diff --git "a/sft_devstral_24B_v2/checkpoints/checkpoint-1500/trainer_state.json" "b/sft_devstral_24B_v2/checkpoints/checkpoint-1500/trainer_state.json" new file mode 100644--- /dev/null +++ "b/sft_devstral_24B_v2/checkpoints/checkpoint-1500/trainer_state.json" @@ -0,0 +1,5413 @@ +{ + "best_global_step": 1500, + "best_metric": 0.7896141409873962, + "best_model_checkpoint": "task2file/sft_devstral_24B_v2/checkpoints/checkpoint-1500", + "epoch": 0.6329113924050633, + "eval_steps": 100, + "global_step": 1500, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0008438818565400844, + "grad_norm": 1.597854733467102, + "learning_rate": 8.787346221441124e-08, + "loss": 1.3927901983261108, + "step": 2 + }, + { + "epoch": 0.0016877637130801688, + "grad_norm": 1.6547431945800781, + "learning_rate": 2.6362038664323375e-07, + "loss": 1.407160758972168, + "step": 4 + }, + { + "epoch": 0.002531645569620253, + "grad_norm": 1.8221601247787476, + "learning_rate": 4.393673110720563e-07, + "loss": 1.376656174659729, + "step": 6 + }, + { + "epoch": 0.0033755274261603376, + "grad_norm": 1.4831048250198364, + "learning_rate": 6.151142355008788e-07, + "loss": 1.247712254524231, + "step": 8 + }, + { + "epoch": 0.004219409282700422, + "grad_norm": 1.668201208114624, + "learning_rate": 7.908611599297013e-07, + "loss": 1.2685163021087646, + "step": 10 + }, + { + "epoch": 0.005063291139240506, + "grad_norm": 1.67417311668396, + "learning_rate": 9.666080843585237e-07, + "loss": 1.2942761182785034, + "step": 12 + }, + { + "epoch": 0.00590717299578059, + "grad_norm": 1.7154079675674438, + "learning_rate": 1.1423550087873463e-06, + "loss": 1.3638604879379272, + "step": 14 + }, + { + "epoch": 0.006751054852320675, + "grad_norm": 1.729427456855774, + "learning_rate": 1.3181019332161688e-06, + "loss": 1.3476728200912476, + "step": 16 + }, + { + "epoch": 0.007594936708860759, + "grad_norm": 1.3813447952270508, + "learning_rate": 1.4938488576449913e-06, + "loss": 1.3476393222808838, + "step": 18 + }, + { + "epoch": 0.008438818565400843, + "grad_norm": 1.557220458984375, + "learning_rate": 1.6695957820738139e-06, + "loss": 1.2449309825897217, + "step": 20 + }, + { + "epoch": 0.009282700421940928, + "grad_norm": 1.1883500814437866, + "learning_rate": 1.8453427065026362e-06, + "loss": 1.3125361204147339, + "step": 22 + }, + { + "epoch": 0.010126582278481013, + "grad_norm": 1.7290029525756836, + "learning_rate": 2.0210896309314587e-06, + "loss": 1.3724769353866577, + "step": 24 + }, + { + "epoch": 0.010970464135021098, + "grad_norm": 1.5627557039260864, + "learning_rate": 2.1968365553602812e-06, + "loss": 1.3401387929916382, + "step": 26 + }, + { + "epoch": 0.01181434599156118, + "grad_norm": 1.796866774559021, + "learning_rate": 2.3725834797891038e-06, + "loss": 1.365437388420105, + "step": 28 + }, + { + "epoch": 0.012658227848101266, + "grad_norm": 1.7030404806137085, + "learning_rate": 2.5483304042179263e-06, + "loss": 1.2706533670425415, + "step": 30 + }, + { + "epoch": 0.01350210970464135, + "grad_norm": 1.3186293840408325, + "learning_rate": 2.724077328646749e-06, + "loss": 1.3084994554519653, + "step": 32 + }, + { + "epoch": 0.014345991561181435, + "grad_norm": 1.5762513875961304, + "learning_rate": 2.8998242530755714e-06, + "loss": 1.3259696960449219, + "step": 34 + }, + { + "epoch": 0.015189873417721518, + "grad_norm": 1.422295331954956, + "learning_rate": 3.075571177504394e-06, + "loss": 1.3205676078796387, + "step": 36 + }, + { + "epoch": 0.016033755274261603, + "grad_norm": 1.495523452758789, + "learning_rate": 3.2513181019332165e-06, + "loss": 1.3740568161010742, + "step": 38 + }, + { + "epoch": 0.016877637130801686, + "grad_norm": 1.5112254619598389, + "learning_rate": 3.427065026362039e-06, + "loss": 1.321828842163086, + "step": 40 + }, + { + "epoch": 0.017721518987341773, + "grad_norm": 1.4667807817459106, + "learning_rate": 3.602811950790861e-06, + "loss": 1.3673173189163208, + "step": 42 + }, + { + "epoch": 0.018565400843881856, + "grad_norm": 1.6609723567962646, + "learning_rate": 3.7785588752196836e-06, + "loss": 1.3968093395233154, + "step": 44 + }, + { + "epoch": 0.019409282700421943, + "grad_norm": 1.59381103515625, + "learning_rate": 3.954305799648506e-06, + "loss": 1.4295302629470825, + "step": 46 + }, + { + "epoch": 0.020253164556962026, + "grad_norm": 1.1470608711242676, + "learning_rate": 4.130052724077329e-06, + "loss": 1.2536572217941284, + "step": 48 + }, + { + "epoch": 0.02109704641350211, + "grad_norm": 1.2014588117599487, + "learning_rate": 4.305799648506151e-06, + "loss": 1.242217779159546, + "step": 50 + }, + { + "epoch": 0.021940928270042195, + "grad_norm": 1.2327464818954468, + "learning_rate": 4.481546572934974e-06, + "loss": 1.2166963815689087, + "step": 52 + }, + { + "epoch": 0.02278481012658228, + "grad_norm": 1.9708983898162842, + "learning_rate": 4.657293497363796e-06, + "loss": 1.25709867477417, + "step": 54 + }, + { + "epoch": 0.02362869198312236, + "grad_norm": 1.180569052696228, + "learning_rate": 4.833040421792619e-06, + "loss": 1.2886158227920532, + "step": 56 + }, + { + "epoch": 0.024472573839662448, + "grad_norm": 1.5029548406600952, + "learning_rate": 5.008787346221441e-06, + "loss": 1.29886794090271, + "step": 58 + }, + { + "epoch": 0.02531645569620253, + "grad_norm": 1.5380216836929321, + "learning_rate": 5.184534270650264e-06, + "loss": 1.2387628555297852, + "step": 60 + }, + { + "epoch": 0.026160337552742614, + "grad_norm": 1.572144865989685, + "learning_rate": 5.3602811950790864e-06, + "loss": 1.2177000045776367, + "step": 62 + }, + { + "epoch": 0.0270042194092827, + "grad_norm": 1.4882780313491821, + "learning_rate": 5.536028119507909e-06, + "loss": 1.181516170501709, + "step": 64 + }, + { + "epoch": 0.027848101265822784, + "grad_norm": 1.2982488870620728, + "learning_rate": 5.7117750439367315e-06, + "loss": 1.2101733684539795, + "step": 66 + }, + { + "epoch": 0.02869198312236287, + "grad_norm": 1.5236955881118774, + "learning_rate": 5.887521968365554e-06, + "loss": 1.2277681827545166, + "step": 68 + }, + { + "epoch": 0.029535864978902954, + "grad_norm": 1.4521006345748901, + "learning_rate": 6.0632688927943766e-06, + "loss": 1.1688424348831177, + "step": 70 + }, + { + "epoch": 0.030379746835443037, + "grad_norm": 1.2352311611175537, + "learning_rate": 6.239015817223199e-06, + "loss": 1.273059368133545, + "step": 72 + }, + { + "epoch": 0.031223628691983123, + "grad_norm": 1.3438209295272827, + "learning_rate": 6.414762741652021e-06, + "loss": 1.1609034538269043, + "step": 74 + }, + { + "epoch": 0.032067510548523206, + "grad_norm": 1.9009398221969604, + "learning_rate": 6.590509666080843e-06, + "loss": 1.2508260011672974, + "step": 76 + }, + { + "epoch": 0.03291139240506329, + "grad_norm": 1.6718412637710571, + "learning_rate": 6.766256590509666e-06, + "loss": 1.2524956464767456, + "step": 78 + }, + { + "epoch": 0.03375527426160337, + "grad_norm": 1.249891757965088, + "learning_rate": 6.942003514938488e-06, + "loss": 1.1472493410110474, + "step": 80 + }, + { + "epoch": 0.03459915611814346, + "grad_norm": 1.4398653507232666, + "learning_rate": 7.117750439367312e-06, + "loss": 1.0845389366149902, + "step": 82 + }, + { + "epoch": 0.035443037974683546, + "grad_norm": 1.3701167106628418, + "learning_rate": 7.293497363796134e-06, + "loss": 1.1088868379592896, + "step": 84 + }, + { + "epoch": 0.036286919831223625, + "grad_norm": 1.277998924255371, + "learning_rate": 7.469244288224957e-06, + "loss": 1.1513772010803223, + "step": 86 + }, + { + "epoch": 0.03713080168776371, + "grad_norm": 1.4970002174377441, + "learning_rate": 7.644991212653779e-06, + "loss": 1.1385771036148071, + "step": 88 + }, + { + "epoch": 0.0379746835443038, + "grad_norm": 1.3384218215942383, + "learning_rate": 7.820738137082601e-06, + "loss": 1.1632680892944336, + "step": 90 + }, + { + "epoch": 0.038818565400843885, + "grad_norm": 1.4317446947097778, + "learning_rate": 7.996485061511425e-06, + "loss": 1.2256064414978027, + "step": 92 + }, + { + "epoch": 0.039662447257383965, + "grad_norm": 1.8743640184402466, + "learning_rate": 8.172231985940246e-06, + "loss": 1.1935789585113525, + "step": 94 + }, + { + "epoch": 0.04050632911392405, + "grad_norm": 1.4789546728134155, + "learning_rate": 8.347978910369069e-06, + "loss": 1.1429362297058105, + "step": 96 + }, + { + "epoch": 0.04135021097046414, + "grad_norm": 1.658605694770813, + "learning_rate": 8.523725834797891e-06, + "loss": 1.1831508874893188, + "step": 98 + }, + { + "epoch": 0.04219409282700422, + "grad_norm": 1.5077892541885376, + "learning_rate": 8.699472759226714e-06, + "loss": 1.0539867877960205, + "step": 100 + }, + { + "epoch": 0.04219409282700422, + "eval_loss": 1.138856053352356, + "eval_runtime": 859.7128, + "eval_samples_per_second": 2.451, + "eval_steps_per_second": 2.451, + "step": 100 + }, + { + "epoch": 0.043037974683544304, + "grad_norm": 1.4335681200027466, + "learning_rate": 8.875219683655536e-06, + "loss": 1.0719901323318481, + "step": 102 + }, + { + "epoch": 0.04388185654008439, + "grad_norm": 1.7387681007385254, + "learning_rate": 9.050966608084359e-06, + "loss": 1.0654313564300537, + "step": 104 + }, + { + "epoch": 0.04472573839662447, + "grad_norm": 1.6071950197219849, + "learning_rate": 9.226713532513181e-06, + "loss": 1.0752698183059692, + "step": 106 + }, + { + "epoch": 0.04556962025316456, + "grad_norm": 1.40005362033844, + "learning_rate": 9.402460456942004e-06, + "loss": 1.1029763221740723, + "step": 108 + }, + { + "epoch": 0.046413502109704644, + "grad_norm": 2.2338669300079346, + "learning_rate": 9.578207381370826e-06, + "loss": 1.1157960891723633, + "step": 110 + }, + { + "epoch": 0.04725738396624472, + "grad_norm": 1.4972727298736572, + "learning_rate": 9.753954305799649e-06, + "loss": 1.1095420122146606, + "step": 112 + }, + { + "epoch": 0.04810126582278481, + "grad_norm": 1.317979097366333, + "learning_rate": 9.929701230228471e-06, + "loss": 1.109113097190857, + "step": 114 + }, + { + "epoch": 0.048945147679324896, + "grad_norm": 1.496346116065979, + "learning_rate": 1.0105448154657294e-05, + "loss": 1.1055104732513428, + "step": 116 + }, + { + "epoch": 0.049789029535864976, + "grad_norm": 1.385406732559204, + "learning_rate": 1.0281195079086117e-05, + "loss": 1.118395209312439, + "step": 118 + }, + { + "epoch": 0.05063291139240506, + "grad_norm": 1.524222731590271, + "learning_rate": 1.0456942003514939e-05, + "loss": 1.1008446216583252, + "step": 120 + }, + { + "epoch": 0.05147679324894515, + "grad_norm": 1.6308200359344482, + "learning_rate": 1.0632688927943762e-05, + "loss": 1.0891425609588623, + "step": 122 + }, + { + "epoch": 0.05232067510548523, + "grad_norm": 1.3681106567382812, + "learning_rate": 1.0808435852372584e-05, + "loss": 0.9080473184585571, + "step": 124 + }, + { + "epoch": 0.053164556962025315, + "grad_norm": 1.9429908990859985, + "learning_rate": 1.0984182776801407e-05, + "loss": 1.0337369441986084, + "step": 126 + }, + { + "epoch": 0.0540084388185654, + "grad_norm": 1.5830830335617065, + "learning_rate": 1.115992970123023e-05, + "loss": 1.0703333616256714, + "step": 128 + }, + { + "epoch": 0.05485232067510549, + "grad_norm": 1.4792555570602417, + "learning_rate": 1.1335676625659052e-05, + "loss": 1.004652738571167, + "step": 130 + }, + { + "epoch": 0.05569620253164557, + "grad_norm": 1.7196226119995117, + "learning_rate": 1.1511423550087874e-05, + "loss": 0.9798293709754944, + "step": 132 + }, + { + "epoch": 0.056540084388185655, + "grad_norm": 1.8733659982681274, + "learning_rate": 1.1687170474516697e-05, + "loss": 1.0213249921798706, + "step": 134 + }, + { + "epoch": 0.05738396624472574, + "grad_norm": 1.3431142568588257, + "learning_rate": 1.186291739894552e-05, + "loss": 1.0358591079711914, + "step": 136 + }, + { + "epoch": 0.05822784810126582, + "grad_norm": 1.527864933013916, + "learning_rate": 1.2038664323374342e-05, + "loss": 0.9372249841690063, + "step": 138 + }, + { + "epoch": 0.05907172995780591, + "grad_norm": 1.5495563745498657, + "learning_rate": 1.2214411247803164e-05, + "loss": 1.0277758836746216, + "step": 140 + }, + { + "epoch": 0.059915611814345994, + "grad_norm": 1.6792418956756592, + "learning_rate": 1.2390158172231985e-05, + "loss": 1.0349801778793335, + "step": 142 + }, + { + "epoch": 0.060759493670886074, + "grad_norm": 1.6468945741653442, + "learning_rate": 1.256590509666081e-05, + "loss": 0.9578297734260559, + "step": 144 + }, + { + "epoch": 0.06160337552742616, + "grad_norm": 1.7243824005126953, + "learning_rate": 1.2741652021089632e-05, + "loss": 1.0628854036331177, + "step": 146 + }, + { + "epoch": 0.06244725738396625, + "grad_norm": 1.7286981344223022, + "learning_rate": 1.2917398945518455e-05, + "loss": 0.9336449503898621, + "step": 148 + }, + { + "epoch": 0.06329113924050633, + "grad_norm": 1.6411832571029663, + "learning_rate": 1.3093145869947277e-05, + "loss": 0.953730583190918, + "step": 150 + }, + { + "epoch": 0.06413502109704641, + "grad_norm": 1.8297001123428345, + "learning_rate": 1.3268892794376098e-05, + "loss": 1.051239013671875, + "step": 152 + }, + { + "epoch": 0.06497890295358649, + "grad_norm": 1.9660519361495972, + "learning_rate": 1.3444639718804922e-05, + "loss": 0.9955035448074341, + "step": 154 + }, + { + "epoch": 0.06582278481012659, + "grad_norm": 1.8423733711242676, + "learning_rate": 1.3620386643233743e-05, + "loss": 0.913300096988678, + "step": 156 + }, + { + "epoch": 0.06666666666666667, + "grad_norm": 1.9146347045898438, + "learning_rate": 1.3796133567662567e-05, + "loss": 1.0429846048355103, + "step": 158 + }, + { + "epoch": 0.06751054852320675, + "grad_norm": 1.6221821308135986, + "learning_rate": 1.3971880492091388e-05, + "loss": 1.0360238552093506, + "step": 160 + }, + { + "epoch": 0.06835443037974684, + "grad_norm": 2.173283338546753, + "learning_rate": 1.4147627416520212e-05, + "loss": 1.0227266550064087, + "step": 162 + }, + { + "epoch": 0.06919831223628692, + "grad_norm": 1.7091665267944336, + "learning_rate": 1.4323374340949033e-05, + "loss": 1.0075194835662842, + "step": 164 + }, + { + "epoch": 0.070042194092827, + "grad_norm": 1.7219135761260986, + "learning_rate": 1.4499121265377857e-05, + "loss": 1.0044782161712646, + "step": 166 + }, + { + "epoch": 0.07088607594936709, + "grad_norm": 1.6558159589767456, + "learning_rate": 1.4674868189806678e-05, + "loss": 0.9393973350524902, + "step": 168 + }, + { + "epoch": 0.07172995780590717, + "grad_norm": 1.9362739324569702, + "learning_rate": 1.4850615114235502e-05, + "loss": 0.9955337643623352, + "step": 170 + }, + { + "epoch": 0.07257383966244725, + "grad_norm": 1.7792853116989136, + "learning_rate": 1.5026362038664323e-05, + "loss": 0.9659126400947571, + "step": 172 + }, + { + "epoch": 0.07341772151898734, + "grad_norm": 1.7184511423110962, + "learning_rate": 1.5202108963093147e-05, + "loss": 0.9077855348587036, + "step": 174 + }, + { + "epoch": 0.07426160337552742, + "grad_norm": 1.5701428651809692, + "learning_rate": 1.537785588752197e-05, + "loss": 0.9305018782615662, + "step": 176 + }, + { + "epoch": 0.0751054852320675, + "grad_norm": 1.970229148864746, + "learning_rate": 1.555360281195079e-05, + "loss": 1.0211774110794067, + "step": 178 + }, + { + "epoch": 0.0759493670886076, + "grad_norm": 1.8410269021987915, + "learning_rate": 1.5729349736379615e-05, + "loss": 0.9479315876960754, + "step": 180 + }, + { + "epoch": 0.07679324894514768, + "grad_norm": 1.8991246223449707, + "learning_rate": 1.5905096660808434e-05, + "loss": 1.0629050731658936, + "step": 182 + }, + { + "epoch": 0.07763713080168777, + "grad_norm": 1.8052008152008057, + "learning_rate": 1.608084358523726e-05, + "loss": 0.946983814239502, + "step": 184 + }, + { + "epoch": 0.07848101265822785, + "grad_norm": 1.547108769416809, + "learning_rate": 1.625659050966608e-05, + "loss": 0.9413356184959412, + "step": 186 + }, + { + "epoch": 0.07932489451476793, + "grad_norm": 1.9713538885116577, + "learning_rate": 1.6432337434094905e-05, + "loss": 0.9337888956069946, + "step": 188 + }, + { + "epoch": 0.08016877637130802, + "grad_norm": 1.708789348602295, + "learning_rate": 1.6608084358523728e-05, + "loss": 0.9816337823867798, + "step": 190 + }, + { + "epoch": 0.0810126582278481, + "grad_norm": 1.815292477607727, + "learning_rate": 1.678383128295255e-05, + "loss": 1.017122507095337, + "step": 192 + }, + { + "epoch": 0.08185654008438818, + "grad_norm": 1.7950682640075684, + "learning_rate": 1.6959578207381373e-05, + "loss": 0.991599440574646, + "step": 194 + }, + { + "epoch": 0.08270042194092828, + "grad_norm": 1.692512035369873, + "learning_rate": 1.7135325131810195e-05, + "loss": 0.9570834040641785, + "step": 196 + }, + { + "epoch": 0.08354430379746836, + "grad_norm": 2.056089162826538, + "learning_rate": 1.7311072056239018e-05, + "loss": 1.035754919052124, + "step": 198 + }, + { + "epoch": 0.08438818565400844, + "grad_norm": 1.7022203207015991, + "learning_rate": 1.7486818980667837e-05, + "loss": 1.0124205350875854, + "step": 200 + }, + { + "epoch": 0.08438818565400844, + "eval_loss": 0.995743453502655, + "eval_runtime": 846.8257, + "eval_samples_per_second": 2.488, + "eval_steps_per_second": 2.488, + "step": 200 + }, + { + "epoch": 0.08523206751054853, + "grad_norm": 1.6088604927062988, + "learning_rate": 1.7662565905096663e-05, + "loss": 0.8946985006332397, + "step": 202 + }, + { + "epoch": 0.08607594936708861, + "grad_norm": 2.02270770072937, + "learning_rate": 1.7838312829525482e-05, + "loss": 0.976133406162262, + "step": 204 + }, + { + "epoch": 0.08691983122362869, + "grad_norm": 1.7832789421081543, + "learning_rate": 1.8014059753954308e-05, + "loss": 0.9079383611679077, + "step": 206 + }, + { + "epoch": 0.08776371308016878, + "grad_norm": 1.9793545007705688, + "learning_rate": 1.8189806678383127e-05, + "loss": 0.8650367856025696, + "step": 208 + }, + { + "epoch": 0.08860759493670886, + "grad_norm": 1.8124271631240845, + "learning_rate": 1.8365553602811953e-05, + "loss": 0.9327266812324524, + "step": 210 + }, + { + "epoch": 0.08945147679324894, + "grad_norm": 1.8581212759017944, + "learning_rate": 1.8541300527240772e-05, + "loss": 0.9811079502105713, + "step": 212 + }, + { + "epoch": 0.09029535864978903, + "grad_norm": 2.001699447631836, + "learning_rate": 1.8717047451669598e-05, + "loss": 0.9546971321105957, + "step": 214 + }, + { + "epoch": 0.09113924050632911, + "grad_norm": 1.6994978189468384, + "learning_rate": 1.8892794376098417e-05, + "loss": 0.9611319899559021, + "step": 216 + }, + { + "epoch": 0.0919831223628692, + "grad_norm": 2.1379497051239014, + "learning_rate": 1.9068541300527243e-05, + "loss": 0.9781531095504761, + "step": 218 + }, + { + "epoch": 0.09282700421940929, + "grad_norm": 1.8961224555969238, + "learning_rate": 1.9244288224956066e-05, + "loss": 0.9374833106994629, + "step": 220 + }, + { + "epoch": 0.09367088607594937, + "grad_norm": 1.851464033126831, + "learning_rate": 1.9420035149384885e-05, + "loss": 0.9681299328804016, + "step": 222 + }, + { + "epoch": 0.09451476793248945, + "grad_norm": 2.0642266273498535, + "learning_rate": 1.959578207381371e-05, + "loss": 1.0086225271224976, + "step": 224 + }, + { + "epoch": 0.09535864978902954, + "grad_norm": 1.8658756017684937, + "learning_rate": 1.977152899824253e-05, + "loss": 0.9190312623977661, + "step": 226 + }, + { + "epoch": 0.09620253164556962, + "grad_norm": 2.4398674964904785, + "learning_rate": 1.9947275922671356e-05, + "loss": 0.9740874171257019, + "step": 228 + }, + { + "epoch": 0.0970464135021097, + "grad_norm": 1.849183440208435, + "learning_rate": 2.0123022847100175e-05, + "loss": 0.884376049041748, + "step": 230 + }, + { + "epoch": 0.09789029535864979, + "grad_norm": 2.027320384979248, + "learning_rate": 2.0298769771529e-05, + "loss": 0.9116487503051758, + "step": 232 + }, + { + "epoch": 0.09873417721518987, + "grad_norm": 1.6800135374069214, + "learning_rate": 2.047451669595782e-05, + "loss": 0.9035115242004395, + "step": 234 + }, + { + "epoch": 0.09957805907172995, + "grad_norm": 2.2362256050109863, + "learning_rate": 2.0650263620386646e-05, + "loss": 0.9043796062469482, + "step": 236 + }, + { + "epoch": 0.10042194092827005, + "grad_norm": 1.938215970993042, + "learning_rate": 2.0826010544815465e-05, + "loss": 1.0888828039169312, + "step": 238 + }, + { + "epoch": 0.10126582278481013, + "grad_norm": 1.890328049659729, + "learning_rate": 2.100175746924429e-05, + "loss": 0.9960280656814575, + "step": 240 + }, + { + "epoch": 0.1021097046413502, + "grad_norm": 2.021235227584839, + "learning_rate": 2.117750439367311e-05, + "loss": 0.9848901629447937, + "step": 242 + }, + { + "epoch": 0.1029535864978903, + "grad_norm": 2.023920774459839, + "learning_rate": 2.1353251318101936e-05, + "loss": 0.891694188117981, + "step": 244 + }, + { + "epoch": 0.10379746835443038, + "grad_norm": 1.8061069250106812, + "learning_rate": 2.1528998242530755e-05, + "loss": 0.9059976935386658, + "step": 246 + }, + { + "epoch": 0.10464135021097046, + "grad_norm": 2.176302194595337, + "learning_rate": 2.1704745166959578e-05, + "loss": 1.0056109428405762, + "step": 248 + }, + { + "epoch": 0.10548523206751055, + "grad_norm": 1.9820969104766846, + "learning_rate": 2.18804920913884e-05, + "loss": 0.9645357728004456, + "step": 250 + }, + { + "epoch": 0.10632911392405063, + "grad_norm": 1.8764572143554688, + "learning_rate": 2.2056239015817223e-05, + "loss": 1.0178182125091553, + "step": 252 + }, + { + "epoch": 0.10717299578059072, + "grad_norm": 2.56221342086792, + "learning_rate": 2.223198594024605e-05, + "loss": 0.9546761512756348, + "step": 254 + }, + { + "epoch": 0.1080168776371308, + "grad_norm": 2.6779074668884277, + "learning_rate": 2.2407732864674868e-05, + "loss": 0.9300968647003174, + "step": 256 + }, + { + "epoch": 0.10886075949367088, + "grad_norm": 2.140897512435913, + "learning_rate": 2.2583479789103694e-05, + "loss": 0.926638662815094, + "step": 258 + }, + { + "epoch": 0.10970464135021098, + "grad_norm": 2.0880508422851562, + "learning_rate": 2.2759226713532513e-05, + "loss": 1.0681840181350708, + "step": 260 + }, + { + "epoch": 0.11054852320675106, + "grad_norm": 2.7273616790771484, + "learning_rate": 2.293497363796134e-05, + "loss": 1.0840941667556763, + "step": 262 + }, + { + "epoch": 0.11139240506329114, + "grad_norm": 1.6723874807357788, + "learning_rate": 2.3110720562390158e-05, + "loss": 0.8637182116508484, + "step": 264 + }, + { + "epoch": 0.11223628691983123, + "grad_norm": 1.806243896484375, + "learning_rate": 2.3286467486818984e-05, + "loss": 0.9554686546325684, + "step": 266 + }, + { + "epoch": 0.11308016877637131, + "grad_norm": 1.9086743593215942, + "learning_rate": 2.3462214411247803e-05, + "loss": 0.9556593894958496, + "step": 268 + }, + { + "epoch": 0.11392405063291139, + "grad_norm": 2.1822304725646973, + "learning_rate": 2.3637961335676626e-05, + "loss": 0.9177709817886353, + "step": 270 + }, + { + "epoch": 0.11476793248945148, + "grad_norm": 2.1009039878845215, + "learning_rate": 2.3813708260105448e-05, + "loss": 0.9288759827613831, + "step": 272 + }, + { + "epoch": 0.11561181434599156, + "grad_norm": 1.9814810752868652, + "learning_rate": 2.398945518453427e-05, + "loss": 0.9881691932678223, + "step": 274 + }, + { + "epoch": 0.11645569620253164, + "grad_norm": 1.9946284294128418, + "learning_rate": 2.4165202108963093e-05, + "loss": 0.9390727281570435, + "step": 276 + }, + { + "epoch": 0.11729957805907174, + "grad_norm": 2.4489169120788574, + "learning_rate": 2.4340949033391916e-05, + "loss": 0.9625692963600159, + "step": 278 + }, + { + "epoch": 0.11814345991561181, + "grad_norm": 2.0919103622436523, + "learning_rate": 2.451669595782074e-05, + "loss": 0.9304702877998352, + "step": 280 + }, + { + "epoch": 0.1189873417721519, + "grad_norm": 1.912914752960205, + "learning_rate": 2.469244288224956e-05, + "loss": 0.9313994646072388, + "step": 282 + }, + { + "epoch": 0.11983122362869199, + "grad_norm": 2.1553256511688232, + "learning_rate": 2.4868189806678387e-05, + "loss": 1.004011869430542, + "step": 284 + }, + { + "epoch": 0.12067510548523207, + "grad_norm": 2.0129058361053467, + "learning_rate": 2.504393673110721e-05, + "loss": 0.9092531204223633, + "step": 286 + }, + { + "epoch": 0.12151898734177215, + "grad_norm": 2.1632325649261475, + "learning_rate": 2.5219683655536032e-05, + "loss": 0.993347704410553, + "step": 288 + }, + { + "epoch": 0.12236286919831224, + "grad_norm": 2.3072738647460938, + "learning_rate": 2.539543057996485e-05, + "loss": 0.978348433971405, + "step": 290 + }, + { + "epoch": 0.12320675105485232, + "grad_norm": 2.056560516357422, + "learning_rate": 2.5571177504393674e-05, + "loss": 1.0018101930618286, + "step": 292 + }, + { + "epoch": 0.1240506329113924, + "grad_norm": 1.8906747102737427, + "learning_rate": 2.5746924428822493e-05, + "loss": 0.9607775211334229, + "step": 294 + }, + { + "epoch": 0.1248945147679325, + "grad_norm": 2.1375651359558105, + "learning_rate": 2.5922671353251322e-05, + "loss": 0.9259153008460999, + "step": 296 + }, + { + "epoch": 0.1257383966244726, + "grad_norm": 1.9994823932647705, + "learning_rate": 2.609841827768014e-05, + "loss": 0.8524524569511414, + "step": 298 + }, + { + "epoch": 0.12658227848101267, + "grad_norm": 2.2421181201934814, + "learning_rate": 2.6274165202108964e-05, + "loss": 1.0047069787979126, + "step": 300 + }, + { + "epoch": 0.12658227848101267, + "eval_loss": 0.9517185688018799, + "eval_runtime": 860.0287, + "eval_samples_per_second": 2.45, + "eval_steps_per_second": 2.45, + "step": 300 + }, + { + "epoch": 0.12742616033755275, + "grad_norm": 2.1206254959106445, + "learning_rate": 2.6449912126537786e-05, + "loss": 0.8475471138954163, + "step": 302 + }, + { + "epoch": 0.12827004219409283, + "grad_norm": 1.885161280632019, + "learning_rate": 2.6625659050966612e-05, + "loss": 0.8643121123313904, + "step": 304 + }, + { + "epoch": 0.1291139240506329, + "grad_norm": 3.1441781520843506, + "learning_rate": 2.680140597539543e-05, + "loss": 0.8804612159729004, + "step": 306 + }, + { + "epoch": 0.12995780590717299, + "grad_norm": 1.953133225440979, + "learning_rate": 2.6977152899824254e-05, + "loss": 0.8348029255867004, + "step": 308 + }, + { + "epoch": 0.1308016877637131, + "grad_norm": 2.3762667179107666, + "learning_rate": 2.7152899824253076e-05, + "loss": 0.8889057040214539, + "step": 310 + }, + { + "epoch": 0.13164556962025317, + "grad_norm": 2.4651103019714355, + "learning_rate": 2.7328646748681902e-05, + "loss": 1.025565505027771, + "step": 312 + }, + { + "epoch": 0.13248945147679325, + "grad_norm": 1.8522284030914307, + "learning_rate": 2.7504393673110725e-05, + "loss": 0.868915855884552, + "step": 314 + }, + { + "epoch": 0.13333333333333333, + "grad_norm": 1.8048083782196045, + "learning_rate": 2.7680140597539544e-05, + "loss": 0.8821638226509094, + "step": 316 + }, + { + "epoch": 0.1341772151898734, + "grad_norm": 1.9933605194091797, + "learning_rate": 2.7855887521968367e-05, + "loss": 0.8735360503196716, + "step": 318 + }, + { + "epoch": 0.1350210970464135, + "grad_norm": 2.044337034225464, + "learning_rate": 2.8031634446397186e-05, + "loss": 0.8288834691047668, + "step": 320 + }, + { + "epoch": 0.1358649789029536, + "grad_norm": 2.416067361831665, + "learning_rate": 2.8207381370826015e-05, + "loss": 0.9104969501495361, + "step": 322 + }, + { + "epoch": 0.13670886075949368, + "grad_norm": 2.0731265544891357, + "learning_rate": 2.8383128295254834e-05, + "loss": 0.8689924478530884, + "step": 324 + }, + { + "epoch": 0.13755274261603376, + "grad_norm": 2.049126386642456, + "learning_rate": 2.8558875219683657e-05, + "loss": 0.9312222003936768, + "step": 326 + }, + { + "epoch": 0.13839662447257384, + "grad_norm": 2.131026268005371, + "learning_rate": 2.8734622144112476e-05, + "loss": 0.8933501839637756, + "step": 328 + }, + { + "epoch": 0.13924050632911392, + "grad_norm": 1.766754150390625, + "learning_rate": 2.8910369068541305e-05, + "loss": 0.8998261094093323, + "step": 330 + }, + { + "epoch": 0.140084388185654, + "grad_norm": 2.197706460952759, + "learning_rate": 2.9086115992970124e-05, + "loss": 0.8826426267623901, + "step": 332 + }, + { + "epoch": 0.1409282700421941, + "grad_norm": 1.953715443611145, + "learning_rate": 2.9261862917398947e-05, + "loss": 0.8590307831764221, + "step": 334 + }, + { + "epoch": 0.14177215189873418, + "grad_norm": 2.200929880142212, + "learning_rate": 2.943760984182777e-05, + "loss": 0.9317060708999634, + "step": 336 + }, + { + "epoch": 0.14261603375527426, + "grad_norm": 2.1195082664489746, + "learning_rate": 2.961335676625659e-05, + "loss": 0.9965578317642212, + "step": 338 + }, + { + "epoch": 0.14345991561181434, + "grad_norm": 2.3449771404266357, + "learning_rate": 2.9789103690685414e-05, + "loss": 0.8353848457336426, + "step": 340 + }, + { + "epoch": 0.14430379746835442, + "grad_norm": 2.000497579574585, + "learning_rate": 2.9964850615114237e-05, + "loss": 0.9154735803604126, + "step": 342 + }, + { + "epoch": 0.1451476793248945, + "grad_norm": 2.141890525817871, + "learning_rate": 3.014059753954306e-05, + "loss": 0.9530655741691589, + "step": 344 + }, + { + "epoch": 0.1459915611814346, + "grad_norm": 1.7717392444610596, + "learning_rate": 3.031634446397188e-05, + "loss": 0.896998405456543, + "step": 346 + }, + { + "epoch": 0.1468354430379747, + "grad_norm": 1.8796685934066772, + "learning_rate": 3.0492091388400708e-05, + "loss": 0.9084208011627197, + "step": 348 + }, + { + "epoch": 0.14767932489451477, + "grad_norm": 2.0298709869384766, + "learning_rate": 3.066783831282953e-05, + "loss": 0.9183387756347656, + "step": 350 + }, + { + "epoch": 0.14852320675105485, + "grad_norm": 1.9245645999908447, + "learning_rate": 3.084358523725835e-05, + "loss": 0.8624772429466248, + "step": 352 + }, + { + "epoch": 0.14936708860759493, + "grad_norm": 2.325681209564209, + "learning_rate": 3.101933216168717e-05, + "loss": 0.9142400026321411, + "step": 354 + }, + { + "epoch": 0.150210970464135, + "grad_norm": 2.1200530529022217, + "learning_rate": 3.1195079086115995e-05, + "loss": 0.9064018130302429, + "step": 356 + }, + { + "epoch": 0.15105485232067511, + "grad_norm": 1.979314923286438, + "learning_rate": 3.137082601054482e-05, + "loss": 0.9199238419532776, + "step": 358 + }, + { + "epoch": 0.1518987341772152, + "grad_norm": 2.1122689247131348, + "learning_rate": 3.154657293497364e-05, + "loss": 0.8030132055282593, + "step": 360 + }, + { + "epoch": 0.15274261603375527, + "grad_norm": 2.105767250061035, + "learning_rate": 3.172231985940246e-05, + "loss": 0.9185854196548462, + "step": 362 + }, + { + "epoch": 0.15358649789029535, + "grad_norm": 2.179471015930176, + "learning_rate": 3.1898066783831285e-05, + "loss": 0.9365083575248718, + "step": 364 + }, + { + "epoch": 0.15443037974683543, + "grad_norm": 2.1444311141967773, + "learning_rate": 3.207381370826011e-05, + "loss": 0.8965140581130981, + "step": 366 + }, + { + "epoch": 0.15527426160337554, + "grad_norm": 2.4171674251556396, + "learning_rate": 3.224956063268893e-05, + "loss": 0.8787504434585571, + "step": 368 + }, + { + "epoch": 0.15611814345991562, + "grad_norm": 2.418628215789795, + "learning_rate": 3.242530755711775e-05, + "loss": 0.8925284147262573, + "step": 370 + }, + { + "epoch": 0.1569620253164557, + "grad_norm": 2.2228314876556396, + "learning_rate": 3.2601054481546575e-05, + "loss": 0.876179039478302, + "step": 372 + }, + { + "epoch": 0.15780590717299578, + "grad_norm": 2.324237108230591, + "learning_rate": 3.27768014059754e-05, + "loss": 0.8365707993507385, + "step": 374 + }, + { + "epoch": 0.15864978902953586, + "grad_norm": 2.6344552040100098, + "learning_rate": 3.295254833040422e-05, + "loss": 0.7864399552345276, + "step": 376 + }, + { + "epoch": 0.15949367088607594, + "grad_norm": 2.047536611557007, + "learning_rate": 3.312829525483304e-05, + "loss": 0.9271875023841858, + "step": 378 + }, + { + "epoch": 0.16033755274261605, + "grad_norm": 2.120025157928467, + "learning_rate": 3.3304042179261865e-05, + "loss": 0.8799133896827698, + "step": 380 + }, + { + "epoch": 0.16118143459915613, + "grad_norm": 2.363692045211792, + "learning_rate": 3.347978910369069e-05, + "loss": 0.8973530530929565, + "step": 382 + }, + { + "epoch": 0.1620253164556962, + "grad_norm": 2.1796772480010986, + "learning_rate": 3.365553602811951e-05, + "loss": 1.0277652740478516, + "step": 384 + }, + { + "epoch": 0.16286919831223629, + "grad_norm": 1.9192595481872559, + "learning_rate": 3.383128295254833e-05, + "loss": 0.8909643888473511, + "step": 386 + }, + { + "epoch": 0.16371308016877636, + "grad_norm": 1.7874376773834229, + "learning_rate": 3.4007029876977155e-05, + "loss": 0.837049663066864, + "step": 388 + }, + { + "epoch": 0.16455696202531644, + "grad_norm": 2.3402366638183594, + "learning_rate": 3.4182776801405974e-05, + "loss": 0.8625202775001526, + "step": 390 + }, + { + "epoch": 0.16540084388185655, + "grad_norm": 2.1137185096740723, + "learning_rate": 3.43585237258348e-05, + "loss": 0.9288321137428284, + "step": 392 + }, + { + "epoch": 0.16624472573839663, + "grad_norm": 2.3776895999908447, + "learning_rate": 3.453427065026362e-05, + "loss": 0.9328726530075073, + "step": 394 + }, + { + "epoch": 0.1670886075949367, + "grad_norm": 2.34941029548645, + "learning_rate": 3.4710017574692445e-05, + "loss": 0.9273309707641602, + "step": 396 + }, + { + "epoch": 0.1679324894514768, + "grad_norm": 2.1272573471069336, + "learning_rate": 3.4885764499121264e-05, + "loss": 0.8703887462615967, + "step": 398 + }, + { + "epoch": 0.16877637130801687, + "grad_norm": 2.047290802001953, + "learning_rate": 3.506151142355009e-05, + "loss": 0.8808165788650513, + "step": 400 + }, + { + "epoch": 0.16877637130801687, + "eval_loss": 0.9282881617546082, + "eval_runtime": 869.6867, + "eval_samples_per_second": 2.423, + "eval_steps_per_second": 2.423, + "step": 400 + }, + { + "epoch": 0.16962025316455695, + "grad_norm": 1.9874159097671509, + "learning_rate": 3.5237258347978916e-05, + "loss": 0.9643645286560059, + "step": 402 + }, + { + "epoch": 0.17046413502109706, + "grad_norm": 1.9299919605255127, + "learning_rate": 3.5413005272407735e-05, + "loss": 0.9173495769500732, + "step": 404 + }, + { + "epoch": 0.17130801687763714, + "grad_norm": 2.3379697799682617, + "learning_rate": 3.5588752196836555e-05, + "loss": 0.8998411893844604, + "step": 406 + }, + { + "epoch": 0.17215189873417722, + "grad_norm": 2.241370916366577, + "learning_rate": 3.5764499121265374e-05, + "loss": 0.9310802221298218, + "step": 408 + }, + { + "epoch": 0.1729957805907173, + "grad_norm": 2.4490108489990234, + "learning_rate": 3.5940246045694206e-05, + "loss": 0.9605053067207336, + "step": 410 + }, + { + "epoch": 0.17383966244725738, + "grad_norm": 1.8247230052947998, + "learning_rate": 3.6115992970123026e-05, + "loss": 0.8485683798789978, + "step": 412 + }, + { + "epoch": 0.17468354430379746, + "grad_norm": 2.4608843326568604, + "learning_rate": 3.6291739894551845e-05, + "loss": 0.9325968623161316, + "step": 414 + }, + { + "epoch": 0.17552742616033756, + "grad_norm": 1.8923161029815674, + "learning_rate": 3.646748681898067e-05, + "loss": 0.9125096201896667, + "step": 416 + }, + { + "epoch": 0.17637130801687764, + "grad_norm": 1.8502769470214844, + "learning_rate": 3.6643233743409497e-05, + "loss": 0.8852217197418213, + "step": 418 + }, + { + "epoch": 0.17721518987341772, + "grad_norm": 1.9155100584030151, + "learning_rate": 3.6818980667838316e-05, + "loss": 0.9192792773246765, + "step": 420 + }, + { + "epoch": 0.1780590717299578, + "grad_norm": 2.181476593017578, + "learning_rate": 3.6994727592267135e-05, + "loss": 0.8787404298782349, + "step": 422 + }, + { + "epoch": 0.17890295358649788, + "grad_norm": 2.2469847202301025, + "learning_rate": 3.717047451669596e-05, + "loss": 0.9109582901000977, + "step": 424 + }, + { + "epoch": 0.17974683544303796, + "grad_norm": 2.08145809173584, + "learning_rate": 3.734622144112479e-05, + "loss": 0.8560389280319214, + "step": 426 + }, + { + "epoch": 0.18059071729957807, + "grad_norm": 4.121932506561279, + "learning_rate": 3.7521968365553606e-05, + "loss": 0.9456104040145874, + "step": 428 + }, + { + "epoch": 0.18143459915611815, + "grad_norm": 2.177459478378296, + "learning_rate": 3.7697715289982425e-05, + "loss": 0.8421300649642944, + "step": 430 + }, + { + "epoch": 0.18227848101265823, + "grad_norm": 2.324970245361328, + "learning_rate": 3.787346221441125e-05, + "loss": 0.9199858903884888, + "step": 432 + }, + { + "epoch": 0.1831223628691983, + "grad_norm": 2.133718490600586, + "learning_rate": 3.804920913884007e-05, + "loss": 0.8953126668930054, + "step": 434 + }, + { + "epoch": 0.1839662447257384, + "grad_norm": 1.8527995347976685, + "learning_rate": 3.8224956063268896e-05, + "loss": 0.8732239007949829, + "step": 436 + }, + { + "epoch": 0.1848101265822785, + "grad_norm": 1.95817232131958, + "learning_rate": 3.8400702987697715e-05, + "loss": 0.8818746209144592, + "step": 438 + }, + { + "epoch": 0.18565400843881857, + "grad_norm": 2.2107293605804443, + "learning_rate": 3.857644991212654e-05, + "loss": 0.9153507947921753, + "step": 440 + }, + { + "epoch": 0.18649789029535865, + "grad_norm": 2.004754066467285, + "learning_rate": 3.875219683655536e-05, + "loss": 0.8960154056549072, + "step": 442 + }, + { + "epoch": 0.18734177215189873, + "grad_norm": 2.1851706504821777, + "learning_rate": 3.8927943760984186e-05, + "loss": 0.909011721611023, + "step": 444 + }, + { + "epoch": 0.1881856540084388, + "grad_norm": 2.4492485523223877, + "learning_rate": 3.9103690685413005e-05, + "loss": 0.8880158066749573, + "step": 446 + }, + { + "epoch": 0.1890295358649789, + "grad_norm": 2.745453119277954, + "learning_rate": 3.927943760984183e-05, + "loss": 0.8500842452049255, + "step": 448 + }, + { + "epoch": 0.189873417721519, + "grad_norm": 2.1924264430999756, + "learning_rate": 3.945518453427065e-05, + "loss": 0.9004045724868774, + "step": 450 + }, + { + "epoch": 0.19071729957805908, + "grad_norm": 2.4051687717437744, + "learning_rate": 3.9630931458699476e-05, + "loss": 0.9020664095878601, + "step": 452 + }, + { + "epoch": 0.19156118143459916, + "grad_norm": 1.8077667951583862, + "learning_rate": 3.9806678383128295e-05, + "loss": 0.8639500737190247, + "step": 454 + }, + { + "epoch": 0.19240506329113924, + "grad_norm": 2.089043378829956, + "learning_rate": 3.998242530755712e-05, + "loss": 0.8642048239707947, + "step": 456 + }, + { + "epoch": 0.19324894514767932, + "grad_norm": 2.029578447341919, + "learning_rate": 4.015817223198594e-05, + "loss": 0.9371927380561829, + "step": 458 + }, + { + "epoch": 0.1940928270042194, + "grad_norm": 2.26582407951355, + "learning_rate": 4.033391915641476e-05, + "loss": 0.9120588302612305, + "step": 460 + }, + { + "epoch": 0.1949367088607595, + "grad_norm": 1.8671411275863647, + "learning_rate": 4.050966608084359e-05, + "loss": 0.8758644461631775, + "step": 462 + }, + { + "epoch": 0.19578059071729959, + "grad_norm": 1.9403492212295532, + "learning_rate": 4.068541300527241e-05, + "loss": 0.914577305316925, + "step": 464 + }, + { + "epoch": 0.19662447257383966, + "grad_norm": 1.9939641952514648, + "learning_rate": 4.086115992970123e-05, + "loss": 0.8592531681060791, + "step": 466 + }, + { + "epoch": 0.19746835443037974, + "grad_norm": 2.1511380672454834, + "learning_rate": 4.103690685413005e-05, + "loss": 0.9251965880393982, + "step": 468 + }, + { + "epoch": 0.19831223628691982, + "grad_norm": 2.2260982990264893, + "learning_rate": 4.121265377855888e-05, + "loss": 0.8465172052383423, + "step": 470 + }, + { + "epoch": 0.1991561181434599, + "grad_norm": 2.0510010719299316, + "learning_rate": 4.13884007029877e-05, + "loss": 0.8943672180175781, + "step": 472 + }, + { + "epoch": 0.2, + "grad_norm": 2.2040133476257324, + "learning_rate": 4.156414762741652e-05, + "loss": 0.9594319462776184, + "step": 474 + }, + { + "epoch": 0.2008438818565401, + "grad_norm": 2.355181932449341, + "learning_rate": 4.173989455184534e-05, + "loss": 0.9031813144683838, + "step": 476 + }, + { + "epoch": 0.20168776371308017, + "grad_norm": 2.8434665203094482, + "learning_rate": 4.1915641476274166e-05, + "loss": 0.9225798845291138, + "step": 478 + }, + { + "epoch": 0.20253164556962025, + "grad_norm": 2.1715340614318848, + "learning_rate": 4.209138840070299e-05, + "loss": 0.894163966178894, + "step": 480 + }, + { + "epoch": 0.20337552742616033, + "grad_norm": 2.078916072845459, + "learning_rate": 4.226713532513181e-05, + "loss": 0.8424109816551208, + "step": 482 + }, + { + "epoch": 0.2042194092827004, + "grad_norm": 1.9760961532592773, + "learning_rate": 4.244288224956064e-05, + "loss": 0.9102715849876404, + "step": 484 + }, + { + "epoch": 0.20506329113924052, + "grad_norm": 1.9684507846832275, + "learning_rate": 4.2618629173989456e-05, + "loss": 0.8693854808807373, + "step": 486 + }, + { + "epoch": 0.2059071729957806, + "grad_norm": 2.1633450984954834, + "learning_rate": 4.279437609841828e-05, + "loss": 0.8617543578147888, + "step": 488 + }, + { + "epoch": 0.20675105485232068, + "grad_norm": 2.2695257663726807, + "learning_rate": 4.29701230228471e-05, + "loss": 0.9167086482048035, + "step": 490 + }, + { + "epoch": 0.20759493670886076, + "grad_norm": 2.4180049896240234, + "learning_rate": 4.314586994727593e-05, + "loss": 0.8333520889282227, + "step": 492 + }, + { + "epoch": 0.20843881856540084, + "grad_norm": 2.2942769527435303, + "learning_rate": 4.3321616871704746e-05, + "loss": 0.918351411819458, + "step": 494 + }, + { + "epoch": 0.20928270042194091, + "grad_norm": 1.826458215713501, + "learning_rate": 4.349736379613357e-05, + "loss": 0.8565171957015991, + "step": 496 + }, + { + "epoch": 0.21012658227848102, + "grad_norm": 1.9694055318832397, + "learning_rate": 4.367311072056239e-05, + "loss": 0.8684167861938477, + "step": 498 + }, + { + "epoch": 0.2109704641350211, + "grad_norm": 1.892659306526184, + "learning_rate": 4.384885764499122e-05, + "loss": 0.7752788662910461, + "step": 500 + }, + { + "epoch": 0.2109704641350211, + "eval_loss": 0.9080732464790344, + "eval_runtime": 857.0753, + "eval_samples_per_second": 2.458, + "eval_steps_per_second": 2.458, + "step": 500 + }, + { + "epoch": 0.21181434599156118, + "grad_norm": 1.9322253465652466, + "learning_rate": 4.4024604569420036e-05, + "loss": 0.948570728302002, + "step": 502 + }, + { + "epoch": 0.21265822784810126, + "grad_norm": 2.0456058979034424, + "learning_rate": 4.4200351493848855e-05, + "loss": 0.8741024732589722, + "step": 504 + }, + { + "epoch": 0.21350210970464134, + "grad_norm": 2.2406177520751953, + "learning_rate": 4.437609841827768e-05, + "loss": 0.9053841829299927, + "step": 506 + }, + { + "epoch": 0.21434599156118145, + "grad_norm": 2.013934850692749, + "learning_rate": 4.455184534270651e-05, + "loss": 0.8886576294898987, + "step": 508 + }, + { + "epoch": 0.21518987341772153, + "grad_norm": 1.9771125316619873, + "learning_rate": 4.4727592267135326e-05, + "loss": 0.8834167718887329, + "step": 510 + }, + { + "epoch": 0.2160337552742616, + "grad_norm": 1.785905361175537, + "learning_rate": 4.4903339191564146e-05, + "loss": 0.7938863039016724, + "step": 512 + }, + { + "epoch": 0.2168776371308017, + "grad_norm": 1.7946031093597412, + "learning_rate": 4.507908611599297e-05, + "loss": 0.8071596026420593, + "step": 514 + }, + { + "epoch": 0.21772151898734177, + "grad_norm": 2.2217721939086914, + "learning_rate": 4.52548330404218e-05, + "loss": 0.797417163848877, + "step": 516 + }, + { + "epoch": 0.21856540084388185, + "grad_norm": 1.9022471904754639, + "learning_rate": 4.5430579964850617e-05, + "loss": 0.8109536170959473, + "step": 518 + }, + { + "epoch": 0.21940928270042195, + "grad_norm": 1.8988343477249146, + "learning_rate": 4.5606326889279436e-05, + "loss": 0.8647034168243408, + "step": 520 + }, + { + "epoch": 0.22025316455696203, + "grad_norm": 2.6014881134033203, + "learning_rate": 4.578207381370827e-05, + "loss": 0.8763713240623474, + "step": 522 + }, + { + "epoch": 0.2210970464135021, + "grad_norm": 1.9512032270431519, + "learning_rate": 4.595782073813709e-05, + "loss": 0.9525764584541321, + "step": 524 + }, + { + "epoch": 0.2219409282700422, + "grad_norm": 1.9246160984039307, + "learning_rate": 4.613356766256591e-05, + "loss": 0.8839208483695984, + "step": 526 + }, + { + "epoch": 0.22278481012658227, + "grad_norm": 1.9713703393936157, + "learning_rate": 4.6309314586994726e-05, + "loss": 0.8888868093490601, + "step": 528 + }, + { + "epoch": 0.22362869198312235, + "grad_norm": 2.1175239086151123, + "learning_rate": 4.648506151142355e-05, + "loss": 0.8123540878295898, + "step": 530 + }, + { + "epoch": 0.22447257383966246, + "grad_norm": 1.7656135559082031, + "learning_rate": 4.666080843585238e-05, + "loss": 0.7447702884674072, + "step": 532 + }, + { + "epoch": 0.22531645569620254, + "grad_norm": 2.15748929977417, + "learning_rate": 4.68365553602812e-05, + "loss": 0.8778411746025085, + "step": 534 + }, + { + "epoch": 0.22616033755274262, + "grad_norm": 2.1733345985412598, + "learning_rate": 4.7012302284710016e-05, + "loss": 0.8985894918441772, + "step": 536 + }, + { + "epoch": 0.2270042194092827, + "grad_norm": 1.7182204723358154, + "learning_rate": 4.718804920913884e-05, + "loss": 0.8031114339828491, + "step": 538 + }, + { + "epoch": 0.22784810126582278, + "grad_norm": 1.8586329221725464, + "learning_rate": 4.736379613356767e-05, + "loss": 0.9399706721305847, + "step": 540 + }, + { + "epoch": 0.22869198312236286, + "grad_norm": 2.105637311935425, + "learning_rate": 4.753954305799649e-05, + "loss": 0.8672119975090027, + "step": 542 + }, + { + "epoch": 0.22953586497890296, + "grad_norm": 1.760584831237793, + "learning_rate": 4.771528998242531e-05, + "loss": 0.8663905262947083, + "step": 544 + }, + { + "epoch": 0.23037974683544304, + "grad_norm": 1.579990267753601, + "learning_rate": 4.789103690685413e-05, + "loss": 0.8575801849365234, + "step": 546 + }, + { + "epoch": 0.23122362869198312, + "grad_norm": 1.9242485761642456, + "learning_rate": 4.806678383128295e-05, + "loss": 0.828412652015686, + "step": 548 + }, + { + "epoch": 0.2320675105485232, + "grad_norm": 1.812137246131897, + "learning_rate": 4.824253075571178e-05, + "loss": 0.8183464407920837, + "step": 550 + }, + { + "epoch": 0.23291139240506328, + "grad_norm": 1.804733395576477, + "learning_rate": 4.84182776801406e-05, + "loss": 0.7822491526603699, + "step": 552 + }, + { + "epoch": 0.23375527426160336, + "grad_norm": 2.052257537841797, + "learning_rate": 4.859402460456942e-05, + "loss": 0.9050943851470947, + "step": 554 + }, + { + "epoch": 0.23459915611814347, + "grad_norm": 1.9803621768951416, + "learning_rate": 4.876977152899824e-05, + "loss": 0.8846852779388428, + "step": 556 + }, + { + "epoch": 0.23544303797468355, + "grad_norm": 1.820125937461853, + "learning_rate": 4.894551845342707e-05, + "loss": 0.8649531602859497, + "step": 558 + }, + { + "epoch": 0.23628691983122363, + "grad_norm": 2.0963921546936035, + "learning_rate": 4.912126537785589e-05, + "loss": 0.9307748079299927, + "step": 560 + }, + { + "epoch": 0.2371308016877637, + "grad_norm": 2.079697847366333, + "learning_rate": 4.929701230228471e-05, + "loss": 0.9092473387718201, + "step": 562 + }, + { + "epoch": 0.2379746835443038, + "grad_norm": 2.0291287899017334, + "learning_rate": 4.947275922671353e-05, + "loss": 0.8976567983627319, + "step": 564 + }, + { + "epoch": 0.23881856540084387, + "grad_norm": 1.9636707305908203, + "learning_rate": 4.964850615114236e-05, + "loss": 0.8931006193161011, + "step": 566 + }, + { + "epoch": 0.23966244725738398, + "grad_norm": 1.922049880027771, + "learning_rate": 4.982425307557118e-05, + "loss": 0.829562246799469, + "step": 568 + }, + { + "epoch": 0.24050632911392406, + "grad_norm": 2.150334596633911, + "learning_rate": 5e-05, + "loss": 0.8568030595779419, + "step": 570 + }, + { + "epoch": 0.24135021097046414, + "grad_norm": 2.024437427520752, + "learning_rate": 5.017574692442882e-05, + "loss": 0.8623508810997009, + "step": 572 + }, + { + "epoch": 0.24219409282700421, + "grad_norm": 1.8312673568725586, + "learning_rate": 5.035149384885765e-05, + "loss": 0.7853795886039734, + "step": 574 + }, + { + "epoch": 0.2430379746835443, + "grad_norm": 1.9271961450576782, + "learning_rate": 5.0527240773286467e-05, + "loss": 0.9727587103843689, + "step": 576 + }, + { + "epoch": 0.2438818565400844, + "grad_norm": 1.931249976158142, + "learning_rate": 5.0702987697715286e-05, + "loss": 0.8859632015228271, + "step": 578 + }, + { + "epoch": 0.24472573839662448, + "grad_norm": 1.8195210695266724, + "learning_rate": 5.087873462214412e-05, + "loss": 0.8959492444992065, + "step": 580 + }, + { + "epoch": 0.24556962025316456, + "grad_norm": 2.0018749237060547, + "learning_rate": 5.105448154657294e-05, + "loss": 0.8146185874938965, + "step": 582 + }, + { + "epoch": 0.24641350210970464, + "grad_norm": 2.09798526763916, + "learning_rate": 5.1230228471001764e-05, + "loss": 0.8545317053794861, + "step": 584 + }, + { + "epoch": 0.24725738396624472, + "grad_norm": 1.8063944578170776, + "learning_rate": 5.140597539543058e-05, + "loss": 0.8650105595588684, + "step": 586 + }, + { + "epoch": 0.2481012658227848, + "grad_norm": 1.8535740375518799, + "learning_rate": 5.15817223198594e-05, + "loss": 0.8395693302154541, + "step": 588 + }, + { + "epoch": 0.2489451476793249, + "grad_norm": 2.1443960666656494, + "learning_rate": 5.175746924428823e-05, + "loss": 0.8267397284507751, + "step": 590 + }, + { + "epoch": 0.249789029535865, + "grad_norm": 1.9637391567230225, + "learning_rate": 5.193321616871705e-05, + "loss": 0.8500015139579773, + "step": 592 + }, + { + "epoch": 0.25063291139240507, + "grad_norm": 1.9457582235336304, + "learning_rate": 5.2108963093145866e-05, + "loss": 0.887481153011322, + "step": 594 + }, + { + "epoch": 0.2514767932489452, + "grad_norm": 1.7458715438842773, + "learning_rate": 5.228471001757469e-05, + "loss": 0.8444154858589172, + "step": 596 + }, + { + "epoch": 0.2523206751054852, + "grad_norm": 1.8341439962387085, + "learning_rate": 5.2460456942003525e-05, + "loss": 0.8301781415939331, + "step": 598 + }, + { + "epoch": 0.25316455696202533, + "grad_norm": 2.127747058868408, + "learning_rate": 5.2636203866432344e-05, + "loss": 0.8921551704406738, + "step": 600 + }, + { + "epoch": 0.25316455696202533, + "eval_loss": 0.8903881311416626, + "eval_runtime": 845.9969, + "eval_samples_per_second": 2.491, + "eval_steps_per_second": 2.491, + "step": 600 + }, + { + "epoch": 0.2540084388185654, + "grad_norm": 2.421459674835205, + "learning_rate": 5.281195079086116e-05, + "loss": 0.8678019642829895, + "step": 602 + }, + { + "epoch": 0.2548523206751055, + "grad_norm": 1.7736057043075562, + "learning_rate": 5.298769771528999e-05, + "loss": 0.8564275503158569, + "step": 604 + }, + { + "epoch": 0.25569620253164554, + "grad_norm": 2.28430438041687, + "learning_rate": 5.316344463971881e-05, + "loss": 0.8529049158096313, + "step": 606 + }, + { + "epoch": 0.25654008438818565, + "grad_norm": 1.8892366886138916, + "learning_rate": 5.333919156414763e-05, + "loss": 0.8672881126403809, + "step": 608 + }, + { + "epoch": 0.25738396624472576, + "grad_norm": 1.9059702157974243, + "learning_rate": 5.3514938488576446e-05, + "loss": 0.9094445109367371, + "step": 610 + }, + { + "epoch": 0.2582278481012658, + "grad_norm": 2.0657339096069336, + "learning_rate": 5.369068541300527e-05, + "loss": 0.8361946940422058, + "step": 612 + }, + { + "epoch": 0.2590717299578059, + "grad_norm": 1.8987553119659424, + "learning_rate": 5.3866432337434105e-05, + "loss": 0.8319925665855408, + "step": 614 + }, + { + "epoch": 0.25991561181434597, + "grad_norm": 2.1176226139068604, + "learning_rate": 5.4042179261862924e-05, + "loss": 0.9818069934844971, + "step": 616 + }, + { + "epoch": 0.2607594936708861, + "grad_norm": 2.142096519470215, + "learning_rate": 5.421792618629174e-05, + "loss": 0.8675919771194458, + "step": 618 + }, + { + "epoch": 0.2616033755274262, + "grad_norm": 1.9527089595794678, + "learning_rate": 5.439367311072057e-05, + "loss": 0.8845479488372803, + "step": 620 + }, + { + "epoch": 0.26244725738396624, + "grad_norm": 1.7071453332901, + "learning_rate": 5.456942003514939e-05, + "loss": 0.809393048286438, + "step": 622 + }, + { + "epoch": 0.26329113924050634, + "grad_norm": 1.9133527278900146, + "learning_rate": 5.474516695957821e-05, + "loss": 0.8262377977371216, + "step": 624 + }, + { + "epoch": 0.2641350210970464, + "grad_norm": 2.0217554569244385, + "learning_rate": 5.492091388400703e-05, + "loss": 0.9006736278533936, + "step": 626 + }, + { + "epoch": 0.2649789029535865, + "grad_norm": 1.773273229598999, + "learning_rate": 5.509666080843585e-05, + "loss": 0.8243603110313416, + "step": 628 + }, + { + "epoch": 0.26582278481012656, + "grad_norm": 1.6580880880355835, + "learning_rate": 5.527240773286467e-05, + "loss": 0.8112778663635254, + "step": 630 + }, + { + "epoch": 0.26666666666666666, + "grad_norm": 1.8342082500457764, + "learning_rate": 5.5448154657293504e-05, + "loss": 0.8390820622444153, + "step": 632 + }, + { + "epoch": 0.26751054852320677, + "grad_norm": 1.863695502281189, + "learning_rate": 5.5623901581722323e-05, + "loss": 0.8264521360397339, + "step": 634 + }, + { + "epoch": 0.2683544303797468, + "grad_norm": 1.9462928771972656, + "learning_rate": 5.579964850615115e-05, + "loss": 0.9512701630592346, + "step": 636 + }, + { + "epoch": 0.26919831223628693, + "grad_norm": 1.7776058912277222, + "learning_rate": 5.597539543057997e-05, + "loss": 0.9422703981399536, + "step": 638 + }, + { + "epoch": 0.270042194092827, + "grad_norm": 2.9457077980041504, + "learning_rate": 5.615114235500879e-05, + "loss": 0.7991042137145996, + "step": 640 + }, + { + "epoch": 0.2708860759493671, + "grad_norm": 1.445265531539917, + "learning_rate": 5.6326889279437614e-05, + "loss": 0.8188099265098572, + "step": 642 + }, + { + "epoch": 0.2717299578059072, + "grad_norm": 2.063850164413452, + "learning_rate": 5.650263620386643e-05, + "loss": 0.9799772500991821, + "step": 644 + }, + { + "epoch": 0.27257383966244725, + "grad_norm": 2.0488009452819824, + "learning_rate": 5.667838312829525e-05, + "loss": 0.8462742567062378, + "step": 646 + }, + { + "epoch": 0.27341772151898736, + "grad_norm": 1.8747851848602295, + "learning_rate": 5.685413005272408e-05, + "loss": 0.8226412534713745, + "step": 648 + }, + { + "epoch": 0.2742616033755274, + "grad_norm": 1.849074125289917, + "learning_rate": 5.702987697715291e-05, + "loss": 0.9146338105201721, + "step": 650 + }, + { + "epoch": 0.2751054852320675, + "grad_norm": 1.7738500833511353, + "learning_rate": 5.720562390158173e-05, + "loss": 0.7574424147605896, + "step": 652 + }, + { + "epoch": 0.2759493670886076, + "grad_norm": 1.911102294921875, + "learning_rate": 5.738137082601055e-05, + "loss": 0.8930003046989441, + "step": 654 + }, + { + "epoch": 0.2767932489451477, + "grad_norm": 1.5716617107391357, + "learning_rate": 5.755711775043937e-05, + "loss": 0.7578965425491333, + "step": 656 + }, + { + "epoch": 0.2776371308016878, + "grad_norm": 1.789036512374878, + "learning_rate": 5.7732864674868194e-05, + "loss": 0.8149038553237915, + "step": 658 + }, + { + "epoch": 0.27848101265822783, + "grad_norm": 1.68622624874115, + "learning_rate": 5.790861159929701e-05, + "loss": 0.8265765905380249, + "step": 660 + }, + { + "epoch": 0.27932489451476794, + "grad_norm": 2.078423261642456, + "learning_rate": 5.808435852372583e-05, + "loss": 0.9651970267295837, + "step": 662 + }, + { + "epoch": 0.280168776371308, + "grad_norm": 1.7878645658493042, + "learning_rate": 5.826010544815466e-05, + "loss": 0.8295148015022278, + "step": 664 + }, + { + "epoch": 0.2810126582278481, + "grad_norm": 1.970838189125061, + "learning_rate": 5.843585237258348e-05, + "loss": 0.7778491377830505, + "step": 666 + }, + { + "epoch": 0.2818565400843882, + "grad_norm": 1.943596363067627, + "learning_rate": 5.861159929701231e-05, + "loss": 0.9818071722984314, + "step": 668 + }, + { + "epoch": 0.28270042194092826, + "grad_norm": 1.8793812990188599, + "learning_rate": 5.878734622144113e-05, + "loss": 0.9297797083854675, + "step": 670 + }, + { + "epoch": 0.28354430379746837, + "grad_norm": 1.8813483715057373, + "learning_rate": 5.8963093145869955e-05, + "loss": 0.8748109936714172, + "step": 672 + }, + { + "epoch": 0.2843881856540084, + "grad_norm": 1.7658562660217285, + "learning_rate": 5.9138840070298774e-05, + "loss": 0.8505244851112366, + "step": 674 + }, + { + "epoch": 0.2852320675105485, + "grad_norm": 1.6767617464065552, + "learning_rate": 5.931458699472759e-05, + "loss": 0.8476597666740417, + "step": 676 + }, + { + "epoch": 0.28607594936708863, + "grad_norm": 2.703104257583618, + "learning_rate": 5.949033391915641e-05, + "loss": 0.8775192499160767, + "step": 678 + }, + { + "epoch": 0.2869198312236287, + "grad_norm": 1.9959728717803955, + "learning_rate": 5.966608084358524e-05, + "loss": 0.855262279510498, + "step": 680 + }, + { + "epoch": 0.2877637130801688, + "grad_norm": 1.9093716144561768, + "learning_rate": 5.984182776801406e-05, + "loss": 0.7574936151504517, + "step": 682 + }, + { + "epoch": 0.28860759493670884, + "grad_norm": 1.9829599857330322, + "learning_rate": 6.001757469244289e-05, + "loss": 0.8630690574645996, + "step": 684 + }, + { + "epoch": 0.28945147679324895, + "grad_norm": 1.8777490854263306, + "learning_rate": 6.019332161687171e-05, + "loss": 0.8513249158859253, + "step": 686 + }, + { + "epoch": 0.290295358649789, + "grad_norm": 1.9453173875808716, + "learning_rate": 6.0369068541300535e-05, + "loss": 0.9097008109092712, + "step": 688 + }, + { + "epoch": 0.2911392405063291, + "grad_norm": 1.8527908325195312, + "learning_rate": 6.0544815465729354e-05, + "loss": 0.8291722536087036, + "step": 690 + }, + { + "epoch": 0.2919831223628692, + "grad_norm": 1.9255812168121338, + "learning_rate": 6.0720562390158174e-05, + "loss": 0.880009651184082, + "step": 692 + }, + { + "epoch": 0.29282700421940927, + "grad_norm": 1.6637977361679077, + "learning_rate": 6.0896309314587e-05, + "loss": 0.8791794180870056, + "step": 694 + }, + { + "epoch": 0.2936708860759494, + "grad_norm": 1.825940728187561, + "learning_rate": 6.107205623901582e-05, + "loss": 0.8662407398223877, + "step": 696 + }, + { + "epoch": 0.29451476793248943, + "grad_norm": 1.9348198175430298, + "learning_rate": 6.124780316344464e-05, + "loss": 0.8984515070915222, + "step": 698 + }, + { + "epoch": 0.29535864978902954, + "grad_norm": 1.659345030784607, + "learning_rate": 6.142355008787346e-05, + "loss": 0.827385663986206, + "step": 700 + }, + { + "epoch": 0.29535864978902954, + "eval_loss": 0.8730722069740295, + "eval_runtime": 858.184, + "eval_samples_per_second": 2.455, + "eval_steps_per_second": 2.455, + "step": 700 + }, + { + "epoch": 0.29620253164556964, + "grad_norm": 1.6531789302825928, + "learning_rate": 6.159929701230229e-05, + "loss": 0.9337764382362366, + "step": 702 + }, + { + "epoch": 0.2970464135021097, + "grad_norm": 1.8269121646881104, + "learning_rate": 6.177504393673111e-05, + "loss": 0.8250943422317505, + "step": 704 + }, + { + "epoch": 0.2978902953586498, + "grad_norm": 1.692808747291565, + "learning_rate": 6.195079086115994e-05, + "loss": 0.8657428026199341, + "step": 706 + }, + { + "epoch": 0.29873417721518986, + "grad_norm": 1.6736913919448853, + "learning_rate": 6.212653778558876e-05, + "loss": 0.8889590501785278, + "step": 708 + }, + { + "epoch": 0.29957805907172996, + "grad_norm": 1.6841140985488892, + "learning_rate": 6.230228471001758e-05, + "loss": 0.7822914123535156, + "step": 710 + }, + { + "epoch": 0.30042194092827, + "grad_norm": 1.6644599437713623, + "learning_rate": 6.24780316344464e-05, + "loss": 0.8747053742408752, + "step": 712 + }, + { + "epoch": 0.3012658227848101, + "grad_norm": 1.8187819719314575, + "learning_rate": 6.265377855887522e-05, + "loss": 0.8976446390151978, + "step": 714 + }, + { + "epoch": 0.30210970464135023, + "grad_norm": 1.7845178842544556, + "learning_rate": 6.282952548330404e-05, + "loss": 0.9401160478591919, + "step": 716 + }, + { + "epoch": 0.3029535864978903, + "grad_norm": 1.559773564338684, + "learning_rate": 6.300527240773286e-05, + "loss": 0.8754280209541321, + "step": 718 + }, + { + "epoch": 0.3037974683544304, + "grad_norm": 1.5919631719589233, + "learning_rate": 6.318101933216169e-05, + "loss": 0.8278581500053406, + "step": 720 + }, + { + "epoch": 0.30464135021097044, + "grad_norm": 1.8551076650619507, + "learning_rate": 6.335676625659052e-05, + "loss": 0.8868640065193176, + "step": 722 + }, + { + "epoch": 0.30548523206751055, + "grad_norm": 1.6907769441604614, + "learning_rate": 6.353251318101934e-05, + "loss": 0.8631605505943298, + "step": 724 + }, + { + "epoch": 0.30632911392405066, + "grad_norm": 1.820867657661438, + "learning_rate": 6.370826010544816e-05, + "loss": 0.9142873883247375, + "step": 726 + }, + { + "epoch": 0.3071729957805907, + "grad_norm": 1.685154676437378, + "learning_rate": 6.388400702987698e-05, + "loss": 0.8258634805679321, + "step": 728 + }, + { + "epoch": 0.3080168776371308, + "grad_norm": 1.9294627904891968, + "learning_rate": 6.40597539543058e-05, + "loss": 0.9545516967773438, + "step": 730 + }, + { + "epoch": 0.30886075949367087, + "grad_norm": 1.6075409650802612, + "learning_rate": 6.423550087873462e-05, + "loss": 0.8370757699012756, + "step": 732 + }, + { + "epoch": 0.309704641350211, + "grad_norm": 1.635750651359558, + "learning_rate": 6.441124780316345e-05, + "loss": 0.8356084823608398, + "step": 734 + }, + { + "epoch": 0.3105485232067511, + "grad_norm": 1.6376131772994995, + "learning_rate": 6.458699472759227e-05, + "loss": 0.7579531669616699, + "step": 736 + }, + { + "epoch": 0.31139240506329113, + "grad_norm": 1.7135766744613647, + "learning_rate": 6.47627416520211e-05, + "loss": 0.8436318039894104, + "step": 738 + }, + { + "epoch": 0.31223628691983124, + "grad_norm": 1.7095093727111816, + "learning_rate": 6.493848857644992e-05, + "loss": 0.7998805046081543, + "step": 740 + }, + { + "epoch": 0.3130801687763713, + "grad_norm": 1.782615303993225, + "learning_rate": 6.511423550087874e-05, + "loss": 0.915776789188385, + "step": 742 + }, + { + "epoch": 0.3139240506329114, + "grad_norm": 1.8461172580718994, + "learning_rate": 6.528998242530756e-05, + "loss": 0.8300962448120117, + "step": 744 + }, + { + "epoch": 0.31476793248945145, + "grad_norm": 1.5659871101379395, + "learning_rate": 6.546572934973638e-05, + "loss": 0.8239848017692566, + "step": 746 + }, + { + "epoch": 0.31561181434599156, + "grad_norm": 1.9997349977493286, + "learning_rate": 6.56414762741652e-05, + "loss": 0.8236988186836243, + "step": 748 + }, + { + "epoch": 0.31645569620253167, + "grad_norm": 1.9811526536941528, + "learning_rate": 6.581722319859403e-05, + "loss": 0.8516603112220764, + "step": 750 + }, + { + "epoch": 0.3172995780590717, + "grad_norm": 1.9877923727035522, + "learning_rate": 6.599297012302285e-05, + "loss": 0.9037567973136902, + "step": 752 + }, + { + "epoch": 0.3181434599156118, + "grad_norm": 1.6729352474212646, + "learning_rate": 6.616871704745168e-05, + "loss": 0.8350864052772522, + "step": 754 + }, + { + "epoch": 0.3189873417721519, + "grad_norm": 1.9055802822113037, + "learning_rate": 6.63444639718805e-05, + "loss": 0.8246616125106812, + "step": 756 + }, + { + "epoch": 0.319831223628692, + "grad_norm": 1.597999930381775, + "learning_rate": 6.652021089630932e-05, + "loss": 0.8014416098594666, + "step": 758 + }, + { + "epoch": 0.3206751054852321, + "grad_norm": 1.7432531118392944, + "learning_rate": 6.669595782073814e-05, + "loss": 0.9199523329734802, + "step": 760 + }, + { + "epoch": 0.32151898734177214, + "grad_norm": 1.820164442062378, + "learning_rate": 6.687170474516696e-05, + "loss": 0.7764829397201538, + "step": 762 + }, + { + "epoch": 0.32236286919831225, + "grad_norm": 1.6408652067184448, + "learning_rate": 6.704745166959578e-05, + "loss": 0.8072620630264282, + "step": 764 + }, + { + "epoch": 0.3232067510548523, + "grad_norm": 1.8894155025482178, + "learning_rate": 6.722319859402461e-05, + "loss": 0.9006885886192322, + "step": 766 + }, + { + "epoch": 0.3240506329113924, + "grad_norm": 1.6903613805770874, + "learning_rate": 6.739894551845343e-05, + "loss": 0.7772189378738403, + "step": 768 + }, + { + "epoch": 0.32489451476793246, + "grad_norm": 1.7540696859359741, + "learning_rate": 6.757469244288225e-05, + "loss": 0.8825590014457703, + "step": 770 + }, + { + "epoch": 0.32573839662447257, + "grad_norm": 1.603008508682251, + "learning_rate": 6.775043936731108e-05, + "loss": 0.8376453518867493, + "step": 772 + }, + { + "epoch": 0.3265822784810127, + "grad_norm": 1.5381462574005127, + "learning_rate": 6.79261862917399e-05, + "loss": 0.92608243227005, + "step": 774 + }, + { + "epoch": 0.32742616033755273, + "grad_norm": 1.4815537929534912, + "learning_rate": 6.810193321616872e-05, + "loss": 0.6842183470726013, + "step": 776 + }, + { + "epoch": 0.32827004219409284, + "grad_norm": 1.8543411493301392, + "learning_rate": 6.827768014059754e-05, + "loss": 0.8868235349655151, + "step": 778 + }, + { + "epoch": 0.3291139240506329, + "grad_norm": 1.8895748853683472, + "learning_rate": 6.845342706502637e-05, + "loss": 0.8148112297058105, + "step": 780 + }, + { + "epoch": 0.329957805907173, + "grad_norm": 1.8150591850280762, + "learning_rate": 6.862917398945519e-05, + "loss": 0.8760337829589844, + "step": 782 + }, + { + "epoch": 0.3308016877637131, + "grad_norm": 1.6661378145217896, + "learning_rate": 6.880492091388401e-05, + "loss": 0.8266322612762451, + "step": 784 + }, + { + "epoch": 0.33164556962025316, + "grad_norm": 2.2849128246307373, + "learning_rate": 6.898066783831283e-05, + "loss": 0.8599053025245667, + "step": 786 + }, + { + "epoch": 0.33248945147679326, + "grad_norm": 1.7233171463012695, + "learning_rate": 6.915641476274165e-05, + "loss": 0.8312317132949829, + "step": 788 + }, + { + "epoch": 0.3333333333333333, + "grad_norm": 1.7637618780136108, + "learning_rate": 6.933216168717048e-05, + "loss": 0.8379700779914856, + "step": 790 + }, + { + "epoch": 0.3341772151898734, + "grad_norm": 1.7780474424362183, + "learning_rate": 6.95079086115993e-05, + "loss": 0.8994934558868408, + "step": 792 + }, + { + "epoch": 0.33502109704641353, + "grad_norm": 1.5798883438110352, + "learning_rate": 6.968365553602812e-05, + "loss": 0.8021857738494873, + "step": 794 + }, + { + "epoch": 0.3358649789029536, + "grad_norm": 1.7316070795059204, + "learning_rate": 6.985940246045695e-05, + "loss": 0.8814419507980347, + "step": 796 + }, + { + "epoch": 0.3367088607594937, + "grad_norm": 1.711315631866455, + "learning_rate": 7.003514938488577e-05, + "loss": 0.8545029163360596, + "step": 798 + }, + { + "epoch": 0.33755274261603374, + "grad_norm": 1.5023137331008911, + "learning_rate": 7.021089630931459e-05, + "loss": 0.8006189465522766, + "step": 800 + }, + { + "epoch": 0.33755274261603374, + "eval_loss": 0.8635594248771667, + "eval_runtime": 865.9348, + "eval_samples_per_second": 2.433, + "eval_steps_per_second": 2.433, + "step": 800 + }, + { + "epoch": 0.33839662447257385, + "grad_norm": 1.8377124071121216, + "learning_rate": 7.038664323374341e-05, + "loss": 0.7625874280929565, + "step": 802 + }, + { + "epoch": 0.3392405063291139, + "grad_norm": 1.5361332893371582, + "learning_rate": 7.056239015817223e-05, + "loss": 0.8490484356880188, + "step": 804 + }, + { + "epoch": 0.340084388185654, + "grad_norm": 1.8727388381958008, + "learning_rate": 7.073813708260105e-05, + "loss": 0.8915753364562988, + "step": 806 + }, + { + "epoch": 0.3409282700421941, + "grad_norm": 1.567700743675232, + "learning_rate": 7.091388400702988e-05, + "loss": 0.8902620077133179, + "step": 808 + }, + { + "epoch": 0.34177215189873417, + "grad_norm": 1.5302914381027222, + "learning_rate": 7.10896309314587e-05, + "loss": 0.7897103428840637, + "step": 810 + }, + { + "epoch": 0.3426160337552743, + "grad_norm": 1.8819153308868408, + "learning_rate": 7.126537785588753e-05, + "loss": 0.8648831248283386, + "step": 812 + }, + { + "epoch": 0.3434599156118143, + "grad_norm": 1.5671379566192627, + "learning_rate": 7.144112478031635e-05, + "loss": 0.8449499607086182, + "step": 814 + }, + { + "epoch": 0.34430379746835443, + "grad_norm": 1.6570971012115479, + "learning_rate": 7.161687170474517e-05, + "loss": 0.848559558391571, + "step": 816 + }, + { + "epoch": 0.34514767932489454, + "grad_norm": 1.9108437299728394, + "learning_rate": 7.179261862917399e-05, + "loss": 0.8847543597221375, + "step": 818 + }, + { + "epoch": 0.3459915611814346, + "grad_norm": 1.4909496307373047, + "learning_rate": 7.196836555360281e-05, + "loss": 0.7642563581466675, + "step": 820 + }, + { + "epoch": 0.3468354430379747, + "grad_norm": 1.768518328666687, + "learning_rate": 7.214411247803163e-05, + "loss": 0.8714305758476257, + "step": 822 + }, + { + "epoch": 0.34767932489451475, + "grad_norm": 1.715343952178955, + "learning_rate": 7.231985940246046e-05, + "loss": 0.7712987661361694, + "step": 824 + }, + { + "epoch": 0.34852320675105486, + "grad_norm": 1.6687803268432617, + "learning_rate": 7.24956063268893e-05, + "loss": 0.8122798204421997, + "step": 826 + }, + { + "epoch": 0.3493670886075949, + "grad_norm": 1.5160514116287231, + "learning_rate": 7.267135325131811e-05, + "loss": 0.793245792388916, + "step": 828 + }, + { + "epoch": 0.350210970464135, + "grad_norm": 1.6449401378631592, + "learning_rate": 7.284710017574693e-05, + "loss": 0.8747497200965881, + "step": 830 + }, + { + "epoch": 0.3510548523206751, + "grad_norm": 1.3907722234725952, + "learning_rate": 7.302284710017575e-05, + "loss": 0.6743978261947632, + "step": 832 + }, + { + "epoch": 0.3518987341772152, + "grad_norm": 1.633555293083191, + "learning_rate": 7.319859402460457e-05, + "loss": 0.8524789214134216, + "step": 834 + }, + { + "epoch": 0.3527426160337553, + "grad_norm": 1.5414257049560547, + "learning_rate": 7.337434094903339e-05, + "loss": 0.8045110702514648, + "step": 836 + }, + { + "epoch": 0.35358649789029534, + "grad_norm": 1.8520616292953491, + "learning_rate": 7.355008787346221e-05, + "loss": 0.8319593071937561, + "step": 838 + }, + { + "epoch": 0.35443037974683544, + "grad_norm": 1.6629763841629028, + "learning_rate": 7.372583479789104e-05, + "loss": 0.8188939094543457, + "step": 840 + }, + { + "epoch": 0.35527426160337555, + "grad_norm": 1.804087519645691, + "learning_rate": 7.390158172231987e-05, + "loss": 0.8875360488891602, + "step": 842 + }, + { + "epoch": 0.3561181434599156, + "grad_norm": 1.6031663417816162, + "learning_rate": 7.407732864674869e-05, + "loss": 0.8159612417221069, + "step": 844 + }, + { + "epoch": 0.3569620253164557, + "grad_norm": 1.7413033246994019, + "learning_rate": 7.425307557117751e-05, + "loss": 0.8422684669494629, + "step": 846 + }, + { + "epoch": 0.35780590717299576, + "grad_norm": 1.7699719667434692, + "learning_rate": 7.442882249560633e-05, + "loss": 0.9343502521514893, + "step": 848 + }, + { + "epoch": 0.35864978902953587, + "grad_norm": 1.4613301753997803, + "learning_rate": 7.460456942003515e-05, + "loss": 0.8168979287147522, + "step": 850 + }, + { + "epoch": 0.3594936708860759, + "grad_norm": 1.542431354522705, + "learning_rate": 7.478031634446397e-05, + "loss": 0.9014382362365723, + "step": 852 + }, + { + "epoch": 0.36033755274261603, + "grad_norm": 1.6070159673690796, + "learning_rate": 7.49560632688928e-05, + "loss": 0.8162738084793091, + "step": 854 + }, + { + "epoch": 0.36118143459915614, + "grad_norm": 1.7979451417922974, + "learning_rate": 7.513181019332162e-05, + "loss": 0.8354527950286865, + "step": 856 + }, + { + "epoch": 0.3620253164556962, + "grad_norm": 2.327045202255249, + "learning_rate": 7.530755711775044e-05, + "loss": 0.8214042782783508, + "step": 858 + }, + { + "epoch": 0.3628691983122363, + "grad_norm": 1.5085111856460571, + "learning_rate": 7.548330404217927e-05, + "loss": 0.7472147941589355, + "step": 860 + }, + { + "epoch": 0.36371308016877635, + "grad_norm": 1.6006290912628174, + "learning_rate": 7.565905096660809e-05, + "loss": 0.7586950063705444, + "step": 862 + }, + { + "epoch": 0.36455696202531646, + "grad_norm": 1.5170620679855347, + "learning_rate": 7.583479789103691e-05, + "loss": 0.8169914484024048, + "step": 864 + }, + { + "epoch": 0.36540084388185656, + "grad_norm": 1.5848352909088135, + "learning_rate": 7.601054481546573e-05, + "loss": 0.8263922929763794, + "step": 866 + }, + { + "epoch": 0.3662447257383966, + "grad_norm": 1.8502342700958252, + "learning_rate": 7.618629173989455e-05, + "loss": 0.8726240992546082, + "step": 868 + }, + { + "epoch": 0.3670886075949367, + "grad_norm": 1.506847620010376, + "learning_rate": 7.636203866432338e-05, + "loss": 0.7220374941825867, + "step": 870 + }, + { + "epoch": 0.3679324894514768, + "grad_norm": 1.5350452661514282, + "learning_rate": 7.65377855887522e-05, + "loss": 0.8028547167778015, + "step": 872 + }, + { + "epoch": 0.3687763713080169, + "grad_norm": 1.5011043548583984, + "learning_rate": 7.671353251318102e-05, + "loss": 0.7659649848937988, + "step": 874 + }, + { + "epoch": 0.369620253164557, + "grad_norm": 1.7019832134246826, + "learning_rate": 7.688927943760984e-05, + "loss": 0.8773653507232666, + "step": 876 + }, + { + "epoch": 0.37046413502109704, + "grad_norm": 1.4918498992919922, + "learning_rate": 7.706502636203867e-05, + "loss": 0.7977569103240967, + "step": 878 + }, + { + "epoch": 0.37130801687763715, + "grad_norm": 1.6422638893127441, + "learning_rate": 7.724077328646749e-05, + "loss": 0.7491976022720337, + "step": 880 + }, + { + "epoch": 0.3721518987341772, + "grad_norm": 1.7590434551239014, + "learning_rate": 7.741652021089631e-05, + "loss": 0.8754181265830994, + "step": 882 + }, + { + "epoch": 0.3729957805907173, + "grad_norm": 3.868894100189209, + "learning_rate": 7.759226713532513e-05, + "loss": 0.8482301235198975, + "step": 884 + }, + { + "epoch": 0.37383966244725736, + "grad_norm": 2.111875534057617, + "learning_rate": 7.776801405975396e-05, + "loss": 0.8109031915664673, + "step": 886 + }, + { + "epoch": 0.37468354430379747, + "grad_norm": 2.0838418006896973, + "learning_rate": 7.794376098418278e-05, + "loss": 0.8660775423049927, + "step": 888 + }, + { + "epoch": 0.3755274261603376, + "grad_norm": 1.553022027015686, + "learning_rate": 7.81195079086116e-05, + "loss": 0.8418024778366089, + "step": 890 + }, + { + "epoch": 0.3763713080168776, + "grad_norm": 1.334747314453125, + "learning_rate": 7.829525483304042e-05, + "loss": 0.7764869928359985, + "step": 892 + }, + { + "epoch": 0.37721518987341773, + "grad_norm": 1.4692286252975464, + "learning_rate": 7.847100175746925e-05, + "loss": 0.7460401654243469, + "step": 894 + }, + { + "epoch": 0.3780590717299578, + "grad_norm": 1.5374023914337158, + "learning_rate": 7.864674868189807e-05, + "loss": 0.7662873268127441, + "step": 896 + }, + { + "epoch": 0.3789029535864979, + "grad_norm": 1.5662524700164795, + "learning_rate": 7.882249560632689e-05, + "loss": 0.8165306448936462, + "step": 898 + }, + { + "epoch": 0.379746835443038, + "grad_norm": 4.498590469360352, + "learning_rate": 7.899824253075572e-05, + "loss": 0.7913232445716858, + "step": 900 + }, + { + "epoch": 0.379746835443038, + "eval_loss": 0.8491304516792297, + "eval_runtime": 852.6211, + "eval_samples_per_second": 2.471, + "eval_steps_per_second": 2.471, + "step": 900 + }, + { + "epoch": 0.38059071729957805, + "grad_norm": 1.6320613622665405, + "learning_rate": 7.917398945518454e-05, + "loss": 0.8097161054611206, + "step": 902 + }, + { + "epoch": 0.38143459915611816, + "grad_norm": 1.2562934160232544, + "learning_rate": 7.934973637961336e-05, + "loss": 0.786399781703949, + "step": 904 + }, + { + "epoch": 0.3822784810126582, + "grad_norm": 1.6957594156265259, + "learning_rate": 7.952548330404218e-05, + "loss": 0.8385500311851501, + "step": 906 + }, + { + "epoch": 0.3831223628691983, + "grad_norm": 1.6662386655807495, + "learning_rate": 7.9701230228471e-05, + "loss": 0.8157848715782166, + "step": 908 + }, + { + "epoch": 0.38396624472573837, + "grad_norm": 1.6717777252197266, + "learning_rate": 7.987697715289982e-05, + "loss": 0.7937968373298645, + "step": 910 + }, + { + "epoch": 0.3848101265822785, + "grad_norm": 1.399484395980835, + "learning_rate": 8.005272407732865e-05, + "loss": 0.7800109386444092, + "step": 912 + }, + { + "epoch": 0.3856540084388186, + "grad_norm": 1.5671080350875854, + "learning_rate": 8.022847100175747e-05, + "loss": 0.8135939240455627, + "step": 914 + }, + { + "epoch": 0.38649789029535864, + "grad_norm": 1.4427763223648071, + "learning_rate": 8.04042179261863e-05, + "loss": 0.7482035160064697, + "step": 916 + }, + { + "epoch": 0.38734177215189874, + "grad_norm": 1.3314121961593628, + "learning_rate": 8.057996485061512e-05, + "loss": 0.7201873064041138, + "step": 918 + }, + { + "epoch": 0.3881856540084388, + "grad_norm": 1.5695286989212036, + "learning_rate": 8.075571177504394e-05, + "loss": 0.7933040857315063, + "step": 920 + }, + { + "epoch": 0.3890295358649789, + "grad_norm": 1.5091747045516968, + "learning_rate": 8.093145869947276e-05, + "loss": 0.8058338165283203, + "step": 922 + }, + { + "epoch": 0.389873417721519, + "grad_norm": 1.6287630796432495, + "learning_rate": 8.110720562390158e-05, + "loss": 0.7617828249931335, + "step": 924 + }, + { + "epoch": 0.39071729957805906, + "grad_norm": 1.6129482984542847, + "learning_rate": 8.12829525483304e-05, + "loss": 0.8710150122642517, + "step": 926 + }, + { + "epoch": 0.39156118143459917, + "grad_norm": 1.6457173824310303, + "learning_rate": 8.145869947275922e-05, + "loss": 0.9122233390808105, + "step": 928 + }, + { + "epoch": 0.3924050632911392, + "grad_norm": 1.6768827438354492, + "learning_rate": 8.163444639718805e-05, + "loss": 0.8339303731918335, + "step": 930 + }, + { + "epoch": 0.39324894514767933, + "grad_norm": 1.5419740676879883, + "learning_rate": 8.181019332161688e-05, + "loss": 0.8220396041870117, + "step": 932 + }, + { + "epoch": 0.39409282700421944, + "grad_norm": 1.4563747644424438, + "learning_rate": 8.19859402460457e-05, + "loss": 0.8531478047370911, + "step": 934 + }, + { + "epoch": 0.3949367088607595, + "grad_norm": 1.6208328008651733, + "learning_rate": 8.216168717047452e-05, + "loss": 0.8330869078636169, + "step": 936 + }, + { + "epoch": 0.3957805907172996, + "grad_norm": 1.6492482423782349, + "learning_rate": 8.233743409490334e-05, + "loss": 0.8011296987533569, + "step": 938 + }, + { + "epoch": 0.39662447257383965, + "grad_norm": 2.1611905097961426, + "learning_rate": 8.251318101933216e-05, + "loss": 0.8111353516578674, + "step": 940 + }, + { + "epoch": 0.39746835443037976, + "grad_norm": 1.7108231782913208, + "learning_rate": 8.268892794376098e-05, + "loss": 0.8282017111778259, + "step": 942 + }, + { + "epoch": 0.3983122362869198, + "grad_norm": 1.543465495109558, + "learning_rate": 8.286467486818981e-05, + "loss": 0.7770059704780579, + "step": 944 + }, + { + "epoch": 0.3991561181434599, + "grad_norm": 1.419969081878662, + "learning_rate": 8.304042179261863e-05, + "loss": 0.8646430373191833, + "step": 946 + }, + { + "epoch": 0.4, + "grad_norm": 1.5002100467681885, + "learning_rate": 8.321616871704746e-05, + "loss": 0.7949403524398804, + "step": 948 + }, + { + "epoch": 0.4008438818565401, + "grad_norm": 1.38933265209198, + "learning_rate": 8.339191564147628e-05, + "loss": 0.8124079704284668, + "step": 950 + }, + { + "epoch": 0.4016877637130802, + "grad_norm": 1.5948443412780762, + "learning_rate": 8.35676625659051e-05, + "loss": 0.8634148836135864, + "step": 952 + }, + { + "epoch": 0.40253164556962023, + "grad_norm": 1.4437624216079712, + "learning_rate": 8.374340949033392e-05, + "loss": 0.7410681247711182, + "step": 954 + }, + { + "epoch": 0.40337552742616034, + "grad_norm": 1.3457095623016357, + "learning_rate": 8.391915641476274e-05, + "loss": 0.7680280208587646, + "step": 956 + }, + { + "epoch": 0.40421940928270045, + "grad_norm": 1.610288143157959, + "learning_rate": 8.409490333919156e-05, + "loss": 0.7921904921531677, + "step": 958 + }, + { + "epoch": 0.4050632911392405, + "grad_norm": 1.5321530103683472, + "learning_rate": 8.427065026362039e-05, + "loss": 0.8320037126541138, + "step": 960 + }, + { + "epoch": 0.4059071729957806, + "grad_norm": 1.699881672859192, + "learning_rate": 8.444639718804921e-05, + "loss": 0.8303092122077942, + "step": 962 + }, + { + "epoch": 0.40675105485232066, + "grad_norm": 1.591515064239502, + "learning_rate": 8.462214411247804e-05, + "loss": 0.9029796719551086, + "step": 964 + }, + { + "epoch": 0.40759493670886077, + "grad_norm": 1.5930429697036743, + "learning_rate": 8.479789103690686e-05, + "loss": 0.8165359497070312, + "step": 966 + }, + { + "epoch": 0.4084388185654008, + "grad_norm": 1.509774923324585, + "learning_rate": 8.497363796133568e-05, + "loss": 0.8276026248931885, + "step": 968 + }, + { + "epoch": 0.4092827004219409, + "grad_norm": 1.3617016077041626, + "learning_rate": 8.51493848857645e-05, + "loss": 0.8159419894218445, + "step": 970 + }, + { + "epoch": 0.41012658227848103, + "grad_norm": 1.3580708503723145, + "learning_rate": 8.532513181019332e-05, + "loss": 0.7882336378097534, + "step": 972 + }, + { + "epoch": 0.4109704641350211, + "grad_norm": 1.3337358236312866, + "learning_rate": 8.550087873462214e-05, + "loss": 0.7462319731712341, + "step": 974 + }, + { + "epoch": 0.4118143459915612, + "grad_norm": 1.450363278388977, + "learning_rate": 8.567662565905097e-05, + "loss": 0.7500866651535034, + "step": 976 + }, + { + "epoch": 0.41265822784810124, + "grad_norm": 1.5305321216583252, + "learning_rate": 8.585237258347979e-05, + "loss": 0.8432503342628479, + "step": 978 + }, + { + "epoch": 0.41350210970464135, + "grad_norm": 1.2097326517105103, + "learning_rate": 8.602811950790861e-05, + "loss": 0.8330482840538025, + "step": 980 + }, + { + "epoch": 0.41434599156118146, + "grad_norm": 1.3916101455688477, + "learning_rate": 8.620386643233744e-05, + "loss": 0.8137149810791016, + "step": 982 + }, + { + "epoch": 0.4151898734177215, + "grad_norm": 1.6411453485488892, + "learning_rate": 8.637961335676626e-05, + "loss": 0.8273854851722717, + "step": 984 + }, + { + "epoch": 0.4160337552742616, + "grad_norm": 1.6734566688537598, + "learning_rate": 8.655536028119508e-05, + "loss": 0.794026255607605, + "step": 986 + }, + { + "epoch": 0.41687763713080167, + "grad_norm": 1.352325677871704, + "learning_rate": 8.67311072056239e-05, + "loss": 0.7721655368804932, + "step": 988 + }, + { + "epoch": 0.4177215189873418, + "grad_norm": 1.5368729829788208, + "learning_rate": 8.690685413005273e-05, + "loss": 0.8123438954353333, + "step": 990 + }, + { + "epoch": 0.41856540084388183, + "grad_norm": 1.4903568029403687, + "learning_rate": 8.708260105448155e-05, + "loss": 0.8370974659919739, + "step": 992 + }, + { + "epoch": 0.41940928270042194, + "grad_norm": 1.3405622243881226, + "learning_rate": 8.725834797891037e-05, + "loss": 0.780426561832428, + "step": 994 + }, + { + "epoch": 0.42025316455696204, + "grad_norm": 1.4761021137237549, + "learning_rate": 8.743409490333919e-05, + "loss": 0.8304934501647949, + "step": 996 + }, + { + "epoch": 0.4210970464135021, + "grad_norm": 1.520033359527588, + "learning_rate": 8.760984182776801e-05, + "loss": 0.7960568070411682, + "step": 998 + }, + { + "epoch": 0.4219409282700422, + "grad_norm": 1.6916255950927734, + "learning_rate": 8.778558875219684e-05, + "loss": 0.7884663939476013, + "step": 1000 + }, + { + "epoch": 0.4219409282700422, + "eval_loss": 0.8388314247131348, + "eval_runtime": 847.4828, + "eval_samples_per_second": 2.486, + "eval_steps_per_second": 2.486, + "step": 1000 + }, + { + "epoch": 0.42278481012658226, + "grad_norm": 1.6796396970748901, + "learning_rate": 8.796133567662566e-05, + "loss": 0.7930826544761658, + "step": 1002 + }, + { + "epoch": 0.42362869198312236, + "grad_norm": 1.4480048418045044, + "learning_rate": 8.813708260105448e-05, + "loss": 0.7138194441795349, + "step": 1004 + }, + { + "epoch": 0.42447257383966247, + "grad_norm": 1.2499021291732788, + "learning_rate": 8.831282952548331e-05, + "loss": 0.7367453575134277, + "step": 1006 + }, + { + "epoch": 0.4253164556962025, + "grad_norm": 1.6906769275665283, + "learning_rate": 8.848857644991213e-05, + "loss": 0.9051005244255066, + "step": 1008 + }, + { + "epoch": 0.42616033755274263, + "grad_norm": 1.4196792840957642, + "learning_rate": 8.866432337434095e-05, + "loss": 0.7469457387924194, + "step": 1010 + }, + { + "epoch": 0.4270042194092827, + "grad_norm": 1.5132776498794556, + "learning_rate": 8.884007029876977e-05, + "loss": 0.7443049550056458, + "step": 1012 + }, + { + "epoch": 0.4278481012658228, + "grad_norm": 1.335705280303955, + "learning_rate": 8.901581722319859e-05, + "loss": 0.784084677696228, + "step": 1014 + }, + { + "epoch": 0.4286919831223629, + "grad_norm": 1.6510252952575684, + "learning_rate": 8.919156414762741e-05, + "loss": 0.8603647947311401, + "step": 1016 + }, + { + "epoch": 0.42953586497890295, + "grad_norm": 1.35535728931427, + "learning_rate": 8.936731107205624e-05, + "loss": 0.7921645641326904, + "step": 1018 + }, + { + "epoch": 0.43037974683544306, + "grad_norm": 1.4952049255371094, + "learning_rate": 8.954305799648506e-05, + "loss": 0.799993634223938, + "step": 1020 + }, + { + "epoch": 0.4312236286919831, + "grad_norm": 1.5026042461395264, + "learning_rate": 8.97188049209139e-05, + "loss": 0.7697094082832336, + "step": 1022 + }, + { + "epoch": 0.4320675105485232, + "grad_norm": 1.5424275398254395, + "learning_rate": 8.989455184534271e-05, + "loss": 0.7988215684890747, + "step": 1024 + }, + { + "epoch": 0.43291139240506327, + "grad_norm": 1.438716173171997, + "learning_rate": 9.007029876977153e-05, + "loss": 0.7841635942459106, + "step": 1026 + }, + { + "epoch": 0.4337552742616034, + "grad_norm": 1.5040369033813477, + "learning_rate": 9.024604569420035e-05, + "loss": 0.7485025525093079, + "step": 1028 + }, + { + "epoch": 0.4345991561181435, + "grad_norm": 1.4354394674301147, + "learning_rate": 9.042179261862917e-05, + "loss": 0.7735623121261597, + "step": 1030 + }, + { + "epoch": 0.43544303797468353, + "grad_norm": 1.4841680526733398, + "learning_rate": 9.059753954305799e-05, + "loss": 0.8918828964233398, + "step": 1032 + }, + { + "epoch": 0.43628691983122364, + "grad_norm": 1.428813099861145, + "learning_rate": 9.077328646748682e-05, + "loss": 0.835110068321228, + "step": 1034 + }, + { + "epoch": 0.4371308016877637, + "grad_norm": 1.559020757675171, + "learning_rate": 9.094903339191566e-05, + "loss": 0.746295690536499, + "step": 1036 + }, + { + "epoch": 0.4379746835443038, + "grad_norm": 1.6996115446090698, + "learning_rate": 9.112478031634448e-05, + "loss": 0.8089123368263245, + "step": 1038 + }, + { + "epoch": 0.4388185654008439, + "grad_norm": 1.6615465879440308, + "learning_rate": 9.13005272407733e-05, + "loss": 0.8807073831558228, + "step": 1040 + }, + { + "epoch": 0.43966244725738396, + "grad_norm": 1.239142894744873, + "learning_rate": 9.147627416520211e-05, + "loss": 0.7638427019119263, + "step": 1042 + }, + { + "epoch": 0.44050632911392407, + "grad_norm": 1.1915178298950195, + "learning_rate": 9.165202108963093e-05, + "loss": 0.7817409634590149, + "step": 1044 + }, + { + "epoch": 0.4413502109704641, + "grad_norm": 1.6276934146881104, + "learning_rate": 9.182776801405975e-05, + "loss": 0.8586427569389343, + "step": 1046 + }, + { + "epoch": 0.4421940928270042, + "grad_norm": 1.480345606803894, + "learning_rate": 9.200351493848857e-05, + "loss": 0.7481811046600342, + "step": 1048 + }, + { + "epoch": 0.4430379746835443, + "grad_norm": 1.308419108390808, + "learning_rate": 9.21792618629174e-05, + "loss": 0.8074686527252197, + "step": 1050 + }, + { + "epoch": 0.4438818565400844, + "grad_norm": 1.6167182922363281, + "learning_rate": 9.235500878734624e-05, + "loss": 0.8455166816711426, + "step": 1052 + }, + { + "epoch": 0.4447257383966245, + "grad_norm": 1.6058826446533203, + "learning_rate": 9.253075571177506e-05, + "loss": 0.7255295515060425, + "step": 1054 + }, + { + "epoch": 0.44556962025316454, + "grad_norm": 1.6745728254318237, + "learning_rate": 9.270650263620387e-05, + "loss": 0.8329368233680725, + "step": 1056 + }, + { + "epoch": 0.44641350210970465, + "grad_norm": 1.5657380819320679, + "learning_rate": 9.28822495606327e-05, + "loss": 0.8583613634109497, + "step": 1058 + }, + { + "epoch": 0.4472573839662447, + "grad_norm": 1.5052601099014282, + "learning_rate": 9.305799648506151e-05, + "loss": 0.8546127080917358, + "step": 1060 + }, + { + "epoch": 0.4481012658227848, + "grad_norm": 1.510636806488037, + "learning_rate": 9.323374340949033e-05, + "loss": 0.8416863679885864, + "step": 1062 + }, + { + "epoch": 0.4489451476793249, + "grad_norm": 1.4446617364883423, + "learning_rate": 9.340949033391916e-05, + "loss": 0.830390453338623, + "step": 1064 + }, + { + "epoch": 0.44978902953586497, + "grad_norm": 1.6032582521438599, + "learning_rate": 9.358523725834798e-05, + "loss": 0.8000447154045105, + "step": 1066 + }, + { + "epoch": 0.4506329113924051, + "grad_norm": 1.5295692682266235, + "learning_rate": 9.37609841827768e-05, + "loss": 0.8310818672180176, + "step": 1068 + }, + { + "epoch": 0.45147679324894513, + "grad_norm": 1.3161942958831787, + "learning_rate": 9.393673110720564e-05, + "loss": 0.8377846479415894, + "step": 1070 + }, + { + "epoch": 0.45232067510548524, + "grad_norm": 1.4101601839065552, + "learning_rate": 9.411247803163445e-05, + "loss": 0.7852389216423035, + "step": 1072 + }, + { + "epoch": 0.4531645569620253, + "grad_norm": 1.4352775812149048, + "learning_rate": 9.428822495606327e-05, + "loss": 0.8763723969459534, + "step": 1074 + }, + { + "epoch": 0.4540084388185654, + "grad_norm": 1.4584673643112183, + "learning_rate": 9.44639718804921e-05, + "loss": 0.8177199363708496, + "step": 1076 + }, + { + "epoch": 0.4548523206751055, + "grad_norm": 1.6470575332641602, + "learning_rate": 9.463971880492091e-05, + "loss": 0.8333053588867188, + "step": 1078 + }, + { + "epoch": 0.45569620253164556, + "grad_norm": 1.4429512023925781, + "learning_rate": 9.481546572934975e-05, + "loss": 0.8546649217605591, + "step": 1080 + }, + { + "epoch": 0.45654008438818566, + "grad_norm": 1.4885371923446655, + "learning_rate": 9.499121265377856e-05, + "loss": 0.838036298751831, + "step": 1082 + }, + { + "epoch": 0.4573839662447257, + "grad_norm": 1.4601678848266602, + "learning_rate": 9.516695957820738e-05, + "loss": 0.7295010089874268, + "step": 1084 + }, + { + "epoch": 0.4582278481012658, + "grad_norm": 1.2399365901947021, + "learning_rate": 9.53427065026362e-05, + "loss": 0.6990782618522644, + "step": 1086 + }, + { + "epoch": 0.45907172995780593, + "grad_norm": 1.2936921119689941, + "learning_rate": 9.551845342706504e-05, + "loss": 0.7790928483009338, + "step": 1088 + }, + { + "epoch": 0.459915611814346, + "grad_norm": 1.3408331871032715, + "learning_rate": 9.569420035149385e-05, + "loss": 0.8061056733131409, + "step": 1090 + }, + { + "epoch": 0.4607594936708861, + "grad_norm": 1.5525178909301758, + "learning_rate": 9.586994727592267e-05, + "loss": 0.856796383857727, + "step": 1092 + }, + { + "epoch": 0.46160337552742614, + "grad_norm": 1.2944618463516235, + "learning_rate": 9.604569420035149e-05, + "loss": 0.7626663446426392, + "step": 1094 + }, + { + "epoch": 0.46244725738396625, + "grad_norm": 1.412204623222351, + "learning_rate": 9.622144112478033e-05, + "loss": 0.7524681091308594, + "step": 1096 + }, + { + "epoch": 0.46329113924050636, + "grad_norm": 1.4851596355438232, + "learning_rate": 9.639718804920914e-05, + "loss": 0.8430375456809998, + "step": 1098 + }, + { + "epoch": 0.4641350210970464, + "grad_norm": 1.831943154335022, + "learning_rate": 9.657293497363796e-05, + "loss": 0.8374918103218079, + "step": 1100 + }, + { + "epoch": 0.4641350210970464, + "eval_loss": 0.8283821940422058, + "eval_runtime": 861.0464, + "eval_samples_per_second": 2.447, + "eval_steps_per_second": 2.447, + "step": 1100 + }, + { + "epoch": 0.4649789029535865, + "grad_norm": 1.4989945888519287, + "learning_rate": 9.674868189806678e-05, + "loss": 0.8063139915466309, + "step": 1102 + }, + { + "epoch": 0.46582278481012657, + "grad_norm": 1.3772722482681274, + "learning_rate": 9.692442882249562e-05, + "loss": 0.8109207153320312, + "step": 1104 + }, + { + "epoch": 0.4666666666666667, + "grad_norm": 1.4963124990463257, + "learning_rate": 9.710017574692443e-05, + "loss": 0.8667853474617004, + "step": 1106 + }, + { + "epoch": 0.4675105485232067, + "grad_norm": 1.4250836372375488, + "learning_rate": 9.727592267135325e-05, + "loss": 0.8020523190498352, + "step": 1108 + }, + { + "epoch": 0.46835443037974683, + "grad_norm": 1.475599765777588, + "learning_rate": 9.745166959578209e-05, + "loss": 0.8271048069000244, + "step": 1110 + }, + { + "epoch": 0.46919831223628694, + "grad_norm": 1.3727436065673828, + "learning_rate": 9.76274165202109e-05, + "loss": 0.7615619897842407, + "step": 1112 + }, + { + "epoch": 0.470042194092827, + "grad_norm": 1.2233914136886597, + "learning_rate": 9.780316344463972e-05, + "loss": 0.7843242883682251, + "step": 1114 + }, + { + "epoch": 0.4708860759493671, + "grad_norm": 1.5734832286834717, + "learning_rate": 9.797891036906854e-05, + "loss": 0.834839940071106, + "step": 1116 + }, + { + "epoch": 0.47172995780590715, + "grad_norm": 1.3778531551361084, + "learning_rate": 9.815465729349736e-05, + "loss": 0.7584373950958252, + "step": 1118 + }, + { + "epoch": 0.47257383966244726, + "grad_norm": 1.5535035133361816, + "learning_rate": 9.833040421792618e-05, + "loss": 0.8204697370529175, + "step": 1120 + }, + { + "epoch": 0.47341772151898737, + "grad_norm": 1.4743636846542358, + "learning_rate": 9.850615114235501e-05, + "loss": 0.9012852311134338, + "step": 1122 + }, + { + "epoch": 0.4742616033755274, + "grad_norm": 1.4134864807128906, + "learning_rate": 9.868189806678383e-05, + "loss": 0.8392805457115173, + "step": 1124 + }, + { + "epoch": 0.4751054852320675, + "grad_norm": 1.3308019638061523, + "learning_rate": 9.885764499121267e-05, + "loss": 0.7135441303253174, + "step": 1126 + }, + { + "epoch": 0.4759493670886076, + "grad_norm": 1.5354844331741333, + "learning_rate": 9.903339191564149e-05, + "loss": 0.8464727401733398, + "step": 1128 + }, + { + "epoch": 0.4767932489451477, + "grad_norm": 1.2730523347854614, + "learning_rate": 9.92091388400703e-05, + "loss": 0.7691597938537598, + "step": 1130 + }, + { + "epoch": 0.47763713080168774, + "grad_norm": 1.5459758043289185, + "learning_rate": 9.938488576449912e-05, + "loss": 0.8068788647651672, + "step": 1132 + }, + { + "epoch": 0.47848101265822784, + "grad_norm": 1.345678687095642, + "learning_rate": 9.956063268892794e-05, + "loss": 0.8091006278991699, + "step": 1134 + }, + { + "epoch": 0.47932489451476795, + "grad_norm": 1.317076563835144, + "learning_rate": 9.973637961335676e-05, + "loss": 0.735533595085144, + "step": 1136 + }, + { + "epoch": 0.480168776371308, + "grad_norm": 1.5011168718338013, + "learning_rate": 9.99121265377856e-05, + "loss": 0.7935182452201843, + "step": 1138 + }, + { + "epoch": 0.4810126582278481, + "grad_norm": 1.673899531364441, + "learning_rate": 9.999999855824502e-05, + "loss": 0.8203520774841309, + "step": 1140 + }, + { + "epoch": 0.48185654008438816, + "grad_norm": 1.344337821006775, + "learning_rate": 9.999998702420562e-05, + "loss": 0.7233241200447083, + "step": 1142 + }, + { + "epoch": 0.48270042194092827, + "grad_norm": 1.5819076299667358, + "learning_rate": 9.999996395612948e-05, + "loss": 0.8795552849769592, + "step": 1144 + }, + { + "epoch": 0.4835443037974684, + "grad_norm": 1.7427241802215576, + "learning_rate": 9.999992935402192e-05, + "loss": 0.8482733964920044, + "step": 1146 + }, + { + "epoch": 0.48438818565400843, + "grad_norm": 1.2877503633499146, + "learning_rate": 9.999988321789093e-05, + "loss": 0.7905706167221069, + "step": 1148 + }, + { + "epoch": 0.48523206751054854, + "grad_norm": 1.4887222051620483, + "learning_rate": 9.999982554774715e-05, + "loss": 0.8609708547592163, + "step": 1150 + }, + { + "epoch": 0.4860759493670886, + "grad_norm": 1.3625136613845825, + "learning_rate": 9.999975634360388e-05, + "loss": 0.7890065908432007, + "step": 1152 + }, + { + "epoch": 0.4869198312236287, + "grad_norm": 1.3631492853164673, + "learning_rate": 9.999967560547708e-05, + "loss": 0.7908958196640015, + "step": 1154 + }, + { + "epoch": 0.4877637130801688, + "grad_norm": 1.5244156122207642, + "learning_rate": 9.99995833333854e-05, + "loss": 0.8509655594825745, + "step": 1156 + }, + { + "epoch": 0.48860759493670886, + "grad_norm": 1.2513200044631958, + "learning_rate": 9.999947952735007e-05, + "loss": 0.7329106330871582, + "step": 1158 + }, + { + "epoch": 0.48945147679324896, + "grad_norm": 1.1539413928985596, + "learning_rate": 9.99993641873951e-05, + "loss": 0.7237489223480225, + "step": 1160 + }, + { + "epoch": 0.490295358649789, + "grad_norm": 1.3859314918518066, + "learning_rate": 9.999923731354706e-05, + "loss": 0.8650591373443604, + "step": 1162 + }, + { + "epoch": 0.4911392405063291, + "grad_norm": 1.2910805940628052, + "learning_rate": 9.999909890583521e-05, + "loss": 0.7516807913780212, + "step": 1164 + }, + { + "epoch": 0.4919831223628692, + "grad_norm": 1.6100077629089355, + "learning_rate": 9.999894896429152e-05, + "loss": 0.7082475423812866, + "step": 1166 + }, + { + "epoch": 0.4928270042194093, + "grad_norm": 1.2313556671142578, + "learning_rate": 9.999878748895053e-05, + "loss": 0.8403750658035278, + "step": 1168 + }, + { + "epoch": 0.4936708860759494, + "grad_norm": 1.3402830362319946, + "learning_rate": 9.999861447984952e-05, + "loss": 0.8083041906356812, + "step": 1170 + }, + { + "epoch": 0.49451476793248944, + "grad_norm": 1.516775131225586, + "learning_rate": 9.999842993702839e-05, + "loss": 0.8339354991912842, + "step": 1172 + }, + { + "epoch": 0.49535864978902955, + "grad_norm": 1.2698423862457275, + "learning_rate": 9.999823386052971e-05, + "loss": 0.7708724141120911, + "step": 1174 + }, + { + "epoch": 0.4962025316455696, + "grad_norm": 1.339390516281128, + "learning_rate": 9.999802625039872e-05, + "loss": 0.7589715719223022, + "step": 1176 + }, + { + "epoch": 0.4970464135021097, + "grad_norm": 1.4618452787399292, + "learning_rate": 9.99978071066833e-05, + "loss": 0.8523206114768982, + "step": 1178 + }, + { + "epoch": 0.4978902953586498, + "grad_norm": 1.4812564849853516, + "learning_rate": 9.9997576429434e-05, + "loss": 0.8143196105957031, + "step": 1180 + }, + { + "epoch": 0.49873417721518987, + "grad_norm": 1.5720716714859009, + "learning_rate": 9.999733421870405e-05, + "loss": 0.800125002861023, + "step": 1182 + }, + { + "epoch": 0.49957805907173, + "grad_norm": 1.4421230554580688, + "learning_rate": 9.99970804745493e-05, + "loss": 0.7618259191513062, + "step": 1184 + }, + { + "epoch": 0.5004219409282701, + "grad_norm": 1.5794934034347534, + "learning_rate": 9.99968151970283e-05, + "loss": 0.7162163853645325, + "step": 1186 + }, + { + "epoch": 0.5012658227848101, + "grad_norm": 1.8590432405471802, + "learning_rate": 9.999653838620225e-05, + "loss": 0.8089820146560669, + "step": 1188 + }, + { + "epoch": 0.5021097046413502, + "grad_norm": 1.5194507837295532, + "learning_rate": 9.999625004213498e-05, + "loss": 0.8011203408241272, + "step": 1190 + }, + { + "epoch": 0.5029535864978903, + "grad_norm": 1.6986470222473145, + "learning_rate": 9.999595016489303e-05, + "loss": 0.761158287525177, + "step": 1192 + }, + { + "epoch": 0.5037974683544304, + "grad_norm": 1.4413946866989136, + "learning_rate": 9.999563875454559e-05, + "loss": 0.7898027300834656, + "step": 1194 + }, + { + "epoch": 0.5046413502109705, + "grad_norm": 1.4509994983673096, + "learning_rate": 9.999531581116443e-05, + "loss": 0.8018442392349243, + "step": 1196 + }, + { + "epoch": 0.5054852320675105, + "grad_norm": 1.400659441947937, + "learning_rate": 9.999498133482412e-05, + "loss": 0.7804076075553894, + "step": 1198 + }, + { + "epoch": 0.5063291139240507, + "grad_norm": 1.486840009689331, + "learning_rate": 9.999463532560178e-05, + "loss": 0.82496178150177, + "step": 1200 + }, + { + "epoch": 0.5063291139240507, + "eval_loss": 0.8186545968055725, + "eval_runtime": 862.1638, + "eval_samples_per_second": 2.444, + "eval_steps_per_second": 2.444, + "step": 1200 + }, + { + "epoch": 0.5071729957805907, + "grad_norm": 1.2770357131958008, + "learning_rate": 9.999427778357723e-05, + "loss": 0.8037722706794739, + "step": 1202 + }, + { + "epoch": 0.5080168776371308, + "grad_norm": 1.4540977478027344, + "learning_rate": 9.999390870883297e-05, + "loss": 0.7329373359680176, + "step": 1204 + }, + { + "epoch": 0.5088607594936709, + "grad_norm": 1.4469913244247437, + "learning_rate": 9.999352810145412e-05, + "loss": 0.8224589824676514, + "step": 1206 + }, + { + "epoch": 0.509704641350211, + "grad_norm": 1.46500563621521, + "learning_rate": 9.999313596152847e-05, + "loss": 0.8106292486190796, + "step": 1208 + }, + { + "epoch": 0.510548523206751, + "grad_norm": 1.3526637554168701, + "learning_rate": 9.999273228914649e-05, + "loss": 0.747698187828064, + "step": 1210 + }, + { + "epoch": 0.5113924050632911, + "grad_norm": 1.28840172290802, + "learning_rate": 9.999231708440131e-05, + "loss": 0.7612425684928894, + "step": 1212 + }, + { + "epoch": 0.5122362869198313, + "grad_norm": 1.0283230543136597, + "learning_rate": 9.99918903473887e-05, + "loss": 0.6839463710784912, + "step": 1214 + }, + { + "epoch": 0.5130801687763713, + "grad_norm": 1.5231431722640991, + "learning_rate": 9.999145207820708e-05, + "loss": 0.8539203405380249, + "step": 1216 + }, + { + "epoch": 0.5139240506329114, + "grad_norm": 1.3289231061935425, + "learning_rate": 9.999100227695758e-05, + "loss": 0.7960102558135986, + "step": 1218 + }, + { + "epoch": 0.5147679324894515, + "grad_norm": 1.3770930767059326, + "learning_rate": 9.999054094374396e-05, + "loss": 0.7639255523681641, + "step": 1220 + }, + { + "epoch": 0.5156118143459916, + "grad_norm": 1.3028030395507812, + "learning_rate": 9.999006807867262e-05, + "loss": 0.7743061780929565, + "step": 1222 + }, + { + "epoch": 0.5164556962025316, + "grad_norm": 1.1827034950256348, + "learning_rate": 9.998958368185265e-05, + "loss": 0.7922407984733582, + "step": 1224 + }, + { + "epoch": 0.5172995780590718, + "grad_norm": 1.2973705530166626, + "learning_rate": 9.99890877533958e-05, + "loss": 0.7671286463737488, + "step": 1226 + }, + { + "epoch": 0.5181434599156118, + "grad_norm": 1.5820153951644897, + "learning_rate": 9.998858029341646e-05, + "loss": 0.7546951174736023, + "step": 1228 + }, + { + "epoch": 0.5189873417721519, + "grad_norm": 1.6140317916870117, + "learning_rate": 9.99880613020317e-05, + "loss": 0.8734183311462402, + "step": 1230 + }, + { + "epoch": 0.5198312236286919, + "grad_norm": 1.1190184354782104, + "learning_rate": 9.998753077936122e-05, + "loss": 0.8410643339157104, + "step": 1232 + }, + { + "epoch": 0.5206751054852321, + "grad_norm": 1.3876196146011353, + "learning_rate": 9.998698872552744e-05, + "loss": 0.7769841551780701, + "step": 1234 + }, + { + "epoch": 0.5215189873417722, + "grad_norm": 1.699522852897644, + "learning_rate": 9.998643514065535e-05, + "loss": 0.8846109509468079, + "step": 1236 + }, + { + "epoch": 0.5223628691983122, + "grad_norm": 1.3805134296417236, + "learning_rate": 9.998587002487271e-05, + "loss": 0.7664945125579834, + "step": 1238 + }, + { + "epoch": 0.5232067510548524, + "grad_norm": 1.3679476976394653, + "learning_rate": 9.998529337830984e-05, + "loss": 0.7243514060974121, + "step": 1240 + }, + { + "epoch": 0.5240506329113924, + "grad_norm": 1.399200677871704, + "learning_rate": 9.998470520109977e-05, + "loss": 0.8061941862106323, + "step": 1242 + }, + { + "epoch": 0.5248945147679325, + "grad_norm": 1.3441044092178345, + "learning_rate": 9.99841054933782e-05, + "loss": 0.7741840481758118, + "step": 1244 + }, + { + "epoch": 0.5257383966244725, + "grad_norm": 1.3375325202941895, + "learning_rate": 9.998349425528344e-05, + "loss": 0.7619491815567017, + "step": 1246 + }, + { + "epoch": 0.5265822784810127, + "grad_norm": 1.5517847537994385, + "learning_rate": 9.998287148695651e-05, + "loss": 0.8315094113349915, + "step": 1248 + }, + { + "epoch": 0.5274261603375527, + "grad_norm": 1.244997501373291, + "learning_rate": 9.998223718854107e-05, + "loss": 0.7536082863807678, + "step": 1250 + }, + { + "epoch": 0.5282700421940928, + "grad_norm": 1.3190033435821533, + "learning_rate": 9.998159136018344e-05, + "loss": 0.826419472694397, + "step": 1252 + }, + { + "epoch": 0.529113924050633, + "grad_norm": 1.2750061750411987, + "learning_rate": 9.998093400203259e-05, + "loss": 0.7866435647010803, + "step": 1254 + }, + { + "epoch": 0.529957805907173, + "grad_norm": 1.422908067703247, + "learning_rate": 9.998026511424017e-05, + "loss": 0.7796626687049866, + "step": 1256 + }, + { + "epoch": 0.5308016877637131, + "grad_norm": 1.435552954673767, + "learning_rate": 9.997958469696048e-05, + "loss": 0.815027117729187, + "step": 1258 + }, + { + "epoch": 0.5316455696202531, + "grad_norm": 1.1950994729995728, + "learning_rate": 9.997889275035049e-05, + "loss": 0.6925795674324036, + "step": 1260 + }, + { + "epoch": 0.5324894514767933, + "grad_norm": 1.3049622774124146, + "learning_rate": 9.997818927456978e-05, + "loss": 0.822464108467102, + "step": 1262 + }, + { + "epoch": 0.5333333333333333, + "grad_norm": 1.2197340726852417, + "learning_rate": 9.997747426978066e-05, + "loss": 0.7955381274223328, + "step": 1264 + }, + { + "epoch": 0.5341772151898734, + "grad_norm": 1.2463661432266235, + "learning_rate": 9.997674773614807e-05, + "loss": 0.8642181754112244, + "step": 1266 + }, + { + "epoch": 0.5350210970464135, + "grad_norm": 1.421393871307373, + "learning_rate": 9.99760096738396e-05, + "loss": 0.8776891827583313, + "step": 1268 + }, + { + "epoch": 0.5358649789029536, + "grad_norm": 1.4347561597824097, + "learning_rate": 9.997526008302549e-05, + "loss": 0.7446491122245789, + "step": 1270 + }, + { + "epoch": 0.5367088607594936, + "grad_norm": 1.2056710720062256, + "learning_rate": 9.99744989638787e-05, + "loss": 0.8581281304359436, + "step": 1272 + }, + { + "epoch": 0.5375527426160338, + "grad_norm": 1.1672608852386475, + "learning_rate": 9.997372631657475e-05, + "loss": 0.7386330366134644, + "step": 1274 + }, + { + "epoch": 0.5383966244725739, + "grad_norm": 1.4313966035842896, + "learning_rate": 9.997294214129191e-05, + "loss": 0.7806804776191711, + "step": 1276 + }, + { + "epoch": 0.5392405063291139, + "grad_norm": 1.1666971445083618, + "learning_rate": 9.997214643821107e-05, + "loss": 0.6830351948738098, + "step": 1278 + }, + { + "epoch": 0.540084388185654, + "grad_norm": 1.491783857345581, + "learning_rate": 9.997133920751578e-05, + "loss": 0.8570694327354431, + "step": 1280 + }, + { + "epoch": 0.5409282700421941, + "grad_norm": 1.1879212856292725, + "learning_rate": 9.997052044939226e-05, + "loss": 0.7016772031784058, + "step": 1282 + }, + { + "epoch": 0.5417721518987342, + "grad_norm": 1.2692012786865234, + "learning_rate": 9.996969016402935e-05, + "loss": 0.7711107134819031, + "step": 1284 + }, + { + "epoch": 0.5426160337552742, + "grad_norm": 1.3318448066711426, + "learning_rate": 9.996884835161863e-05, + "loss": 0.7807164788246155, + "step": 1286 + }, + { + "epoch": 0.5434599156118144, + "grad_norm": 1.1786744594573975, + "learning_rate": 9.996799501235425e-05, + "loss": 0.7331319451332092, + "step": 1288 + }, + { + "epoch": 0.5443037974683544, + "grad_norm": 1.4092369079589844, + "learning_rate": 9.996713014643309e-05, + "loss": 0.7191547155380249, + "step": 1290 + }, + { + "epoch": 0.5451476793248945, + "grad_norm": 1.377099633216858, + "learning_rate": 9.996625375405463e-05, + "loss": 0.7233871221542358, + "step": 1292 + }, + { + "epoch": 0.5459915611814345, + "grad_norm": 1.404945969581604, + "learning_rate": 9.996536583542105e-05, + "loss": 0.7925472855567932, + "step": 1294 + }, + { + "epoch": 0.5468354430379747, + "grad_norm": 1.2555286884307861, + "learning_rate": 9.996446639073718e-05, + "loss": 0.7749786376953125, + "step": 1296 + }, + { + "epoch": 0.5476793248945148, + "grad_norm": 1.2577459812164307, + "learning_rate": 9.996355542021048e-05, + "loss": 0.7647517919540405, + "step": 1298 + }, + { + "epoch": 0.5485232067510548, + "grad_norm": 1.3587758541107178, + "learning_rate": 9.996263292405113e-05, + "loss": 0.8621891140937805, + "step": 1300 + }, + { + "epoch": 0.5485232067510548, + "eval_loss": 0.808323085308075, + "eval_runtime": 853.577, + "eval_samples_per_second": 2.468, + "eval_steps_per_second": 2.468, + "step": 1300 + }, + { + "epoch": 0.549367088607595, + "grad_norm": 1.327125906944275, + "learning_rate": 9.996169890247191e-05, + "loss": 0.749254584312439, + "step": 1302 + }, + { + "epoch": 0.550210970464135, + "grad_norm": 1.4620670080184937, + "learning_rate": 9.99607533556883e-05, + "loss": 0.7362856268882751, + "step": 1304 + }, + { + "epoch": 0.5510548523206751, + "grad_norm": 1.4119454622268677, + "learning_rate": 9.99597962839184e-05, + "loss": 0.7918445467948914, + "step": 1306 + }, + { + "epoch": 0.5518987341772152, + "grad_norm": 1.497522234916687, + "learning_rate": 9.995882768738298e-05, + "loss": 0.7348005175590515, + "step": 1308 + }, + { + "epoch": 0.5527426160337553, + "grad_norm": 1.535741925239563, + "learning_rate": 9.99578475663055e-05, + "loss": 0.8310725688934326, + "step": 1310 + }, + { + "epoch": 0.5535864978902953, + "grad_norm": 1.4606215953826904, + "learning_rate": 9.995685592091204e-05, + "loss": 0.8232766389846802, + "step": 1312 + }, + { + "epoch": 0.5544303797468354, + "grad_norm": 1.2442357540130615, + "learning_rate": 9.995585275143136e-05, + "loss": 0.8273071050643921, + "step": 1314 + }, + { + "epoch": 0.5552742616033756, + "grad_norm": 1.5128520727157593, + "learning_rate": 9.995483805809487e-05, + "loss": 0.7518656253814697, + "step": 1316 + }, + { + "epoch": 0.5561181434599156, + "grad_norm": 1.340149998664856, + "learning_rate": 9.995381184113664e-05, + "loss": 0.8261662721633911, + "step": 1318 + }, + { + "epoch": 0.5569620253164557, + "grad_norm": 1.1409451961517334, + "learning_rate": 9.99527741007934e-05, + "loss": 0.5775256156921387, + "step": 1320 + }, + { + "epoch": 0.5578059071729958, + "grad_norm": 1.3489247560501099, + "learning_rate": 9.995172483730455e-05, + "loss": 0.7698423862457275, + "step": 1322 + }, + { + "epoch": 0.5586497890295359, + "grad_norm": 1.4950530529022217, + "learning_rate": 9.995066405091211e-05, + "loss": 0.8053334355354309, + "step": 1324 + }, + { + "epoch": 0.5594936708860759, + "grad_norm": 1.3814653158187866, + "learning_rate": 9.994959174186078e-05, + "loss": 0.7826266288757324, + "step": 1326 + }, + { + "epoch": 0.560337552742616, + "grad_norm": 1.3383625745773315, + "learning_rate": 9.994850791039796e-05, + "loss": 0.7862131595611572, + "step": 1328 + }, + { + "epoch": 0.5611814345991561, + "grad_norm": 1.3529670238494873, + "learning_rate": 9.994741255677363e-05, + "loss": 0.8428501486778259, + "step": 1330 + }, + { + "epoch": 0.5620253164556962, + "grad_norm": 1.254215121269226, + "learning_rate": 9.994630568124049e-05, + "loss": 0.7340869307518005, + "step": 1332 + }, + { + "epoch": 0.5628691983122363, + "grad_norm": 1.2869828939437866, + "learning_rate": 9.994518728405386e-05, + "loss": 0.7052226662635803, + "step": 1334 + }, + { + "epoch": 0.5637130801687764, + "grad_norm": 1.4321808815002441, + "learning_rate": 9.994405736547174e-05, + "loss": 0.8297074437141418, + "step": 1336 + }, + { + "epoch": 0.5645569620253165, + "grad_norm": 1.4638891220092773, + "learning_rate": 9.994291592575478e-05, + "loss": 0.7183220982551575, + "step": 1338 + }, + { + "epoch": 0.5654008438818565, + "grad_norm": 1.4947413206100464, + "learning_rate": 9.994176296516628e-05, + "loss": 0.8146093487739563, + "step": 1340 + }, + { + "epoch": 0.5662447257383966, + "grad_norm": 1.343862533569336, + "learning_rate": 9.994059848397221e-05, + "loss": 0.7583593130111694, + "step": 1342 + }, + { + "epoch": 0.5670886075949367, + "grad_norm": 1.203550100326538, + "learning_rate": 9.993942248244121e-05, + "loss": 0.7682924270629883, + "step": 1344 + }, + { + "epoch": 0.5679324894514768, + "grad_norm": 1.287660002708435, + "learning_rate": 9.993823496084455e-05, + "loss": 0.8139828443527222, + "step": 1346 + }, + { + "epoch": 0.5687763713080168, + "grad_norm": 1.3326014280319214, + "learning_rate": 9.993703591945616e-05, + "loss": 0.7529099583625793, + "step": 1348 + }, + { + "epoch": 0.569620253164557, + "grad_norm": 1.2441487312316895, + "learning_rate": 9.993582535855263e-05, + "loss": 0.6997471451759338, + "step": 1350 + }, + { + "epoch": 0.570464135021097, + "grad_norm": 1.2647649049758911, + "learning_rate": 9.993460327841325e-05, + "loss": 0.7421218752861023, + "step": 1352 + }, + { + "epoch": 0.5713080168776371, + "grad_norm": 1.146399974822998, + "learning_rate": 9.99333696793199e-05, + "loss": 0.7342398166656494, + "step": 1354 + }, + { + "epoch": 0.5721518987341773, + "grad_norm": 1.3346691131591797, + "learning_rate": 9.993212456155715e-05, + "loss": 0.7175891399383545, + "step": 1356 + }, + { + "epoch": 0.5729957805907173, + "grad_norm": 1.3950672149658203, + "learning_rate": 9.993086792541222e-05, + "loss": 0.8108891248703003, + "step": 1358 + }, + { + "epoch": 0.5738396624472574, + "grad_norm": 1.339931845664978, + "learning_rate": 9.992959977117502e-05, + "loss": 0.6979889273643494, + "step": 1360 + }, + { + "epoch": 0.5746835443037974, + "grad_norm": 1.3276840448379517, + "learning_rate": 9.992832009913806e-05, + "loss": 0.7635799050331116, + "step": 1362 + }, + { + "epoch": 0.5755274261603376, + "grad_norm": 1.5015610456466675, + "learning_rate": 9.992702890959653e-05, + "loss": 0.7575043439865112, + "step": 1364 + }, + { + "epoch": 0.5763713080168776, + "grad_norm": 1.4755414724349976, + "learning_rate": 9.99257262028483e-05, + "loss": 0.8134847283363342, + "step": 1366 + }, + { + "epoch": 0.5772151898734177, + "grad_norm": 1.3788783550262451, + "learning_rate": 9.992441197919388e-05, + "loss": 0.7663828134536743, + "step": 1368 + }, + { + "epoch": 0.5780590717299579, + "grad_norm": 1.2814711332321167, + "learning_rate": 9.992308623893644e-05, + "loss": 0.6711251735687256, + "step": 1370 + }, + { + "epoch": 0.5789029535864979, + "grad_norm": 1.5343635082244873, + "learning_rate": 9.99217489823818e-05, + "loss": 0.8097200393676758, + "step": 1372 + }, + { + "epoch": 0.579746835443038, + "grad_norm": 1.3029557466506958, + "learning_rate": 9.992040020983843e-05, + "loss": 0.8274240493774414, + "step": 1374 + }, + { + "epoch": 0.580590717299578, + "grad_norm": 1.4034144878387451, + "learning_rate": 9.991903992161746e-05, + "loss": 0.7758964896202087, + "step": 1376 + }, + { + "epoch": 0.5814345991561182, + "grad_norm": 1.2340021133422852, + "learning_rate": 9.991766811803271e-05, + "loss": 0.6571930050849915, + "step": 1378 + }, + { + "epoch": 0.5822784810126582, + "grad_norm": 1.3082842826843262, + "learning_rate": 9.991628479940061e-05, + "loss": 0.7381542921066284, + "step": 1380 + }, + { + "epoch": 0.5831223628691983, + "grad_norm": 1.8134801387786865, + "learning_rate": 9.991488996604025e-05, + "loss": 0.8081237077713013, + "step": 1382 + }, + { + "epoch": 0.5839662447257384, + "grad_norm": 1.4598309993743896, + "learning_rate": 9.991348361827343e-05, + "loss": 0.7761610746383667, + "step": 1384 + }, + { + "epoch": 0.5848101265822785, + "grad_norm": 1.2974225282669067, + "learning_rate": 9.991206575642453e-05, + "loss": 0.6872953176498413, + "step": 1386 + }, + { + "epoch": 0.5856540084388185, + "grad_norm": 1.24009370803833, + "learning_rate": 9.991063638082065e-05, + "loss": 0.7601345777511597, + "step": 1388 + }, + { + "epoch": 0.5864978902953587, + "grad_norm": 1.176713228225708, + "learning_rate": 9.99091954917915e-05, + "loss": 0.7138593792915344, + "step": 1390 + }, + { + "epoch": 0.5873417721518988, + "grad_norm": 1.1056525707244873, + "learning_rate": 9.990774308966949e-05, + "loss": 0.7730305194854736, + "step": 1392 + }, + { + "epoch": 0.5881856540084388, + "grad_norm": 1.382847547531128, + "learning_rate": 9.990627917478962e-05, + "loss": 0.7076689600944519, + "step": 1394 + }, + { + "epoch": 0.5890295358649789, + "grad_norm": 1.2507930994033813, + "learning_rate": 9.990480374748964e-05, + "loss": 0.7970513105392456, + "step": 1396 + }, + { + "epoch": 0.589873417721519, + "grad_norm": 1.2266724109649658, + "learning_rate": 9.990331680810987e-05, + "loss": 0.7906717658042908, + "step": 1398 + }, + { + "epoch": 0.5907172995780591, + "grad_norm": 1.299920916557312, + "learning_rate": 9.99018183569933e-05, + "loss": 0.853204607963562, + "step": 1400 + }, + { + "epoch": 0.5907172995780591, + "eval_loss": 0.8009664416313171, + "eval_runtime": 851.9417, + "eval_samples_per_second": 2.473, + "eval_steps_per_second": 2.473, + "step": 1400 + }, + { + "epoch": 0.5915611814345991, + "grad_norm": 1.2114863395690918, + "learning_rate": 9.990030839448564e-05, + "loss": 0.8140703439712524, + "step": 1402 + }, + { + "epoch": 0.5924050632911393, + "grad_norm": 1.3301794528961182, + "learning_rate": 9.989878692093518e-05, + "loss": 0.7471320629119873, + "step": 1404 + }, + { + "epoch": 0.5932489451476793, + "grad_norm": 1.2611899375915527, + "learning_rate": 9.98972539366929e-05, + "loss": 0.7307024002075195, + "step": 1406 + }, + { + "epoch": 0.5940928270042194, + "grad_norm": 1.1717802286148071, + "learning_rate": 9.989570944211244e-05, + "loss": 0.6843112111091614, + "step": 1408 + }, + { + "epoch": 0.5949367088607594, + "grad_norm": 1.3323513269424438, + "learning_rate": 9.989415343755006e-05, + "loss": 0.7025372385978699, + "step": 1410 + }, + { + "epoch": 0.5957805907172996, + "grad_norm": 1.4225109815597534, + "learning_rate": 9.989258592336473e-05, + "loss": 0.7792683839797974, + "step": 1412 + }, + { + "epoch": 0.5966244725738397, + "grad_norm": 1.2878522872924805, + "learning_rate": 9.989100689991804e-05, + "loss": 0.8328315019607544, + "step": 1414 + }, + { + "epoch": 0.5974683544303797, + "grad_norm": 1.2067214250564575, + "learning_rate": 9.988941636757421e-05, + "loss": 0.7700617909431458, + "step": 1416 + }, + { + "epoch": 0.5983122362869199, + "grad_norm": 1.1213195323944092, + "learning_rate": 9.988781432670019e-05, + "loss": 0.6872363090515137, + "step": 1418 + }, + { + "epoch": 0.5991561181434599, + "grad_norm": 1.3211694955825806, + "learning_rate": 9.98862007776655e-05, + "loss": 0.7184111475944519, + "step": 1420 + }, + { + "epoch": 0.6, + "grad_norm": 1.1916998624801636, + "learning_rate": 9.98845757208424e-05, + "loss": 0.8120859265327454, + "step": 1422 + }, + { + "epoch": 0.60084388185654, + "grad_norm": 1.2772804498672485, + "learning_rate": 9.988293915660572e-05, + "loss": 0.7586462497711182, + "step": 1424 + }, + { + "epoch": 0.6016877637130802, + "grad_norm": 1.4139106273651123, + "learning_rate": 9.988129108533299e-05, + "loss": 0.8175994157791138, + "step": 1426 + }, + { + "epoch": 0.6025316455696202, + "grad_norm": 1.4481157064437866, + "learning_rate": 9.987963150740439e-05, + "loss": 0.7662636041641235, + "step": 1428 + }, + { + "epoch": 0.6033755274261603, + "grad_norm": 1.6000999212265015, + "learning_rate": 9.987796042320277e-05, + "loss": 0.7477837800979614, + "step": 1430 + }, + { + "epoch": 0.6042194092827005, + "grad_norm": 1.26194429397583, + "learning_rate": 9.98762778331136e-05, + "loss": 0.7392798662185669, + "step": 1432 + }, + { + "epoch": 0.6050632911392405, + "grad_norm": 1.2370645999908447, + "learning_rate": 9.987458373752503e-05, + "loss": 0.7795998454093933, + "step": 1434 + }, + { + "epoch": 0.6059071729957806, + "grad_norm": 1.4908311367034912, + "learning_rate": 9.987287813682784e-05, + "loss": 0.7833777070045471, + "step": 1436 + }, + { + "epoch": 0.6067510548523207, + "grad_norm": 1.2918652296066284, + "learning_rate": 9.987116103141549e-05, + "loss": 0.7269768118858337, + "step": 1438 + }, + { + "epoch": 0.6075949367088608, + "grad_norm": 1.2170461416244507, + "learning_rate": 9.98694324216841e-05, + "loss": 0.7599279284477234, + "step": 1440 + }, + { + "epoch": 0.6084388185654008, + "grad_norm": 1.4373505115509033, + "learning_rate": 9.98676923080324e-05, + "loss": 0.8256514668464661, + "step": 1442 + }, + { + "epoch": 0.6092827004219409, + "grad_norm": 1.3523614406585693, + "learning_rate": 9.986594069086181e-05, + "loss": 0.8462428450584412, + "step": 1444 + }, + { + "epoch": 0.610126582278481, + "grad_norm": 1.5131851434707642, + "learning_rate": 9.98641775705764e-05, + "loss": 0.8402239084243774, + "step": 1446 + }, + { + "epoch": 0.6109704641350211, + "grad_norm": 1.3518229722976685, + "learning_rate": 9.98624029475829e-05, + "loss": 0.7585759162902832, + "step": 1448 + }, + { + "epoch": 0.6118143459915611, + "grad_norm": 1.3403998613357544, + "learning_rate": 9.986061682229064e-05, + "loss": 0.773881733417511, + "step": 1450 + }, + { + "epoch": 0.6126582278481013, + "grad_norm": 1.1835366487503052, + "learning_rate": 9.985881919511168e-05, + "loss": 0.6770316958427429, + "step": 1452 + }, + { + "epoch": 0.6135021097046414, + "grad_norm": 1.1825730800628662, + "learning_rate": 9.985701006646069e-05, + "loss": 0.7081645727157593, + "step": 1454 + }, + { + "epoch": 0.6143459915611814, + "grad_norm": 1.378994345664978, + "learning_rate": 9.9855189436755e-05, + "loss": 0.7750917673110962, + "step": 1456 + }, + { + "epoch": 0.6151898734177215, + "grad_norm": 1.4208749532699585, + "learning_rate": 9.985335730641458e-05, + "loss": 0.7517801523208618, + "step": 1458 + }, + { + "epoch": 0.6160337552742616, + "grad_norm": 1.1413639783859253, + "learning_rate": 9.98515136758621e-05, + "loss": 0.712832510471344, + "step": 1460 + }, + { + "epoch": 0.6168776371308017, + "grad_norm": 1.3949562311172485, + "learning_rate": 9.984965854552283e-05, + "loss": 0.7884142994880676, + "step": 1462 + }, + { + "epoch": 0.6177215189873417, + "grad_norm": 1.4057096242904663, + "learning_rate": 9.984779191582471e-05, + "loss": 0.796623706817627, + "step": 1464 + }, + { + "epoch": 0.6185654008438819, + "grad_norm": 1.1681689023971558, + "learning_rate": 9.984591378719834e-05, + "loss": 0.7862933874130249, + "step": 1466 + }, + { + "epoch": 0.619409282700422, + "grad_norm": 1.2585291862487793, + "learning_rate": 9.984402416007696e-05, + "loss": 0.7889828681945801, + "step": 1468 + }, + { + "epoch": 0.620253164556962, + "grad_norm": 1.2598098516464233, + "learning_rate": 9.984212303489649e-05, + "loss": 0.7375997304916382, + "step": 1470 + }, + { + "epoch": 0.6210970464135022, + "grad_norm": 1.4628467559814453, + "learning_rate": 9.984021041209547e-05, + "loss": 0.7839564085006714, + "step": 1472 + }, + { + "epoch": 0.6219409282700422, + "grad_norm": 1.3606770038604736, + "learning_rate": 9.983828629211511e-05, + "loss": 0.7566051483154297, + "step": 1474 + }, + { + "epoch": 0.6227848101265823, + "grad_norm": 1.182644248008728, + "learning_rate": 9.983635067539927e-05, + "loss": 0.6638457179069519, + "step": 1476 + }, + { + "epoch": 0.6236286919831223, + "grad_norm": 1.5617793798446655, + "learning_rate": 9.983440356239445e-05, + "loss": 0.8227225542068481, + "step": 1478 + }, + { + "epoch": 0.6244725738396625, + "grad_norm": 1.2290058135986328, + "learning_rate": 9.98324449535498e-05, + "loss": 0.7086431980133057, + "step": 1480 + }, + { + "epoch": 0.6253164556962025, + "grad_norm": 1.3822678327560425, + "learning_rate": 9.983047484931716e-05, + "loss": 0.8076596856117249, + "step": 1482 + }, + { + "epoch": 0.6261603375527426, + "grad_norm": 1.163699746131897, + "learning_rate": 9.982849325015098e-05, + "loss": 0.7514539361000061, + "step": 1484 + }, + { + "epoch": 0.6270042194092827, + "grad_norm": 1.2635631561279297, + "learning_rate": 9.982650015650839e-05, + "loss": 0.7298142910003662, + "step": 1486 + }, + { + "epoch": 0.6278481012658228, + "grad_norm": 1.3135387897491455, + "learning_rate": 9.982449556884914e-05, + "loss": 0.8092831373214722, + "step": 1488 + }, + { + "epoch": 0.6286919831223629, + "grad_norm": 1.3577877283096313, + "learning_rate": 9.982247948763567e-05, + "loss": 0.7934147715568542, + "step": 1490 + }, + { + "epoch": 0.6295358649789029, + "grad_norm": 1.1482092142105103, + "learning_rate": 9.982045191333304e-05, + "loss": 0.789363443851471, + "step": 1492 + }, + { + "epoch": 0.6303797468354431, + "grad_norm": 1.189771056175232, + "learning_rate": 9.981841284640895e-05, + "loss": 0.7458413243293762, + "step": 1494 + }, + { + "epoch": 0.6312236286919831, + "grad_norm": 1.2815836668014526, + "learning_rate": 9.981636228733383e-05, + "loss": 0.7299918532371521, + "step": 1496 + }, + { + "epoch": 0.6320675105485232, + "grad_norm": 1.36761474609375, + "learning_rate": 9.981430023658068e-05, + "loss": 0.7545169591903687, + "step": 1498 + }, + { + "epoch": 0.6329113924050633, + "grad_norm": 1.2594345808029175, + "learning_rate": 9.981222669462513e-05, + "loss": 0.7358481884002686, + "step": 1500 + }, + { + "epoch": 0.6329113924050633, + "eval_loss": 0.7896141409873962, + "eval_runtime": 865.9069, + "eval_samples_per_second": 2.433, + "eval_steps_per_second": 2.433, + "step": 1500 + } + ], + "logging_steps": 2, + "max_steps": 14220, + "num_input_tokens_seen": 0, + "num_train_epochs": 6, + "save_steps": 500, + "stateful_callbacks": { + "EarlyStoppingCallback": { + "args": { + "early_stopping_patience": 5, + "early_stopping_threshold": 0.001 + }, + "attributes": { + "early_stopping_patience_counter": 0 + } + }, + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 1.5578111292507464e+18, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +}