| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 7.0, |
| "eval_steps": 500, |
| "global_step": 1722, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.040774719673802244, |
| "grad_norm": 18.189022064208984, |
| "learning_rate": 9.000000000000001e-07, |
| "loss": 0.8515, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.08154943934760449, |
| "grad_norm": 9.354066848754883, |
| "learning_rate": 1.9000000000000002e-06, |
| "loss": 0.905, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.12232415902140673, |
| "grad_norm": 10.508342742919922, |
| "learning_rate": 2.9e-06, |
| "loss": 0.8879, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.16309887869520898, |
| "grad_norm": 9.291099548339844, |
| "learning_rate": 3.900000000000001e-06, |
| "loss": 0.8905, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.2038735983690112, |
| "grad_norm": 13.063141822814941, |
| "learning_rate": 4.9000000000000005e-06, |
| "loss": 0.8604, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.24464831804281345, |
| "grad_norm": 7.156795024871826, |
| "learning_rate": 5.9e-06, |
| "loss": 0.7997, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.2854230377166157, |
| "grad_norm": 12.09440803527832, |
| "learning_rate": 6.9e-06, |
| "loss": 0.7996, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.32619775739041795, |
| "grad_norm": 14.44044303894043, |
| "learning_rate": 7.9e-06, |
| "loss": 0.8254, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.3669724770642202, |
| "grad_norm": 12.786622047424316, |
| "learning_rate": 8.900000000000001e-06, |
| "loss": 0.8999, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.4077471967380224, |
| "grad_norm": 14.936609268188477, |
| "learning_rate": 9.9e-06, |
| "loss": 0.7893, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.44852191641182465, |
| "grad_norm": 12.025712013244629, |
| "learning_rate": 9.961864406779662e-06, |
| "loss": 0.8599, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.4892966360856269, |
| "grad_norm": 18.530860900878906, |
| "learning_rate": 9.91949152542373e-06, |
| "loss": 0.7895, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.5300713557594292, |
| "grad_norm": 10.44506549835205, |
| "learning_rate": 9.877118644067798e-06, |
| "loss": 0.8066, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.5708460754332314, |
| "grad_norm": 18.924278259277344, |
| "learning_rate": 9.834745762711865e-06, |
| "loss": 0.8618, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.6116207951070336, |
| "grad_norm": 18.275020599365234, |
| "learning_rate": 9.792372881355933e-06, |
| "loss": 0.8419, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.6523955147808359, |
| "grad_norm": 11.579925537109375, |
| "learning_rate": 9.75e-06, |
| "loss": 0.8823, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.6931702344546381, |
| "grad_norm": 13.954208374023438, |
| "learning_rate": 9.707627118644068e-06, |
| "loss": 0.8893, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.7339449541284404, |
| "grad_norm": 11.016424179077148, |
| "learning_rate": 9.665254237288136e-06, |
| "loss": 0.9281, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.7747196738022426, |
| "grad_norm": 9.621957778930664, |
| "learning_rate": 9.622881355932205e-06, |
| "loss": 0.8009, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.8154943934760448, |
| "grad_norm": 17.310779571533203, |
| "learning_rate": 9.580508474576273e-06, |
| "loss": 0.8456, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.8562691131498471, |
| "grad_norm": 10.48460578918457, |
| "learning_rate": 9.53813559322034e-06, |
| "loss": 0.7, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.8970438328236493, |
| "grad_norm": 10.495179176330566, |
| "learning_rate": 9.495762711864408e-06, |
| "loss": 0.6497, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.9378185524974516, |
| "grad_norm": 19.09280776977539, |
| "learning_rate": 9.453389830508474e-06, |
| "loss": 0.8303, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.9785932721712538, |
| "grad_norm": 18.26506233215332, |
| "learning_rate": 9.411016949152543e-06, |
| "loss": 0.736, |
| "step": 240 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_general_loss": 0.7072561979293823, |
| "eval_general_runtime": 258.0411, |
| "eval_general_samples_per_second": 3.546, |
| "eval_general_steps_per_second": 0.887, |
| "step": 246 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_code_loss": 0.8012194037437439, |
| "eval_code_runtime": 301.035, |
| "eval_code_samples_per_second": 3.049, |
| "eval_code_steps_per_second": 0.764, |
| "step": 246 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_stem_loss": 0.7626175284385681, |
| "eval_stem_runtime": 254.1723, |
| "eval_stem_samples_per_second": 3.596, |
| "eval_stem_steps_per_second": 0.901, |
| "step": 246 |
| }, |
| { |
| "epoch": 1.016309887869521, |
| "grad_norm": 14.707352638244629, |
| "learning_rate": 9.368644067796611e-06, |
| "loss": 0.6495, |
| "step": 250 |
| }, |
| { |
| "epoch": 1.0570846075433231, |
| "grad_norm": 14.996277809143066, |
| "learning_rate": 9.32627118644068e-06, |
| "loss": 0.6753, |
| "step": 260 |
| }, |
| { |
| "epoch": 1.0978593272171253, |
| "grad_norm": 16.44031524658203, |
| "learning_rate": 9.283898305084746e-06, |
| "loss": 0.6465, |
| "step": 270 |
| }, |
| { |
| "epoch": 1.1386340468909277, |
| "grad_norm": 12.770790100097656, |
| "learning_rate": 9.241525423728814e-06, |
| "loss": 0.7006, |
| "step": 280 |
| }, |
| { |
| "epoch": 1.17940876656473, |
| "grad_norm": 9.644815444946289, |
| "learning_rate": 9.199152542372882e-06, |
| "loss": 0.7345, |
| "step": 290 |
| }, |
| { |
| "epoch": 1.2201834862385321, |
| "grad_norm": 12.410354614257812, |
| "learning_rate": 9.15677966101695e-06, |
| "loss": 0.5812, |
| "step": 300 |
| }, |
| { |
| "epoch": 1.2609582059123343, |
| "grad_norm": 14.592584609985352, |
| "learning_rate": 9.114406779661017e-06, |
| "loss": 0.7479, |
| "step": 310 |
| }, |
| { |
| "epoch": 1.3017329255861365, |
| "grad_norm": 8.983531951904297, |
| "learning_rate": 9.072033898305086e-06, |
| "loss": 0.6735, |
| "step": 320 |
| }, |
| { |
| "epoch": 1.3425076452599387, |
| "grad_norm": 10.594444274902344, |
| "learning_rate": 9.029661016949152e-06, |
| "loss": 0.6125, |
| "step": 330 |
| }, |
| { |
| "epoch": 1.3832823649337411, |
| "grad_norm": 8.827781677246094, |
| "learning_rate": 8.98728813559322e-06, |
| "loss": 0.5886, |
| "step": 340 |
| }, |
| { |
| "epoch": 1.4240570846075433, |
| "grad_norm": 8.48524284362793, |
| "learning_rate": 8.944915254237289e-06, |
| "loss": 0.6677, |
| "step": 350 |
| }, |
| { |
| "epoch": 1.4648318042813455, |
| "grad_norm": 14.892068862915039, |
| "learning_rate": 8.902542372881357e-06, |
| "loss": 0.6921, |
| "step": 360 |
| }, |
| { |
| "epoch": 1.505606523955148, |
| "grad_norm": 13.356966018676758, |
| "learning_rate": 8.860169491525424e-06, |
| "loss": 0.7095, |
| "step": 370 |
| }, |
| { |
| "epoch": 1.5463812436289501, |
| "grad_norm": 10.22076416015625, |
| "learning_rate": 8.817796610169492e-06, |
| "loss": 0.6031, |
| "step": 380 |
| }, |
| { |
| "epoch": 1.5871559633027523, |
| "grad_norm": 9.876967430114746, |
| "learning_rate": 8.77542372881356e-06, |
| "loss": 0.5818, |
| "step": 390 |
| }, |
| { |
| "epoch": 1.6279306829765545, |
| "grad_norm": 9.625609397888184, |
| "learning_rate": 8.733050847457629e-06, |
| "loss": 0.5913, |
| "step": 400 |
| }, |
| { |
| "epoch": 1.6687054026503567, |
| "grad_norm": 8.925414085388184, |
| "learning_rate": 8.690677966101695e-06, |
| "loss": 0.578, |
| "step": 410 |
| }, |
| { |
| "epoch": 1.709480122324159, |
| "grad_norm": 7.93916130065918, |
| "learning_rate": 8.648305084745763e-06, |
| "loss": 0.6955, |
| "step": 420 |
| }, |
| { |
| "epoch": 1.750254841997961, |
| "grad_norm": 7.942169666290283, |
| "learning_rate": 8.60593220338983e-06, |
| "loss": 0.5284, |
| "step": 430 |
| }, |
| { |
| "epoch": 1.7910295616717635, |
| "grad_norm": 9.36421012878418, |
| "learning_rate": 8.563559322033898e-06, |
| "loss": 0.7499, |
| "step": 440 |
| }, |
| { |
| "epoch": 1.8318042813455657, |
| "grad_norm": 12.293156623840332, |
| "learning_rate": 8.521186440677967e-06, |
| "loss": 0.6923, |
| "step": 450 |
| }, |
| { |
| "epoch": 1.8725790010193681, |
| "grad_norm": 9.337946891784668, |
| "learning_rate": 8.478813559322035e-06, |
| "loss": 0.6138, |
| "step": 460 |
| }, |
| { |
| "epoch": 1.9133537206931703, |
| "grad_norm": 10.548377990722656, |
| "learning_rate": 8.436440677966102e-06, |
| "loss": 0.6029, |
| "step": 470 |
| }, |
| { |
| "epoch": 1.9541284403669725, |
| "grad_norm": 17.341978073120117, |
| "learning_rate": 8.39406779661017e-06, |
| "loss": 0.624, |
| "step": 480 |
| }, |
| { |
| "epoch": 1.9949031600407747, |
| "grad_norm": 13.546788215637207, |
| "learning_rate": 8.351694915254238e-06, |
| "loss": 0.6197, |
| "step": 490 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_general_loss": 0.6272298097610474, |
| "eval_general_runtime": 255.3018, |
| "eval_general_samples_per_second": 3.584, |
| "eval_general_steps_per_second": 0.897, |
| "step": 492 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_code_loss": 0.69377201795578, |
| "eval_code_runtime": 298.208, |
| "eval_code_samples_per_second": 3.078, |
| "eval_code_steps_per_second": 0.771, |
| "step": 492 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_stem_loss": 0.7002437710762024, |
| "eval_stem_runtime": 251.7134, |
| "eval_stem_samples_per_second": 3.631, |
| "eval_stem_steps_per_second": 0.91, |
| "step": 492 |
| }, |
| { |
| "epoch": 2.032619775739042, |
| "grad_norm": 8.038460731506348, |
| "learning_rate": 8.309322033898306e-06, |
| "loss": 0.4989, |
| "step": 500 |
| }, |
| { |
| "epoch": 2.073394495412844, |
| "grad_norm": 11.866241455078125, |
| "learning_rate": 8.266949152542375e-06, |
| "loss": 0.6914, |
| "step": 510 |
| }, |
| { |
| "epoch": 2.1141692150866462, |
| "grad_norm": 6.622210502624512, |
| "learning_rate": 8.224576271186441e-06, |
| "loss": 0.5043, |
| "step": 520 |
| }, |
| { |
| "epoch": 2.1549439347604484, |
| "grad_norm": 11.639506340026855, |
| "learning_rate": 8.182203389830508e-06, |
| "loss": 0.6054, |
| "step": 530 |
| }, |
| { |
| "epoch": 2.1957186544342506, |
| "grad_norm": 8.884720802307129, |
| "learning_rate": 8.139830508474576e-06, |
| "loss": 0.5524, |
| "step": 540 |
| }, |
| { |
| "epoch": 2.236493374108053, |
| "grad_norm": 12.228666305541992, |
| "learning_rate": 8.097457627118644e-06, |
| "loss": 0.6011, |
| "step": 550 |
| }, |
| { |
| "epoch": 2.2772680937818555, |
| "grad_norm": 10.770685195922852, |
| "learning_rate": 8.055084745762713e-06, |
| "loss": 0.5481, |
| "step": 560 |
| }, |
| { |
| "epoch": 2.3180428134556577, |
| "grad_norm": 12.777567863464355, |
| "learning_rate": 8.012711864406781e-06, |
| "loss": 0.496, |
| "step": 570 |
| }, |
| { |
| "epoch": 2.35881753312946, |
| "grad_norm": 12.713486671447754, |
| "learning_rate": 7.970338983050848e-06, |
| "loss": 0.4764, |
| "step": 580 |
| }, |
| { |
| "epoch": 2.399592252803262, |
| "grad_norm": 15.633949279785156, |
| "learning_rate": 7.927966101694916e-06, |
| "loss": 0.6179, |
| "step": 590 |
| }, |
| { |
| "epoch": 2.4403669724770642, |
| "grad_norm": 10.710966110229492, |
| "learning_rate": 7.885593220338984e-06, |
| "loss": 0.5726, |
| "step": 600 |
| }, |
| { |
| "epoch": 2.4811416921508664, |
| "grad_norm": 10.519601821899414, |
| "learning_rate": 7.843220338983052e-06, |
| "loss": 0.4988, |
| "step": 610 |
| }, |
| { |
| "epoch": 2.5219164118246686, |
| "grad_norm": 7.1472392082214355, |
| "learning_rate": 7.800847457627119e-06, |
| "loss": 0.5648, |
| "step": 620 |
| }, |
| { |
| "epoch": 2.562691131498471, |
| "grad_norm": 13.704323768615723, |
| "learning_rate": 7.758474576271187e-06, |
| "loss": 0.6488, |
| "step": 630 |
| }, |
| { |
| "epoch": 2.603465851172273, |
| "grad_norm": 8.187666893005371, |
| "learning_rate": 7.716101694915254e-06, |
| "loss": 0.5047, |
| "step": 640 |
| }, |
| { |
| "epoch": 2.6442405708460752, |
| "grad_norm": 8.175298690795898, |
| "learning_rate": 7.673728813559322e-06, |
| "loss": 0.5948, |
| "step": 650 |
| }, |
| { |
| "epoch": 2.6850152905198774, |
| "grad_norm": 11.140478134155273, |
| "learning_rate": 7.63135593220339e-06, |
| "loss": 0.5212, |
| "step": 660 |
| }, |
| { |
| "epoch": 2.72579001019368, |
| "grad_norm": 10.118830680847168, |
| "learning_rate": 7.588983050847458e-06, |
| "loss": 0.506, |
| "step": 670 |
| }, |
| { |
| "epoch": 2.7665647298674823, |
| "grad_norm": 12.593578338623047, |
| "learning_rate": 7.546610169491526e-06, |
| "loss": 0.5584, |
| "step": 680 |
| }, |
| { |
| "epoch": 2.8073394495412844, |
| "grad_norm": 8.544402122497559, |
| "learning_rate": 7.504237288135594e-06, |
| "loss": 0.5202, |
| "step": 690 |
| }, |
| { |
| "epoch": 2.8481141692150866, |
| "grad_norm": 11.403071403503418, |
| "learning_rate": 7.461864406779662e-06, |
| "loss": 0.6049, |
| "step": 700 |
| }, |
| { |
| "epoch": 2.888888888888889, |
| "grad_norm": 12.620019912719727, |
| "learning_rate": 7.41949152542373e-06, |
| "loss": 0.5484, |
| "step": 710 |
| }, |
| { |
| "epoch": 2.929663608562691, |
| "grad_norm": 8.751862525939941, |
| "learning_rate": 7.377118644067798e-06, |
| "loss": 0.5454, |
| "step": 720 |
| }, |
| { |
| "epoch": 2.9704383282364932, |
| "grad_norm": 10.521064758300781, |
| "learning_rate": 7.334745762711864e-06, |
| "loss": 0.5242, |
| "step": 730 |
| }, |
| { |
| "epoch": 3.0, |
| "eval_general_loss": 0.5933007597923279, |
| "eval_general_runtime": 255.4722, |
| "eval_general_samples_per_second": 3.582, |
| "eval_general_steps_per_second": 0.896, |
| "step": 738 |
| }, |
| { |
| "epoch": 3.0, |
| "eval_code_loss": 0.6812524795532227, |
| "eval_code_runtime": 299.2052, |
| "eval_code_samples_per_second": 3.068, |
| "eval_code_steps_per_second": 0.769, |
| "step": 738 |
| }, |
| { |
| "epoch": 3.0, |
| "eval_stem_loss": 0.6883611679077148, |
| "eval_stem_runtime": 251.705, |
| "eval_stem_samples_per_second": 3.631, |
| "eval_stem_steps_per_second": 0.91, |
| "step": 738 |
| }, |
| { |
| "epoch": 3.0081549439347604, |
| "grad_norm": 5.164834022521973, |
| "learning_rate": 7.292372881355933e-06, |
| "loss": 0.5569, |
| "step": 740 |
| }, |
| { |
| "epoch": 3.0489296636085625, |
| "grad_norm": 8.926407814025879, |
| "learning_rate": 7.25e-06, |
| "loss": 0.512, |
| "step": 750 |
| }, |
| { |
| "epoch": 3.0897043832823647, |
| "grad_norm": 6.960042953491211, |
| "learning_rate": 7.207627118644068e-06, |
| "loss": 0.4701, |
| "step": 760 |
| }, |
| { |
| "epoch": 3.1304791029561674, |
| "grad_norm": 12.42642879486084, |
| "learning_rate": 7.165254237288136e-06, |
| "loss": 0.568, |
| "step": 770 |
| }, |
| { |
| "epoch": 3.1712538226299696, |
| "grad_norm": 13.820547103881836, |
| "learning_rate": 7.122881355932204e-06, |
| "loss": 0.5463, |
| "step": 780 |
| }, |
| { |
| "epoch": 3.2120285423037718, |
| "grad_norm": 10.348419189453125, |
| "learning_rate": 7.080508474576272e-06, |
| "loss": 0.5706, |
| "step": 790 |
| }, |
| { |
| "epoch": 3.252803261977574, |
| "grad_norm": 9.376842498779297, |
| "learning_rate": 7.03813559322034e-06, |
| "loss": 0.5124, |
| "step": 800 |
| }, |
| { |
| "epoch": 3.293577981651376, |
| "grad_norm": 9.71473217010498, |
| "learning_rate": 6.995762711864408e-06, |
| "loss": 0.4291, |
| "step": 810 |
| }, |
| { |
| "epoch": 3.3343527013251784, |
| "grad_norm": 11.891929626464844, |
| "learning_rate": 6.953389830508475e-06, |
| "loss": 0.5006, |
| "step": 820 |
| }, |
| { |
| "epoch": 3.3751274209989806, |
| "grad_norm": 7.028322219848633, |
| "learning_rate": 6.911016949152542e-06, |
| "loss": 0.4249, |
| "step": 830 |
| }, |
| { |
| "epoch": 3.4159021406727827, |
| "grad_norm": 6.9452009201049805, |
| "learning_rate": 6.86864406779661e-06, |
| "loss": 0.4576, |
| "step": 840 |
| }, |
| { |
| "epoch": 3.456676860346585, |
| "grad_norm": 9.390569686889648, |
| "learning_rate": 6.826271186440679e-06, |
| "loss": 0.4251, |
| "step": 850 |
| }, |
| { |
| "epoch": 3.497451580020387, |
| "grad_norm": 8.582491874694824, |
| "learning_rate": 6.783898305084746e-06, |
| "loss": 0.4872, |
| "step": 860 |
| }, |
| { |
| "epoch": 3.5382262996941893, |
| "grad_norm": 11.553034782409668, |
| "learning_rate": 6.741525423728814e-06, |
| "loss": 0.4284, |
| "step": 870 |
| }, |
| { |
| "epoch": 3.579001019367992, |
| "grad_norm": 13.446784019470215, |
| "learning_rate": 6.699152542372882e-06, |
| "loss": 0.4923, |
| "step": 880 |
| }, |
| { |
| "epoch": 3.619775739041794, |
| "grad_norm": 8.266268730163574, |
| "learning_rate": 6.65677966101695e-06, |
| "loss": 0.4186, |
| "step": 890 |
| }, |
| { |
| "epoch": 3.6605504587155964, |
| "grad_norm": 20.191940307617188, |
| "learning_rate": 6.614406779661018e-06, |
| "loss": 0.5954, |
| "step": 900 |
| }, |
| { |
| "epoch": 3.7013251783893986, |
| "grad_norm": 8.65824031829834, |
| "learning_rate": 6.572033898305086e-06, |
| "loss": 0.5261, |
| "step": 910 |
| }, |
| { |
| "epoch": 3.7420998980632008, |
| "grad_norm": 13.471404075622559, |
| "learning_rate": 6.5296610169491525e-06, |
| "loss": 0.4982, |
| "step": 920 |
| }, |
| { |
| "epoch": 3.782874617737003, |
| "grad_norm": 17.626903533935547, |
| "learning_rate": 6.487288135593221e-06, |
| "loss": 0.4446, |
| "step": 930 |
| }, |
| { |
| "epoch": 3.823649337410805, |
| "grad_norm": 18.55869483947754, |
| "learning_rate": 6.444915254237288e-06, |
| "loss": 0.4997, |
| "step": 940 |
| }, |
| { |
| "epoch": 3.864424057084608, |
| "grad_norm": 12.997821807861328, |
| "learning_rate": 6.4025423728813565e-06, |
| "loss": 0.5789, |
| "step": 950 |
| }, |
| { |
| "epoch": 3.90519877675841, |
| "grad_norm": 16.951148986816406, |
| "learning_rate": 6.360169491525424e-06, |
| "loss": 0.5246, |
| "step": 960 |
| }, |
| { |
| "epoch": 3.945973496432212, |
| "grad_norm": 12.54166030883789, |
| "learning_rate": 6.317796610169492e-06, |
| "loss": 0.4787, |
| "step": 970 |
| }, |
| { |
| "epoch": 3.9867482161060144, |
| "grad_norm": 13.518985748291016, |
| "learning_rate": 6.27542372881356e-06, |
| "loss": 0.5098, |
| "step": 980 |
| }, |
| { |
| "epoch": 4.0, |
| "eval_general_loss": 0.5695427060127258, |
| "eval_general_runtime": 255.8406, |
| "eval_general_samples_per_second": 3.576, |
| "eval_general_steps_per_second": 0.895, |
| "step": 984 |
| }, |
| { |
| "epoch": 4.0, |
| "eval_code_loss": 0.6757102608680725, |
| "eval_code_runtime": 297.9071, |
| "eval_code_samples_per_second": 3.081, |
| "eval_code_steps_per_second": 0.772, |
| "step": 984 |
| }, |
| { |
| "epoch": 4.0, |
| "eval_stem_loss": 0.6891591548919678, |
| "eval_stem_runtime": 251.3672, |
| "eval_stem_samples_per_second": 3.636, |
| "eval_stem_steps_per_second": 0.911, |
| "step": 984 |
| }, |
| { |
| "epoch": 4.0244648318042815, |
| "grad_norm": 10.329503059387207, |
| "learning_rate": 6.233050847457628e-06, |
| "loss": 0.4314, |
| "step": 990 |
| }, |
| { |
| "epoch": 4.065239551478084, |
| "grad_norm": 13.447497367858887, |
| "learning_rate": 6.190677966101695e-06, |
| "loss": 0.4775, |
| "step": 1000 |
| }, |
| { |
| "epoch": 4.106014271151886, |
| "grad_norm": 6.44377326965332, |
| "learning_rate": 6.148305084745764e-06, |
| "loss": 0.4268, |
| "step": 1010 |
| }, |
| { |
| "epoch": 4.146788990825688, |
| "grad_norm": 8.970619201660156, |
| "learning_rate": 6.10593220338983e-06, |
| "loss": 0.4198, |
| "step": 1020 |
| }, |
| { |
| "epoch": 4.18756371049949, |
| "grad_norm": 5.479137420654297, |
| "learning_rate": 6.0635593220338986e-06, |
| "loss": 0.4173, |
| "step": 1030 |
| }, |
| { |
| "epoch": 4.2283384301732925, |
| "grad_norm": 8.389227867126465, |
| "learning_rate": 6.021186440677966e-06, |
| "loss": 0.4917, |
| "step": 1040 |
| }, |
| { |
| "epoch": 4.269113149847095, |
| "grad_norm": 6.150132179260254, |
| "learning_rate": 5.978813559322034e-06, |
| "loss": 0.4613, |
| "step": 1050 |
| }, |
| { |
| "epoch": 4.309887869520897, |
| "grad_norm": 11.207158088684082, |
| "learning_rate": 5.936440677966102e-06, |
| "loss": 0.3489, |
| "step": 1060 |
| }, |
| { |
| "epoch": 4.350662589194699, |
| "grad_norm": 7.1269145011901855, |
| "learning_rate": 5.89406779661017e-06, |
| "loss": 0.4335, |
| "step": 1070 |
| }, |
| { |
| "epoch": 4.391437308868501, |
| "grad_norm": 6.830160140991211, |
| "learning_rate": 5.851694915254238e-06, |
| "loss": 0.4218, |
| "step": 1080 |
| }, |
| { |
| "epoch": 4.4322120285423035, |
| "grad_norm": 11.747178077697754, |
| "learning_rate": 5.809322033898306e-06, |
| "loss": 0.5124, |
| "step": 1090 |
| }, |
| { |
| "epoch": 4.472986748216106, |
| "grad_norm": 10.610580444335938, |
| "learning_rate": 5.766949152542374e-06, |
| "loss": 0.3823, |
| "step": 1100 |
| }, |
| { |
| "epoch": 4.513761467889909, |
| "grad_norm": 10.419934272766113, |
| "learning_rate": 5.7245762711864415e-06, |
| "loss": 0.4551, |
| "step": 1110 |
| }, |
| { |
| "epoch": 4.554536187563711, |
| "grad_norm": 9.977579116821289, |
| "learning_rate": 5.682203389830508e-06, |
| "loss": 0.3395, |
| "step": 1120 |
| }, |
| { |
| "epoch": 4.595310907237513, |
| "grad_norm": 22.077423095703125, |
| "learning_rate": 5.639830508474576e-06, |
| "loss": 0.5031, |
| "step": 1130 |
| }, |
| { |
| "epoch": 4.636085626911315, |
| "grad_norm": 8.453546524047852, |
| "learning_rate": 5.597457627118645e-06, |
| "loss": 0.3867, |
| "step": 1140 |
| }, |
| { |
| "epoch": 4.6768603465851175, |
| "grad_norm": 11.514775276184082, |
| "learning_rate": 5.555084745762712e-06, |
| "loss": 0.3771, |
| "step": 1150 |
| }, |
| { |
| "epoch": 4.71763506625892, |
| "grad_norm": 9.698599815368652, |
| "learning_rate": 5.51271186440678e-06, |
| "loss": 0.5053, |
| "step": 1160 |
| }, |
| { |
| "epoch": 4.758409785932722, |
| "grad_norm": 8.122666358947754, |
| "learning_rate": 5.470338983050848e-06, |
| "loss": 0.4136, |
| "step": 1170 |
| }, |
| { |
| "epoch": 4.799184505606524, |
| "grad_norm": 16.033802032470703, |
| "learning_rate": 5.427966101694916e-06, |
| "loss": 0.4073, |
| "step": 1180 |
| }, |
| { |
| "epoch": 4.839959225280326, |
| "grad_norm": 12.248979568481445, |
| "learning_rate": 5.3855932203389836e-06, |
| "loss": 0.4709, |
| "step": 1190 |
| }, |
| { |
| "epoch": 4.8807339449541285, |
| "grad_norm": 12.649367332458496, |
| "learning_rate": 5.343220338983052e-06, |
| "loss": 0.5022, |
| "step": 1200 |
| }, |
| { |
| "epoch": 4.921508664627931, |
| "grad_norm": 15.413296699523926, |
| "learning_rate": 5.300847457627119e-06, |
| "loss": 0.4231, |
| "step": 1210 |
| }, |
| { |
| "epoch": 4.962283384301733, |
| "grad_norm": 14.376687049865723, |
| "learning_rate": 5.258474576271187e-06, |
| "loss": 0.4172, |
| "step": 1220 |
| }, |
| { |
| "epoch": 5.0, |
| "grad_norm": 73.38561248779297, |
| "learning_rate": 5.216101694915254e-06, |
| "loss": 0.5843, |
| "step": 1230 |
| }, |
| { |
| "epoch": 5.0, |
| "eval_general_loss": 0.5886819958686829, |
| "eval_general_runtime": 256.5693, |
| "eval_general_samples_per_second": 3.566, |
| "eval_general_steps_per_second": 0.893, |
| "step": 1230 |
| }, |
| { |
| "epoch": 5.0, |
| "eval_code_loss": 0.667267918586731, |
| "eval_code_runtime": 298.4051, |
| "eval_code_samples_per_second": 3.076, |
| "eval_code_steps_per_second": 0.771, |
| "step": 1230 |
| }, |
| { |
| "epoch": 5.0, |
| "eval_stem_loss": 0.7440467476844788, |
| "eval_stem_runtime": 252.1598, |
| "eval_stem_samples_per_second": 3.625, |
| "eval_stem_steps_per_second": 0.908, |
| "step": 1230 |
| }, |
| { |
| "epoch": 5.040774719673802, |
| "grad_norm": 11.078204154968262, |
| "learning_rate": 5.1737288135593225e-06, |
| "loss": 0.3649, |
| "step": 1240 |
| }, |
| { |
| "epoch": 5.081549439347604, |
| "grad_norm": 17.9708251953125, |
| "learning_rate": 5.13135593220339e-06, |
| "loss": 0.3542, |
| "step": 1250 |
| }, |
| { |
| "epoch": 5.122324159021407, |
| "grad_norm": 23.31520652770996, |
| "learning_rate": 5.088983050847458e-06, |
| "loss": 0.4107, |
| "step": 1260 |
| }, |
| { |
| "epoch": 5.163098878695209, |
| "grad_norm": 11.54787826538086, |
| "learning_rate": 5.046610169491526e-06, |
| "loss": 0.3702, |
| "step": 1270 |
| }, |
| { |
| "epoch": 5.203873598369011, |
| "grad_norm": 14.472805976867676, |
| "learning_rate": 5.004237288135594e-06, |
| "loss": 0.4591, |
| "step": 1280 |
| }, |
| { |
| "epoch": 5.244648318042813, |
| "grad_norm": 8.058154106140137, |
| "learning_rate": 4.961864406779661e-06, |
| "loss": 0.3983, |
| "step": 1290 |
| }, |
| { |
| "epoch": 5.285423037716615, |
| "grad_norm": 10.769429206848145, |
| "learning_rate": 4.919491525423729e-06, |
| "loss": 0.3268, |
| "step": 1300 |
| }, |
| { |
| "epoch": 5.326197757390418, |
| "grad_norm": 18.2336368560791, |
| "learning_rate": 4.877118644067797e-06, |
| "loss": 0.2733, |
| "step": 1310 |
| }, |
| { |
| "epoch": 5.36697247706422, |
| "grad_norm": 9.624357223510742, |
| "learning_rate": 4.8347457627118645e-06, |
| "loss": 0.3423, |
| "step": 1320 |
| }, |
| { |
| "epoch": 5.407747196738022, |
| "grad_norm": 20.92825698852539, |
| "learning_rate": 4.792372881355933e-06, |
| "loss": 0.4394, |
| "step": 1330 |
| }, |
| { |
| "epoch": 5.448521916411825, |
| "grad_norm": 20.59296226501465, |
| "learning_rate": 4.75e-06, |
| "loss": 0.3442, |
| "step": 1340 |
| }, |
| { |
| "epoch": 5.489296636085627, |
| "grad_norm": 11.993790626525879, |
| "learning_rate": 4.707627118644068e-06, |
| "loss": 0.3717, |
| "step": 1350 |
| }, |
| { |
| "epoch": 5.530071355759429, |
| "grad_norm": 14.13767147064209, |
| "learning_rate": 4.665254237288136e-06, |
| "loss": 0.465, |
| "step": 1360 |
| }, |
| { |
| "epoch": 5.570846075433232, |
| "grad_norm": 17.664396286010742, |
| "learning_rate": 4.622881355932204e-06, |
| "loss": 0.3928, |
| "step": 1370 |
| }, |
| { |
| "epoch": 5.611620795107034, |
| "grad_norm": 8.398447036743164, |
| "learning_rate": 4.580508474576272e-06, |
| "loss": 0.3448, |
| "step": 1380 |
| }, |
| { |
| "epoch": 5.652395514780836, |
| "grad_norm": 15.034809112548828, |
| "learning_rate": 4.538135593220339e-06, |
| "loss": 0.3503, |
| "step": 1390 |
| }, |
| { |
| "epoch": 5.693170234454638, |
| "grad_norm": 13.326154708862305, |
| "learning_rate": 4.4957627118644075e-06, |
| "loss": 0.4093, |
| "step": 1400 |
| }, |
| { |
| "epoch": 5.73394495412844, |
| "grad_norm": 35.410621643066406, |
| "learning_rate": 4.453389830508475e-06, |
| "loss": 0.4609, |
| "step": 1410 |
| }, |
| { |
| "epoch": 5.774719673802243, |
| "grad_norm": 10.093385696411133, |
| "learning_rate": 4.411016949152543e-06, |
| "loss": 0.3377, |
| "step": 1420 |
| }, |
| { |
| "epoch": 5.815494393476045, |
| "grad_norm": 4.060236930847168, |
| "learning_rate": 4.368644067796611e-06, |
| "loss": 0.3625, |
| "step": 1430 |
| }, |
| { |
| "epoch": 5.856269113149847, |
| "grad_norm": 12.651063919067383, |
| "learning_rate": 4.326271186440678e-06, |
| "loss": 0.3744, |
| "step": 1440 |
| }, |
| { |
| "epoch": 5.897043832823649, |
| "grad_norm": 17.561542510986328, |
| "learning_rate": 4.283898305084746e-06, |
| "loss": 0.3624, |
| "step": 1450 |
| }, |
| { |
| "epoch": 5.937818552497451, |
| "grad_norm": 17.572458267211914, |
| "learning_rate": 4.241525423728814e-06, |
| "loss": 0.4267, |
| "step": 1460 |
| }, |
| { |
| "epoch": 5.978593272171254, |
| "grad_norm": 15.854689598083496, |
| "learning_rate": 4.199152542372882e-06, |
| "loss": 0.3535, |
| "step": 1470 |
| }, |
| { |
| "epoch": 6.0, |
| "eval_general_loss": 0.586328387260437, |
| "eval_general_runtime": 256.6674, |
| "eval_general_samples_per_second": 3.565, |
| "eval_general_steps_per_second": 0.892, |
| "step": 1476 |
| }, |
| { |
| "epoch": 6.0, |
| "eval_code_loss": 0.6865141987800598, |
| "eval_code_runtime": 299.8173, |
| "eval_code_samples_per_second": 3.062, |
| "eval_code_steps_per_second": 0.767, |
| "step": 1476 |
| }, |
| { |
| "epoch": 6.0, |
| "eval_stem_loss": 0.7701802253723145, |
| "eval_stem_runtime": 253.0522, |
| "eval_stem_samples_per_second": 3.612, |
| "eval_stem_steps_per_second": 0.905, |
| "step": 1476 |
| }, |
| { |
| "epoch": 6.016309887869521, |
| "grad_norm": 21.478351593017578, |
| "learning_rate": 4.1567796610169495e-06, |
| "loss": 0.3105, |
| "step": 1480 |
| }, |
| { |
| "epoch": 6.057084607543323, |
| "grad_norm": 14.138855934143066, |
| "learning_rate": 4.114406779661017e-06, |
| "loss": 0.2904, |
| "step": 1490 |
| }, |
| { |
| "epoch": 6.097859327217125, |
| "grad_norm": 4.692451000213623, |
| "learning_rate": 4.072033898305085e-06, |
| "loss": 0.2857, |
| "step": 1500 |
| }, |
| { |
| "epoch": 6.138634046890927, |
| "grad_norm": 12.039435386657715, |
| "learning_rate": 4.029661016949153e-06, |
| "loss": 0.2959, |
| "step": 1510 |
| }, |
| { |
| "epoch": 6.1794087665647295, |
| "grad_norm": 15.02013111114502, |
| "learning_rate": 3.987288135593221e-06, |
| "loss": 0.2821, |
| "step": 1520 |
| }, |
| { |
| "epoch": 6.220183486238533, |
| "grad_norm": 9.34999942779541, |
| "learning_rate": 3.9449152542372884e-06, |
| "loss": 0.3188, |
| "step": 1530 |
| }, |
| { |
| "epoch": 6.260958205912335, |
| "grad_norm": 6.76429557800293, |
| "learning_rate": 3.902542372881356e-06, |
| "loss": 0.3847, |
| "step": 1540 |
| }, |
| { |
| "epoch": 6.301732925586137, |
| "grad_norm": 18.529054641723633, |
| "learning_rate": 3.860169491525424e-06, |
| "loss": 0.3834, |
| "step": 1550 |
| }, |
| { |
| "epoch": 6.342507645259939, |
| "grad_norm": 18.2543888092041, |
| "learning_rate": 3.817796610169492e-06, |
| "loss": 0.3482, |
| "step": 1560 |
| }, |
| { |
| "epoch": 6.383282364933741, |
| "grad_norm": 13.644851684570312, |
| "learning_rate": 3.77542372881356e-06, |
| "loss": 0.3079, |
| "step": 1570 |
| }, |
| { |
| "epoch": 6.4240570846075435, |
| "grad_norm": 15.071892738342285, |
| "learning_rate": 3.7330508474576277e-06, |
| "loss": 0.3469, |
| "step": 1580 |
| }, |
| { |
| "epoch": 6.464831804281346, |
| "grad_norm": 19.419057846069336, |
| "learning_rate": 3.690677966101695e-06, |
| "loss": 0.2463, |
| "step": 1590 |
| }, |
| { |
| "epoch": 6.505606523955148, |
| "grad_norm": 7.33676290512085, |
| "learning_rate": 3.648305084745763e-06, |
| "loss": 0.3431, |
| "step": 1600 |
| }, |
| { |
| "epoch": 6.54638124362895, |
| "grad_norm": 9.915693283081055, |
| "learning_rate": 3.605932203389831e-06, |
| "loss": 0.3703, |
| "step": 1610 |
| }, |
| { |
| "epoch": 6.587155963302752, |
| "grad_norm": 21.167028427124023, |
| "learning_rate": 3.5635593220338988e-06, |
| "loss": 0.3688, |
| "step": 1620 |
| }, |
| { |
| "epoch": 6.6279306829765545, |
| "grad_norm": 9.088415145874023, |
| "learning_rate": 3.5211864406779667e-06, |
| "loss": 0.3394, |
| "step": 1630 |
| }, |
| { |
| "epoch": 6.668705402650357, |
| "grad_norm": 22.053159713745117, |
| "learning_rate": 3.478813559322034e-06, |
| "loss": 0.3261, |
| "step": 1640 |
| }, |
| { |
| "epoch": 6.709480122324159, |
| "grad_norm": 12.980785369873047, |
| "learning_rate": 3.436440677966102e-06, |
| "loss": 0.3494, |
| "step": 1650 |
| }, |
| { |
| "epoch": 6.750254841997961, |
| "grad_norm": 17.203041076660156, |
| "learning_rate": 3.39406779661017e-06, |
| "loss": 0.3862, |
| "step": 1660 |
| }, |
| { |
| "epoch": 6.791029561671763, |
| "grad_norm": 14.678689002990723, |
| "learning_rate": 3.3516949152542377e-06, |
| "loss": 0.3752, |
| "step": 1670 |
| }, |
| { |
| "epoch": 6.8318042813455655, |
| "grad_norm": 9.484582901000977, |
| "learning_rate": 3.309322033898305e-06, |
| "loss": 0.2588, |
| "step": 1680 |
| }, |
| { |
| "epoch": 6.872579001019368, |
| "grad_norm": 6.880331516265869, |
| "learning_rate": 3.266949152542373e-06, |
| "loss": 0.2594, |
| "step": 1690 |
| }, |
| { |
| "epoch": 6.91335372069317, |
| "grad_norm": 15.112305641174316, |
| "learning_rate": 3.224576271186441e-06, |
| "loss": 0.3514, |
| "step": 1700 |
| }, |
| { |
| "epoch": 6.954128440366972, |
| "grad_norm": 12.322484016418457, |
| "learning_rate": 3.1822033898305087e-06, |
| "loss": 0.2714, |
| "step": 1710 |
| }, |
| { |
| "epoch": 6.994903160040774, |
| "grad_norm": 13.882013320922852, |
| "learning_rate": 3.1398305084745766e-06, |
| "loss": 0.2907, |
| "step": 1720 |
| }, |
| { |
| "epoch": 7.0, |
| "eval_general_loss": 0.6201872229576111, |
| "eval_general_runtime": 255.9089, |
| "eval_general_samples_per_second": 3.575, |
| "eval_general_steps_per_second": 0.895, |
| "step": 1722 |
| }, |
| { |
| "epoch": 7.0, |
| "eval_code_loss": 0.7187511920928955, |
| "eval_code_runtime": 298.4259, |
| "eval_code_samples_per_second": 3.076, |
| "eval_code_steps_per_second": 0.771, |
| "step": 1722 |
| }, |
| { |
| "epoch": 7.0, |
| "eval_stem_loss": 0.8482697606086731, |
| "eval_stem_runtime": 251.9587, |
| "eval_stem_samples_per_second": 3.628, |
| "eval_stem_steps_per_second": 0.909, |
| "step": 1722 |
| } |
| ], |
| "logging_steps": 10, |
| "max_steps": 2460, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 10, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 3.7744572785853727e+18, |
| "train_batch_size": 4, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
