 |
| }, |
| { |
| "epoch": 1.0829413789147642, |
| "eval_loss": 0.41336381435394287, |
| "eval_runtime": 12.9704, |
| "eval_samples_per_second": 507.617, |
| "eval_steps_per_second": 7.941, |
| "step": 9440 |
| }, |
| { |
| "epoch": 1.0840885625788688, |
| "eval_loss": 0.4134499728679657, |
| "eval_runtime": 12.9698, |
| "eval_samples_per_second": 507.642, |
| "eval_steps_per_second": 7.942, |
| "step": 9450 |
| }, |
| { |
| "epoch": 1.0852357462429736, |
| "eval_loss": 0.41325509548187256, |
| "eval_runtime": 12.979, |
| "eval_samples_per_second": 507.282, |
| "eval_steps_per_second": 7.936, |
| "step": 9460 |
| }, |
| { |
| "epoch": 1.0863829299070782, |
| "eval_loss": 0.41316714882850647, |
| "eval_runtime": 12.9865, |
| "eval_samples_per_second": 506.989, |
| "eval_steps_per_second": 7.931, |
| "step": 9470 |
| }, |
| { |
| "epoch": 1.0875301135711828, |
| "eval_loss": 0.4132201075553894, |
| "eval_runtime": 13.1011, |
| "eval_samples_per_second": 502.553, |
| "eval_steps_per_second": 7.862, |
| "step": 9480 |
| }, |
| { |
| "epoch": 1.0886772972352874, |
| "eval_loss": 0.41336166858673096, |
| "eval_runtime": 13.0164, |
| "eval_samples_per_second": 505.823, |
| "eval_steps_per_second": 7.913, |
| "step": 9490 |
| }, |
| { |
| "epoch": 1.089824480899392, |
| "grad_norm": 2.4363036155700684, |
| "learning_rate": 7.074994688761419e-06, |
| "loss": 0.7433, |
| "step": 9500 |
| }, |
| { |
| "epoch": 1.089824480899392, |
| "eval_loss": 0.41334328055381775, |
| "eval_runtime": 12.9884, |
| "eval_samples_per_second": 506.915, |
| "eval_steps_per_second": 7.93, |
| "step": 9500 |
| }, |
| { |
| "epoch": 1.0909716645634966, |
| "eval_loss": 0.41326838731765747, |
| "eval_runtime": 12.9711, |
| "eval_samples_per_second": 507.591, |
| "eval_steps_per_second": 7.941, |
| "step": 9510 |
| }, |
| { |
| "epoch": 1.0921188482276012, |
| "eval_loss": 0.4132576882839203, |
| "eval_runtime": 13.1151, |
| "eval_samples_per_second": 502.015, |
| "eval_steps_per_second": 7.854, |
| "step": 9520 |
| }, |
| { |
| "epoch": 1.0932660318917058, |
| "eval_loss": 0.41321423649787903, |
| "eval_runtime": 13.116, |
| "eval_samples_per_second": 501.981, |
| "eval_steps_per_second": 7.853, |
| "step": 9530 |
| }, |
| { |
| "epoch": 1.0944132155558104, |
| "eval_loss": 0.413114458322525, |
| "eval_runtime": 12.9655, |
| "eval_samples_per_second": 507.807, |
| "eval_steps_per_second": 7.944, |
| "step": 9540 |
| }, |
| { |
| "epoch": 1.0955603992199152, |
| "eval_loss": 0.4129582345485687, |
| "eval_runtime": 12.9669, |
| "eval_samples_per_second": 507.754, |
| "eval_steps_per_second": 7.943, |
| "step": 9550 |
| }, |
| { |
| "epoch": 1.0967075828840198, |
| "eval_loss": 0.41295140981674194, |
| "eval_runtime": 13.1202, |
| "eval_samples_per_second": 501.824, |
| "eval_steps_per_second": 7.851, |
| "step": 9560 |
| }, |
| { |
| "epoch": 1.0978547665481244, |
| "eval_loss": 0.41260066628456116, |
| "eval_runtime": 13.1374, |
| "eval_samples_per_second": 501.165, |
| "eval_steps_per_second": 7.84, |
| "step": 9570 |
| }, |
| { |
| "epoch": 1.099001950212229, |
| "eval_loss": 0.412467360496521, |
| "eval_runtime": 13.1531, |
| "eval_samples_per_second": 500.567, |
| "eval_steps_per_second": 7.831, |
| "step": 9580 |
| }, |
| { |
| "epoch": 1.1001491338763336, |
| "eval_loss": 0.4120550751686096, |
| "eval_runtime": 13.1555, |
| "eval_samples_per_second": 500.476, |
| "eval_steps_per_second": 7.829, |
| "step": 9590 |
| }, |
| { |
| "epoch": 1.1012963175404382, |
| "grad_norm": 10.157108306884766, |
| "learning_rate": 7.032504780114723e-06, |
| "loss": 0.746, |
| "step": 9600 |
| }, |
| { |
| "epoch": 1.1012963175404382, |
| "eval_loss": 0.4119147062301636, |
| "eval_runtime": 13.1713, |
| "eval_samples_per_second": 499.873, |
| "eval_steps_per_second": 7.82, |
| "step": 9600 |
| }, |
| { |
| "epoch": 1.1024435012045428, |
| "eval_loss": 0.41167396306991577, |
| "eval_runtime": 13.0777, |
| "eval_samples_per_second": 503.453, |
| "eval_steps_per_second": 7.876, |
| "step": 9610 |
| }, |
| { |
| "epoch": 1.1035906848686474, |
| "eval_loss": 0.41122013330459595, |
| "eval_runtime": 12.9711, |
| "eval_samples_per_second": 507.589, |
| "eval_steps_per_second": 7.941, |
| "step": 9620 |
| }, |
| { |
| "epoch": 1.104737868532752, |
| "eval_loss": 0.4109072983264923, |
| "eval_runtime": 12.9819, |
| "eval_samples_per_second": 507.168, |
| "eval_steps_per_second": 7.934, |
| "step": 9630 |
| }, |
| { |
| "epoch": 1.1058850521968566, |
| "eval_loss": 0.4105525016784668, |
| "eval_runtime": 12.9649, |
| "eval_samples_per_second": 507.834, |
| "eval_steps_per_second": 7.945, |
| "step": 9640 |
| }, |
| { |
| "epoch": 1.1070322358609612, |
| "eval_loss": 0.4100634455680847, |
| "eval_runtime": 12.9793, |
| "eval_samples_per_second": 507.268, |
| "eval_steps_per_second": 7.936, |
| "step": 9650 |
| }, |
| { |
| "epoch": 1.108179419525066, |
| "eval_loss": 0.4101249873638153, |
| "eval_runtime": 12.9715, |
| "eval_samples_per_second": 507.573, |
| "eval_steps_per_second": 7.94, |
| "step": 9660 |
| }, |
| { |
| "epoch": 1.1093266031891706, |
| "eval_loss": 0.4101654589176178, |
| "eval_runtime": 13.0384, |
| "eval_samples_per_second": 504.969, |
| "eval_steps_per_second": 7.9, |
| "step": 9670 |
| }, |
| { |
| "epoch": 1.1104737868532752, |
| "eval_loss": 0.41017717123031616, |
| "eval_runtime": 12.958, |
| "eval_samples_per_second": 508.103, |
| "eval_steps_per_second": 7.949, |
| "step": 9680 |
| }, |
| { |
| "epoch": 1.1116209705173798, |
| "eval_loss": 0.41011014580726624, |
| "eval_runtime": 12.9681, |
| "eval_samples_per_second": 507.707, |
| "eval_steps_per_second": 7.943, |
| "step": 9690 |
| }, |
| { |
| "epoch": 1.1127681541814844, |
| "grad_norm": 4.019915580749512, |
| "learning_rate": 6.9900148714680275e-06, |
| "loss": 0.7447, |
| "step": 9700 |
| }, |
| { |
| "epoch": 1.1127681541814844, |
| "eval_loss": 0.40987932682037354, |
| "eval_runtime": 12.9711, |
| "eval_samples_per_second": 507.59, |
| "eval_steps_per_second": 7.941, |
| "step": 9700 |
| }, |
| { |
| "epoch": 1.113915337845589, |
| "eval_loss": 0.4099966287612915, |
| "eval_runtime": 12.9678, |
| "eval_samples_per_second": 507.717, |
| "eval_steps_per_second": 7.943, |
| "step": 9710 |
| }, |
| { |
| "epoch": 1.1150625215096936, |
| "eval_loss": 0.4098173677921295, |
| "eval_runtime": 13.0864, |
| "eval_samples_per_second": 503.119, |
| "eval_steps_per_second": 7.871, |
| "step": 9720 |
| }, |
| { |
| "epoch": 1.1162097051737983, |
| "eval_loss": 0.4097200334072113, |
| "eval_runtime": 13.0025, |
| "eval_samples_per_second": 506.364, |
| "eval_steps_per_second": 7.922, |
| "step": 9730 |
| }, |
| { |
| "epoch": 1.1173568888379029, |
| "eval_loss": 0.4094499945640564, |
| "eval_runtime": 12.9682, |
| "eval_samples_per_second": 507.702, |
| "eval_steps_per_second": 7.942, |
| "step": 9740 |
| }, |
| { |
| "epoch": 1.1185040725020077, |
| "eval_loss": 0.4097111225128174, |
| "eval_runtime": 13.0802, |
| "eval_samples_per_second": 503.355, |
| "eval_steps_per_second": 7.874, |
| "step": 9750 |
| }, |
| { |
| "epoch": 1.1196512561661123, |
| "eval_loss": 0.40956640243530273, |
| "eval_runtime": 12.9973, |
| "eval_samples_per_second": 506.565, |
| "eval_steps_per_second": 7.925, |
| "step": 9760 |
| }, |
| { |
| "epoch": 1.1207984398302169, |
| "eval_loss": 0.4096319377422333, |
| "eval_runtime": 12.9889, |
| "eval_samples_per_second": 506.893, |
| "eval_steps_per_second": 7.93, |
| "step": 9770 |
| }, |
| { |
| "epoch": 1.1219456234943215, |
| "eval_loss": 0.40973737835884094, |
| "eval_runtime": 13.0276, |
| "eval_samples_per_second": 505.387, |
| "eval_steps_per_second": 7.906, |
| "step": 9780 |
| }, |
| { |
| "epoch": 1.123092807158426, |
| "eval_loss": 0.4097270667552948, |
| "eval_runtime": 13.0363, |
| "eval_samples_per_second": 505.052, |
| "eval_steps_per_second": 7.901, |
| "step": 9790 |
| }, |
| { |
| "epoch": 1.1242399908225307, |
| "grad_norm": 2.9748287200927734, |
| "learning_rate": 6.94752496282133e-06, |
| "loss": 0.7234, |
| "step": 9800 |
| }, |
| { |
| "epoch": 1.1242399908225307, |
| "eval_loss": 0.4094075858592987, |
| "eval_runtime": 12.966, |
| "eval_samples_per_second": 507.791, |
| "eval_steps_per_second": 7.944, |
| "step": 9800 |
| }, |
| { |
| "epoch": 1.1253871744866353, |
| "eval_loss": 0.4090143144130707, |
| "eval_runtime": 12.9724, |
| "eval_samples_per_second": 507.539, |
| "eval_steps_per_second": 7.94, |
| "step": 9810 |
| }, |
| { |
| "epoch": 1.1265343581507399, |
| "eval_loss": 0.4090229868888855, |
| "eval_runtime": 12.968, |
| "eval_samples_per_second": 507.71, |
| "eval_steps_per_second": 7.943, |
| "step": 9820 |
| }, |
| { |
| "epoch": 1.1276815418148445, |
| "eval_loss": 0.4091443121433258, |
| "eval_runtime": 12.9606, |
| "eval_samples_per_second": 508.0, |
| "eval_steps_per_second": 7.947, |
| "step": 9830 |
| }, |
| { |
| "epoch": 1.1288287254789493, |
| "eval_loss": 0.40912991762161255, |
| "eval_runtime": 12.9798, |
| "eval_samples_per_second": 507.249, |
| "eval_steps_per_second": 7.935, |
| "step": 9840 |
| }, |
| { |
| "epoch": 1.1299759091430537, |
| "eval_loss": 0.40901440382003784, |
| "eval_runtime": 13.0934, |
| "eval_samples_per_second": 502.849, |
| "eval_steps_per_second": 7.867, |
| "step": 9850 |
| }, |
| { |
| "epoch": 1.1311230928071585, |
| "eval_loss": 0.4088425040245056, |
| "eval_runtime": 13.0824, |
| "eval_samples_per_second": 503.271, |
| "eval_steps_per_second": 7.873, |
| "step": 9860 |
| }, |
| { |
| "epoch": 1.132270276471263, |
| "eval_loss": 0.4087802469730377, |
| "eval_runtime": 12.9962, |
| "eval_samples_per_second": 506.611, |
| "eval_steps_per_second": 7.925, |
| "step": 9870 |
| }, |
| { |
| "epoch": 1.1334174601353677, |
| "eval_loss": 0.4084618091583252, |
| "eval_runtime": 12.9906, |
| "eval_samples_per_second": 506.829, |
| "eval_steps_per_second": 7.929, |
| "step": 9880 |
| }, |
| { |
| "epoch": 1.1345646437994723, |
| "eval_loss": 0.408491313457489, |
| "eval_runtime": 12.9856, |
| "eval_samples_per_second": 507.024, |
| "eval_steps_per_second": 7.932, |
| "step": 9890 |
| }, |
| { |
| "epoch": 1.135711827463577, |
| "grad_norm": 3.249502658843994, |
| "learning_rate": 6.905035054174634e-06, |
| "loss": 0.7054, |
| "step": 9900 |
| }, |
| { |
| "epoch": 1.135711827463577, |
| "eval_loss": 0.4083896279335022, |
| "eval_runtime": 12.9718, |
| "eval_samples_per_second": 507.564, |
| "eval_steps_per_second": 7.94, |
| "step": 9900 |
| }, |
| { |
| "epoch": 1.1368590111276815, |
| "eval_loss": 0.40869393944740295, |
| "eval_runtime": 12.9713, |
| "eval_samples_per_second": 507.584, |
| "eval_steps_per_second": 7.941, |
| "step": 9910 |
| }, |
| { |
| "epoch": 1.138006194791786, |
| "eval_loss": 0.40894752740859985, |
| "eval_runtime": 12.9686, |
| "eval_samples_per_second": 507.688, |
| "eval_steps_per_second": 7.942, |
| "step": 9920 |
| }, |
| { |
| "epoch": 1.1391533784558907, |
| "eval_loss": 0.4088941514492035, |
| "eval_runtime": 12.9681, |
| "eval_samples_per_second": 507.706, |
| "eval_steps_per_second": 7.943, |
| "step": 9930 |
| }, |
| { |
| "epoch": 1.1403005621199953, |
| "eval_loss": 0.4087817668914795, |
| "eval_runtime": 12.9793, |
| "eval_samples_per_second": 507.268, |
| "eval_steps_per_second": 7.936, |
| "step": 9940 |
| }, |
| { |
| "epoch": 1.1414477457841001, |
| "eval_loss": 0.40913549065589905, |
| "eval_runtime": 12.9726, |
| "eval_samples_per_second": 507.53, |
| "eval_steps_per_second": 7.94, |
| "step": 9950 |
| }, |
| { |
| "epoch": 1.1425949294482047, |
| "eval_loss": 0.4088318645954132, |
| "eval_runtime": 12.9782, |
| "eval_samples_per_second": 507.312, |
| "eval_steps_per_second": 7.936, |
| "step": 9960 |
| }, |
| { |
| "epoch": 1.1437421131123093, |
| "eval_loss": 0.40860241651535034, |
| "eval_runtime": 12.9643, |
| "eval_samples_per_second": 507.857, |
| "eval_steps_per_second": 7.945, |
| "step": 9970 |
| }, |
| { |
| "epoch": 1.144889296776414, |
| "eval_loss": 0.40839141607284546, |
| "eval_runtime": 12.9763, |
| "eval_samples_per_second": 507.387, |
| "eval_steps_per_second": 7.938, |
| "step": 9980 |
| }, |
| { |
| "epoch": 1.1460364804405185, |
| "eval_loss": 0.4088607430458069, |
| "eval_runtime": 13.0947, |
| "eval_samples_per_second": 502.799, |
| "eval_steps_per_second": 7.866, |
| "step": 9990 |
| }, |
| { |
| "epoch": 1.1471836641046231, |
| "grad_norm": 3.114164352416992, |
| "learning_rate": 6.8625451455279375e-06, |
| "loss": 0.7071, |
| "step": 10000 |
| }, |
| { |
| "epoch": 1.1471836641046231, |
| "eval_loss": 0.40881696343421936, |
| "eval_runtime": 12.99, |
| "eval_samples_per_second": 506.85, |
| "eval_steps_per_second": 7.929, |
| "step": 10000 |
| }, |
| { |
| "epoch": 1.1483308477687277, |
| "eval_loss": 0.4086223840713501, |
| "eval_runtime": 13.0722, |
| "eval_samples_per_second": 503.664, |
| "eval_steps_per_second": 7.879, |
| "step": 10010 |
| }, |
| { |
| "epoch": 1.1494780314328323, |
| "eval_loss": 0.4080895483493805, |
| "eval_runtime": 13.1475, |
| "eval_samples_per_second": 500.78, |
| "eval_steps_per_second": 7.834, |
| "step": 10020 |
| }, |
| { |
| "epoch": 1.150625215096937, |
| "eval_loss": 0.4078758955001831, |
| "eval_runtime": 13.1448, |
| "eval_samples_per_second": 500.883, |
| "eval_steps_per_second": 7.836, |
| "step": 10030 |
| }, |
| { |
| "epoch": 1.1517723987610418, |
| "eval_loss": 0.407888263463974, |
| "eval_runtime": 12.9859, |
| "eval_samples_per_second": 507.012, |
| "eval_steps_per_second": 7.932, |
| "step": 10040 |
| }, |
| { |
| "epoch": 1.1529195824251464, |
| "eval_loss": 0.4081161320209503, |
| "eval_runtime": 12.9877, |
| "eval_samples_per_second": 506.94, |
| "eval_steps_per_second": 7.931, |
| "step": 10050 |
| }, |
| { |
| "epoch": 1.154066766089251, |
| "eval_loss": 0.40805235505104065, |
| "eval_runtime": 12.9851, |
| "eval_samples_per_second": 507.041, |
| "eval_steps_per_second": 7.932, |
| "step": 10060 |
| }, |
| { |
| "epoch": 1.1552139497533556, |
| "eval_loss": 0.4079611301422119, |
| "eval_runtime": 12.9949, |
| "eval_samples_per_second": 506.659, |
| "eval_steps_per_second": 7.926, |
| "step": 10070 |
| }, |
| { |
| "epoch": 1.1563611334174602, |
| "eval_loss": 0.4078901708126068, |
| "eval_runtime": 12.9898, |
| "eval_samples_per_second": 506.861, |
| "eval_steps_per_second": 7.929, |
| "step": 10080 |
| }, |
| { |
| "epoch": 1.1575083170815648, |
| "eval_loss": 0.40782755613327026, |
| "eval_runtime": 12.9965, |
| "eval_samples_per_second": 506.599, |
| "eval_steps_per_second": 7.925, |
| "step": 10090 |
| }, |
| { |
| "epoch": 1.1586555007456694, |
| "grad_norm": 3.363734722137451, |
| "learning_rate": 6.820055236881241e-06, |
| "loss": 0.7289, |
| "step": 10100 |
| }, |
| { |
| "epoch": 1.1586555007456694, |
| "eval_loss": 0.40747976303100586, |
| "eval_runtime": 12.9969, |
| "eval_samples_per_second": 506.58, |
| "eval_steps_per_second": 7.925, |
| "step": 10100 |
| }, |
| { |
| "epoch": 1.159802684409774, |
| "eval_loss": 0.40719860792160034, |
| "eval_runtime": 12.9988, |
| "eval_samples_per_second": 506.51, |
| "eval_steps_per_second": 7.924, |
| "step": 10110 |
| }, |
| { |
| "epoch": 1.1609498680738786, |
| "eval_loss": 0.40698835253715515, |
| "eval_runtime": 12.9892, |
| "eval_samples_per_second": 506.881, |
| "eval_steps_per_second": 7.93, |
| "step": 10120 |
| }, |
| { |
| "epoch": 1.1620970517379832, |
| "eval_loss": 0.4069991409778595, |
| "eval_runtime": 13.0211, |
| "eval_samples_per_second": 505.641, |
| "eval_steps_per_second": 7.91, |
| "step": 10130 |
| }, |
| { |
| "epoch": 1.1632442354020878, |
| "eval_loss": 0.4073794186115265, |
| "eval_runtime": 12.9743, |
| "eval_samples_per_second": 507.464, |
| "eval_steps_per_second": 7.939, |
| "step": 10140 |
| }, |
| { |
| "epoch": 1.1643914190661926, |
| "eval_loss": 0.40735331177711487, |
| "eval_runtime": 13.0957, |
| "eval_samples_per_second": 502.76, |
| "eval_steps_per_second": 7.865, |
| "step": 10150 |
| }, |
| { |
| "epoch": 1.1655386027302972, |
| "eval_loss": 0.40725043416023254, |
| "eval_runtime": 13.0358, |
| "eval_samples_per_second": 505.072, |
| "eval_steps_per_second": 7.901, |
| "step": 10160 |
| }, |
| { |
| "epoch": 1.1666857863944018, |
| "eval_loss": 0.4072812795639038, |
| "eval_runtime": 12.9686, |
| "eval_samples_per_second": 507.686, |
| "eval_steps_per_second": 7.942, |
| "step": 10170 |
| }, |
| { |
| "epoch": 1.1678329700585064, |
| "eval_loss": 0.40720149874687195, |
| "eval_runtime": 12.9804, |
| "eval_samples_per_second": 507.228, |
| "eval_steps_per_second": 7.935, |
| "step": 10180 |
| }, |
| { |
| "epoch": 1.168980153722611, |
| "eval_loss": 0.40726402401924133, |
| "eval_runtime": 12.9556, |
| "eval_samples_per_second": 508.196, |
| "eval_steps_per_second": 7.95, |
| "step": 10190 |
| }, |
| { |
| "epoch": 1.1701273373867156, |
| "grad_norm": 3.229079246520996, |
| "learning_rate": 6.777565328234544e-06, |
| "loss": 0.758, |
| "step": 10200 |
| }, |
| { |
| "epoch": 1.1701273373867156, |
| "eval_loss": 0.40712031722068787, |
| "eval_runtime": 12.9761, |
| "eval_samples_per_second": 507.395, |
| "eval_steps_per_second": 7.938, |
| "step": 10200 |
| }, |
| { |
| "epoch": 1.1712745210508202, |
| "eval_loss": 0.4070831835269928, |
| "eval_runtime": 12.9807, |
| "eval_samples_per_second": 507.214, |
| "eval_steps_per_second": 7.935, |
| "step": 10210 |
| }, |
| { |
| "epoch": 1.1724217047149248, |
| "eval_loss": 0.40712282061576843, |
| "eval_runtime": 12.9719, |
| "eval_samples_per_second": 507.557, |
| "eval_steps_per_second": 7.94, |
| "step": 10220 |
| }, |
| { |
| "epoch": 1.1735688883790294, |
| "eval_loss": 0.40677547454833984, |
| "eval_runtime": 13.0644, |
| "eval_samples_per_second": 503.964, |
| "eval_steps_per_second": 7.884, |
| "step": 10230 |
| }, |
| { |
| "epoch": 1.1747160720431342, |
| "eval_loss": 0.406320184469223, |
| "eval_runtime": 13.0322, |
| "eval_samples_per_second": 505.21, |
| "eval_steps_per_second": 7.903, |
| "step": 10240 |
| }, |
| { |
| "epoch": 1.1758632557072388, |
| "eval_loss": 0.4062091112136841, |
| "eval_runtime": 12.9674, |
| "eval_samples_per_second": 507.733, |
| "eval_steps_per_second": 7.943, |
| "step": 10250 |
| }, |
| { |
| "epoch": 1.1770104393713434, |
| "eval_loss": 0.40635228157043457, |
| "eval_runtime": 12.9838, |
| "eval_samples_per_second": 507.095, |
| "eval_steps_per_second": 7.933, |
| "step": 10260 |
| }, |
| { |
| "epoch": 1.178157623035448, |
| "eval_loss": 0.4064924418926239, |
| "eval_runtime": 13.0812, |
| "eval_samples_per_second": 503.319, |
| "eval_steps_per_second": 7.874, |
| "step": 10270 |
| }, |
| { |
| "epoch": 1.1793048066995526, |
| "eval_loss": 0.40626075863838196, |
| "eval_runtime": 12.9882, |
| "eval_samples_per_second": 506.922, |
| "eval_steps_per_second": 7.93, |
| "step": 10280 |
| }, |
| { |
| "epoch": 1.1804519903636572, |
| "eval_loss": 0.4064696729183197, |
| "eval_runtime": 12.9717, |
| "eval_samples_per_second": 507.565, |
| "eval_steps_per_second": 7.94, |
| "step": 10290 |
| }, |
| { |
| "epoch": 1.1815991740277618, |
| "grad_norm": 2.9564294815063477, |
| "learning_rate": 6.7350754195878485e-06, |
| "loss": 0.7322, |
| "step": 10300 |
| }, |
| { |
| "epoch": 1.1815991740277618, |
| "eval_loss": 0.4065658450126648, |
| "eval_runtime": 12.9857, |
| "eval_samples_per_second": 507.021, |
| "eval_steps_per_second": 7.932, |
| "step": 10300 |
| }, |
| { |
| "epoch": 1.1827463576918664, |
| "eval_loss": 0.40649843215942383, |
| "eval_runtime": 12.9714, |
| "eval_samples_per_second": 507.578, |
| "eval_steps_per_second": 7.941, |
| "step": 10310 |
| }, |
| { |
| "epoch": 1.183893541355971, |
| "eval_loss": 0.40647172927856445, |
| "eval_runtime": 12.9687, |
| "eval_samples_per_second": 507.685, |
| "eval_steps_per_second": 7.942, |
| "step": 10320 |
| }, |
| { |
| "epoch": 1.1850407250200756, |
| "eval_loss": 0.40608131885528564, |
| "eval_runtime": 12.966, |
| "eval_samples_per_second": 507.79, |
| "eval_steps_per_second": 7.944, |
| "step": 10330 |
| }, |
| { |
| "epoch": 1.1861879086841802, |
| "eval_loss": 0.40596771240234375, |
| "eval_runtime": 12.962, |
| "eval_samples_per_second": 507.946, |
| "eval_steps_per_second": 7.946, |
| "step": 10340 |
| }, |
| { |
| "epoch": 1.187335092348285, |
| "eval_loss": 0.405696839094162, |
| "eval_runtime": 12.9686, |
| "eval_samples_per_second": 507.689, |
| "eval_steps_per_second": 7.942, |
| "step": 10350 |
| }, |
| { |
| "epoch": 1.1884822760123896, |
| "eval_loss": 0.4055940508842468, |
| "eval_runtime": 13.0291, |
| "eval_samples_per_second": 505.33, |
| "eval_steps_per_second": 7.905, |
| "step": 10360 |
| }, |
| { |
| "epoch": 1.1896294596764942, |
| "eval_loss": 0.4056401550769806, |
| "eval_runtime": 12.9997, |
| "eval_samples_per_second": 506.472, |
| "eval_steps_per_second": 7.923, |
| "step": 10370 |
| }, |
| { |
| "epoch": 1.1907766433405989, |
| "eval_loss": 0.40586522221565247, |
| "eval_runtime": 12.9701, |
| "eval_samples_per_second": 507.629, |
| "eval_steps_per_second": 7.941, |
| "step": 10380 |
| }, |
| { |
| "epoch": 1.1919238270047035, |
| "eval_loss": 0.4060729742050171, |
| "eval_runtime": 13.1164, |
| "eval_samples_per_second": 501.967, |
| "eval_steps_per_second": 7.853, |
| "step": 10390 |
| }, |
| { |
| "epoch": 1.193071010668808, |
| "grad_norm": 3.28369140625, |
| "learning_rate": 6.692585510941152e-06, |
| "loss": 0.6948, |
| "step": 10400 |
| }, |
| { |
| "epoch": 1.193071010668808, |
| "eval_loss": 0.40594592690467834, |
| "eval_runtime": 12.966, |
| "eval_samples_per_second": 507.791, |
| "eval_steps_per_second": 7.944, |
| "step": 10400 |
| }, |
| { |
| "epoch": 1.1942181943329127, |
| "eval_loss": 0.4059278070926666, |
| "eval_runtime": 13.083, |
| "eval_samples_per_second": 503.25, |
| "eval_steps_per_second": 7.873, |
| "step": 10410 |
| }, |
| { |
| "epoch": 1.1953653779970173, |
| "eval_loss": 0.40602797269821167, |
| "eval_runtime": 12.9673, |
| "eval_samples_per_second": 507.739, |
| "eval_steps_per_second": 7.943, |
| "step": 10420 |
| }, |
| { |
| "epoch": 1.1965125616611219, |
| "eval_loss": 0.4057934284210205, |
| "eval_runtime": 12.9682, |
| "eval_samples_per_second": 507.705, |
| "eval_steps_per_second": 7.943, |
| "step": 10430 |
| }, |
| { |
| "epoch": 1.1976597453252267, |
| "eval_loss": 0.40573570132255554, |
| "eval_runtime": 12.9647, |
| "eval_samples_per_second": 507.841, |
| "eval_steps_per_second": 7.945, |
| "step": 10440 |
| }, |
| { |
| "epoch": 1.1988069289893313, |
| "eval_loss": 0.4055929183959961, |
| "eval_runtime": 13.0998, |
| "eval_samples_per_second": 502.602, |
| "eval_steps_per_second": 7.863, |
| "step": 10450 |
| }, |
| { |
| "epoch": 1.1999541126534359, |
| "eval_loss": 0.40564846992492676, |
| "eval_runtime": 13.0983, |
| "eval_samples_per_second": 502.662, |
| "eval_steps_per_second": 7.864, |
| "step": 10460 |
| }, |
| { |
| "epoch": 1.2011012963175405, |
| "eval_loss": 0.4055902063846588, |
| "eval_runtime": 13.0995, |
| "eval_samples_per_second": 502.613, |
| "eval_steps_per_second": 7.863, |
| "step": 10470 |
| }, |
| { |
| "epoch": 1.202248479981645, |
| "eval_loss": 0.40567445755004883, |
| "eval_runtime": 13.0889, |
| "eval_samples_per_second": 503.021, |
| "eval_steps_per_second": 7.869, |
| "step": 10480 |
| }, |
| { |
| "epoch": 1.2033956636457497, |
| "eval_loss": 0.4056432545185089, |
| "eval_runtime": 12.9688, |
| "eval_samples_per_second": 507.682, |
| "eval_steps_per_second": 7.942, |
| "step": 10490 |
| }, |
| { |
| "epoch": 1.2045428473098543, |
| "grad_norm": 2.8693108558654785, |
| "learning_rate": 6.650095602294455e-06, |
| "loss": 0.7185, |
| "step": 10500 |
| }, |
| { |
| "epoch": 1.2045428473098543, |
| "eval_loss": 0.40548571944236755, |
| "eval_runtime": 12.97, |
| "eval_samples_per_second": 507.633, |
| "eval_steps_per_second": 7.941, |
| "step": 10500 |
| }, |
| { |
| "epoch": 1.2056900309739589, |
| "eval_loss": 0.40558117628097534, |
| "eval_runtime": 12.9805, |
| "eval_samples_per_second": 507.223, |
| "eval_steps_per_second": 7.935, |
| "step": 10510 |
| }, |
| { |
| "epoch": 1.2068372146380635, |
| "eval_loss": 0.405443012714386, |
| "eval_runtime": 12.9751, |
| "eval_samples_per_second": 507.435, |
| "eval_steps_per_second": 7.938, |
| "step": 10520 |
| }, |
| { |
| "epoch": 1.207984398302168, |
| "eval_loss": 0.4053213596343994, |
| "eval_runtime": 12.9895, |
| "eval_samples_per_second": 506.871, |
| "eval_steps_per_second": 7.929, |
| "step": 10530 |
| }, |
| { |
| "epoch": 1.2091315819662727, |
| "eval_loss": 0.40506511926651, |
| "eval_runtime": 12.9667, |
| "eval_samples_per_second": 507.763, |
| "eval_steps_per_second": 7.943, |
| "step": 10540 |
| }, |
| { |
| "epoch": 1.2102787656303775, |
| "eval_loss": 0.4049833416938782, |
| "eval_runtime": 12.9524, |
| "eval_samples_per_second": 508.323, |
| "eval_steps_per_second": 7.952, |
| "step": 10550 |
| }, |
| { |
| "epoch": 1.211425949294482, |
| "eval_loss": 0.40509331226348877, |
| "eval_runtime": 13.0843, |
| "eval_samples_per_second": 503.197, |
| "eval_steps_per_second": 7.872, |
| "step": 10560 |
| }, |
| { |
| "epoch": 1.2125731329585867, |
| "eval_loss": 0.40517449378967285, |
| "eval_runtime": 12.9925, |
| "eval_samples_per_second": 506.755, |
| "eval_steps_per_second": 7.928, |
| "step": 10570 |
| }, |
| { |
| "epoch": 1.2137203166226913, |
| "eval_loss": 0.4053093194961548, |
| "eval_runtime": 12.9885, |
| "eval_samples_per_second": 506.912, |
| "eval_steps_per_second": 7.93, |
| "step": 10580 |
| }, |
| { |
| "epoch": 1.214867500286796, |
| "eval_loss": 0.40528470277786255, |
| "eval_runtime": 13.0899, |
| "eval_samples_per_second": 502.982, |
| "eval_steps_per_second": 7.869, |
| "step": 10590 |
| }, |
| { |
| "epoch": 1.2160146839509005, |
| "grad_norm": 3.2297863960266113, |
| "learning_rate": 6.607605693647759e-06, |
| "loss": 0.7039, |
| "step": 10600 |
| }, |
| { |
| "epoch": 1.2160146839509005, |
| "eval_loss": 0.4053327441215515, |
| "eval_runtime": 13.0145, |
| "eval_samples_per_second": 505.899, |
| "eval_steps_per_second": 7.914, |
| "step": 10600 |
| }, |
| { |
| "epoch": 1.217161867615005, |
| "eval_loss": 0.4054310619831085, |
| "eval_runtime": 12.973, |
| "eval_samples_per_second": 507.515, |
| "eval_steps_per_second": 7.94, |
| "step": 10610 |
| }, |
| { |
| "epoch": 1.2183090512791097, |
| "eval_loss": 0.40506982803344727, |
| "eval_runtime": 12.9721, |
| "eval_samples_per_second": 507.55, |
| "eval_steps_per_second": 7.94, |
| "step": 10620 |
| }, |
| { |
| "epoch": 1.2194562349432143, |
| "eval_loss": 0.4049602746963501, |
| "eval_runtime": 13.0034, |
| "eval_samples_per_second": 506.33, |
| "eval_steps_per_second": 7.921, |
| "step": 10630 |
| }, |
| { |
| "epoch": 1.2206034186073191, |
| "eval_loss": 0.4047999083995819, |
| "eval_runtime": 12.9807, |
| "eval_samples_per_second": 507.215, |
| "eval_steps_per_second": 7.935, |
| "step": 10640 |
| }, |
| { |
| "epoch": 1.2217506022714237, |
| "eval_loss": 0.4044307470321655, |
| "eval_runtime": 13.065, |
| "eval_samples_per_second": 503.942, |
| "eval_steps_per_second": 7.884, |
| "step": 10650 |
| }, |
| { |
| "epoch": 1.2228977859355283, |
| "eval_loss": 0.40463724732398987, |
| "eval_runtime": 12.958, |
| "eval_samples_per_second": 508.104, |
| "eval_steps_per_second": 7.949, |
| "step": 10660 |
| }, |
| { |
| "epoch": 1.224044969599633, |
| "eval_loss": 0.4044046401977539, |
| "eval_runtime": 13.0802, |
| "eval_samples_per_second": 503.356, |
| "eval_steps_per_second": 7.874, |
| "step": 10670 |
| }, |
| { |
| "epoch": 1.2251921532637375, |
| "eval_loss": 0.4041491746902466, |
| "eval_runtime": 13.1424, |
| "eval_samples_per_second": 500.973, |
| "eval_steps_per_second": 7.837, |
| "step": 10680 |
| }, |
| { |
| "epoch": 1.2263393369278421, |
| "eval_loss": 0.40389564633369446, |
| "eval_runtime": 12.9899, |
| "eval_samples_per_second": 506.854, |
| "eval_steps_per_second": 7.929, |
| "step": 10690 |
| }, |
| { |
| "epoch": 1.2274865205919467, |
| "grad_norm": 2.6197776794433594, |
| "learning_rate": 6.565115785001063e-06, |
| "loss": 0.6969, |
| "step": 10700 |
| }, |
| { |
| "epoch": 1.2274865205919467, |
| "eval_loss": 0.40372923016548157, |
| "eval_runtime": 13.0976, |
| "eval_samples_per_second": 502.688, |
| "eval_steps_per_second": 7.864, |
| "step": 10700 |
| }, |
| { |
| "epoch": 1.2286337042560513, |
| "eval_loss": 0.40366119146347046, |
| "eval_runtime": 13.1012, |
| "eval_samples_per_second": 502.548, |
| "eval_steps_per_second": 7.862, |
| "step": 10710 |
| }, |
| { |
| "epoch": 1.229780887920156, |
| "eval_loss": 0.4035404920578003, |
| "eval_runtime": 13.1117, |
| "eval_samples_per_second": 502.146, |
| "eval_steps_per_second": 7.856, |
| "step": 10720 |
| }, |
| { |
| "epoch": 1.2309280715842608, |
| "eval_loss": 0.4036332368850708, |
| "eval_runtime": 12.9567, |
| "eval_samples_per_second": 508.155, |
| "eval_steps_per_second": 7.95, |
| "step": 10730 |
| }, |
| { |
| "epoch": 1.2320752552483651, |
| "eval_loss": 0.40354517102241516, |
| "eval_runtime": 13.1022, |
| "eval_samples_per_second": 502.51, |
| "eval_steps_per_second": 7.861, |
| "step": 10740 |
| }, |
| { |
| "epoch": 1.23322243891247, |
| "eval_loss": 0.4037734270095825, |
| "eval_runtime": 12.9663, |
| "eval_samples_per_second": 507.778, |
| "eval_steps_per_second": 7.944, |
| "step": 10750 |
| }, |
| { |
| "epoch": 1.2343696225765746, |
| "eval_loss": 0.4038192629814148, |
| "eval_runtime": 13.0833, |
| "eval_samples_per_second": 503.238, |
| "eval_steps_per_second": 7.873, |
| "step": 10760 |
| }, |
| { |
| "epoch": 1.2355168062406792, |
| "eval_loss": 0.4037155210971832, |
| "eval_runtime": 12.9697, |
| "eval_samples_per_second": 507.645, |
| "eval_steps_per_second": 7.942, |
| "step": 10770 |
| }, |
| { |
| "epoch": 1.2366639899047838, |
| "eval_loss": 0.4036622643470764, |
| "eval_runtime": 13.0999, |
| "eval_samples_per_second": 502.599, |
| "eval_steps_per_second": 7.863, |
| "step": 10780 |
| }, |
| { |
| "epoch": 1.2378111735688884, |
| "eval_loss": 0.4037114381790161, |
| "eval_runtime": 13.1006, |
| "eval_samples_per_second": 502.571, |
| "eval_steps_per_second": 7.862, |
| "step": 10790 |
| }, |
| { |
| "epoch": 1.238958357232993, |
| "grad_norm": 2.4447872638702393, |
| "learning_rate": 6.522625876354366e-06, |
| "loss": 0.6921, |
| "step": 10800 |
| }, |
| { |
| "epoch": 1.238958357232993, |
| "eval_loss": 0.40383169054985046, |
| "eval_runtime": 12.9711, |
| "eval_samples_per_second": 507.591, |
| "eval_steps_per_second": 7.941, |
| "step": 10800 |
| }, |
| { |
| "epoch": 1.2401055408970976, |
| "eval_loss": 0.40391016006469727, |
| "eval_runtime": 13.096, |
| "eval_samples_per_second": 502.75, |
| "eval_steps_per_second": 7.865, |
| "step": 10810 |
| }, |
| { |
| "epoch": 1.2412527245612022, |
| "eval_loss": 0.4037761986255646, |
| "eval_runtime": 12.9752, |
| "eval_samples_per_second": 507.428, |
| "eval_steps_per_second": 7.938, |
| "step": 10820 |
| }, |
| { |
| "epoch": 1.2423999082253068, |
| "eval_loss": 0.4037272334098816, |
| "eval_runtime": 12.9591, |
| "eval_samples_per_second": 508.059, |
| "eval_steps_per_second": 7.948, |
| "step": 10830 |
| }, |
| { |
| "epoch": 1.2435470918894116, |
| "eval_loss": 0.40398961305618286, |
| "eval_runtime": 12.9728, |
| "eval_samples_per_second": 507.523, |
| "eval_steps_per_second": 7.94, |
| "step": 10840 |
| }, |
| { |
| "epoch": 1.2446942755535162, |
| "eval_loss": 0.40415656566619873, |
| "eval_runtime": 12.967, |
| "eval_samples_per_second": 507.75, |
| "eval_steps_per_second": 7.943, |
| "step": 10850 |
| }, |
| { |
| "epoch": 1.2458414592176208, |
| "eval_loss": 0.40438222885131836, |
| "eval_runtime": 12.9639, |
| "eval_samples_per_second": 507.873, |
| "eval_steps_per_second": 7.945, |
| "step": 10860 |
| }, |
| { |
| "epoch": 1.2469886428817254, |
| "eval_loss": 0.4042970538139343, |
| "eval_runtime": 12.985, |
| "eval_samples_per_second": 507.045, |
| "eval_steps_per_second": 7.932, |
| "step": 10870 |
| }, |
| { |
| "epoch": 1.24813582654583, |
| "eval_loss": 0.40430957078933716, |
| "eval_runtime": 12.9651, |
| "eval_samples_per_second": 507.826, |
| "eval_steps_per_second": 7.944, |
| "step": 10880 |
| }, |
| { |
| "epoch": 1.2492830102099346, |
| "eval_loss": 0.40437033772468567, |
| "eval_runtime": 12.9617, |
| "eval_samples_per_second": 507.959, |
| "eval_steps_per_second": 7.947, |
| "step": 10890 |
| }, |
| { |
| "epoch": 1.2504301938740392, |
| "grad_norm": 4.182785987854004, |
| "learning_rate": 6.48013596770767e-06, |
| "loss": 0.728, |
| "step": 10900 |
| }, |
| { |
| "epoch": 1.2504301938740392, |
| "eval_loss": 0.4041915535926819, |
| "eval_runtime": 12.9637, |
| "eval_samples_per_second": 507.879, |
| "eval_steps_per_second": 7.945, |
| "step": 10900 |
| }, |
| { |
| "epoch": 1.2515773775381438, |
| "eval_loss": 0.40437549352645874, |
| "eval_runtime": 13.1594, |
| "eval_samples_per_second": 500.326, |
| "eval_steps_per_second": 7.827, |
| "step": 10910 |
| }, |
| { |
| "epoch": 1.2527245612022484, |
| "eval_loss": 0.4042831063270569, |
| "eval_runtime": 12.9641, |
| "eval_samples_per_second": 507.864, |
| "eval_steps_per_second": 7.945, |
| "step": 10920 |
| }, |
| { |
| "epoch": 1.2538717448663532, |
| "eval_loss": 0.40394535660743713, |
| "eval_runtime": 13.0706, |
| "eval_samples_per_second": 503.728, |
| "eval_steps_per_second": 7.88, |
| "step": 10930 |
| }, |
| { |
| "epoch": 1.2550189285304576, |
| "eval_loss": 0.4037913382053375, |
| "eval_runtime": 12.9769, |
| "eval_samples_per_second": 507.362, |
| "eval_steps_per_second": 7.937, |
| "step": 10940 |
| }, |
| { |
| "epoch": 1.2561661121945624, |
| "eval_loss": 0.40372398495674133, |
| "eval_runtime": 12.9645, |
| "eval_samples_per_second": 507.849, |
| "eval_steps_per_second": 7.945, |
| "step": 10950 |
| }, |
| { |
| "epoch": 1.257313295858667, |
| "eval_loss": 0.4034961462020874, |
| "eval_runtime": 13.0823, |
| "eval_samples_per_second": 503.274, |
| "eval_steps_per_second": 7.873, |
| "step": 10960 |
| }, |
| { |
| "epoch": 1.2584604795227716, |
| "eval_loss": 0.40317344665527344, |
| "eval_runtime": 13.0799, |
| "eval_samples_per_second": 503.366, |
| "eval_steps_per_second": 7.875, |
| "step": 10970 |
| }, |
| { |
| "epoch": 1.2596076631868762, |
| "eval_loss": 0.4024331867694855, |
| "eval_runtime": 12.9932, |
| "eval_samples_per_second": 506.727, |
| "eval_steps_per_second": 7.927, |
| "step": 10980 |
| }, |
| { |
| "epoch": 1.2607548468509808, |
| "eval_loss": 0.401925653219223, |
| "eval_runtime": 13.0961, |
| "eval_samples_per_second": 502.743, |
| "eval_steps_per_second": 7.865, |
| "step": 10990 |
| }, |
| { |
| "epoch": 1.2619020305150854, |
| "grad_norm": 18.197425842285156, |
| "learning_rate": 6.437646059060973e-06, |
| "loss": 0.713, |
| "step": 11000 |
| }, |
| { |
| "epoch": 1.2619020305150854, |
| "eval_loss": 0.4017943739891052, |
| "eval_runtime": 12.9661, |
| "eval_samples_per_second": 507.784, |
| "eval_steps_per_second": 7.944, |
| "step": 11000 |
| }, |
| { |
| "epoch": 1.26304921417919, |
| "eval_loss": 0.4015377461910248, |
| "eval_runtime": 12.9836, |
| "eval_samples_per_second": 507.101, |
| "eval_steps_per_second": 7.933, |
| "step": 11010 |
| }, |
| { |
| "epoch": 1.2641963978432949, |
| "eval_loss": 0.40145203471183777, |
| "eval_runtime": 12.9705, |
| "eval_samples_per_second": 507.615, |
| "eval_steps_per_second": 7.941, |
| "step": 11020 |
| }, |
| { |
| "epoch": 1.2653435815073992, |
| "eval_loss": 0.40141260623931885, |
| "eval_runtime": 12.9828, |
| "eval_samples_per_second": 507.132, |
| "eval_steps_per_second": 7.934, |
| "step": 11030 |
| }, |
| { |
| "epoch": 1.266490765171504, |
| "eval_loss": 0.40151217579841614, |
| "eval_runtime": 12.9733, |
| "eval_samples_per_second": 507.502, |
| "eval_steps_per_second": 7.939, |
| "step": 11040 |
| }, |
| { |
| "epoch": 1.2676379488356087, |
| "eval_loss": 0.4014013111591339, |
| "eval_runtime": 12.9667, |
| "eval_samples_per_second": 507.763, |
| "eval_steps_per_second": 7.943, |
| "step": 11050 |
| }, |
| { |
| "epoch": 1.2687851324997133, |
| "eval_loss": 0.40132954716682434, |
| "eval_runtime": 13.0465, |
| "eval_samples_per_second": 504.655, |
| "eval_steps_per_second": 7.895, |
| "step": 11060 |
| }, |
| { |
| "epoch": 1.2699323161638179, |
| "eval_loss": 0.40150803327560425, |
| "eval_runtime": 12.9649, |
| "eval_samples_per_second": 507.834, |
| "eval_steps_per_second": 7.945, |
| "step": 11070 |
| }, |
| { |
| "epoch": 1.2710794998279225, |
| "eval_loss": 0.401555597782135, |
| "eval_runtime": 13.0395, |
| "eval_samples_per_second": 504.929, |
| "eval_steps_per_second": 7.899, |
| "step": 11080 |
| }, |
| { |
| "epoch": 1.272226683492027, |
| "eval_loss": 0.40172475576400757, |
| "eval_runtime": 13.1435, |
| "eval_samples_per_second": 500.932, |
| "eval_steps_per_second": 7.837, |
| "step": 11090 |
| }, |
| { |
| "epoch": 1.2733738671561317, |
| "grad_norm": 5.314532279968262, |
| "learning_rate": 6.395156150414277e-06, |
| "loss": 0.668, |
| "step": 11100 |
| }, |
| { |
| "epoch": 1.2733738671561317, |
| "eval_loss": 0.40174537897109985, |
| "eval_runtime": 12.9903, |
| "eval_samples_per_second": 506.84, |
| "eval_steps_per_second": 7.929, |
| "step": 11100 |
| }, |
| { |
| "epoch": 1.2745210508202363, |
| "eval_loss": 0.40156856179237366, |
| "eval_runtime": 12.9845, |
| "eval_samples_per_second": 507.065, |
| "eval_steps_per_second": 7.933, |
| "step": 11110 |
| }, |
| { |
| "epoch": 1.2756682344843409, |
| "eval_loss": 0.40162721276283264, |
| "eval_runtime": 12.9833, |
| "eval_samples_per_second": 507.113, |
| "eval_steps_per_second": 7.933, |
| "step": 11120 |
| }, |
| { |
| "epoch": 1.2768154181484457, |
| "eval_loss": 0.40191736817359924, |
| "eval_runtime": 12.9745, |
| "eval_samples_per_second": 507.456, |
| "eval_steps_per_second": 7.939, |
| "step": 11130 |
| }, |
| { |
| "epoch": 1.27796260181255, |
| "eval_loss": 0.4020519256591797, |
| "eval_runtime": 13.0133, |
| "eval_samples_per_second": 505.942, |
| "eval_steps_per_second": 7.915, |
| "step": 11140 |
| }, |
| { |
| "epoch": 1.2791097854766549, |
| "eval_loss": 0.40185847878456116, |
| "eval_runtime": 12.9821, |
| "eval_samples_per_second": 507.16, |
| "eval_steps_per_second": 7.934, |
| "step": 11150 |
| }, |
| { |
| "epoch": 1.2802569691407595, |
| "eval_loss": 0.4017007648944855, |
| "eval_runtime": 12.9713, |
| "eval_samples_per_second": 507.583, |
| "eval_steps_per_second": 7.941, |
| "step": 11160 |
| }, |
| { |
| "epoch": 1.281404152804864, |
| "eval_loss": 0.4016687273979187, |
| "eval_runtime": 12.9988, |
| "eval_samples_per_second": 506.507, |
| "eval_steps_per_second": 7.924, |
| "step": 11170 |
| }, |
| { |
| "epoch": 1.2825513364689687, |
| "eval_loss": 0.40175148844718933, |
| "eval_runtime": 12.9745, |
| "eval_samples_per_second": 507.455, |
| "eval_steps_per_second": 7.939, |
| "step": 11180 |
| }, |
| { |
| "epoch": 1.2836985201330733, |
| "eval_loss": 0.40132349729537964, |
| "eval_runtime": 12.9777, |
| "eval_samples_per_second": 507.333, |
| "eval_steps_per_second": 7.937, |
| "step": 11190 |
| }, |
| { |
| "epoch": 1.2848457037971779, |
| "grad_norm": 2.2018253803253174, |
| "learning_rate": 6.352666241767581e-06, |
| "loss": 0.7101, |
| "step": 11200 |
| }, |
| { |
| "epoch": 1.2848457037971779, |
| "eval_loss": 0.40107598900794983, |
| "eval_runtime": 13.0059, |
| "eval_samples_per_second": 506.234, |
| "eval_steps_per_second": 7.92, |
| "step": 11200 |
| }, |
| { |
| "epoch": 1.2859928874612825, |
| "eval_loss": 0.40110349655151367, |
| "eval_runtime": 12.9791, |
| "eval_samples_per_second": 507.279, |
| "eval_steps_per_second": 7.936, |
| "step": 11210 |
| }, |
| { |
| "epoch": 1.2871400711253873, |
| "eval_loss": 0.40136098861694336, |
| "eval_runtime": 13.0847, |
| "eval_samples_per_second": 503.182, |
| "eval_steps_per_second": 7.872, |
| "step": 11220 |
| }, |
| { |
| "epoch": 1.2882872547894917, |
| "eval_loss": 0.4015278220176697, |
| "eval_runtime": 13.1023, |
| "eval_samples_per_second": 502.508, |
| "eval_steps_per_second": 7.861, |
| "step": 11230 |
| }, |
| { |
| "epoch": 1.2894344384535965, |
| "eval_loss": 0.4010304808616638, |
| "eval_runtime": 13.1018, |
| "eval_samples_per_second": 502.526, |
| "eval_steps_per_second": 7.862, |
| "step": 11240 |
| }, |
| { |
| "epoch": 1.290581622117701, |
| "eval_loss": 0.4012353718280792, |
| "eval_runtime": 12.9666, |
| "eval_samples_per_second": 507.765, |
| "eval_steps_per_second": 7.943, |
| "step": 11250 |
| }, |
| { |
| "epoch": 1.2917288057818057, |
| "eval_loss": 0.40129220485687256, |
| "eval_runtime": 12.966, |
| "eval_samples_per_second": 507.791, |
| "eval_steps_per_second": 7.944, |
| "step": 11260 |
| }, |
| { |
| "epoch": 1.2928759894459103, |
| "eval_loss": 0.4010326564311981, |
| "eval_runtime": 12.9631, |
| "eval_samples_per_second": 507.902, |
| "eval_steps_per_second": 7.946, |
| "step": 11270 |
| }, |
| { |
| "epoch": 1.294023173110015, |
| "eval_loss": 0.4006401002407074, |
| "eval_runtime": 12.9593, |
| "eval_samples_per_second": 508.054, |
| "eval_steps_per_second": 7.948, |
| "step": 11280 |
| }, |
| { |
| "epoch": 1.2951703567741195, |
| "eval_loss": 0.4004891812801361, |
| "eval_runtime": 12.9698, |
| "eval_samples_per_second": 507.643, |
| "eval_steps_per_second": 7.942, |
| "step": 11290 |
| }, |
| { |
| "epoch": 1.2963175404382241, |
| "grad_norm": 2.9533190727233887, |
| "learning_rate": 6.310176333120884e-06, |
| "loss": 0.6963, |
| "step": 11300 |
| }, |
| { |
| "epoch": 1.2963175404382241, |
| "eval_loss": 0.4003821015357971, |
| "eval_runtime": 12.9636, |
| "eval_samples_per_second": 507.882, |
| "eval_steps_per_second": 7.945, |
| "step": 11300 |
| }, |
| { |
| "epoch": 1.2974647241023287, |
| "eval_loss": 0.40033960342407227, |
| "eval_runtime": 12.9927, |
| "eval_samples_per_second": 506.745, |
| "eval_steps_per_second": 7.928, |
| "step": 11310 |
| }, |
| { |
| "epoch": 1.2986119077664333, |
| "eval_loss": 0.4003883898258209, |
| "eval_runtime": 13.0034, |
| "eval_samples_per_second": 506.33, |
| "eval_steps_per_second": 7.921, |
| "step": 11320 |
| }, |
| { |
| "epoch": 1.2997590914305381, |
| "eval_loss": 0.4002770185470581, |
| "eval_runtime": 12.9617, |
| "eval_samples_per_second": 507.957, |
| "eval_steps_per_second": 7.946, |
| "step": 11330 |
| }, |
| { |
| "epoch": 1.3009062750946425, |
| "eval_loss": 0.3999290466308594, |
| "eval_runtime": 12.9526, |
| "eval_samples_per_second": 508.313, |
| "eval_steps_per_second": 7.952, |
| "step": 11340 |
| }, |
| { |
| "epoch": 1.3020534587587473, |
| "eval_loss": 0.3997403681278229, |
| "eval_runtime": 13.1193, |
| "eval_samples_per_second": 501.856, |
| "eval_steps_per_second": 7.851, |
| "step": 11350 |
| }, |
| { |
| "epoch": 1.303200642422852, |
| "eval_loss": 0.39959993958473206, |
| "eval_runtime": 12.9503, |
| "eval_samples_per_second": 508.407, |
| "eval_steps_per_second": 7.954, |
| "step": 11360 |
| }, |
| { |
| "epoch": 1.3043478260869565, |
| "eval_loss": 0.3997476398944855, |
| "eval_runtime": 12.9729, |
| "eval_samples_per_second": 507.519, |
| "eval_steps_per_second": 7.94, |
| "step": 11370 |
| }, |
| { |
| "epoch": 1.3054950097510611, |
| "eval_loss": 0.39957594871520996, |
| "eval_runtime": 13.1077, |
| "eval_samples_per_second": 502.3, |
| "eval_steps_per_second": 7.858, |
| "step": 11380 |
| }, |
| { |
| "epoch": 1.3066421934151657, |
| "eval_loss": 0.39939528703689575, |
| "eval_runtime": 13.0747, |
| "eval_samples_per_second": 503.567, |
| "eval_steps_per_second": 7.878, |
| "step": 11390 |
| }, |
| { |
| "epoch": 1.3077893770792703, |
| "grad_norm": 2.4936652183532715, |
| "learning_rate": 6.267686424474188e-06, |
| "loss": 0.6706, |
| "step": 11400 |
| }, |
| { |
| "epoch": 1.3077893770792703, |
| "eval_loss": 0.3992767930030823, |
| "eval_runtime": 13.0013, |
| "eval_samples_per_second": 506.412, |
| "eval_steps_per_second": 7.922, |
| "step": 11400 |
| }, |
| { |
| "epoch": 1.308936560743375, |
| "eval_loss": 0.39906489849090576, |
| "eval_runtime": 12.9612, |
| "eval_samples_per_second": 507.978, |
| "eval_steps_per_second": 7.947, |
| "step": 11410 |
| }, |
| { |
| "epoch": 1.3100837444074798, |
| "eval_loss": 0.3990459144115448, |
| "eval_runtime": 13.0813, |
| "eval_samples_per_second": 503.312, |
| "eval_steps_per_second": 7.874, |
| "step": 11420 |
| }, |
| { |
| "epoch": 1.3112309280715841, |
| "eval_loss": 0.3989590108394623, |
| "eval_runtime": 12.9957, |
| "eval_samples_per_second": 506.629, |
| "eval_steps_per_second": 7.926, |
| "step": 11430 |
| }, |
| { |
| "epoch": 1.312378111735689, |
| "eval_loss": 0.39872854948043823, |
| "eval_runtime": 13.089, |
| "eval_samples_per_second": 503.017, |
| "eval_steps_per_second": 7.869, |
| "step": 11440 |
| }, |
| { |
| "epoch": 1.3135252953997936, |
| "eval_loss": 0.3981229066848755, |
| "eval_runtime": 13.0764, |
| "eval_samples_per_second": 503.504, |
| "eval_steps_per_second": 7.877, |
| "step": 11450 |
| }, |
| { |
| "epoch": 1.3146724790638982, |
| "eval_loss": 0.3978273570537567, |
| "eval_runtime": 12.9697, |
| "eval_samples_per_second": 507.645, |
| "eval_steps_per_second": 7.942, |
| "step": 11460 |
| }, |
| { |
| "epoch": 1.3158196627280028, |
| "eval_loss": 0.39747264981269836, |
| "eval_runtime": 12.9678, |
| "eval_samples_per_second": 507.72, |
| "eval_steps_per_second": 7.943, |
| "step": 11470 |
| }, |
| { |
| "epoch": 1.3169668463921074, |
| "eval_loss": 0.3973662555217743, |
| "eval_runtime": 12.9677, |
| "eval_samples_per_second": 507.721, |
| "eval_steps_per_second": 7.943, |
| "step": 11480 |
| }, |
| { |
| "epoch": 1.318114030056212, |
| "eval_loss": 0.39742839336395264, |
| "eval_runtime": 12.9618, |
| "eval_samples_per_second": 507.955, |
| "eval_steps_per_second": 7.946, |
| "step": 11490 |
| }, |
| { |
| "epoch": 1.3192612137203166, |
| "grad_norm": 9.057049751281738, |
| "learning_rate": 6.225196515827492e-06, |
| "loss": 0.6962, |
| "step": 11500 |
| }, |
| { |
| "epoch": 1.3192612137203166, |
| "eval_loss": 0.3974221646785736, |
| "eval_runtime": 12.9684, |
| "eval_samples_per_second": 507.694, |
| "eval_steps_per_second": 7.942, |
| "step": 11500 |
| }, |
| { |
| "epoch": 1.3204083973844212, |
| "eval_loss": 0.3974750339984894, |
| "eval_runtime": 13.0641, |
| "eval_samples_per_second": 503.978, |
| "eval_steps_per_second": 7.884, |
| "step": 11510 |
| }, |
| { |
| "epoch": 1.3215555810485258, |
| "eval_loss": 0.39754682779312134, |
| "eval_runtime": 13.0993, |
| "eval_samples_per_second": 502.622, |
| "eval_steps_per_second": 7.863, |
| "step": 11520 |
| }, |
| { |
| "epoch": 1.3227027647126306, |
| "eval_loss": 0.3975520730018616, |
| "eval_runtime": 12.9683, |
| "eval_samples_per_second": 507.7, |
| "eval_steps_per_second": 7.942, |
| "step": 11530 |
| }, |
| { |
| "epoch": 1.3238499483767352, |
| "eval_loss": 0.3976677358150482, |
| "eval_runtime": 12.9651, |
| "eval_samples_per_second": 507.825, |
| "eval_steps_per_second": 7.944, |
| "step": 11540 |
| }, |
| { |
| "epoch": 1.3249971320408398, |
| "eval_loss": 0.39753204584121704, |
| "eval_runtime": 13.0069, |
| "eval_samples_per_second": 506.191, |
| "eval_steps_per_second": 7.919, |
| "step": 11550 |
| }, |
| { |
| "epoch": 1.3261443157049444, |
| "eval_loss": 0.3974483907222748, |
| "eval_runtime": 12.986, |
| "eval_samples_per_second": 507.008, |
| "eval_steps_per_second": 7.932, |
| "step": 11560 |
| }, |
| { |
| "epoch": 1.327291499369049, |
| "eval_loss": 0.39730414748191833, |
| "eval_runtime": 12.9683, |
| "eval_samples_per_second": 507.7, |
| "eval_steps_per_second": 7.942, |
| "step": 11570 |
| }, |
| { |
| "epoch": 1.3284386830331536, |
| "eval_loss": 0.39712825417518616, |
| "eval_runtime": 12.9766, |
| "eval_samples_per_second": 507.376, |
| "eval_steps_per_second": 7.937, |
| "step": 11580 |
| }, |
| { |
| "epoch": 1.3295858666972582, |
| "eval_loss": 0.39687350392341614, |
| "eval_runtime": 12.9703, |
| "eval_samples_per_second": 507.622, |
| "eval_steps_per_second": 7.941, |
| "step": 11590 |
| }, |
| { |
| "epoch": 1.3307330503613628, |
| "grad_norm": 6.837019920349121, |
| "learning_rate": 6.182706607180795e-06, |
| "loss": 0.7083, |
| "step": 11600 |
| }, |
| { |
| "epoch": 1.3307330503613628, |
| "eval_loss": 0.39695462584495544, |
| "eval_runtime": 12.9673, |
| "eval_samples_per_second": 507.739, |
| "eval_steps_per_second": 7.943, |
| "step": 11600 |
| }, |
| { |
| "epoch": 1.3318802340254674, |
| "eval_loss": 0.3969835638999939, |
| "eval_runtime": 12.974, |
| "eval_samples_per_second": 507.477, |
| "eval_steps_per_second": 7.939, |
| "step": 11610 |
| }, |
| { |
| "epoch": 1.3330274176895722, |
| "eval_loss": 0.3970675468444824, |
| "eval_runtime": 12.969, |
| "eval_samples_per_second": 507.672, |
| "eval_steps_per_second": 7.942, |
| "step": 11620 |
| }, |
| { |
| "epoch": 1.3341746013536766, |
| "eval_loss": 0.39733174443244934, |
| "eval_runtime": 12.969, |
| "eval_samples_per_second": 507.674, |
| "eval_steps_per_second": 7.942, |
| "step": 11630 |
| }, |
| { |
| "epoch": 1.3353217850177814, |
| "eval_loss": 0.3975182771682739, |
| "eval_runtime": 12.9727, |
| "eval_samples_per_second": 507.528, |
| "eval_steps_per_second": 7.94, |
| "step": 11640 |
| }, |
| { |
| "epoch": 1.336468968681886, |
| "eval_loss": 0.39726847410202026, |
| "eval_runtime": 12.9649, |
| "eval_samples_per_second": 507.834, |
| "eval_steps_per_second": 7.945, |
| "step": 11650 |
| }, |
| { |
| "epoch": 1.3376161523459906, |
| "eval_loss": 0.39728638529777527, |
| "eval_runtime": 12.9706, |
| "eval_samples_per_second": 507.608, |
| "eval_steps_per_second": 7.941, |
| "step": 11660 |
| }, |
| { |
| "epoch": 1.3387633360100952, |
| "eval_loss": 0.39728325605392456, |
| "eval_runtime": 12.9684, |
| "eval_samples_per_second": 507.694, |
| "eval_steps_per_second": 7.942, |
| "step": 11670 |
| }, |
| { |
| "epoch": 1.3399105196741998, |
| "eval_loss": 0.39760005474090576, |
| "eval_runtime": 13.0176, |
| "eval_samples_per_second": 505.775, |
| "eval_steps_per_second": 7.912, |
| "step": 11680 |
| }, |
| { |
| "epoch": 1.3410577033383044, |
| "eval_loss": 0.3975573480129242, |
| "eval_runtime": 12.9898, |
| "eval_samples_per_second": 506.861, |
| "eval_steps_per_second": 7.929, |
| "step": 11690 |
| }, |
| { |
| "epoch": 1.342204887002409, |
| "grad_norm": 2.889777898788452, |
| "learning_rate": 6.140216698534099e-06, |
| "loss": 0.6757, |
| "step": 11700 |
| }, |
| { |
| "epoch": 1.342204887002409, |
| "eval_loss": 0.3976421058177948, |
| "eval_runtime": 12.9799, |
| "eval_samples_per_second": 507.247, |
| "eval_steps_per_second": 7.935, |
| "step": 11700 |
| }, |
| { |
| "epoch": 1.3433520706665136, |
| "eval_loss": 0.39746442437171936, |
| "eval_runtime": 12.9841, |
| "eval_samples_per_second": 507.082, |
| "eval_steps_per_second": 7.933, |
| "step": 11710 |
| }, |
| { |
| "epoch": 1.3444992543306182, |
| "eval_loss": 0.3973475396633148, |
| "eval_runtime": 13.1014, |
| "eval_samples_per_second": 502.542, |
| "eval_steps_per_second": 7.862, |
| "step": 11720 |
| }, |
| { |
| "epoch": 1.345646437994723, |
| "eval_loss": 0.397087037563324, |
| "eval_runtime": 12.9954, |
| "eval_samples_per_second": 506.641, |
| "eval_steps_per_second": 7.926, |
| "step": 11730 |
| }, |
| { |
| "epoch": 1.3467936216588277, |
| "eval_loss": 0.39631280303001404, |
| "eval_runtime": 12.9668, |
| "eval_samples_per_second": 507.759, |
| "eval_steps_per_second": 7.943, |
| "step": 11740 |
| }, |
| { |
| "epoch": 1.3479408053229323, |
| "eval_loss": 0.3963833153247833, |
| "eval_runtime": 13.0809, |
| "eval_samples_per_second": 503.328, |
| "eval_steps_per_second": 7.874, |
| "step": 11750 |
| }, |
| { |
| "epoch": 1.3490879889870369, |
| "eval_loss": 0.39647385478019714, |
| "eval_runtime": 12.991, |
| "eval_samples_per_second": 506.812, |
| "eval_steps_per_second": 7.929, |
| "step": 11760 |
| }, |
| { |
| "epoch": 1.3502351726511415, |
| "eval_loss": 0.3966831564903259, |
| "eval_runtime": 13.0658, |
| "eval_samples_per_second": 503.91, |
| "eval_steps_per_second": 7.883, |
| "step": 11770 |
| }, |
| { |
| "epoch": 1.351382356315246, |
| "eval_loss": 0.3966105878353119, |
| "eval_runtime": 13.0945, |
| "eval_samples_per_second": 502.808, |
| "eval_steps_per_second": 7.866, |
| "step": 11780 |
| }, |
| { |
| "epoch": 1.3525295399793507, |
| "eval_loss": 0.3964127004146576, |
| "eval_runtime": 12.9708, |
| "eval_samples_per_second": 507.602, |
| "eval_steps_per_second": 7.941, |
| "step": 11790 |
| }, |
| { |
| "epoch": 1.3536767236434553, |
| "grad_norm": 2.7533326148986816, |
| "learning_rate": 6.097726789887402e-06, |
| "loss": 0.7091, |
| "step": 11800 |
| }, |
| { |
| "epoch": 1.3536767236434553, |
| "eval_loss": 0.3965069055557251, |
| "eval_runtime": 13.0934, |
| "eval_samples_per_second": 502.85, |
| "eval_steps_per_second": 7.867, |
| "step": 11800 |
| }, |
| { |
| "epoch": 1.3548239073075599, |
| "eval_loss": 0.39640066027641296, |
| "eval_runtime": 12.9784, |
| "eval_samples_per_second": 507.304, |
| "eval_steps_per_second": 7.936, |
| "step": 11810 |
| }, |
| { |
| "epoch": 1.3559710909716647, |
| "eval_loss": 0.3964119255542755, |
| "eval_runtime": 12.9731, |
| "eval_samples_per_second": 507.513, |
| "eval_steps_per_second": 7.94, |
| "step": 11820 |
| }, |
| { |
| "epoch": 1.357118274635769, |
| "eval_loss": 0.3963049054145813, |
| "eval_runtime": 13.0434, |
| "eval_samples_per_second": 504.777, |
| "eval_steps_per_second": 7.897, |
| "step": 11830 |
| }, |
| { |
| "epoch": 1.3582654582998739, |
| "eval_loss": 0.39616140723228455, |
| "eval_runtime": 12.9606, |
| "eval_samples_per_second": 508.0, |
| "eval_steps_per_second": 7.947, |
| "step": 11840 |
| }, |
| { |
| "epoch": 1.3594126419639785, |
| "eval_loss": 0.39606061577796936, |
| "eval_runtime": 12.9599, |
| "eval_samples_per_second": 508.027, |
| "eval_steps_per_second": 7.948, |
| "step": 11850 |
| }, |
| { |
| "epoch": 1.360559825628083, |
| "eval_loss": 0.39563143253326416, |
| "eval_runtime": 12.9693, |
| "eval_samples_per_second": 507.66, |
| "eval_steps_per_second": 7.942, |
| "step": 11860 |
| }, |
| { |
| "epoch": 1.3617070092921877, |
| "eval_loss": 0.3956204652786255, |
| "eval_runtime": 13.0053, |
| "eval_samples_per_second": 506.254, |
| "eval_steps_per_second": 7.92, |
| "step": 11870 |
| }, |
| { |
| "epoch": 1.3628541929562923, |
| "eval_loss": 0.3960905373096466, |
| "eval_runtime": 12.9588, |
| "eval_samples_per_second": 508.072, |
| "eval_steps_per_second": 7.948, |
| "step": 11880 |
| }, |
| { |
| "epoch": 1.3640013766203969, |
| "eval_loss": 0.39633554220199585, |
| "eval_runtime": 12.9603, |
| "eval_samples_per_second": 508.014, |
| "eval_steps_per_second": 7.947, |
| "step": 11890 |
| }, |
| { |
| "epoch": 1.3651485602845015, |
| "grad_norm": 3.27715802192688, |
| "learning_rate": 6.055236881240706e-06, |
| "loss": 0.6977, |
| "step": 11900 |
| }, |
| { |
| "epoch": 1.3651485602845015, |
| "eval_loss": 0.3961818814277649, |
| "eval_runtime": 12.9653, |
| "eval_samples_per_second": 507.816, |
| "eval_steps_per_second": 7.944, |
| "step": 11900 |
| }, |
| { |
| "epoch": 1.366295743948606, |
| "eval_loss": 0.3958257734775543, |
| "eval_runtime": 12.9637, |
| "eval_samples_per_second": 507.879, |
| "eval_steps_per_second": 7.945, |
| "step": 11910 |
| }, |
| { |
| "epoch": 1.3674429276127107, |
| "eval_loss": 0.3959980905056, |
| "eval_runtime": 12.9635, |
| "eval_samples_per_second": 507.888, |
| "eval_steps_per_second": 7.945, |
| "step": 11920 |
| }, |
| { |
| "epoch": 1.3685901112768155, |
| "eval_loss": 0.3963344991207123, |
| "eval_runtime": 12.9838, |
| "eval_samples_per_second": 507.093, |
| "eval_steps_per_second": 7.933, |
| "step": 11930 |
| }, |
| { |
| "epoch": 1.3697372949409201, |
| "eval_loss": 0.39638736844062805, |
| "eval_runtime": 13.0044, |
| "eval_samples_per_second": 506.291, |
| "eval_steps_per_second": 7.92, |
| "step": 11940 |
| }, |
| { |
| "epoch": 1.3708844786050247, |
| "eval_loss": 0.3961441218852997, |
| "eval_runtime": 12.9756, |
| "eval_samples_per_second": 507.414, |
| "eval_steps_per_second": 7.938, |
| "step": 11950 |
| }, |
| { |
| "epoch": 1.3720316622691293, |
| "eval_loss": 0.39601022005081177, |
| "eval_runtime": 12.9612, |
| "eval_samples_per_second": 507.978, |
| "eval_steps_per_second": 7.947, |
| "step": 11960 |
| }, |
| { |
| "epoch": 1.373178845933234, |
| "eval_loss": 0.395821750164032, |
| "eval_runtime": 13.0916, |
| "eval_samples_per_second": 502.918, |
| "eval_steps_per_second": 7.868, |
| "step": 11970 |
| }, |
| { |
| "epoch": 1.3743260295973385, |
| "eval_loss": 0.3953630030155182, |
| "eval_runtime": 12.9615, |
| "eval_samples_per_second": 507.966, |
| "eval_steps_per_second": 7.947, |
| "step": 11980 |
| }, |
| { |
| "epoch": 1.3754732132614431, |
| "eval_loss": 0.3947514295578003, |
| "eval_runtime": 13.0081, |
| "eval_samples_per_second": 506.146, |
| "eval_steps_per_second": 7.918, |
| "step": 11990 |
| }, |
| { |
| "epoch": 1.3766203969255477, |
| "grad_norm": 2.9197328090667725, |
| "learning_rate": 6.012746972594009e-06, |
| "loss": 0.7003, |
| "step": 12000 |
| }, |
| { |
| "epoch": 1.3766203969255477, |
| "eval_loss": 0.39438483119010925, |
| "eval_runtime": 12.9743, |
| "eval_samples_per_second": 507.463, |
| "eval_steps_per_second": 7.939, |
| "step": 12000 |
| }, |
| { |
| "epoch": 1.3777675805896523, |
| "eval_loss": 0.3940276801586151, |
| "eval_runtime": 12.9612, |
| "eval_samples_per_second": 507.978, |
| "eval_steps_per_second": 7.947, |
| "step": 12010 |
| }, |
| { |
| "epoch": 1.3789147642537571, |
| "eval_loss": 0.39397919178009033, |
| "eval_runtime": 12.9582, |
| "eval_samples_per_second": 508.094, |
| "eval_steps_per_second": 7.949, |
| "step": 12020 |
| }, |
| { |
| "epoch": 1.3800619479178615, |
| "eval_loss": 0.39384689927101135, |
| "eval_runtime": 12.9628, |
| "eval_samples_per_second": 507.915, |
| "eval_steps_per_second": 7.946, |
| "step": 12030 |
| }, |
| { |
| "epoch": 1.3812091315819663, |
| "eval_loss": 0.39393553137779236, |
| "eval_runtime": 13.6524, |
| "eval_samples_per_second": 482.259, |
| "eval_steps_per_second": 7.544, |
| "step": 12040 |
| }, |
| { |
| "epoch": 1.382356315246071, |
| "eval_loss": 0.39427098631858826, |
| "eval_runtime": 12.9859, |
| "eval_samples_per_second": 507.01, |
| "eval_steps_per_second": 7.932, |
| "step": 12050 |
| }, |
| { |
| "epoch": 1.3835034989101755, |
| "eval_loss": 0.39460989832878113, |
| "eval_runtime": 12.9534, |
| "eval_samples_per_second": 508.282, |
| "eval_steps_per_second": 7.952, |
| "step": 12060 |
| }, |
| { |
| "epoch": 1.3846506825742801, |
| "eval_loss": 0.3946506977081299, |
| "eval_runtime": 12.9612, |
| "eval_samples_per_second": 507.977, |
| "eval_steps_per_second": 7.947, |
| "step": 12070 |
| }, |
| { |
| "epoch": 1.3857978662383847, |
| "eval_loss": 0.3943174481391907, |
| "eval_runtime": 12.9601, |
| "eval_samples_per_second": 508.019, |
| "eval_steps_per_second": 7.947, |
| "step": 12080 |
| }, |
| { |
| "epoch": 1.3869450499024893, |
| "eval_loss": 0.3940339982509613, |
| "eval_runtime": 12.9587, |
| "eval_samples_per_second": 508.074, |
| "eval_steps_per_second": 7.948, |
| "step": 12090 |
| }, |
| { |
| "epoch": 1.388092233566594, |
| "grad_norm": 2.296860933303833, |
| "learning_rate": 5.970257063947313e-06, |
| "loss": 0.679, |
| "step": 12100 |
| }, |
| { |
| "epoch": 1.388092233566594, |
| "eval_loss": 0.39428094029426575, |
| "eval_runtime": 13.0286, |
| "eval_samples_per_second": 505.35, |
| "eval_steps_per_second": 7.906, |
| "step": 12100 |
| }, |
| { |
| "epoch": 1.3892394172306988, |
| "eval_loss": 0.39449721574783325, |
| "eval_runtime": 13.0214, |
| "eval_samples_per_second": 505.629, |
| "eval_steps_per_second": 7.91, |
| "step": 12110 |
| }, |
| { |
| "epoch": 1.3903866008948031, |
| "eval_loss": 0.39464056491851807, |
| "eval_runtime": 13.0558, |
| "eval_samples_per_second": 504.295, |
| "eval_steps_per_second": 7.889, |
| "step": 12120 |
| }, |
| { |
| "epoch": 1.391533784558908, |
| "eval_loss": 0.39441561698913574, |
| "eval_runtime": 12.9688, |
| "eval_samples_per_second": 507.679, |
| "eval_steps_per_second": 7.942, |
| "step": 12130 |
| }, |
| { |
| "epoch": 1.3926809682230126, |
| "eval_loss": 0.3941043019294739, |
| "eval_runtime": 12.9725, |
| "eval_samples_per_second": 507.535, |
| "eval_steps_per_second": 7.94, |
| "step": 12140 |
| }, |
| { |
| "epoch": 1.3938281518871172, |
| "eval_loss": 0.3941352665424347, |
| "eval_runtime": 13.0119, |
| "eval_samples_per_second": 505.998, |
| "eval_steps_per_second": 7.916, |
| "step": 12150 |
| }, |
| { |
| "epoch": 1.3949753355512218, |
| "eval_loss": 0.3940858244895935, |
| "eval_runtime": 12.9717, |
| "eval_samples_per_second": 507.567, |
| "eval_steps_per_second": 7.94, |
| "step": 12160 |
| }, |
| { |
| "epoch": 1.3961225192153264, |
| "eval_loss": 0.3939216732978821, |
| "eval_runtime": 12.9712, |
| "eval_samples_per_second": 507.585, |
| "eval_steps_per_second": 7.941, |
| "step": 12170 |
| }, |
| { |
| "epoch": 1.397269702879431, |
| "eval_loss": 0.3939203917980194, |
| "eval_runtime": 13.1042, |
| "eval_samples_per_second": 502.433, |
| "eval_steps_per_second": 7.86, |
| "step": 12180 |
| }, |
| { |
| "epoch": 1.3984168865435356, |
| "eval_loss": 0.39389124512672424, |
| "eval_runtime": 13.0688, |
| "eval_samples_per_second": 503.795, |
| "eval_steps_per_second": 7.881, |
| "step": 12190 |
| }, |
| { |
| "epoch": 1.3995640702076402, |
| "grad_norm": 4.0118255615234375, |
| "learning_rate": 5.927767155300617e-06, |
| "loss": 0.692, |
| "step": 12200 |
| }, |
| { |
| "epoch": 1.3995640702076402, |
| "eval_loss": 0.3937877118587494, |
| "eval_runtime": 13.1601, |
| "eval_samples_per_second": 500.299, |
| "eval_steps_per_second": 7.827, |
| "step": 12200 |
| }, |
| { |
| "epoch": 1.4007112538717448, |
| "eval_loss": 0.3937278687953949, |
| "eval_runtime": 13.0991, |
| "eval_samples_per_second": 502.63, |
| "eval_steps_per_second": 7.863, |
| "step": 12210 |
| }, |
| { |
| "epoch": 1.4018584375358496, |
| "eval_loss": 0.393211305141449, |
| "eval_runtime": 12.9743, |
| "eval_samples_per_second": 507.464, |
| "eval_steps_per_second": 7.939, |
| "step": 12220 |
| }, |
| { |
| "epoch": 1.403005621199954, |
| "eval_loss": 0.3928295969963074, |
| "eval_runtime": 13.0835, |
| "eval_samples_per_second": 503.229, |
| "eval_steps_per_second": 7.873, |
| "step": 12230 |
| }, |
| { |
| "epoch": 1.4041528048640588, |
| "eval_loss": 0.39248955249786377, |
| "eval_runtime": 13.0865, |
| "eval_samples_per_second": 503.113, |
| "eval_steps_per_second": 7.871, |
| "step": 12240 |
| }, |
| { |
| "epoch": 1.4052999885281634, |
| "eval_loss": 0.39223751425743103, |
| "eval_runtime": 13.0006, |
| "eval_samples_per_second": 506.437, |
| "eval_steps_per_second": 7.923, |
| "step": 12250 |
| }, |
| { |
| "epoch": 1.406447172192268, |
| "eval_loss": 0.3923845589160919, |
| "eval_runtime": 12.9673, |
| "eval_samples_per_second": 507.738, |
| "eval_steps_per_second": 7.943, |
| "step": 12260 |
| }, |
| { |
| "epoch": 1.4075943558563726, |
| "eval_loss": 0.39231061935424805, |
| "eval_runtime": 12.9641, |
| "eval_samples_per_second": 507.863, |
| "eval_steps_per_second": 7.945, |
| "step": 12270 |
| }, |
| { |
| "epoch": 1.4087415395204772, |
| "eval_loss": 0.39256516098976135, |
| "eval_runtime": 13.0105, |
| "eval_samples_per_second": 506.052, |
| "eval_steps_per_second": 7.917, |
| "step": 12280 |
| }, |
| { |
| "epoch": 1.4098887231845818, |
| "eval_loss": 0.3923798203468323, |
| "eval_runtime": 12.9761, |
| "eval_samples_per_second": 507.394, |
| "eval_steps_per_second": 7.938, |
| "step": 12290 |
| }, |
| { |
| "epoch": 1.4110359068486864, |
| "grad_norm": 3.407397747039795, |
| "learning_rate": 5.88527724665392e-06, |
| "loss": 0.6677, |
| "step": 12300 |
| }, |
| { |
| "epoch": 1.4110359068486864, |
| "eval_loss": 0.3925481140613556, |
| "eval_runtime": 13.023, |
| "eval_samples_per_second": 505.567, |
| "eval_steps_per_second": 7.909, |
| "step": 12300 |
| }, |
| { |
| "epoch": 1.4121830905127912, |
| "eval_loss": 0.3925573527812958, |
| "eval_runtime": 13.0503, |
| "eval_samples_per_second": 504.509, |
| "eval_steps_per_second": 7.893, |
| "step": 12310 |
| }, |
| { |
| "epoch": 1.4133302741768956, |
| "eval_loss": 0.39271554350852966, |
| "eval_runtime": 12.978, |
| "eval_samples_per_second": 507.319, |
| "eval_steps_per_second": 7.936, |
| "step": 12320 |
| }, |
| { |
| "epoch": 1.4144774578410004, |
| "eval_loss": 0.39276665449142456, |
| "eval_runtime": 12.9774, |
| "eval_samples_per_second": 507.345, |
| "eval_steps_per_second": 7.937, |
| "step": 12330 |
| }, |
| { |
| "epoch": 1.415624641505105, |
| "eval_loss": 0.3927762806415558, |
| "eval_runtime": 12.973, |
| "eval_samples_per_second": 507.517, |
| "eval_steps_per_second": 7.94, |
| "step": 12340 |
| }, |
| { |
| "epoch": 1.4167718251692096, |
| "eval_loss": 0.3928840756416321, |
| "eval_runtime": 12.9618, |
| "eval_samples_per_second": 507.954, |
| "eval_steps_per_second": 7.946, |
| "step": 12350 |
| }, |
| { |
| "epoch": 1.4179190088333142, |
| "eval_loss": 0.3932951092720032, |
| "eval_runtime": 12.9713, |
| "eval_samples_per_second": 507.583, |
| "eval_steps_per_second": 7.941, |
| "step": 12360 |
| }, |
| { |
| "epoch": 1.4190661924974188, |
| "eval_loss": 0.39335715770721436, |
| "eval_runtime": 12.9806, |
| "eval_samples_per_second": 507.217, |
| "eval_steps_per_second": 7.935, |
| "step": 12370 |
| }, |
| { |
| "epoch": 1.4202133761615234, |
| "eval_loss": 0.393307626247406, |
| "eval_runtime": 12.9654, |
| "eval_samples_per_second": 507.811, |
| "eval_steps_per_second": 7.944, |
| "step": 12380 |
| }, |
| { |
| "epoch": 1.421360559825628, |
| "eval_loss": 0.3932768702507019, |
| "eval_runtime": 12.9695, |
| "eval_samples_per_second": 507.653, |
| "eval_steps_per_second": 7.942, |
| "step": 12390 |
| }, |
| { |
| "epoch": 1.4225077434897326, |
| "grad_norm": 2.3648879528045654, |
| "learning_rate": 5.842787338007224e-06, |
| "loss": 0.6892, |
| "step": 12400 |
| }, |
| { |
| "epoch": 1.4225077434897326, |
| "eval_loss": 0.39297202229499817, |
| "eval_runtime": 13.115, |
| "eval_samples_per_second": 502.02, |
| "eval_steps_per_second": 7.854, |
| "step": 12400 |
| }, |
| { |
| "epoch": 1.4236549271538372, |
| "eval_loss": 0.39275166392326355, |
| "eval_runtime": 13.0524, |
| "eval_samples_per_second": 504.43, |
| "eval_steps_per_second": 7.891, |
| "step": 12410 |
| }, |
| { |
| "epoch": 1.424802110817942, |
| "eval_loss": 0.3927864730358124, |
| "eval_runtime": 13.0275, |
| "eval_samples_per_second": 505.393, |
| "eval_steps_per_second": 7.906, |
| "step": 12420 |
| }, |
| { |
| "epoch": 1.4259492944820467, |
| "eval_loss": 0.3926820158958435, |
| "eval_runtime": 12.9648, |
| "eval_samples_per_second": 507.836, |
| "eval_steps_per_second": 7.945, |
| "step": 12430 |
| }, |
| { |
| "epoch": 1.4270964781461513, |
| "eval_loss": 0.39270150661468506, |
| "eval_runtime": 13.1021, |
| "eval_samples_per_second": 502.513, |
| "eval_steps_per_second": 7.861, |
| "step": 12440 |
| }, |
| { |
| "epoch": 1.4282436618102559, |
| "eval_loss": 0.39239150285720825, |
| "eval_runtime": 13.0233, |
| "eval_samples_per_second": 505.554, |
| "eval_steps_per_second": 7.909, |
| "step": 12450 |
| }, |
| { |
| "epoch": 1.4293908454743605, |
| "eval_loss": 0.39238443970680237, |
| "eval_runtime": 12.9704, |
| "eval_samples_per_second": 507.617, |
| "eval_steps_per_second": 7.941, |
| "step": 12460 |
| }, |
| { |
| "epoch": 1.430538029138465, |
| "eval_loss": 0.39221516251564026, |
| "eval_runtime": 12.9619, |
| "eval_samples_per_second": 507.951, |
| "eval_steps_per_second": 7.946, |
| "step": 12470 |
| }, |
| { |
| "epoch": 1.4316852128025697, |
| "eval_loss": 0.39195549488067627, |
| "eval_runtime": 13.1302, |
| "eval_samples_per_second": 501.439, |
| "eval_steps_per_second": 7.845, |
| "step": 12480 |
| }, |
| { |
| "epoch": 1.4328323964666743, |
| "eval_loss": 0.391914963722229, |
| "eval_runtime": 12.9526, |
| "eval_samples_per_second": 508.313, |
| "eval_steps_per_second": 7.952, |
| "step": 12490 |
| }, |
| { |
| "epoch": 1.4339795801307789, |
| "grad_norm": 3.077915668487549, |
| "learning_rate": 5.800297429360528e-06, |
| "loss": 0.7016, |
| "step": 12500 |
| }, |
| { |
| "epoch": 1.4339795801307789, |
| "eval_loss": 0.39166682958602905, |
| "eval_runtime": 12.9728, |
| "eval_samples_per_second": 507.525, |
| "eval_steps_per_second": 7.94, |
| "step": 12500 |
| }, |
| { |
| "epoch": 1.4351267637948837, |
| "eval_loss": 0.39142662286758423, |
| "eval_runtime": 12.973, |
| "eval_samples_per_second": 507.516, |
| "eval_steps_per_second": 7.94, |
| "step": 12510 |
| }, |
| { |
| "epoch": 1.436273947458988, |
| "eval_loss": 0.3911990821361542, |
| "eval_runtime": 12.993, |
| "eval_samples_per_second": 506.735, |
| "eval_steps_per_second": 7.927, |
| "step": 12520 |
| }, |
| { |
| "epoch": 1.4374211311230929, |
| "eval_loss": 0.39141345024108887, |
| "eval_runtime": 12.9679, |
| "eval_samples_per_second": 507.717, |
| "eval_steps_per_second": 7.943, |
| "step": 12530 |
| }, |
| { |
| "epoch": 1.4385683147871975, |
| "eval_loss": 0.39145925641059875, |
| "eval_runtime": 13.014, |
| "eval_samples_per_second": 505.917, |
| "eval_steps_per_second": 7.915, |
| "step": 12540 |
| }, |
| { |
| "epoch": 1.439715498451302, |
| "eval_loss": 0.39148348569869995, |
| "eval_runtime": 13.0497, |
| "eval_samples_per_second": 504.533, |
| "eval_steps_per_second": 7.893, |
| "step": 12550 |
| }, |
| { |
| "epoch": 1.4408626821154067, |
| "eval_loss": 0.39154401421546936, |
| "eval_runtime": 13.2068, |
| "eval_samples_per_second": 498.532, |
| "eval_steps_per_second": 7.799, |
| "step": 12560 |
| }, |
| { |
| "epoch": 1.4420098657795113, |
| "eval_loss": 0.3912140727043152, |
| "eval_runtime": 12.9588, |
| "eval_samples_per_second": 508.07, |
| "eval_steps_per_second": 7.948, |
| "step": 12570 |
| }, |
| { |
| "epoch": 1.443157049443616, |
| "eval_loss": 0.3910275101661682, |
| "eval_runtime": 12.9792, |
| "eval_samples_per_second": 507.274, |
| "eval_steps_per_second": 7.936, |
| "step": 12580 |
| }, |
| { |
| "epoch": 1.4443042331077205, |
| "eval_loss": 0.39096659421920776, |
| "eval_runtime": 12.9619, |
| "eval_samples_per_second": 507.949, |
| "eval_steps_per_second": 7.946, |
| "step": 12590 |
| }, |
| { |
| "epoch": 1.445451416771825, |
| "grad_norm": 3.3011014461517334, |
| "learning_rate": 5.757807520713831e-06, |
| "loss": 0.68, |
| "step": 12600 |
| }, |
| { |
| "epoch": 1.445451416771825, |
| "eval_loss": 0.3906503915786743, |
| "eval_runtime": 12.9606, |
| "eval_samples_per_second": 508.003, |
| "eval_steps_per_second": 7.947, |
| "step": 12600 |
| }, |
| { |
| "epoch": 1.4465986004359297, |
| "eval_loss": 0.39063605666160583, |
| "eval_runtime": 12.9641, |
| "eval_samples_per_second": 507.863, |
| "eval_steps_per_second": 7.945, |
| "step": 12610 |
| }, |
| { |
| "epoch": 1.4477457841000345, |
| "eval_loss": 0.39036861062049866, |
| "eval_runtime": 12.9653, |
| "eval_samples_per_second": 507.818, |
| "eval_steps_per_second": 7.944, |
| "step": 12620 |
| }, |
| { |
| "epoch": 1.4488929677641391, |
| "eval_loss": 0.3902880847454071, |
| "eval_runtime": 12.9617, |
| "eval_samples_per_second": 507.958, |
| "eval_steps_per_second": 7.946, |
| "step": 12630 |
| }, |
| { |
| "epoch": 1.4500401514282437, |
| "eval_loss": 0.39019864797592163, |
| "eval_runtime": 12.9715, |
| "eval_samples_per_second": 507.576, |
| "eval_steps_per_second": 7.941, |
| "step": 12640 |
| }, |
| { |
| "epoch": 1.4511873350923483, |
| "eval_loss": 0.38993674516677856, |
| "eval_runtime": 13.0977, |
| "eval_samples_per_second": 502.682, |
| "eval_steps_per_second": 7.864, |
| "step": 12650 |
| }, |
| { |
| "epoch": 1.452334518756453, |
| "eval_loss": 0.3897811770439148, |
| "eval_runtime": 13.08, |
| "eval_samples_per_second": 503.364, |
| "eval_steps_per_second": 7.875, |
| "step": 12660 |
| }, |
| { |
| "epoch": 1.4534817024205575, |
| "eval_loss": 0.38978081941604614, |
| "eval_runtime": 12.9828, |
| "eval_samples_per_second": 507.133, |
| "eval_steps_per_second": 7.934, |
| "step": 12670 |
| }, |
| { |
| "epoch": 1.4546288860846621, |
| "eval_loss": 0.3897199332714081, |
| "eval_runtime": 13.0044, |
| "eval_samples_per_second": 506.289, |
| "eval_steps_per_second": 7.92, |
| "step": 12680 |
| }, |
| { |
| "epoch": 1.4557760697487667, |
| "eval_loss": 0.38960719108581543, |
| "eval_runtime": 13.0553, |
| "eval_samples_per_second": 504.315, |
| "eval_steps_per_second": 7.89, |
| "step": 12690 |
| }, |
| { |
| "epoch": 1.4569232534128713, |
| "grad_norm": 2.6866369247436523, |
| "learning_rate": 5.715317612067135e-06, |
| "loss": 0.681, |
| "step": 12700 |
| }, |
| { |
| "epoch": 1.4569232534128713, |
| "eval_loss": 0.389370322227478, |
| "eval_runtime": 12.9886, |
| "eval_samples_per_second": 506.905, |
| "eval_steps_per_second": 7.93, |
| "step": 12700 |
| }, |
| { |
| "epoch": 1.4580704370769761, |
| "eval_loss": 0.38921910524368286, |
| "eval_runtime": 12.9647, |
| "eval_samples_per_second": 507.84, |
| "eval_steps_per_second": 7.945, |
| "step": 12710 |
| }, |
| { |
| "epoch": 1.4592176207410805, |
| "eval_loss": 0.38923558592796326, |
| "eval_runtime": 12.9732, |
| "eval_samples_per_second": 507.509, |
| "eval_steps_per_second": 7.939, |
| "step": 12720 |
| }, |
| { |
| "epoch": 1.4603648044051853, |
| "eval_loss": 0.38930267095565796, |
| "eval_runtime": 13.0784, |
| "eval_samples_per_second": 503.425, |
| "eval_steps_per_second": 7.876, |
| "step": 12730 |
| }, |
| { |
| "epoch": 1.46151198806929, |
| "eval_loss": 0.38959333300590515, |
| "eval_runtime": 12.9719, |
| "eval_samples_per_second": 507.558, |
| "eval_steps_per_second": 7.94, |
| "step": 12740 |
| }, |
| { |
| "epoch": 1.4626591717333945, |
| "eval_loss": 0.3897537589073181, |
| "eval_runtime": 13.0613, |
| "eval_samples_per_second": 504.086, |
| "eval_steps_per_second": 7.886, |
| "step": 12750 |
| }, |
| { |
| "epoch": 1.4638063553974991, |
| "eval_loss": 0.3899790644645691, |
| "eval_runtime": 12.958, |
| "eval_samples_per_second": 508.103, |
| "eval_steps_per_second": 7.949, |
| "step": 12760 |
| }, |
| { |
| "epoch": 1.4649535390616037, |
| "eval_loss": 0.38985395431518555, |
| "eval_runtime": 13.0397, |
| "eval_samples_per_second": 504.918, |
| "eval_steps_per_second": 7.899, |
| "step": 12770 |
| }, |
| { |
| "epoch": 1.4661007227257083, |
| "eval_loss": 0.38976746797561646, |
| "eval_runtime": 12.9572, |
| "eval_samples_per_second": 508.133, |
| "eval_steps_per_second": 7.949, |
| "step": 12780 |
| }, |
| { |
| "epoch": 1.467247906389813, |
| "eval_loss": 0.3899408280849457, |
| "eval_runtime": 12.963, |
| "eval_samples_per_second": 507.906, |
| "eval_steps_per_second": 7.946, |
| "step": 12790 |
| }, |
| { |
| "epoch": 1.4683950900539176, |
| "grad_norm": 10.585307121276855, |
| "learning_rate": 5.672827703420438e-06, |
| "loss": 0.6816, |
| "step": 12800 |
| }, |
| { |
| "epoch": 1.4683950900539176, |
| "eval_loss": 0.3900618851184845, |
| "eval_runtime": 13.0538, |
| "eval_samples_per_second": 504.375, |
| "eval_steps_per_second": 7.89, |
| "step": 12800 |
| }, |
| { |
| "epoch": 1.4695422737180222, |
| "eval_loss": 0.3901263475418091, |
| "eval_runtime": 12.9484, |
| "eval_samples_per_second": 508.479, |
| "eval_steps_per_second": 7.955, |
| "step": 12810 |
| }, |
| { |
| "epoch": 1.470689457382127, |
| "eval_loss": 0.3900851905345917, |
| "eval_runtime": 12.9637, |
| "eval_samples_per_second": 507.879, |
| "eval_steps_per_second": 7.945, |
| "step": 12820 |
| }, |
| { |
| "epoch": 1.4718366410462316, |
| "eval_loss": 0.3898305892944336, |
| "eval_runtime": 13.0968, |
| "eval_samples_per_second": 502.719, |
| "eval_steps_per_second": 7.865, |
| "step": 12830 |
| }, |
| { |
| "epoch": 1.4729838247103362, |
| "eval_loss": 0.3896641731262207, |
| "eval_runtime": 13.0016, |
| "eval_samples_per_second": 506.4, |
| "eval_steps_per_second": 7.922, |
| "step": 12840 |
| }, |
| { |
| "epoch": 1.4741310083744408, |
| "eval_loss": 0.3897162675857544, |
| "eval_runtime": 12.9628, |
| "eval_samples_per_second": 507.914, |
| "eval_steps_per_second": 7.946, |
| "step": 12850 |
| }, |
| { |
| "epoch": 1.4752781920385454, |
| "eval_loss": 0.38947173953056335, |
| "eval_runtime": 13.0682, |
| "eval_samples_per_second": 503.819, |
| "eval_steps_per_second": 7.882, |
| "step": 12860 |
| }, |
| { |
| "epoch": 1.47642537570265, |
| "eval_loss": 0.3895629346370697, |
| "eval_runtime": 13.0868, |
| "eval_samples_per_second": 503.1, |
| "eval_steps_per_second": 7.87, |
| "step": 12870 |
| }, |
| { |
| "epoch": 1.4775725593667546, |
| "eval_loss": 0.3895267844200134, |
| "eval_runtime": 12.9656, |
| "eval_samples_per_second": 507.804, |
| "eval_steps_per_second": 7.944, |
| "step": 12880 |
| }, |
| { |
| "epoch": 1.4787197430308592, |
| "eval_loss": 0.38955989480018616, |
| "eval_runtime": 13.0895, |
| "eval_samples_per_second": 502.997, |
| "eval_steps_per_second": 7.869, |
| "step": 12890 |
| }, |
| { |
| "epoch": 1.4798669266949638, |
| "grad_norm": 3.164057970046997, |
| "learning_rate": 5.6303377947737415e-06, |
| "loss": 0.6635, |
| "step": 12900 |
| }, |
| { |
| "epoch": 1.4798669266949638, |
| "eval_loss": 0.3896629214286804, |
| "eval_runtime": 12.9639, |
| "eval_samples_per_second": 507.871, |
| "eval_steps_per_second": 7.945, |
| "step": 12900 |
| }, |
| { |
| "epoch": 1.4810141103590686, |
| "eval_loss": 0.38974663615226746, |
| "eval_runtime": 12.9726, |
| "eval_samples_per_second": 507.533, |
| "eval_steps_per_second": 7.94, |
| "step": 12910 |
| }, |
| { |
| "epoch": 1.482161294023173, |
| "eval_loss": 0.3899209797382355, |
| "eval_runtime": 13.092, |
| "eval_samples_per_second": 502.903, |
| "eval_steps_per_second": 7.867, |
| "step": 12920 |
| }, |
| { |
| "epoch": 1.4833084776872778, |
| "eval_loss": 0.3899885416030884, |
| "eval_runtime": 12.9663, |
| "eval_samples_per_second": 507.776, |
| "eval_steps_per_second": 7.944, |
| "step": 12930 |
| }, |
| { |
| "epoch": 1.4844556613513824, |
| "eval_loss": 0.38974055647850037, |
| "eval_runtime": 12.9709, |
| "eval_samples_per_second": 507.598, |
| "eval_steps_per_second": 7.941, |
| "step": 12940 |
| }, |
| { |
| "epoch": 1.485602845015487, |
| "eval_loss": 0.38984495401382446, |
| "eval_runtime": 13.0302, |
| "eval_samples_per_second": 505.286, |
| "eval_steps_per_second": 7.905, |
| "step": 12950 |
| }, |
| { |
| "epoch": 1.4867500286795916, |
| "eval_loss": 0.38993626832962036, |
| "eval_runtime": 13.0827, |
| "eval_samples_per_second": 503.258, |
| "eval_steps_per_second": 7.873, |
| "step": 12960 |
| }, |
| { |
| "epoch": 1.4878972123436962, |
| "eval_loss": 0.38980451226234436, |
| "eval_runtime": 13.1684, |
| "eval_samples_per_second": 499.985, |
| "eval_steps_per_second": 7.822, |
| "step": 12970 |
| }, |
| { |
| "epoch": 1.4890443960078008, |
| "eval_loss": 0.38976091146469116, |
| "eval_runtime": 13.1601, |
| "eval_samples_per_second": 500.299, |
| "eval_steps_per_second": 7.827, |
| "step": 12980 |
| }, |
| { |
| "epoch": 1.4901915796719054, |
| "eval_loss": 0.3894117474555969, |
| "eval_runtime": 13.1572, |
| "eval_samples_per_second": 500.412, |
| "eval_steps_per_second": 7.828, |
| "step": 12990 |
| }, |
| { |
| "epoch": 1.4913387633360102, |
| "grad_norm": 3.515048027038574, |
| "learning_rate": 5.587847886127046e-06, |
| "loss": 0.6698, |
| "step": 13000 |
| }, |
| { |
| "epoch": 1.4913387633360102, |
| "eval_loss": 0.38920843601226807, |
| "eval_runtime": 13.096, |
| "eval_samples_per_second": 502.749, |
| "eval_steps_per_second": 7.865, |
| "step": 13000 |
| }, |
| { |
| "epoch": 1.4924859470001146, |
| "eval_loss": 0.3892129361629486, |
| "eval_runtime": 12.96, |
| "eval_samples_per_second": 508.026, |
| "eval_steps_per_second": 7.948, |
| "step": 13010 |
| }, |
| { |
| "epoch": 1.4936331306642194, |
| "eval_loss": 0.3893972635269165, |
| "eval_runtime": 12.9665, |
| "eval_samples_per_second": 507.772, |
| "eval_steps_per_second": 7.944, |
| "step": 13020 |
| }, |
| { |
| "epoch": 1.494780314328324, |
| "eval_loss": 0.3892994225025177, |
| "eval_runtime": 13.0908, |
| "eval_samples_per_second": 502.947, |
| "eval_steps_per_second": 7.868, |
| "step": 13030 |
| }, |
| { |
| "epoch": 1.4959274979924286, |
| "eval_loss": 0.3893571197986603, |
| "eval_runtime": 13.0295, |
| "eval_samples_per_second": 505.315, |
| "eval_steps_per_second": 7.905, |
| "step": 13040 |
| }, |
| { |
| "epoch": 1.4970746816565332, |
| "eval_loss": 0.3892894983291626, |
| "eval_runtime": 13.1028, |
| "eval_samples_per_second": 502.488, |
| "eval_steps_per_second": 7.861, |
| "step": 13050 |
| }, |
| { |
| "epoch": 1.4982218653206378, |
| "eval_loss": 0.3894209861755371, |
| "eval_runtime": 12.9701, |
| "eval_samples_per_second": 507.627, |
| "eval_steps_per_second": 7.941, |
| "step": 13060 |
| }, |
| { |
| "epoch": 1.4993690489847424, |
| "eval_loss": 0.38930463790893555, |
| "eval_runtime": 12.9825, |
| "eval_samples_per_second": 507.146, |
| "eval_steps_per_second": 7.934, |
| "step": 13070 |
| }, |
| { |
| "epoch": 1.500516232648847, |
| "eval_loss": 0.3894728720188141, |
| "eval_runtime": 12.9678, |
| "eval_samples_per_second": 507.719, |
| "eval_steps_per_second": 7.943, |
| "step": 13080 |
| }, |
| { |
| "epoch": 1.5016634163129519, |
| "eval_loss": 0.3894925117492676, |
| "eval_runtime": 12.9709, |
| "eval_samples_per_second": 507.597, |
| "eval_steps_per_second": 7.941, |
| "step": 13090 |
| }, |
| { |
| "epoch": 1.5028105999770562, |
| "grad_norm": 2.5569674968719482, |
| "learning_rate": 5.545357977480349e-06, |
| "loss": 0.6757, |
| "step": 13100 |
| }, |
| { |
| "epoch": 1.5028105999770562, |
| "eval_loss": 0.38975808024406433, |
| "eval_runtime": 12.9663, |
| "eval_samples_per_second": 507.778, |
| "eval_steps_per_second": 7.944, |
| "step": 13100 |
| }, |
| { |
| "epoch": 1.503957783641161, |
| "eval_loss": 0.38975581526756287, |
| "eval_runtime": 12.9764, |
| "eval_samples_per_second": 507.382, |
| "eval_steps_per_second": 7.937, |
| "step": 13110 |
| }, |
| { |
| "epoch": 1.5051049673052654, |
| "eval_loss": 0.38966700434684753, |
| "eval_runtime": 12.9624, |
| "eval_samples_per_second": 507.929, |
| "eval_steps_per_second": 7.946, |
| "step": 13120 |
| }, |
| { |
| "epoch": 1.5062521509693703, |
| "eval_loss": 0.38971978425979614, |
| "eval_runtime": 13.0971, |
| "eval_samples_per_second": 502.708, |
| "eval_steps_per_second": 7.864, |
| "step": 13130 |
| }, |
| { |
| "epoch": 1.5073993346334749, |
| "eval_loss": 0.3897435963153839, |
| "eval_runtime": 12.9671, |
| "eval_samples_per_second": 507.748, |
| "eval_steps_per_second": 7.943, |
| "step": 13140 |
| }, |
| { |
| "epoch": 1.5085465182975795, |
| "eval_loss": 0.3898574709892273, |
| "eval_runtime": 12.9686, |
| "eval_samples_per_second": 507.687, |
| "eval_steps_per_second": 7.942, |
| "step": 13150 |
| }, |
| { |
| "epoch": 1.509693701961684, |
| "eval_loss": 0.39010557532310486, |
| "eval_runtime": 13.0379, |
| "eval_samples_per_second": 504.991, |
| "eval_steps_per_second": 7.9, |
| "step": 13160 |
| }, |
| { |
| "epoch": 1.5108408856257887, |
| "eval_loss": 0.39012542366981506, |
| "eval_runtime": 13.0626, |
| "eval_samples_per_second": 504.035, |
| "eval_steps_per_second": 7.885, |
| "step": 13170 |
| }, |
| { |
| "epoch": 1.5119880692898933, |
| "eval_loss": 0.3902994692325592, |
| "eval_runtime": 13.0923, |
| "eval_samples_per_second": 502.893, |
| "eval_steps_per_second": 7.867, |
| "step": 13180 |
| }, |
| { |
| "epoch": 1.5131352529539979, |
| "eval_loss": 0.3900642991065979, |
| "eval_runtime": 12.9755, |
| "eval_samples_per_second": 507.419, |
| "eval_steps_per_second": 7.938, |
| "step": 13190 |
| }, |
| { |
| "epoch": 1.5142824366181027, |
| "grad_norm": 2.3656346797943115, |
| "learning_rate": 5.5028680688336524e-06, |
| "loss": 0.6483, |
| "step": 13200 |
| }, |
| { |
| "epoch": 1.5142824366181027, |
| "eval_loss": 0.39013996720314026, |
| "eval_runtime": 12.9668, |
| "eval_samples_per_second": 507.759, |
| "eval_steps_per_second": 7.943, |
| "step": 13200 |
| }, |
| { |
| "epoch": 1.515429620282207, |
| "eval_loss": 0.3904174566268921, |
| "eval_runtime": 12.9611, |
| "eval_samples_per_second": 507.983, |
| "eval_steps_per_second": 7.947, |
| "step": 13210 |
| }, |
| { |
| "epoch": 1.516576803946312, |
| "eval_loss": 0.3904300332069397, |
| "eval_runtime": 12.9624, |
| "eval_samples_per_second": 507.932, |
| "eval_steps_per_second": 7.946, |
| "step": 13220 |
| }, |
| { |
| "epoch": 1.5177239876104163, |
| "eval_loss": 0.39029890298843384, |
| "eval_runtime": 13.0538, |
| "eval_samples_per_second": 504.374, |
| "eval_steps_per_second": 7.89, |
| "step": 13230 |
| }, |
| { |
| "epoch": 1.518871171274521, |
| "eval_loss": 0.3900206685066223, |
| "eval_runtime": 13.0801, |
| "eval_samples_per_second": 503.362, |
| "eval_steps_per_second": 7.875, |
| "step": 13240 |
| }, |
| { |
| "epoch": 1.5200183549386257, |
| "eval_loss": 0.38975393772125244, |
| "eval_runtime": 13.006, |
| "eval_samples_per_second": 506.228, |
| "eval_steps_per_second": 7.919, |
| "step": 13250 |
| }, |
| { |
| "epoch": 1.5211655386027303, |
| "eval_loss": 0.3894404172897339, |
| "eval_runtime": 13.0886, |
| "eval_samples_per_second": 503.035, |
| "eval_steps_per_second": 7.869, |
| "step": 13260 |
| }, |
| { |
| "epoch": 1.522312722266835, |
| "eval_loss": 0.38918736577033997, |
| "eval_runtime": 13.0263, |
| "eval_samples_per_second": 505.441, |
| "eval_steps_per_second": 7.907, |
| "step": 13270 |
| }, |
| { |
| "epoch": 1.5234599059309395, |
| "eval_loss": 0.3890722990036011, |
| "eval_runtime": 12.9501, |
| "eval_samples_per_second": 508.412, |
| "eval_steps_per_second": 7.954, |
| "step": 13280 |
| }, |
| { |
| "epoch": 1.5246070895950443, |
| "eval_loss": 0.3890245258808136, |
| "eval_runtime": 13.0758, |
| "eval_samples_per_second": 503.527, |
| "eval_steps_per_second": 7.877, |
| "step": 13290 |
| }, |
| { |
| "epoch": 1.5257542732591487, |
| "grad_norm": 3.821657180786133, |
| "learning_rate": 5.460378160186957e-06, |
| "loss": 0.686, |
| "step": 13300 |
| }, |
| { |
| "epoch": 1.5257542732591487, |
| "eval_loss": 0.3892159163951874, |
| "eval_runtime": 12.9771, |
| "eval_samples_per_second": 507.354, |
| "eval_steps_per_second": 7.937, |
| "step": 13300 |
| }, |
| { |
| "epoch": 1.5269014569232535, |
| "eval_loss": 0.38923677802085876, |
| "eval_runtime": 13.0857, |
| "eval_samples_per_second": 503.145, |
| "eval_steps_per_second": 7.871, |
| "step": 13310 |
| }, |
| { |
| "epoch": 1.528048640587358, |
| "eval_loss": 0.38923653960227966, |
| "eval_runtime": 13.0991, |
| "eval_samples_per_second": 502.63, |
| "eval_steps_per_second": 7.863, |
| "step": 13320 |
| }, |
| { |
| "epoch": 1.5291958242514627, |
| "eval_loss": 0.3890940546989441, |
| "eval_runtime": 12.9572, |
| "eval_samples_per_second": 508.134, |
| "eval_steps_per_second": 7.949, |
| "step": 13330 |
| }, |
| { |
| "epoch": 1.5303430079155673, |
| "eval_loss": 0.3890242874622345, |
| "eval_runtime": 12.9711, |
| "eval_samples_per_second": 507.591, |
| "eval_steps_per_second": 7.941, |
| "step": 13340 |
| }, |
| { |
| "epoch": 1.531490191579672, |
| "eval_loss": 0.3894191384315491, |
| "eval_runtime": 12.9626, |
| "eval_samples_per_second": 507.922, |
| "eval_steps_per_second": 7.946, |
| "step": 13350 |
| }, |
| { |
| "epoch": 1.5326373752437765, |
| "eval_loss": 0.389509916305542, |
| "eval_runtime": 12.9621, |
| "eval_samples_per_second": 507.942, |
| "eval_steps_per_second": 7.946, |
| "step": 13360 |
| }, |
| { |
| "epoch": 1.5337845589078811, |
| "eval_loss": 0.3895265758037567, |
| "eval_runtime": 12.9819, |
| "eval_samples_per_second": 507.166, |
| "eval_steps_per_second": 7.934, |
| "step": 13370 |
| }, |
| { |
| "epoch": 1.5349317425719857, |
| "eval_loss": 0.389409601688385, |
| "eval_runtime": 12.9697, |
| "eval_samples_per_second": 507.644, |
| "eval_steps_per_second": 7.942, |
| "step": 13380 |
| }, |
| { |
| "epoch": 1.5360789262360903, |
| "eval_loss": 0.3894564211368561, |
| "eval_runtime": 12.9656, |
| "eval_samples_per_second": 507.805, |
| "eval_steps_per_second": 7.944, |
| "step": 13390 |
| }, |
| { |
| "epoch": 1.5372261099001951, |
| "grad_norm": 3.710810661315918, |
| "learning_rate": 5.41788825154026e-06, |
| "loss": 0.6901, |
| "step": 13400 |
| }, |
| { |
| "epoch": 1.5372261099001951, |
| "eval_loss": 0.38964158296585083, |
| "eval_runtime": 12.9557, |
| "eval_samples_per_second": 508.195, |
| "eval_steps_per_second": 7.95, |
| "step": 13400 |
| }, |
| { |
| "epoch": 1.5383732935642995, |
| "eval_loss": 0.3894788324832916, |
| "eval_runtime": 12.9588, |
| "eval_samples_per_second": 508.074, |
| "eval_steps_per_second": 7.948, |
| "step": 13410 |
| }, |
| { |
| "epoch": 1.5395204772284043, |
| "eval_loss": 0.38910993933677673, |
| "eval_runtime": 12.9613, |
| "eval_samples_per_second": 507.974, |
| "eval_steps_per_second": 7.947, |
| "step": 13420 |
| }, |
| { |
| "epoch": 1.5406676608925087, |
| "eval_loss": 0.3891458511352539, |
| "eval_runtime": 12.9664, |
| "eval_samples_per_second": 507.775, |
| "eval_steps_per_second": 7.944, |
| "step": 13430 |
| }, |
| { |
| "epoch": 1.5418148445566136, |
| "eval_loss": 0.38902804255485535, |
| "eval_runtime": 12.948, |
| "eval_samples_per_second": 508.496, |
| "eval_steps_per_second": 7.955, |
| "step": 13440 |
| }, |
| { |
| "epoch": 1.5429620282207182, |
| "eval_loss": 0.3888970911502838, |
| "eval_runtime": 12.9611, |
| "eval_samples_per_second": 507.98, |
| "eval_steps_per_second": 7.947, |
| "step": 13450 |
| }, |
| { |
| "epoch": 1.5441092118848228, |
| "eval_loss": 0.38868385553359985, |
| "eval_runtime": 12.9633, |
| "eval_samples_per_second": 507.894, |
| "eval_steps_per_second": 7.945, |
| "step": 13460 |
| }, |
| { |
| "epoch": 1.5452563955489274, |
| "eval_loss": 0.3885091245174408, |
| "eval_runtime": 12.9823, |
| "eval_samples_per_second": 507.153, |
| "eval_steps_per_second": 7.934, |
| "step": 13470 |
| }, |
| { |
| "epoch": 1.546403579213032, |
| "eval_loss": 0.38850271701812744, |
| "eval_runtime": 12.9631, |
| "eval_samples_per_second": 507.902, |
| "eval_steps_per_second": 7.946, |
| "step": 13480 |
| }, |
| { |
| "epoch": 1.5475507628771368, |
| "eval_loss": 0.3886268138885498, |
| "eval_runtime": 13.0746, |
| "eval_samples_per_second": 503.57, |
| "eval_steps_per_second": 7.878, |
| "step": 13490 |
| }, |
| { |
| "epoch": 1.5486979465412412, |
| "grad_norm": 3.1383347511291504, |
| "learning_rate": 5.375398342893563e-06, |
| "loss": 0.6568, |
| "step": 13500 |
| }, |
| { |
| "epoch": 1.5486979465412412, |
| "eval_loss": 0.3886937201023102, |
| "eval_runtime": 12.9626, |
| "eval_samples_per_second": 507.924, |
| "eval_steps_per_second": 7.946, |
| "step": 13500 |
| }, |
| { |
| "epoch": 1.549845130205346, |
| "eval_loss": 0.38839656114578247, |
| "eval_runtime": 12.9616, |
| "eval_samples_per_second": 507.961, |
| "eval_steps_per_second": 7.947, |
| "step": 13510 |
| }, |
| { |
| "epoch": 1.5509923138694504, |
| "eval_loss": 0.38791465759277344, |
| "eval_runtime": 12.966, |
| "eval_samples_per_second": 507.788, |
| "eval_steps_per_second": 7.944, |
| "step": 13520 |
| }, |
| { |
| "epoch": 1.5521394975335552, |
| "eval_loss": 0.38740622997283936, |
| "eval_runtime": 12.9773, |
| "eval_samples_per_second": 507.346, |
| "eval_steps_per_second": 7.937, |
| "step": 13530 |
| }, |
| { |
| "epoch": 1.5532866811976598, |
| "eval_loss": 0.38698500394821167, |
| "eval_runtime": 12.9644, |
| "eval_samples_per_second": 507.851, |
| "eval_steps_per_second": 7.945, |
| "step": 13540 |
| }, |
| { |
| "epoch": 1.5544338648617644, |
| "eval_loss": 0.386811226606369, |
| "eval_runtime": 12.976, |
| "eval_samples_per_second": 507.397, |
| "eval_steps_per_second": 7.938, |
| "step": 13550 |
| }, |
| { |
| "epoch": 1.555581048525869, |
| "eval_loss": 0.3869019150733948, |
| "eval_runtime": 13.1091, |
| "eval_samples_per_second": 502.248, |
| "eval_steps_per_second": 7.857, |
| "step": 13560 |
| }, |
| { |
| "epoch": 1.5567282321899736, |
| "eval_loss": 0.3871867060661316, |
| "eval_runtime": 12.9849, |
| "eval_samples_per_second": 507.051, |
| "eval_steps_per_second": 7.932, |
| "step": 13570 |
| }, |
| { |
| "epoch": 1.5578754158540782, |
| "eval_loss": 0.3873494565486908, |
| "eval_runtime": 12.9548, |
| "eval_samples_per_second": 508.229, |
| "eval_steps_per_second": 7.951, |
| "step": 13580 |
| }, |
| { |
| "epoch": 1.5590225995181828, |
| "eval_loss": 0.3873807191848755, |
| "eval_runtime": 13.099, |
| "eval_samples_per_second": 502.632, |
| "eval_steps_per_second": 7.863, |
| "step": 13590 |
| }, |
| { |
| "epoch": 1.5601697831822876, |
| "grad_norm": 3.738719940185547, |
| "learning_rate": 5.332908434246867e-06, |
| "loss": 0.6665, |
| "step": 13600 |
| }, |
| { |
| "epoch": 1.5601697831822876, |
| "eval_loss": 0.3875296115875244, |
| "eval_runtime": 13.0515, |
| "eval_samples_per_second": 504.464, |
| "eval_steps_per_second": 7.892, |
| "step": 13600 |
| }, |
| { |
| "epoch": 1.561316966846392, |
| "eval_loss": 0.38764673471450806, |
| "eval_runtime": 12.9945, |
| "eval_samples_per_second": 506.675, |
| "eval_steps_per_second": 7.926, |
| "step": 13610 |
| }, |
| { |
| "epoch": 1.5624641505104968, |
| "eval_loss": 0.38748273253440857, |
| "eval_runtime": 12.9655, |
| "eval_samples_per_second": 507.808, |
| "eval_steps_per_second": 7.944, |
| "step": 13620 |
| }, |
| { |
| "epoch": 1.5636113341746014, |
| "eval_loss": 0.3872354328632355, |
| "eval_runtime": 13.0501, |
| "eval_samples_per_second": 504.516, |
| "eval_steps_per_second": 7.893, |
| "step": 13630 |
| }, |
| { |
| "epoch": 1.564758517838706, |
| "eval_loss": 0.3872641324996948, |
| "eval_runtime": 13.094, |
| "eval_samples_per_second": 502.828, |
| "eval_steps_per_second": 7.866, |
| "step": 13640 |
| }, |
| { |
| "epoch": 1.5659057015028106, |
| "eval_loss": 0.38719263672828674, |
| "eval_runtime": 12.9663, |
| "eval_samples_per_second": 507.777, |
| "eval_steps_per_second": 7.944, |
| "step": 13650 |
| }, |
| { |
| "epoch": 1.5670528851669152, |
| "eval_loss": 0.38694635033607483, |
| "eval_runtime": 13.0975, |
| "eval_samples_per_second": 502.692, |
| "eval_steps_per_second": 7.864, |
| "step": 13660 |
| }, |
| { |
| "epoch": 1.5682000688310198, |
| "eval_loss": 0.38668397068977356, |
| "eval_runtime": 13.0928, |
| "eval_samples_per_second": 502.871, |
| "eval_steps_per_second": 7.867, |
| "step": 13670 |
| }, |
| { |
| "epoch": 1.5693472524951244, |
| "eval_loss": 0.38638022541999817, |
| "eval_runtime": 13.0859, |
| "eval_samples_per_second": 503.138, |
| "eval_steps_per_second": 7.871, |
| "step": 13680 |
| }, |
| { |
| "epoch": 1.5704944361592292, |
| "eval_loss": 0.38605645298957825, |
| "eval_runtime": 12.9769, |
| "eval_samples_per_second": 507.362, |
| "eval_steps_per_second": 7.937, |
| "step": 13690 |
| }, |
| { |
| "epoch": 1.5716416198233336, |
| "grad_norm": 2.6057910919189453, |
| "learning_rate": 5.290418525600171e-06, |
| "loss": 0.6795, |
| "step": 13700 |
| }, |
| { |
| "epoch": 1.5716416198233336, |
| "eval_loss": 0.3859562277793884, |
| "eval_runtime": 12.9572, |
| "eval_samples_per_second": 508.136, |
| "eval_steps_per_second": 7.949, |
| "step": 13700 |
| }, |
| { |
| "epoch": 1.5727888034874384, |
| "eval_loss": 0.38584911823272705, |
| "eval_runtime": 13.1023, |
| "eval_samples_per_second": 502.508, |
| "eval_steps_per_second": 7.861, |
| "step": 13710 |
| }, |
| { |
| "epoch": 1.5739359871515428, |
| "eval_loss": 0.38575005531311035, |
| "eval_runtime": 13.0365, |
| "eval_samples_per_second": 505.044, |
| "eval_steps_per_second": 7.901, |
| "step": 13720 |
| }, |
| { |
| "epoch": 1.5750831708156476, |
| "eval_loss": 0.3853883743286133, |
| "eval_runtime": 12.9679, |
| "eval_samples_per_second": 507.717, |
| "eval_steps_per_second": 7.943, |
| "step": 13730 |
| }, |
| { |
| "epoch": 1.5762303544797522, |
| "eval_loss": 0.3850836455821991, |
| "eval_runtime": 13.069, |
| "eval_samples_per_second": 503.787, |
| "eval_steps_per_second": 7.881, |
| "step": 13740 |
| }, |
| { |
| "epoch": 1.5773775381438568, |
| "eval_loss": 0.38496798276901245, |
| "eval_runtime": 13.0957, |
| "eval_samples_per_second": 502.762, |
| "eval_steps_per_second": 7.865, |
| "step": 13750 |
| }, |
| { |
| "epoch": 1.5785247218079614, |
| "eval_loss": 0.3848700523376465, |
| "eval_runtime": 12.9588, |
| "eval_samples_per_second": 508.072, |
| "eval_steps_per_second": 7.948, |
| "step": 13760 |
| }, |
| { |
| "epoch": 1.579671905472066, |
| "eval_loss": 0.38486871123313904, |
| "eval_runtime": 13.11, |
| "eval_samples_per_second": 502.211, |
| "eval_steps_per_second": 7.857, |
| "step": 13770 |
| }, |
| { |
| "epoch": 1.5808190891361709, |
| "eval_loss": 0.3848922848701477, |
| "eval_runtime": 12.9647, |
| "eval_samples_per_second": 507.842, |
| "eval_steps_per_second": 7.945, |
| "step": 13780 |
| }, |
| { |
| "epoch": 1.5819662728002752, |
| "eval_loss": 0.38480743765830994, |
| "eval_runtime": 12.9296, |
| "eval_samples_per_second": 509.221, |
| "eval_steps_per_second": 7.966, |
| "step": 13790 |
| }, |
| { |
| "epoch": 1.58311345646438, |
| "grad_norm": 2.840847969055176, |
| "learning_rate": 5.2479286169534735e-06, |
| "loss": 0.6894, |
| "step": 13800 |
| }, |
| { |
| "epoch": 1.58311345646438, |
| "eval_loss": 0.3848100006580353, |
| "eval_runtime": 12.9813, |
| "eval_samples_per_second": 507.19, |
| "eval_steps_per_second": 7.934, |
| "step": 13800 |
| }, |
| { |
| "epoch": 1.5842606401284844, |
| "eval_loss": 0.3845812678337097, |
| "eval_runtime": 13.0882, |
| "eval_samples_per_second": 503.049, |
| "eval_steps_per_second": 7.87, |
| "step": 13810 |
| }, |
| { |
| "epoch": 1.5854078237925893, |
| "eval_loss": 0.38454440236091614, |
| "eval_runtime": 12.9632, |
| "eval_samples_per_second": 507.9, |
| "eval_steps_per_second": 7.946, |
| "step": 13820 |
| }, |
| { |
| "epoch": 1.5865550074566939, |
| "eval_loss": 0.3846888542175293, |
| "eval_runtime": 12.9566, |
| "eval_samples_per_second": 508.158, |
| "eval_steps_per_second": 7.95, |
| "step": 13830 |
| }, |
| { |
| "epoch": 1.5877021911207985, |
| "eval_loss": 0.3848173916339874, |
| "eval_runtime": 12.9642, |
| "eval_samples_per_second": 507.859, |
| "eval_steps_per_second": 7.945, |
| "step": 13840 |
| }, |
| { |
| "epoch": 1.588849374784903, |
| "eval_loss": 0.38487833738327026, |
| "eval_runtime": 13.1067, |
| "eval_samples_per_second": 502.339, |
| "eval_steps_per_second": 7.859, |
| "step": 13850 |
| }, |
| { |
| "epoch": 1.5899965584490077, |
| "eval_loss": 0.384766548871994, |
| "eval_runtime": 12.9627, |
| "eval_samples_per_second": 507.92, |
| "eval_steps_per_second": 7.946, |
| "step": 13860 |
| }, |
| { |
| "epoch": 1.5911437421131123, |
| "eval_loss": 0.3845934271812439, |
| "eval_runtime": 13.0544, |
| "eval_samples_per_second": 504.35, |
| "eval_steps_per_second": 7.89, |
| "step": 13870 |
| }, |
| { |
| "epoch": 1.5922909257772169, |
| "eval_loss": 0.38453155755996704, |
| "eval_runtime": 13.0444, |
| "eval_samples_per_second": 504.739, |
| "eval_steps_per_second": 7.896, |
| "step": 13880 |
| }, |
| { |
| "epoch": 1.5934381094413217, |
| "eval_loss": 0.38435566425323486, |
| "eval_runtime": 12.9528, |
| "eval_samples_per_second": 508.307, |
| "eval_steps_per_second": 7.952, |
| "step": 13890 |
| }, |
| { |
| "epoch": 1.594585293105426, |
| "grad_norm": 2.864393472671509, |
| "learning_rate": 5.205438708306778e-06, |
| "loss": 0.6483, |
| "step": 13900 |
| }, |
| { |
| "epoch": 1.594585293105426, |
| "eval_loss": 0.38446468114852905, |
| "eval_runtime": 12.9637, |
| "eval_samples_per_second": 507.879, |
| "eval_steps_per_second": 7.945, |
| "step": 13900 |
| }, |
| { |
| "epoch": 1.595732476769531, |
| "eval_loss": 0.3841317296028137, |
| "eval_runtime": 12.9591, |
| "eval_samples_per_second": 508.061, |
| "eval_steps_per_second": 7.948, |
| "step": 13910 |
| }, |
| { |
| "epoch": 1.5968796604336353, |
| "eval_loss": 0.3841482996940613, |
| "eval_runtime": 12.9682, |
| "eval_samples_per_second": 507.705, |
| "eval_steps_per_second": 7.943, |
| "step": 13920 |
| }, |
| { |
| "epoch": 1.59802684409774, |
| "eval_loss": 0.3841838240623474, |
| "eval_runtime": 12.9647, |
| "eval_samples_per_second": 507.841, |
| "eval_steps_per_second": 7.945, |
| "step": 13930 |
| }, |
| { |
| "epoch": 1.5991740277618447, |
| "eval_loss": 0.38424983620643616, |
| "eval_runtime": 13.0821, |
| "eval_samples_per_second": 503.282, |
| "eval_steps_per_second": 7.873, |
| "step": 13940 |
| }, |
| { |
| "epoch": 1.6003212114259493, |
| "eval_loss": 0.38426274061203003, |
| "eval_runtime": 12.9605, |
| "eval_samples_per_second": 508.006, |
| "eval_steps_per_second": 7.947, |
| "step": 13950 |
| }, |
| { |
| "epoch": 1.601468395090054, |
| "eval_loss": 0.3843201994895935, |
| "eval_runtime": 13.0602, |
| "eval_samples_per_second": 504.128, |
| "eval_steps_per_second": 7.887, |
| "step": 13960 |
| }, |
| { |
| "epoch": 1.6026155787541585, |
| "eval_loss": 0.3843342959880829, |
| "eval_runtime": 13.085, |
| "eval_samples_per_second": 503.171, |
| "eval_steps_per_second": 7.872, |
| "step": 13970 |
| }, |
| { |
| "epoch": 1.6037627624182633, |
| "eval_loss": 0.3844236731529236, |
| "eval_runtime": 12.9522, |
| "eval_samples_per_second": 508.329, |
| "eval_steps_per_second": 7.952, |
| "step": 13980 |
| }, |
| { |
| "epoch": 1.6049099460823677, |
| "eval_loss": 0.3844641447067261, |
| "eval_runtime": 12.9669, |
| "eval_samples_per_second": 507.755, |
| "eval_steps_per_second": 7.943, |
| "step": 13990 |
| }, |
| { |
| "epoch": 1.6060571297464725, |
| "grad_norm": 3.092078685760498, |
| "learning_rate": 5.162948799660082e-06, |
| "loss": 0.6856, |
| "step": 14000 |
| }, |
| { |
| "epoch": 1.6060571297464725, |
| "eval_loss": 0.3846040964126587, |
| "eval_runtime": 12.9582, |
| "eval_samples_per_second": 508.094, |
| "eval_steps_per_second": 7.949, |
| "step": 14000 |
| }, |
| { |
| "epoch": 1.607204313410577, |
| "eval_loss": 0.38453853130340576, |
| "eval_runtime": 12.953, |
| "eval_samples_per_second": 508.3, |
| "eval_steps_per_second": 7.952, |
| "step": 14010 |
| }, |
| { |
| "epoch": 1.6083514970746817, |
| "eval_loss": 0.3846151530742645, |
| "eval_runtime": 13.097, |
| "eval_samples_per_second": 502.709, |
| "eval_steps_per_second": 7.864, |
| "step": 14020 |
| }, |
| { |
| "epoch": 1.6094986807387863, |
| "eval_loss": 0.3845311999320984, |
| "eval_runtime": 13.0896, |
| "eval_samples_per_second": 502.996, |
| "eval_steps_per_second": 7.869, |
| "step": 14030 |
| }, |
| { |
| "epoch": 1.610645864402891, |
| "eval_loss": 0.38428565859794617, |
| "eval_runtime": 13.098, |
| "eval_samples_per_second": 502.674, |
| "eval_steps_per_second": 7.864, |
| "step": 14040 |
| }, |
| { |
| "epoch": 1.6117930480669955, |
| "eval_loss": 0.3841724097728729, |
| "eval_runtime": 12.9612, |
| "eval_samples_per_second": 507.978, |
| "eval_steps_per_second": 7.947, |
| "step": 14050 |
| }, |
| { |
| "epoch": 1.6129402317311001, |
| "eval_loss": 0.38405221700668335, |
| "eval_runtime": 12.9652, |
| "eval_samples_per_second": 507.822, |
| "eval_steps_per_second": 7.944, |
| "step": 14060 |
| }, |
| { |
| "epoch": 1.6140874153952047, |
| "eval_loss": 0.3841976225376129, |
| "eval_runtime": 12.964, |
| "eval_samples_per_second": 507.868, |
| "eval_steps_per_second": 7.945, |
| "step": 14070 |
| }, |
| { |
| "epoch": 1.6152345990593093, |
| "eval_loss": 0.3842831254005432, |
| "eval_runtime": 13.0, |
| "eval_samples_per_second": 506.461, |
| "eval_steps_per_second": 7.923, |
| "step": 14080 |
| }, |
| { |
| "epoch": 1.6163817827234142, |
| "eval_loss": 0.38453409075737, |
| "eval_runtime": 12.9554, |
| "eval_samples_per_second": 508.207, |
| "eval_steps_per_second": 7.95, |
| "step": 14090 |
| }, |
| { |
| "epoch": 1.6175289663875185, |
| "grad_norm": 2.7563345432281494, |
| "learning_rate": 5.120458891013384e-06, |
| "loss": 0.6797, |
| "step": 14100 |
| }, |
| { |
| "epoch": 1.6175289663875185, |
| "eval_loss": 0.3844517767429352, |
| "eval_runtime": 12.97, |
| "eval_samples_per_second": 507.631, |
| "eval_steps_per_second": 7.941, |
| "step": 14100 |
| }, |
| { |
| "epoch": 1.6186761500516234, |
| "eval_loss": 0.3844999074935913, |
| "eval_runtime": 13.0562, |
| "eval_samples_per_second": 504.28, |
| "eval_steps_per_second": 7.889, |
| "step": 14110 |
| }, |
| { |
| "epoch": 1.6198233337157277, |
| "eval_loss": 0.38437753915786743, |
| "eval_runtime": 12.9753, |
| "eval_samples_per_second": 507.426, |
| "eval_steps_per_second": 7.938, |
| "step": 14120 |
| }, |
| { |
| "epoch": 1.6209705173798326, |
| "eval_loss": 0.38423284888267517, |
| "eval_runtime": 13.1084, |
| "eval_samples_per_second": 502.274, |
| "eval_steps_per_second": 7.858, |
| "step": 14130 |
| }, |
| { |
| "epoch": 1.6221177010439372, |
| "eval_loss": 0.38408252596855164, |
| "eval_runtime": 12.9795, |
| "eval_samples_per_second": 507.261, |
| "eval_steps_per_second": 7.936, |
| "step": 14140 |
| }, |
| { |
| "epoch": 1.6232648847080418, |
| "eval_loss": 0.38377270102500916, |
| "eval_runtime": 12.9965, |
| "eval_samples_per_second": 506.598, |
| "eval_steps_per_second": 7.925, |
| "step": 14150 |
| }, |
| { |
| "epoch": 1.6244120683721464, |
| "eval_loss": 0.38364750146865845, |
| "eval_runtime": 12.9604, |
| "eval_samples_per_second": 508.011, |
| "eval_steps_per_second": 7.947, |
| "step": 14160 |
| }, |
| { |
| "epoch": 1.625559252036251, |
| "eval_loss": 0.383523553609848, |
| "eval_runtime": 12.9732, |
| "eval_samples_per_second": 507.509, |
| "eval_steps_per_second": 7.939, |
| "step": 14170 |
| }, |
| { |
| "epoch": 1.6267064357003558, |
| "eval_loss": 0.3833784759044647, |
| "eval_runtime": 12.9603, |
| "eval_samples_per_second": 508.012, |
| "eval_steps_per_second": 7.947, |
| "step": 14180 |
| }, |
| { |
| "epoch": 1.6278536193644602, |
| "eval_loss": 0.3830562233924866, |
| "eval_runtime": 12.969, |
| "eval_samples_per_second": 507.674, |
| "eval_steps_per_second": 7.942, |
| "step": 14190 |
| }, |
| { |
| "epoch": 1.629000803028565, |
| "grad_norm": 2.473512649536133, |
| "learning_rate": 5.077968982366688e-06, |
| "loss": 0.7057, |
| "step": 14200 |
| }, |
| { |
| "epoch": 1.629000803028565, |
| "eval_loss": 0.3828422427177429, |
| "eval_runtime": 12.9537, |
| "eval_samples_per_second": 508.272, |
| "eval_steps_per_second": 7.951, |
| "step": 14200 |
| }, |
| { |
| "epoch": 1.6301479866926694, |
| "eval_loss": 0.38250473141670227, |
| "eval_runtime": 12.9581, |
| "eval_samples_per_second": 508.098, |
| "eval_steps_per_second": 7.949, |
| "step": 14210 |
| }, |
| { |
| "epoch": 1.6312951703567742, |
| "eval_loss": 0.38224315643310547, |
| "eval_runtime": 13.0436, |
| "eval_samples_per_second": 504.767, |
| "eval_steps_per_second": 7.897, |
| "step": 14220 |
| }, |
| { |
| "epoch": 1.6324423540208788, |
| "eval_loss": 0.3821259140968323, |
| "eval_runtime": 13.0024, |
| "eval_samples_per_second": 506.369, |
| "eval_steps_per_second": 7.922, |
| "step": 14230 |
| }, |
| { |
| "epoch": 1.6335895376849834, |
| "eval_loss": 0.38201645016670227, |
| "eval_runtime": 13.028, |
| "eval_samples_per_second": 505.372, |
| "eval_steps_per_second": 7.906, |
| "step": 14240 |
| }, |
| { |
| "epoch": 1.634736721349088, |
| "eval_loss": 0.3822481334209442, |
| "eval_runtime": 12.964, |
| "eval_samples_per_second": 507.867, |
| "eval_steps_per_second": 7.945, |
| "step": 14250 |
| }, |
| { |
| "epoch": 1.6358839050131926, |
| "eval_loss": 0.3822289705276489, |
| "eval_runtime": 13.0599, |
| "eval_samples_per_second": 504.137, |
| "eval_steps_per_second": 7.887, |
| "step": 14260 |
| }, |
| { |
| "epoch": 1.6370310886772972, |
| "eval_loss": 0.38217175006866455, |
| "eval_runtime": 13.005, |
| "eval_samples_per_second": 506.268, |
| "eval_steps_per_second": 7.92, |
| "step": 14270 |
| }, |
| { |
| "epoch": 1.6381782723414018, |
| "eval_loss": 0.3820948004722595, |
| "eval_runtime": 13.0391, |
| "eval_samples_per_second": 504.944, |
| "eval_steps_per_second": 7.899, |
| "step": 14280 |
| }, |
| { |
| "epoch": 1.6393254560055066, |
| "eval_loss": 0.38222813606262207, |
| "eval_runtime": 13.101, |
| "eval_samples_per_second": 502.557, |
| "eval_steps_per_second": 7.862, |
| "step": 14290 |
| }, |
| { |
| "epoch": 1.640472639669611, |
| "grad_norm": 3.483987808227539, |
| "learning_rate": 5.0354790737199925e-06, |
| "loss": 0.6699, |
| "step": 14300 |
| }, |
| { |
| "epoch": 1.640472639669611, |
| "eval_loss": 0.38273701071739197, |
| "eval_runtime": 13.1111, |
| "eval_samples_per_second": 502.169, |
| "eval_steps_per_second": 7.856, |
| "step": 14300 |
| }, |
| { |
| "epoch": 1.6416198233337158, |
| "eval_loss": 0.3827689588069916, |
| "eval_runtime": 12.9555, |
| "eval_samples_per_second": 508.203, |
| "eval_steps_per_second": 7.95, |
| "step": 14310 |
| }, |
| { |
| "epoch": 1.6427670069978202, |
| "eval_loss": 0.3826591968536377, |
| "eval_runtime": 13.0336, |
| "eval_samples_per_second": 505.156, |
| "eval_steps_per_second": 7.903, |
| "step": 14320 |
| }, |
| { |
| "epoch": 1.643914190661925, |
| "eval_loss": 0.38226479291915894, |
| "eval_runtime": 13.13, |
| "eval_samples_per_second": 501.448, |
| "eval_steps_per_second": 7.845, |
| "step": 14330 |
| }, |
| { |
| "epoch": 1.6450613743260296, |
| "eval_loss": 0.38224324584007263, |
| "eval_runtime": 12.9703, |
| "eval_samples_per_second": 507.623, |
| "eval_steps_per_second": 7.941, |
| "step": 14340 |
| }, |
| { |
| "epoch": 1.6462085579901342, |
| "eval_loss": 0.3823966383934021, |
| "eval_runtime": 13.1055, |
| "eval_samples_per_second": 502.385, |
| "eval_steps_per_second": 7.859, |
| "step": 14350 |
| }, |
| { |
| "epoch": 1.6473557416542388, |
| "eval_loss": 0.3825666606426239, |
| "eval_runtime": 12.9613, |
| "eval_samples_per_second": 507.974, |
| "eval_steps_per_second": 7.947, |
| "step": 14360 |
| }, |
| { |
| "epoch": 1.6485029253183434, |
| "eval_loss": 0.3826480209827423, |
| "eval_runtime": 12.9585, |
| "eval_samples_per_second": 508.082, |
| "eval_steps_per_second": 7.948, |
| "step": 14370 |
| }, |
| { |
| "epoch": 1.6496501089824482, |
| "eval_loss": 0.38273876905441284, |
| "eval_runtime": 12.9694, |
| "eval_samples_per_second": 507.656, |
| "eval_steps_per_second": 7.942, |
| "step": 14380 |
| }, |
| { |
| "epoch": 1.6507972926465526, |
| "eval_loss": 0.3827420473098755, |
| "eval_runtime": 12.9925, |
| "eval_samples_per_second": 506.754, |
| "eval_steps_per_second": 7.928, |
| "step": 14390 |
| }, |
| { |
| "epoch": 1.6519444763106574, |
| "grad_norm": 3.916184425354004, |
| "learning_rate": 4.992989165073295e-06, |
| "loss": 0.6615, |
| "step": 14400 |
| }, |
| { |
| "epoch": 1.6519444763106574, |
| "eval_loss": 0.3827430009841919, |
| "eval_runtime": 13.0985, |
| "eval_samples_per_second": 502.652, |
| "eval_steps_per_second": 7.863, |
| "step": 14400 |
| }, |
| { |
| "epoch": 1.6530916599747618, |
| "eval_loss": 0.3827821612358093, |
| "eval_runtime": 12.9458, |
| "eval_samples_per_second": 508.584, |
| "eval_steps_per_second": 7.956, |
| "step": 14410 |
| }, |
| { |
| "epoch": 1.6542388436388666, |
| "eval_loss": 0.3825940191745758, |
| "eval_runtime": 13.0603, |
| "eval_samples_per_second": 504.123, |
| "eval_steps_per_second": 7.886, |
| "step": 14420 |
| }, |
| { |
| "epoch": 1.6553860273029712, |
| "eval_loss": 0.382512629032135, |
| "eval_runtime": 12.9555, |
| "eval_samples_per_second": 508.202, |
| "eval_steps_per_second": 7.95, |
| "step": 14430 |
| }, |
| { |
| "epoch": 1.6565332109670758, |
| "eval_loss": 0.38264140486717224, |
| "eval_runtime": 12.9739, |
| "eval_samples_per_second": 507.48, |
| "eval_steps_per_second": 7.939, |
| "step": 14440 |
| }, |
| { |
| "epoch": 1.6576803946311804, |
| "eval_loss": 0.38300880789756775, |
| "eval_runtime": 13.1471, |
| "eval_samples_per_second": 500.795, |
| "eval_steps_per_second": 7.834, |
| "step": 14450 |
| }, |
| { |
| "epoch": 1.658827578295285, |
| "eval_loss": 0.3830065131187439, |
| "eval_runtime": 12.9524, |
| "eval_samples_per_second": 508.322, |
| "eval_steps_per_second": 7.952, |
| "step": 14460 |
| }, |
| { |
| "epoch": 1.6599747619593896, |
| "eval_loss": 0.3830389976501465, |
| "eval_runtime": 13.1053, |
| "eval_samples_per_second": 502.392, |
| "eval_steps_per_second": 7.859, |
| "step": 14470 |
| }, |
| { |
| "epoch": 1.6611219456234942, |
| "eval_loss": 0.3830309808254242, |
| "eval_runtime": 12.963, |
| "eval_samples_per_second": 507.906, |
| "eval_steps_per_second": 7.946, |
| "step": 14480 |
| }, |
| { |
| "epoch": 1.662269129287599, |
| "eval_loss": 0.3830114006996155, |
| "eval_runtime": 12.9804, |
| "eval_samples_per_second": 507.228, |
| "eval_steps_per_second": 7.935, |
| "step": 14490 |
| }, |
| { |
| "epoch": 1.6634163129517034, |
| "grad_norm": 3.4649813175201416, |
| "learning_rate": 4.950499256426599e-06, |
| "loss": 0.6628, |
| "step": 14500 |
| }, |
| { |
| "epoch": 1.6634163129517034, |
| "eval_loss": 0.38285231590270996, |
| "eval_runtime": 12.9706, |
| "eval_samples_per_second": 507.609, |
| "eval_steps_per_second": 7.941, |
| "step": 14500 |
| }, |
| { |
| "epoch": 1.6645634966158083, |
| "eval_loss": 0.38294288516044617, |
| "eval_runtime": 12.97, |
| "eval_samples_per_second": 507.634, |
| "eval_steps_per_second": 7.941, |
| "step": 14510 |
| }, |
| { |
| "epoch": 1.6657106802799126, |
| "eval_loss": 0.3828694820404053, |
| "eval_runtime": 12.964, |
| "eval_samples_per_second": 507.869, |
| "eval_steps_per_second": 7.945, |
| "step": 14520 |
| }, |
| { |
| "epoch": 1.6668578639440175, |
| "eval_loss": 0.3828655183315277, |
| "eval_runtime": 13.004, |
| "eval_samples_per_second": 506.306, |
| "eval_steps_per_second": 7.921, |
| "step": 14530 |
| }, |
| { |
| "epoch": 1.668005047608122, |
| "eval_loss": 0.3829069435596466, |
| "eval_runtime": 12.9668, |
| "eval_samples_per_second": 507.757, |
| "eval_steps_per_second": 7.943, |
| "step": 14540 |
| }, |
| { |
| "epoch": 1.6691522312722267, |
| "eval_loss": 0.3828739821910858, |
| "eval_runtime": 13.0454, |
| "eval_samples_per_second": 504.698, |
| "eval_steps_per_second": 7.895, |
| "step": 14550 |
| }, |
| { |
| "epoch": 1.6702994149363313, |
| "eval_loss": 0.38302430510520935, |
| "eval_runtime": 12.9576, |
| "eval_samples_per_second": 508.119, |
| "eval_steps_per_second": 7.949, |
| "step": 14560 |
| }, |
| { |
| "epoch": 1.6714465986004359, |
| "eval_loss": 0.3827770948410034, |
| "eval_runtime": 12.9639, |
| "eval_samples_per_second": 507.872, |
| "eval_steps_per_second": 7.945, |
| "step": 14570 |
| }, |
| { |
| "epoch": 1.6725937822645407, |
| "eval_loss": 0.38250812888145447, |
| "eval_runtime": 12.9893, |
| "eval_samples_per_second": 506.88, |
| "eval_steps_per_second": 7.93, |
| "step": 14580 |
| }, |
| { |
| "epoch": 1.673740965928645, |
| "eval_loss": 0.38223034143447876, |
| "eval_runtime": 12.9511, |
| "eval_samples_per_second": 508.374, |
| "eval_steps_per_second": 7.953, |
| "step": 14590 |
| }, |
| { |
| "epoch": 1.67488814959275, |
| "grad_norm": 2.9048054218292236, |
| "learning_rate": 4.908009347779903e-06, |
| "loss": 0.6728, |
| "step": 14600 |
| }, |
| { |
| "epoch": 1.67488814959275, |
| "eval_loss": 0.3818991482257843, |
| "eval_runtime": 12.9661, |
| "eval_samples_per_second": 507.786, |
| "eval_steps_per_second": 7.944, |
| "step": 14600 |
| }, |
| { |
| "epoch": 1.6760353332568543, |
| "eval_loss": 0.3817276358604431, |
| "eval_runtime": 12.958, |
| "eval_samples_per_second": 508.103, |
| "eval_steps_per_second": 7.949, |
| "step": 14610 |
| }, |
| { |
| "epoch": 1.677182516920959, |
| "eval_loss": 0.381703644990921, |
| "eval_runtime": 13.0018, |
| "eval_samples_per_second": 506.392, |
| "eval_steps_per_second": 7.922, |
| "step": 14620 |
| }, |
| { |
| "epoch": 1.6783297005850637, |
| "eval_loss": 0.381527304649353, |
| "eval_runtime": 12.9663, |
| "eval_samples_per_second": 507.777, |
| "eval_steps_per_second": 7.944, |
| "step": 14630 |
| }, |
| { |
| "epoch": 1.6794768842491683, |
| "eval_loss": 0.38126641511917114, |
| "eval_runtime": 12.9768, |
| "eval_samples_per_second": 507.366, |
| "eval_steps_per_second": 7.937, |
| "step": 14640 |
| }, |
| { |
| "epoch": 1.680624067913273, |
| "eval_loss": 0.38146719336509705, |
| "eval_runtime": 12.9671, |
| "eval_samples_per_second": 507.746, |
| "eval_steps_per_second": 7.943, |
| "step": 14650 |
| }, |
| { |
| "epoch": 1.6817712515773775, |
| "eval_loss": 0.3814009130001068, |
| "eval_runtime": 12.9659, |
| "eval_samples_per_second": 507.795, |
| "eval_steps_per_second": 7.944, |
| "step": 14660 |
| }, |
| { |
| "epoch": 1.6829184352414823, |
| "eval_loss": 0.38138487935066223, |
| "eval_runtime": 13.0548, |
| "eval_samples_per_second": 504.337, |
| "eval_steps_per_second": 7.89, |
| "step": 14670 |
| }, |
| { |
| "epoch": 1.6840656189055867, |
| "eval_loss": 0.3811706304550171, |
| "eval_runtime": 13.0051, |
| "eval_samples_per_second": 506.263, |
| "eval_steps_per_second": 7.92, |
| "step": 14680 |
| }, |
| { |
| "epoch": 1.6852128025696915, |
| "eval_loss": 0.38091224431991577, |
| "eval_runtime": 13.1142, |
| "eval_samples_per_second": 502.052, |
| "eval_steps_per_second": 7.854, |
| "step": 14690 |
| }, |
| { |
| "epoch": 1.686359986233796, |
| "grad_norm": 7.453961372375488, |
| "learning_rate": 4.865519439133206e-06, |
| "loss": 0.6852, |
| "step": 14700 |
| }, |
| { |
| "epoch": 1.686359986233796, |
| "eval_loss": 0.3808043897151947, |
| "eval_runtime": 13.1647, |
| "eval_samples_per_second": 500.127, |
| "eval_steps_per_second": 7.824, |
| "step": 14700 |
| }, |
| { |
| "epoch": 1.6875071698979007, |
| "eval_loss": 0.3806935250759125, |
| "eval_runtime": 13.1399, |
| "eval_samples_per_second": 501.069, |
| "eval_steps_per_second": 7.839, |
| "step": 14710 |
| }, |
| { |
| "epoch": 1.6886543535620053, |
| "eval_loss": 0.38041648268699646, |
| "eval_runtime": 13.1475, |
| "eval_samples_per_second": 500.779, |
| "eval_steps_per_second": 7.834, |
| "step": 14720 |
| }, |
| { |
| "epoch": 1.68980153722611, |
| "eval_loss": 0.3801563084125519, |
| "eval_runtime": 13.1111, |
| "eval_samples_per_second": 502.168, |
| "eval_steps_per_second": 7.856, |
| "step": 14730 |
| }, |
| { |
| "epoch": 1.6909487208902145, |
| "eval_loss": 0.37989431619644165, |
| "eval_runtime": 12.9361, |
| "eval_samples_per_second": 508.964, |
| "eval_steps_per_second": 7.962, |
| "step": 14740 |
| }, |
| { |
| "epoch": 1.6920959045543191, |
| "eval_loss": 0.37977877259254456, |
| "eval_runtime": 12.9645, |
| "eval_samples_per_second": 507.85, |
| "eval_steps_per_second": 7.945, |
| "step": 14750 |
| }, |
| { |
| "epoch": 1.6932430882184237, |
| "eval_loss": 0.37966009974479675, |
| "eval_runtime": 13.1245, |
| "eval_samples_per_second": 501.657, |
| "eval_steps_per_second": 7.848, |
| "step": 14760 |
| }, |
| { |
| "epoch": 1.6943902718825283, |
| "eval_loss": 0.37949711084365845, |
| "eval_runtime": 12.9703, |
| "eval_samples_per_second": 507.621, |
| "eval_steps_per_second": 7.941, |
| "step": 14770 |
| }, |
| { |
| "epoch": 1.6955374555466332, |
| "eval_loss": 0.37965020537376404, |
| "eval_runtime": 12.9812, |
| "eval_samples_per_second": 507.196, |
| "eval_steps_per_second": 7.935, |
| "step": 14780 |
| }, |
| { |
| "epoch": 1.6966846392107375, |
| "eval_loss": 0.37967824935913086, |
| "eval_runtime": 13.1525, |
| "eval_samples_per_second": 500.588, |
| "eval_steps_per_second": 7.831, |
| "step": 14790 |
| }, |
| { |
| "epoch": 1.6978318228748424, |
| "grad_norm": 3.1470093727111816, |
| "learning_rate": 4.82302953048651e-06, |
| "loss": 0.6585, |
| "step": 14800 |
| }, |
| { |
| "epoch": 1.6978318228748424, |
| "eval_loss": 0.3796652853488922, |
| "eval_runtime": 12.9576, |
| "eval_samples_per_second": 508.118, |
| "eval_steps_per_second": 7.949, |
| "step": 14800 |
| }, |
| { |
| "epoch": 1.6989790065389467, |
| "eval_loss": 0.37974977493286133, |
| "eval_runtime": 12.9883, |
| "eval_samples_per_second": 506.916, |
| "eval_steps_per_second": 7.93, |
| "step": 14810 |
| }, |
| { |
| "epoch": 1.7001261902030516, |
| "eval_loss": 0.37977394461631775, |
| "eval_runtime": 13.0415, |
| "eval_samples_per_second": 504.851, |
| "eval_steps_per_second": 7.898, |
| "step": 14820 |
| }, |
| { |
| "epoch": 1.7012733738671562, |
| "eval_loss": 0.37995630502700806, |
| "eval_runtime": 12.9589, |
| "eval_samples_per_second": 508.066, |
| "eval_steps_per_second": 7.948, |
| "step": 14830 |
| }, |
| { |
| "epoch": 1.7024205575312608, |
| "eval_loss": 0.3800138533115387, |
| "eval_runtime": 13.0014, |
| "eval_samples_per_second": 506.406, |
| "eval_steps_per_second": 7.922, |
| "step": 14840 |
| }, |
| { |
| "epoch": 1.7035677411953654, |
| "eval_loss": 0.3798474073410034, |
| "eval_runtime": 13.0115, |
| "eval_samples_per_second": 506.012, |
| "eval_steps_per_second": 7.916, |
| "step": 14850 |
| }, |
| { |
| "epoch": 1.70471492485947, |
| "eval_loss": 0.3798816204071045, |
| "eval_runtime": 13.007, |
| "eval_samples_per_second": 506.187, |
| "eval_steps_per_second": 7.919, |
| "step": 14860 |
| }, |
| { |
| "epoch": 1.7058621085235748, |
| "eval_loss": 0.37989550828933716, |
| "eval_runtime": 13.0121, |
| "eval_samples_per_second": 505.989, |
| "eval_steps_per_second": 7.916, |
| "step": 14870 |
| }, |
| { |
| "epoch": 1.7070092921876792, |
| "eval_loss": 0.37979933619499207, |
| "eval_runtime": 13.0621, |
| "eval_samples_per_second": 504.054, |
| "eval_steps_per_second": 7.885, |
| "step": 14880 |
| }, |
| { |
| "epoch": 1.708156475851784, |
| "eval_loss": 0.3800634741783142, |
| "eval_runtime": 13.0733, |
| "eval_samples_per_second": 503.62, |
| "eval_steps_per_second": 7.879, |
| "step": 14890 |
| }, |
| { |
| "epoch": 1.7093036595158884, |
| "grad_norm": 2.8225584030151367, |
| "learning_rate": 4.7805396218398135e-06, |
| "loss": 0.6691, |
| "step": 14900 |
| }, |
| { |
| "epoch": 1.7093036595158884, |
| "eval_loss": 0.3800990581512451, |
| "eval_runtime": 13.0903, |
| "eval_samples_per_second": 502.969, |
| "eval_steps_per_second": 7.868, |
| "step": 14900 |
| }, |
| { |
| "epoch": 1.7104508431799932, |
| "eval_loss": 0.37998729944229126, |
| "eval_runtime": 13.1808, |
| "eval_samples_per_second": 499.515, |
| "eval_steps_per_second": 7.814, |
| "step": 14910 |
| }, |
| { |
| "epoch": 1.7115980268440978, |
| "eval_loss": 0.3797684609889984, |
| "eval_runtime": 12.9921, |
| "eval_samples_per_second": 506.77, |
| "eval_steps_per_second": 7.928, |
| "step": 14920 |
| }, |
| { |
| "epoch": 1.7127452105082024, |
| "eval_loss": 0.37945589423179626, |
| "eval_runtime": 13.0664, |
| "eval_samples_per_second": 503.888, |
| "eval_steps_per_second": 7.883, |
| "step": 14930 |
| }, |
| { |
| "epoch": 1.713892394172307, |
| "eval_loss": 0.37920668721199036, |
| "eval_runtime": 12.995, |
| "eval_samples_per_second": 506.655, |
| "eval_steps_per_second": 7.926, |
| "step": 14940 |
| }, |
| { |
| "epoch": 1.7150395778364116, |
| "eval_loss": 0.3791378438472748, |
| "eval_runtime": 13.0937, |
| "eval_samples_per_second": 502.837, |
| "eval_steps_per_second": 7.866, |
| "step": 14950 |
| }, |
| { |
| "epoch": 1.7161867615005162, |
| "eval_loss": 0.3789806663990021, |
| "eval_runtime": 13.1261, |
| "eval_samples_per_second": 501.595, |
| "eval_steps_per_second": 7.847, |
| "step": 14960 |
| }, |
| { |
| "epoch": 1.7173339451646208, |
| "eval_loss": 0.37904536724090576, |
| "eval_runtime": 12.9834, |
| "eval_samples_per_second": 507.108, |
| "eval_steps_per_second": 7.933, |
| "step": 14970 |
| }, |
| { |
| "epoch": 1.7184811288287256, |
| "eval_loss": 0.3793109953403473, |
| "eval_runtime": 12.9987, |
| "eval_samples_per_second": 506.511, |
| "eval_steps_per_second": 7.924, |
| "step": 14980 |
| }, |
| { |
| "epoch": 1.71962831249283, |
| "eval_loss": 0.3793599605560303, |
| "eval_runtime": 13.0796, |
| "eval_samples_per_second": 503.381, |
| "eval_steps_per_second": 7.875, |
| "step": 14990 |
| }, |
| { |
| "epoch": 1.7207754961569348, |
| "grad_norm": 2.551968812942505, |
| "learning_rate": 4.738049713193117e-06, |
| "loss": 0.6676, |
| "step": 15000 |
| }, |
| { |
| "epoch": 1.7207754961569348, |
| "eval_loss": 0.37940821051597595, |
| "eval_runtime": 13.0354, |
| "eval_samples_per_second": 505.086, |
| "eval_steps_per_second": 7.902, |
| "step": 15000 |
| }, |
| { |
| "epoch": 1.7219226798210392, |
| "eval_loss": 0.37939274311065674, |
| "eval_runtime": 12.9847, |
| "eval_samples_per_second": 507.058, |
| "eval_steps_per_second": 7.932, |
| "step": 15010 |
| }, |
| { |
| "epoch": 1.723069863485144, |
| "eval_loss": 0.37943676114082336, |
| "eval_runtime": 13.0329, |
| "eval_samples_per_second": 505.183, |
| "eval_steps_per_second": 7.903, |
| "step": 15020 |
| }, |
| { |
| "epoch": 1.7242170471492486, |
| "eval_loss": 0.37933647632598877, |
| "eval_runtime": 13.0479, |
| "eval_samples_per_second": 504.602, |
| "eval_steps_per_second": 7.894, |
| "step": 15030 |
| }, |
| { |
| "epoch": 1.7253642308133532, |
| "eval_loss": 0.37906157970428467, |
| "eval_runtime": 13.1336, |
| "eval_samples_per_second": 501.31, |
| "eval_steps_per_second": 7.842, |
| "step": 15040 |
| }, |
| { |
| "epoch": 1.7265114144774578, |
| "eval_loss": 0.37899768352508545, |
| "eval_runtime": 13.0861, |
| "eval_samples_per_second": 503.128, |
| "eval_steps_per_second": 7.871, |
| "step": 15050 |
| }, |
| { |
| "epoch": 1.7276585981415624, |
| "eval_loss": 0.3787804841995239, |
| "eval_runtime": 13.1253, |
| "eval_samples_per_second": 501.627, |
| "eval_steps_per_second": 7.847, |
| "step": 15060 |
| }, |
| { |
| "epoch": 1.7288057818056672, |
| "eval_loss": 0.3787018656730652, |
| "eval_runtime": 13.0265, |
| "eval_samples_per_second": 505.431, |
| "eval_steps_per_second": 7.907, |
| "step": 15070 |
| }, |
| { |
| "epoch": 1.7299529654697716, |
| "eval_loss": 0.37867027521133423, |
| "eval_runtime": 13.0069, |
| "eval_samples_per_second": 506.194, |
| "eval_steps_per_second": 7.919, |
| "step": 15080 |
| }, |
| { |
| "epoch": 1.7311001491338764, |
| "eval_loss": 0.37871935963630676, |
| "eval_runtime": 13.0522, |
| "eval_samples_per_second": 504.435, |
| "eval_steps_per_second": 7.891, |
| "step": 15090 |
| }, |
| { |
| "epoch": 1.7322473327979808, |
| "grad_norm": 4.536778450012207, |
| "learning_rate": 4.695559804546421e-06, |
| "loss": 0.6945, |
| "step": 15100 |
| }, |
| { |
| "epoch": 1.7322473327979808, |
| "eval_loss": 0.37847813963890076, |
| "eval_runtime": 13.025, |
| "eval_samples_per_second": 505.489, |
| "eval_steps_per_second": 7.908, |
| "step": 15100 |
| }, |
| { |
| "epoch": 1.7333945164620856, |
| "eval_loss": 0.3782278895378113, |
| "eval_runtime": 13.1146, |
| "eval_samples_per_second": 502.034, |
| "eval_steps_per_second": 7.854, |
| "step": 15110 |
| }, |
| { |
| "epoch": 1.7345417001261902, |
| "eval_loss": 0.378121018409729, |
| "eval_runtime": 12.999, |
| "eval_samples_per_second": 506.499, |
| "eval_steps_per_second": 7.924, |
| "step": 15120 |
| }, |
| { |
| "epoch": 1.7356888837902948, |
| "eval_loss": 0.37795355916023254, |
| "eval_runtime": 13.1272, |
| "eval_samples_per_second": 501.553, |
| "eval_steps_per_second": 7.846, |
| "step": 15130 |
| }, |
| { |
| "epoch": 1.7368360674543994, |
| "eval_loss": 0.3781651258468628, |
| "eval_runtime": 12.9858, |
| "eval_samples_per_second": 507.014, |
| "eval_steps_per_second": 7.932, |
| "step": 15140 |
| }, |
| { |
| "epoch": 1.737983251118504, |
| "eval_loss": 0.37823063135147095, |
| "eval_runtime": 13.0081, |
| "eval_samples_per_second": 506.147, |
| "eval_steps_per_second": 7.918, |
| "step": 15150 |
| }, |
| { |
| "epoch": 1.7391304347826086, |
| "eval_loss": 0.3781428933143616, |
| "eval_runtime": 13.0233, |
| "eval_samples_per_second": 505.557, |
| "eval_steps_per_second": 7.909, |
| "step": 15160 |
| }, |
| { |
| "epoch": 1.7402776184467132, |
| "eval_loss": 0.37814781069755554, |
| "eval_runtime": 13.044, |
| "eval_samples_per_second": 504.753, |
| "eval_steps_per_second": 7.896, |
| "step": 15170 |
| }, |
| { |
| "epoch": 1.741424802110818, |
| "eval_loss": 0.3780823051929474, |
| "eval_runtime": 13.023, |
| "eval_samples_per_second": 505.567, |
| "eval_steps_per_second": 7.909, |
| "step": 15180 |
| }, |
| { |
| "epoch": 1.7425719857749224, |
| "eval_loss": 0.3784046173095703, |
| "eval_runtime": 13.0494, |
| "eval_samples_per_second": 504.545, |
| "eval_steps_per_second": 7.893, |
| "step": 15190 |
| }, |
| { |
| "epoch": 1.7437191694390273, |
| "grad_norm": 2.781858205795288, |
| "learning_rate": 4.6530698958997245e-06, |
| "loss": 0.6697, |
| "step": 15200 |
| }, |
| { |
| "epoch": 1.7437191694390273, |
| "eval_loss": 0.37867581844329834, |
| "eval_runtime": 12.9898, |
| "eval_samples_per_second": 506.861, |
| "eval_steps_per_second": 7.929, |
| "step": 15200 |
| }, |
| { |
| "epoch": 1.7448663531031317, |
| "eval_loss": 0.3789690136909485, |
| "eval_runtime": 13.074, |
| "eval_samples_per_second": 503.595, |
| "eval_steps_per_second": 7.878, |
| "step": 15210 |
| }, |
| { |
| "epoch": 1.7460135367672365, |
| "eval_loss": 0.3791664242744446, |
| "eval_runtime": 12.9993, |
| "eval_samples_per_second": 506.487, |
| "eval_steps_per_second": 7.923, |
| "step": 15220 |
| }, |
| { |
| "epoch": 1.747160720431341, |
| "eval_loss": 0.3791807293891907, |
| "eval_runtime": 13.0387, |
| "eval_samples_per_second": 504.96, |
| "eval_steps_per_second": 7.9, |
| "step": 15230 |
| }, |
| { |
| "epoch": 1.7483079040954457, |
| "eval_loss": 0.37914615869522095, |
| "eval_runtime": 13.173, |
| "eval_samples_per_second": 499.811, |
| "eval_steps_per_second": 7.819, |
| "step": 15240 |
| }, |
| { |
| "epoch": 1.7494550877595503, |
| "eval_loss": 0.37912288308143616, |
| "eval_runtime": 13.1398, |
| "eval_samples_per_second": 501.072, |
| "eval_steps_per_second": 7.839, |
| "step": 15250 |
| }, |
| { |
| "epoch": 1.7506022714236549, |
| "eval_loss": 0.37882035970687866, |
| "eval_runtime": 13.0039, |
| "eval_samples_per_second": 506.309, |
| "eval_steps_per_second": 7.921, |
| "step": 15260 |
| }, |
| { |
| "epoch": 1.7517494550877597, |
| "eval_loss": 0.37890636920928955, |
| "eval_runtime": 12.9934, |
| "eval_samples_per_second": 506.718, |
| "eval_steps_per_second": 7.927, |
| "step": 15270 |
| }, |
| { |
| "epoch": 1.752896638751864, |
| "eval_loss": 0.3788420259952545, |
| "eval_runtime": 13.0107, |
| "eval_samples_per_second": 506.046, |
| "eval_steps_per_second": 7.917, |
| "step": 15280 |
| }, |
| { |
| "epoch": 1.754043822415969, |
| "eval_loss": 0.37884387373924255, |
| "eval_runtime": 13.0489, |
| "eval_samples_per_second": 504.562, |
| "eval_steps_per_second": 7.893, |
| "step": 15290 |
| }, |
| { |
| "epoch": 1.7551910060800733, |
| "grad_norm": 2.8187098503112793, |
| "learning_rate": 4.610579987253028e-06, |
| "loss": 0.6557, |
| "step": 15300 |
| }, |
| { |
| "epoch": 1.7551910060800733, |
| "eval_loss": 0.3783915638923645, |
| "eval_runtime": 13.1367, |
| "eval_samples_per_second": 501.191, |
| "eval_steps_per_second": 7.841, |
| "step": 15300 |
| }, |
| { |
| "epoch": 1.756338189744178, |
| "eval_loss": 0.37841951847076416, |
| "eval_runtime": 12.9975, |
| "eval_samples_per_second": 506.558, |
| "eval_steps_per_second": 7.925, |
| "step": 15310 |
| }, |
| { |
| "epoch": 1.7574853734082827, |
| "eval_loss": 0.37844744324684143, |
| "eval_runtime": 13.1286, |
| "eval_samples_per_second": 501.501, |
| "eval_steps_per_second": 7.845, |
| "step": 15320 |
| }, |
| { |
| "epoch": 1.7586325570723873, |
| "eval_loss": 0.3785025477409363, |
| "eval_runtime": 12.9969, |
| "eval_samples_per_second": 506.581, |
| "eval_steps_per_second": 7.925, |
| "step": 15330 |
| }, |
| { |
| "epoch": 1.759779740736492, |
| "eval_loss": 0.3788582384586334, |
| "eval_runtime": 13.0143, |
| "eval_samples_per_second": 505.906, |
| "eval_steps_per_second": 7.914, |
| "step": 15340 |
| }, |
| { |
| "epoch": 1.7609269244005965, |
| "eval_loss": 0.3790527284145355, |
| "eval_runtime": 13.0335, |
| "eval_samples_per_second": 505.158, |
| "eval_steps_per_second": 7.903, |
| "step": 15350 |
| }, |
| { |
| "epoch": 1.762074108064701, |
| "eval_loss": 0.3790960907936096, |
| "eval_runtime": 12.9827, |
| "eval_samples_per_second": 507.136, |
| "eval_steps_per_second": 7.934, |
| "step": 15360 |
| }, |
| { |
| "epoch": 1.7632212917288057, |
| "eval_loss": 0.3790503442287445, |
| "eval_runtime": 13.0311, |
| "eval_samples_per_second": 505.251, |
| "eval_steps_per_second": 7.904, |
| "step": 15370 |
| }, |
| { |
| "epoch": 1.7643684753929105, |
| "eval_loss": 0.3789319694042206, |
| "eval_runtime": 13.008, |
| "eval_samples_per_second": 506.151, |
| "eval_steps_per_second": 7.918, |
| "step": 15380 |
| }, |
| { |
| "epoch": 1.765515659057015, |
| "eval_loss": 0.37879255414009094, |
| "eval_runtime": 13.058, |
| "eval_samples_per_second": 504.212, |
| "eval_steps_per_second": 7.888, |
| "step": 15390 |
| }, |
| { |
| "epoch": 1.7666628427211197, |
| "grad_norm": 8.184600830078125, |
| "learning_rate": 4.568090078606332e-06, |
| "loss": 0.6837, |
| "step": 15400 |
| }, |
| { |
| "epoch": 1.7666628427211197, |
| "eval_loss": 0.3788207471370697, |
| "eval_runtime": 13.0399, |
| "eval_samples_per_second": 504.91, |
| "eval_steps_per_second": 7.899, |
| "step": 15400 |
| }, |
| { |
| "epoch": 1.767810026385224, |
| "eval_loss": 0.3787640929222107, |
| "eval_runtime": 13.2284, |
| "eval_samples_per_second": 497.718, |
| "eval_steps_per_second": 7.786, |
| "step": 15410 |
| }, |
| { |
| "epoch": 1.768957210049329, |
| "eval_loss": 0.37880727648735046, |
| "eval_runtime": 13.195, |
| "eval_samples_per_second": 498.975, |
| "eval_steps_per_second": 7.806, |
| "step": 15420 |
| }, |
| { |
| "epoch": 1.7701043937134335, |
| "eval_loss": 0.37865129113197327, |
| "eval_runtime": 13.1286, |
| "eval_samples_per_second": 501.501, |
| "eval_steps_per_second": 7.845, |
| "step": 15430 |
| }, |
| { |
| "epoch": 1.7712515773775381, |
| "eval_loss": 0.37859317660331726, |
| "eval_runtime": 13.0049, |
| "eval_samples_per_second": 506.272, |
| "eval_steps_per_second": 7.92, |
| "step": 15440 |
| }, |
| { |
| "epoch": 1.7723987610416427, |
| "eval_loss": 0.3784623146057129, |
| "eval_runtime": 13.0822, |
| "eval_samples_per_second": 503.28, |
| "eval_steps_per_second": 7.873, |
| "step": 15450 |
| }, |
| { |
| "epoch": 1.7735459447057473, |
| "eval_loss": 0.3783509135246277, |
| "eval_runtime": 13.1198, |
| "eval_samples_per_second": 501.838, |
| "eval_steps_per_second": 7.851, |
| "step": 15460 |
| }, |
| { |
| "epoch": 1.7746931283698522, |
| "eval_loss": 0.3779684603214264, |
| "eval_runtime": 12.9832, |
| "eval_samples_per_second": 507.116, |
| "eval_steps_per_second": 7.933, |
| "step": 15470 |
| }, |
| { |
| "epoch": 1.7758403120339565, |
| "eval_loss": 0.37776073813438416, |
| "eval_runtime": 12.9898, |
| "eval_samples_per_second": 506.857, |
| "eval_steps_per_second": 7.929, |
| "step": 15480 |
| }, |
| { |
| "epoch": 1.7769874956980614, |
| "eval_loss": 0.3778015971183777, |
| "eval_runtime": 12.9889, |
| "eval_samples_per_second": 506.896, |
| "eval_steps_per_second": 7.93, |
| "step": 15490 |
| }, |
| { |
| "epoch": 1.7781346793621657, |
| "grad_norm": 2.779513120651245, |
| "learning_rate": 4.5256001699596345e-06, |
| "loss": 0.6685, |
| "step": 15500 |
| }, |
| { |
| "epoch": 1.7781346793621657, |
| "eval_loss": 0.3778555989265442, |
| "eval_runtime": 12.9832, |
| "eval_samples_per_second": 507.118, |
| "eval_steps_per_second": 7.933, |
| "step": 15500 |
| }, |
| { |
| "epoch": 1.7792818630262706, |
| "eval_loss": 0.37806859612464905, |
| "eval_runtime": 13.0971, |
| "eval_samples_per_second": 502.707, |
| "eval_steps_per_second": 7.864, |
| "step": 15510 |
| }, |
| { |
| "epoch": 1.7804290466903752, |
| "eval_loss": 0.37831228971481323, |
| "eval_runtime": 13.0608, |
| "eval_samples_per_second": 504.103, |
| "eval_steps_per_second": 7.886, |
| "step": 15520 |
| }, |
| { |
| "epoch": 1.7815762303544798, |
| "eval_loss": 0.3784082531929016, |
| "eval_runtime": 13.0713, |
| "eval_samples_per_second": 503.7, |
| "eval_steps_per_second": 7.88, |
| "step": 15530 |
| }, |
| { |
| "epoch": 1.7827234140185844, |
| "eval_loss": 0.3781934082508087, |
| "eval_runtime": 13.0455, |
| "eval_samples_per_second": 504.694, |
| "eval_steps_per_second": 7.895, |
| "step": 15540 |
| }, |
| { |
| "epoch": 1.783870597682689, |
| "eval_loss": 0.3779049813747406, |
| "eval_runtime": 13.0203, |
| "eval_samples_per_second": 505.674, |
| "eval_steps_per_second": 7.911, |
| "step": 15550 |
| }, |
| { |
| "epoch": 1.7850177813467938, |
| "eval_loss": 0.3778923749923706, |
| "eval_runtime": 12.9902, |
| "eval_samples_per_second": 506.843, |
| "eval_steps_per_second": 7.929, |
| "step": 15560 |
| }, |
| { |
| "epoch": 1.7861649650108982, |
| "eval_loss": 0.3781958818435669, |
| "eval_runtime": 13.0084, |
| "eval_samples_per_second": 506.134, |
| "eval_steps_per_second": 7.918, |
| "step": 15570 |
| }, |
| { |
| "epoch": 1.787312148675003, |
| "eval_loss": 0.3785502016544342, |
| "eval_runtime": 13.001, |
| "eval_samples_per_second": 506.424, |
| "eval_steps_per_second": 7.922, |
| "step": 15580 |
| }, |
| { |
| "epoch": 1.7884593323391074, |
| "eval_loss": 0.37848520278930664, |
| "eval_runtime": 12.9918, |
| "eval_samples_per_second": 506.781, |
| "eval_steps_per_second": 7.928, |
| "step": 15590 |
| }, |
| { |
| "epoch": 1.7896065160032122, |
| "grad_norm": 3.62384033203125, |
| "learning_rate": 4.483110261312938e-06, |
| "loss": 0.6521, |
| "step": 15600 |
| }, |
| { |
| "epoch": 1.7896065160032122, |
| "eval_loss": 0.3783304691314697, |
| "eval_runtime": 12.9931, |
| "eval_samples_per_second": 506.73, |
| "eval_steps_per_second": 7.927, |
| "step": 15600 |
| }, |
| { |
| "epoch": 1.7907536996673168, |
| "eval_loss": 0.37847575545310974, |
| "eval_runtime": 13.0385, |
| "eval_samples_per_second": 504.965, |
| "eval_steps_per_second": 7.9, |
| "step": 15610 |
| }, |
| { |
| "epoch": 1.7919008833314214, |
| "eval_loss": 0.3782954812049866, |
| "eval_runtime": 13.1268, |
| "eval_samples_per_second": 501.57, |
| "eval_steps_per_second": 7.847, |
| "step": 15620 |
| }, |
| { |
| "epoch": 1.793048066995526, |
| "eval_loss": 0.37828463315963745, |
| "eval_runtime": 13.268, |
| "eval_samples_per_second": 496.232, |
| "eval_steps_per_second": 7.763, |
| "step": 15630 |
| }, |
| { |
| "epoch": 1.7941952506596306, |
| "eval_loss": 0.3784072995185852, |
| "eval_runtime": 13.103, |
| "eval_samples_per_second": 502.479, |
| "eval_steps_per_second": 7.861, |
| "step": 15640 |
| }, |
| { |
| "epoch": 1.7953424343237352, |
| "eval_loss": 0.37834054231643677, |
| "eval_runtime": 13.0101, |
| "eval_samples_per_second": 506.07, |
| "eval_steps_per_second": 7.917, |
| "step": 15650 |
| }, |
| { |
| "epoch": 1.7964896179878398, |
| "eval_loss": 0.3782414495944977, |
| "eval_runtime": 12.9915, |
| "eval_samples_per_second": 506.791, |
| "eval_steps_per_second": 7.928, |
| "step": 15660 |
| }, |
| { |
| "epoch": 1.7976368016519446, |
| "eval_loss": 0.3780037760734558, |
| "eval_runtime": 13.0901, |
| "eval_samples_per_second": 502.977, |
| "eval_steps_per_second": 7.869, |
| "step": 15670 |
| }, |
| { |
| "epoch": 1.798783985316049, |
| "eval_loss": 0.37788066267967224, |
| "eval_runtime": 13.1795, |
| "eval_samples_per_second": 499.564, |
| "eval_steps_per_second": 7.815, |
| "step": 15680 |
| }, |
| { |
| "epoch": 1.7999311689801538, |
| "eval_loss": 0.3778708875179291, |
| "eval_runtime": 13.2186, |
| "eval_samples_per_second": 498.084, |
| "eval_steps_per_second": 7.792, |
| "step": 15690 |
| }, |
| { |
| "epoch": 1.8010783526442582, |
| "grad_norm": 2.410911798477173, |
| "learning_rate": 4.440620352666242e-06, |
| "loss": 0.649, |
| "step": 15700 |
| }, |
| { |
| "epoch": 1.8010783526442582, |
| "eval_loss": 0.3779171407222748, |
| "eval_runtime": 13.2337, |
| "eval_samples_per_second": 497.518, |
| "eval_steps_per_second": 7.783, |
| "step": 15700 |
| }, |
| { |
| "epoch": 1.802225536308363, |
| "eval_loss": 0.3780859708786011, |
| "eval_runtime": 13.1694, |
| "eval_samples_per_second": 499.946, |
| "eval_steps_per_second": 7.821, |
| "step": 15710 |
| }, |
| { |
| "epoch": 1.8033727199724676, |
| "eval_loss": 0.3781272768974304, |
| "eval_runtime": 13.0269, |
| "eval_samples_per_second": 505.416, |
| "eval_steps_per_second": 7.907, |
| "step": 15720 |
| }, |
| { |
| "epoch": 1.8045199036365722, |
| "eval_loss": 0.37824758887290955, |
| "eval_runtime": 13.0075, |
| "eval_samples_per_second": 506.168, |
| "eval_steps_per_second": 7.918, |
| "step": 15730 |
| }, |
| { |
| "epoch": 1.8056670873006768, |
| "eval_loss": 0.37803974747657776, |
| "eval_runtime": 13.0119, |
| "eval_samples_per_second": 505.997, |
| "eval_steps_per_second": 7.916, |
| "step": 15740 |
| }, |
| { |
| "epoch": 1.8068142709647814, |
| "eval_loss": 0.37797218561172485, |
| "eval_runtime": 13.0081, |
| "eval_samples_per_second": 506.148, |
| "eval_steps_per_second": 7.918, |
| "step": 15750 |
| }, |
| { |
| "epoch": 1.8079614546288862, |
| "eval_loss": 0.3780378997325897, |
| "eval_runtime": 12.996, |
| "eval_samples_per_second": 506.616, |
| "eval_steps_per_second": 7.925, |
| "step": 15760 |
| }, |
| { |
| "epoch": 1.8091086382929906, |
| "eval_loss": 0.37803494930267334, |
| "eval_runtime": 13.0267, |
| "eval_samples_per_second": 505.423, |
| "eval_steps_per_second": 7.907, |
| "step": 15770 |
| }, |
| { |
| "epoch": 1.8102558219570954, |
| "eval_loss": 0.378009557723999, |
| "eval_runtime": 13.2703, |
| "eval_samples_per_second": 496.145, |
| "eval_steps_per_second": 7.762, |
| "step": 15780 |
| }, |
| { |
| "epoch": 1.8114030056211998, |
| "eval_loss": 0.378147691488266, |
| "eval_runtime": 12.9945, |
| "eval_samples_per_second": 506.674, |
| "eval_steps_per_second": 7.926, |
| "step": 15790 |
| }, |
| { |
| "epoch": 1.8125501892853046, |
| "grad_norm": 2.8491785526275635, |
| "learning_rate": 4.3981304440195455e-06, |
| "loss": 0.6673, |
| "step": 15800 |
| }, |
| { |
| "epoch": 1.8125501892853046, |
| "eval_loss": 0.37826037406921387, |
| "eval_runtime": 13.0555, |
| "eval_samples_per_second": 504.309, |
| "eval_steps_per_second": 7.889, |
| "step": 15800 |
| }, |
| { |
| "epoch": 1.8136973729494092, |
| "eval_loss": 0.3781185746192932, |
| "eval_runtime": 13.0622, |
| "eval_samples_per_second": 504.049, |
| "eval_steps_per_second": 7.885, |
| "step": 15810 |
| }, |
| { |
| "epoch": 1.8148445566135138, |
| "eval_loss": 0.3781217634677887, |
| "eval_runtime": 13.0061, |
| "eval_samples_per_second": 506.224, |
| "eval_steps_per_second": 7.919, |
| "step": 15820 |
| }, |
| { |
| "epoch": 1.8159917402776184, |
| "eval_loss": 0.37847980856895447, |
| "eval_runtime": 12.9932, |
| "eval_samples_per_second": 506.726, |
| "eval_steps_per_second": 7.927, |
| "step": 15830 |
| }, |
| { |
| "epoch": 1.817138923941723, |
| "eval_loss": 0.3788329064846039, |
| "eval_runtime": 12.977, |
| "eval_samples_per_second": 507.361, |
| "eval_steps_per_second": 7.937, |
| "step": 15840 |
| }, |
| { |
| "epoch": 1.8182861076058277, |
| "eval_loss": 0.3788602650165558, |
| "eval_runtime": 12.9802, |
| "eval_samples_per_second": 507.233, |
| "eval_steps_per_second": 7.935, |
| "step": 15850 |
| }, |
| { |
| "epoch": 1.8194332912699323, |
| "eval_loss": 0.3787531852722168, |
| "eval_runtime": 12.9911, |
| "eval_samples_per_second": 506.81, |
| "eval_steps_per_second": 7.929, |
| "step": 15860 |
| }, |
| { |
| "epoch": 1.820580474934037, |
| "eval_loss": 0.37862464785575867, |
| "eval_runtime": 12.9846, |
| "eval_samples_per_second": 507.063, |
| "eval_steps_per_second": 7.932, |
| "step": 15870 |
| }, |
| { |
| "epoch": 1.8217276585981415, |
| "eval_loss": 0.3782861828804016, |
| "eval_runtime": 13.0036, |
| "eval_samples_per_second": 506.321, |
| "eval_steps_per_second": 7.921, |
| "step": 15880 |
| }, |
| { |
| "epoch": 1.8228748422622463, |
| "eval_loss": 0.3782361149787903, |
| "eval_runtime": 12.9891, |
| "eval_samples_per_second": 506.885, |
| "eval_steps_per_second": 7.93, |
| "step": 15890 |
| }, |
| { |
| "epoch": 1.8240220259263507, |
| "grad_norm": 3.4326915740966797, |
| "learning_rate": 4.355640535372849e-06, |
| "loss": 0.6902, |
| "step": 15900 |
| }, |
| { |
| "epoch": 1.8240220259263507, |
| "eval_loss": 0.37826868891716003, |
| "eval_runtime": 12.9815, |
| "eval_samples_per_second": 507.184, |
| "eval_steps_per_second": 7.934, |
| "step": 15900 |
| }, |
| { |
| "epoch": 1.8251692095904555, |
| "eval_loss": 0.3781268894672394, |
| "eval_runtime": 12.9888, |
| "eval_samples_per_second": 506.898, |
| "eval_steps_per_second": 7.93, |
| "step": 15910 |
| }, |
| { |
| "epoch": 1.82631639325456, |
| "eval_loss": 0.37785670161247253, |
| "eval_runtime": 12.9961, |
| "eval_samples_per_second": 506.614, |
| "eval_steps_per_second": 7.925, |
| "step": 15920 |
| }, |
| { |
| "epoch": 1.8274635769186647, |
| "eval_loss": 0.3777265250682831, |
| "eval_runtime": 13.0354, |
| "eval_samples_per_second": 505.088, |
| "eval_steps_per_second": 7.902, |
| "step": 15930 |
| }, |
| { |
| "epoch": 1.8286107605827693, |
| "eval_loss": 0.37783369421958923, |
| "eval_runtime": 12.99, |
| "eval_samples_per_second": 506.85, |
| "eval_steps_per_second": 7.929, |
| "step": 15940 |
| }, |
| { |
| "epoch": 1.8297579442468739, |
| "eval_loss": 0.3780220150947571, |
| "eval_runtime": 12.9952, |
| "eval_samples_per_second": 506.649, |
| "eval_steps_per_second": 7.926, |
| "step": 15950 |
| }, |
| { |
| "epoch": 1.8309051279109787, |
| "eval_loss": 0.37800726294517517, |
| "eval_runtime": 13.1587, |
| "eval_samples_per_second": 500.353, |
| "eval_steps_per_second": 7.828, |
| "step": 15960 |
| }, |
| { |
| "epoch": 1.832052311575083, |
| "eval_loss": 0.3780716061592102, |
| "eval_runtime": 13.1275, |
| "eval_samples_per_second": 501.544, |
| "eval_steps_per_second": 7.846, |
| "step": 15970 |
| }, |
| { |
| "epoch": 1.833199495239188, |
| "eval_loss": 0.377960205078125, |
| "eval_runtime": 12.9877, |
| "eval_samples_per_second": 506.943, |
| "eval_steps_per_second": 7.931, |
| "step": 15980 |
| }, |
| { |
| "epoch": 1.8343466789032923, |
| "eval_loss": 0.37785816192626953, |
| "eval_runtime": 13.0073, |
| "eval_samples_per_second": 506.177, |
| "eval_steps_per_second": 7.919, |
| "step": 15990 |
| }, |
| { |
| "epoch": 1.835493862567397, |
| "grad_norm": 6.324818134307861, |
| "learning_rate": 4.313150626726153e-06, |
| "loss": 0.6568, |
| "step": 16000 |
| }, |
| { |
| "epoch": 1.835493862567397, |
| "eval_loss": 0.3778100609779358, |
| "eval_runtime": 13.0429, |
| "eval_samples_per_second": 504.795, |
| "eval_steps_per_second": 7.897, |
| "step": 16000 |
| } |
| ], |
| "logging_steps": 100, |
| "max_steps": 26151, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "save_steps": 10, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 0.0, |
| "train_batch_size": 64, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
