| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 2424, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00041269022440030954, | |
| "grad_norm": 5.357641696929932, | |
| "learning_rate": 0.0, | |
| "loss": 6.1189, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0008253804488006191, | |
| "grad_norm": 5.467103004455566, | |
| "learning_rate": 8.19672131147541e-07, | |
| "loss": 6.1624, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0016507608976012382, | |
| "grad_norm": 5.270254611968994, | |
| "learning_rate": 2.459016393442623e-06, | |
| "loss": 6.1301, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.002476141346401857, | |
| "grad_norm": 4.018209457397461, | |
| "learning_rate": 4.098360655737704e-06, | |
| "loss": 6.1305, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0033015217952024763, | |
| "grad_norm": 3.973541259765625, | |
| "learning_rate": 5.737704918032787e-06, | |
| "loss": 6.0095, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0041269022440030955, | |
| "grad_norm": 3.5741689205169678, | |
| "learning_rate": 7.3770491803278695e-06, | |
| "loss": 5.9256, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.004952282692803714, | |
| "grad_norm": 4.449726581573486, | |
| "learning_rate": 9.016393442622952e-06, | |
| "loss": 5.8059, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.005777663141604333, | |
| "grad_norm": 4.23346471786499, | |
| "learning_rate": 1.0655737704918032e-05, | |
| "loss": 5.6952, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.006603043590404953, | |
| "grad_norm": 2.9237682819366455, | |
| "learning_rate": 1.2295081967213116e-05, | |
| "loss": 5.5323, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.007428424039205571, | |
| "grad_norm": 2.1714608669281006, | |
| "learning_rate": 1.3934426229508196e-05, | |
| "loss": 5.3692, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.008253804488006191, | |
| "grad_norm": 1.9534616470336914, | |
| "learning_rate": 1.557377049180328e-05, | |
| "loss": 5.1891, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.00907918493680681, | |
| "grad_norm": 1.7847157716751099, | |
| "learning_rate": 1.721311475409836e-05, | |
| "loss": 5.0308, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.009904565385607429, | |
| "grad_norm": 1.329651117324829, | |
| "learning_rate": 1.8852459016393442e-05, | |
| "loss": 4.8984, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.010729945834408047, | |
| "grad_norm": 1.3483397960662842, | |
| "learning_rate": 2.0491803278688525e-05, | |
| "loss": 4.7523, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.011555326283208666, | |
| "grad_norm": 1.213932991027832, | |
| "learning_rate": 2.2131147540983607e-05, | |
| "loss": 4.6048, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.012380706732009285, | |
| "grad_norm": 1.2783870697021484, | |
| "learning_rate": 2.377049180327869e-05, | |
| "loss": 4.5203, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.013206087180809905, | |
| "grad_norm": 1.4986013174057007, | |
| "learning_rate": 2.540983606557377e-05, | |
| "loss": 4.3679, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.014031467629610524, | |
| "grad_norm": 0.9324406981468201, | |
| "learning_rate": 2.7049180327868856e-05, | |
| "loss": 4.2537, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.014856848078411143, | |
| "grad_norm": 2.258493185043335, | |
| "learning_rate": 2.8688524590163935e-05, | |
| "loss": 4.1994, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.015682228527211763, | |
| "grad_norm": 2.1433424949645996, | |
| "learning_rate": 3.0327868852459017e-05, | |
| "loss": 4.1349, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.016507608976012382, | |
| "grad_norm": 2.0129520893096924, | |
| "learning_rate": 3.19672131147541e-05, | |
| "loss": 4.0318, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.017332989424813, | |
| "grad_norm": 1.2179840803146362, | |
| "learning_rate": 3.360655737704918e-05, | |
| "loss": 3.9541, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.01815836987361362, | |
| "grad_norm": 1.133366346359253, | |
| "learning_rate": 3.524590163934427e-05, | |
| "loss": 3.8931, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.01898375032241424, | |
| "grad_norm": 1.4476598501205444, | |
| "learning_rate": 3.6885245901639346e-05, | |
| "loss": 3.8386, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.019809130771214857, | |
| "grad_norm": 1.8356342315673828, | |
| "learning_rate": 3.8524590163934424e-05, | |
| "loss": 3.7593, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.020634511220015476, | |
| "grad_norm": 1.1986051797866821, | |
| "learning_rate": 4.016393442622951e-05, | |
| "loss": 3.7367, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.021459891668816095, | |
| "grad_norm": 1.4108922481536865, | |
| "learning_rate": 4.1803278688524595e-05, | |
| "loss": 3.6536, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.022285272117616713, | |
| "grad_norm": 1.194887638092041, | |
| "learning_rate": 4.3442622950819674e-05, | |
| "loss": 3.6377, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.023110652566417332, | |
| "grad_norm": 1.5970392227172852, | |
| "learning_rate": 4.508196721311476e-05, | |
| "loss": 3.592, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.02393603301521795, | |
| "grad_norm": 1.7871198654174805, | |
| "learning_rate": 4.672131147540984e-05, | |
| "loss": 3.5467, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.02476141346401857, | |
| "grad_norm": 2.0405406951904297, | |
| "learning_rate": 4.836065573770492e-05, | |
| "loss": 3.5071, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.025586793912819192, | |
| "grad_norm": 1.6245758533477783, | |
| "learning_rate": 5e-05, | |
| "loss": 3.4754, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.02641217436161981, | |
| "grad_norm": 1.3766052722930908, | |
| "learning_rate": 5.163934426229509e-05, | |
| "loss": 3.4631, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.02723755481042043, | |
| "grad_norm": 1.3058711290359497, | |
| "learning_rate": 5.327868852459017e-05, | |
| "loss": 3.4197, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.028062935259221048, | |
| "grad_norm": 1.545015573501587, | |
| "learning_rate": 5.491803278688525e-05, | |
| "loss": 3.4313, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.028888315708021667, | |
| "grad_norm": 1.439721703529358, | |
| "learning_rate": 5.6557377049180324e-05, | |
| "loss": 3.3894, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.029713696156822286, | |
| "grad_norm": 1.6934937238693237, | |
| "learning_rate": 5.819672131147541e-05, | |
| "loss": 3.3193, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.030539076605622904, | |
| "grad_norm": 1.3454101085662842, | |
| "learning_rate": 5.9836065573770495e-05, | |
| "loss": 3.3252, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.03136445705442353, | |
| "grad_norm": 1.610787272453308, | |
| "learning_rate": 6.147540983606557e-05, | |
| "loss": 3.2966, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.03218983750322414, | |
| "grad_norm": 2.0271148681640625, | |
| "learning_rate": 6.311475409836067e-05, | |
| "loss": 3.2892, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.033015217952024764, | |
| "grad_norm": 2.165980100631714, | |
| "learning_rate": 6.475409836065574e-05, | |
| "loss": 3.2792, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.03384059840082538, | |
| "grad_norm": 1.7957913875579834, | |
| "learning_rate": 6.639344262295082e-05, | |
| "loss": 3.2543, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.034665978849626, | |
| "grad_norm": 1.3472362756729126, | |
| "learning_rate": 6.80327868852459e-05, | |
| "loss": 3.251, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.03549135929842662, | |
| "grad_norm": 1.3264447450637817, | |
| "learning_rate": 6.967213114754098e-05, | |
| "loss": 3.2185, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.03631673974722724, | |
| "grad_norm": 1.5266629457473755, | |
| "learning_rate": 7.131147540983607e-05, | |
| "loss": 3.2058, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.037142120196027854, | |
| "grad_norm": 1.36456298828125, | |
| "learning_rate": 7.295081967213115e-05, | |
| "loss": 3.1753, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.03796750064482848, | |
| "grad_norm": 1.3469734191894531, | |
| "learning_rate": 7.459016393442624e-05, | |
| "loss": 3.1905, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.03879288109362909, | |
| "grad_norm": 1.2221981287002563, | |
| "learning_rate": 7.622950819672131e-05, | |
| "loss": 3.1526, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.039618261542429714, | |
| "grad_norm": 1.3852319717407227, | |
| "learning_rate": 7.78688524590164e-05, | |
| "loss": 3.156, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.04044364199123033, | |
| "grad_norm": 1.3925163745880127, | |
| "learning_rate": 7.950819672131148e-05, | |
| "loss": 3.149, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.04126902244003095, | |
| "grad_norm": 1.4574236869812012, | |
| "learning_rate": 8.114754098360656e-05, | |
| "loss": 3.1252, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.042094402888831574, | |
| "grad_norm": 1.1478595733642578, | |
| "learning_rate": 8.278688524590165e-05, | |
| "loss": 3.1093, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.04291978333763219, | |
| "grad_norm": 1.4830694198608398, | |
| "learning_rate": 8.442622950819673e-05, | |
| "loss": 3.0921, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.04374516378643281, | |
| "grad_norm": 1.4275153875350952, | |
| "learning_rate": 8.606557377049181e-05, | |
| "loss": 3.0861, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.04457054423523343, | |
| "grad_norm": 2.0077579021453857, | |
| "learning_rate": 8.770491803278689e-05, | |
| "loss": 3.0694, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.04539592468403405, | |
| "grad_norm": 1.533471703529358, | |
| "learning_rate": 8.934426229508197e-05, | |
| "loss": 3.0526, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.046221305132834664, | |
| "grad_norm": 1.394168496131897, | |
| "learning_rate": 9.098360655737706e-05, | |
| "loss": 3.0655, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.047046685581635286, | |
| "grad_norm": 1.6954408884048462, | |
| "learning_rate": 9.262295081967214e-05, | |
| "loss": 3.0338, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.0478720660304359, | |
| "grad_norm": 1.4712835550308228, | |
| "learning_rate": 9.426229508196722e-05, | |
| "loss": 3.0511, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.048697446479236524, | |
| "grad_norm": 1.473305344581604, | |
| "learning_rate": 9.59016393442623e-05, | |
| "loss": 3.031, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.04952282692803714, | |
| "grad_norm": 1.5957138538360596, | |
| "learning_rate": 9.754098360655737e-05, | |
| "loss": 3.003, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.05034820737683776, | |
| "grad_norm": 1.7283776998519897, | |
| "learning_rate": 9.918032786885247e-05, | |
| "loss": 3.0025, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.051173587825638384, | |
| "grad_norm": 1.286211609840393, | |
| "learning_rate": 9.999995343827644e-05, | |
| "loss": 3.0046, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.051998968274439, | |
| "grad_norm": 1.612631916999817, | |
| "learning_rate": 9.99995809450083e-05, | |
| "loss": 2.9935, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.05282434872323962, | |
| "grad_norm": 1.7912741899490356, | |
| "learning_rate": 9.9998835961247e-05, | |
| "loss": 3.0016, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.053649729172040236, | |
| "grad_norm": 1.7926831245422363, | |
| "learning_rate": 9.999771849254263e-05, | |
| "loss": 2.9718, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.05447510962084086, | |
| "grad_norm": 1.4038861989974976, | |
| "learning_rate": 9.999622854722017e-05, | |
| "loss": 2.9792, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.055300490069641474, | |
| "grad_norm": 1.7067828178405762, | |
| "learning_rate": 9.99943661363795e-05, | |
| "loss": 2.968, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.056125870518442096, | |
| "grad_norm": 1.5349268913269043, | |
| "learning_rate": 9.999213127389536e-05, | |
| "loss": 2.9373, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.05695125096724271, | |
| "grad_norm": 1.2763527631759644, | |
| "learning_rate": 9.99895239764172e-05, | |
| "loss": 2.9384, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.057776631416043334, | |
| "grad_norm": 1.3789610862731934, | |
| "learning_rate": 9.998654426336905e-05, | |
| "loss": 2.9416, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.05860201186484395, | |
| "grad_norm": 1.6175661087036133, | |
| "learning_rate": 9.998319215694936e-05, | |
| "loss": 2.9323, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.05942739231364457, | |
| "grad_norm": 1.5398344993591309, | |
| "learning_rate": 9.997946768213095e-05, | |
| "loss": 2.9334, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.060252772762445186, | |
| "grad_norm": 1.6788642406463623, | |
| "learning_rate": 9.997537086666063e-05, | |
| "loss": 2.9218, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.06107815321124581, | |
| "grad_norm": 1.4843031167984009, | |
| "learning_rate": 9.997090174105919e-05, | |
| "loss": 2.9366, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.06190353366004643, | |
| "grad_norm": 1.3358060121536255, | |
| "learning_rate": 9.996606033862102e-05, | |
| "loss": 2.9279, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.06272891410884705, | |
| "grad_norm": 1.9375636577606201, | |
| "learning_rate": 9.996084669541397e-05, | |
| "loss": 2.9202, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.06355429455764766, | |
| "grad_norm": 1.3849859237670898, | |
| "learning_rate": 9.9955260850279e-05, | |
| "loss": 2.895, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.06437967500644828, | |
| "grad_norm": 1.3628286123275757, | |
| "learning_rate": 9.994930284482993e-05, | |
| "loss": 2.8983, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.0652050554552489, | |
| "grad_norm": 1.2561815977096558, | |
| "learning_rate": 9.994297272345319e-05, | |
| "loss": 2.9089, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.06603043590404953, | |
| "grad_norm": 1.3705800771713257, | |
| "learning_rate": 9.993627053330732e-05, | |
| "loss": 2.878, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.06685581635285014, | |
| "grad_norm": 1.2955900430679321, | |
| "learning_rate": 9.99291963243228e-05, | |
| "loss": 2.8591, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.06768119680165076, | |
| "grad_norm": 1.231101155281067, | |
| "learning_rate": 9.992175014920161e-05, | |
| "loss": 2.8616, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.06850657725045138, | |
| "grad_norm": 1.1412620544433594, | |
| "learning_rate": 9.991393206341677e-05, | |
| "loss": 2.8353, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.069331957699252, | |
| "grad_norm": 1.102623701095581, | |
| "learning_rate": 9.990574212521205e-05, | |
| "loss": 2.8262, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.07015733814805261, | |
| "grad_norm": 1.0235016345977783, | |
| "learning_rate": 9.98971803956014e-05, | |
| "loss": 2.8581, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.07098271859685323, | |
| "grad_norm": 1.475123643875122, | |
| "learning_rate": 9.988824693836864e-05, | |
| "loss": 2.8709, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.07180809904565386, | |
| "grad_norm": 0.8155277371406555, | |
| "learning_rate": 9.98789418200669e-05, | |
| "loss": 2.8426, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.07263347949445448, | |
| "grad_norm": 1.3113749027252197, | |
| "learning_rate": 9.98692651100181e-05, | |
| "loss": 2.8017, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.0734588599432551, | |
| "grad_norm": 1.2450861930847168, | |
| "learning_rate": 9.985921688031252e-05, | |
| "loss": 2.8317, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.07428424039205571, | |
| "grad_norm": 1.304402470588684, | |
| "learning_rate": 9.984879720580816e-05, | |
| "loss": 2.8157, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.07510962084085633, | |
| "grad_norm": 1.1851410865783691, | |
| "learning_rate": 9.983800616413026e-05, | |
| "loss": 2.8245, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.07593500128965695, | |
| "grad_norm": 1.2967396974563599, | |
| "learning_rate": 9.982684383567071e-05, | |
| "loss": 2.8363, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.07676038173845758, | |
| "grad_norm": 1.2011407613754272, | |
| "learning_rate": 9.981531030358746e-05, | |
| "loss": 2.8142, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.07758576218725818, | |
| "grad_norm": 1.0165106058120728, | |
| "learning_rate": 9.980340565380382e-05, | |
| "loss": 2.7913, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.0784111426360588, | |
| "grad_norm": 1.3044579029083252, | |
| "learning_rate": 9.979112997500792e-05, | |
| "loss": 2.7805, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.07923652308485943, | |
| "grad_norm": 1.1849685907363892, | |
| "learning_rate": 9.9778483358652e-05, | |
| "loss": 2.7707, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.08006190353366005, | |
| "grad_norm": 0.9122027158737183, | |
| "learning_rate": 9.976546589895175e-05, | |
| "loss": 2.7777, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.08088728398246066, | |
| "grad_norm": 1.0830117464065552, | |
| "learning_rate": 9.975207769288556e-05, | |
| "loss": 2.8048, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.08171266443126128, | |
| "grad_norm": 1.1544275283813477, | |
| "learning_rate": 9.973831884019387e-05, | |
| "loss": 2.7761, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.0825380448800619, | |
| "grad_norm": 0.8355935215950012, | |
| "learning_rate": 9.972418944337835e-05, | |
| "loss": 2.7593, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.08336342532886253, | |
| "grad_norm": 1.203262209892273, | |
| "learning_rate": 9.970968960770124e-05, | |
| "loss": 2.7695, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.08418880577766315, | |
| "grad_norm": 1.23800790309906, | |
| "learning_rate": 9.969481944118443e-05, | |
| "loss": 2.7576, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.08501418622646376, | |
| "grad_norm": 0.8839966058731079, | |
| "learning_rate": 9.96795790546088e-05, | |
| "loss": 2.7442, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.08583956667526438, | |
| "grad_norm": 0.9399611949920654, | |
| "learning_rate": 9.966396856151326e-05, | |
| "loss": 2.7402, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.086664947124065, | |
| "grad_norm": 1.1721992492675781, | |
| "learning_rate": 9.964798807819397e-05, | |
| "loss": 2.7378, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.08749032757286562, | |
| "grad_norm": 0.9647835493087769, | |
| "learning_rate": 9.963163772370352e-05, | |
| "loss": 2.7256, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.08831570802166623, | |
| "grad_norm": 0.9155466556549072, | |
| "learning_rate": 9.961491761984996e-05, | |
| "loss": 2.7255, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.08914108847046685, | |
| "grad_norm": 0.9373721480369568, | |
| "learning_rate": 9.959782789119592e-05, | |
| "loss": 2.7544, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.08996646891926748, | |
| "grad_norm": 0.9547314643859863, | |
| "learning_rate": 9.958036866505772e-05, | |
| "loss": 2.7333, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.0907918493680681, | |
| "grad_norm": 1.0028138160705566, | |
| "learning_rate": 9.956254007150432e-05, | |
| "loss": 2.7232, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.0916172298168687, | |
| "grad_norm": 1.2652791738510132, | |
| "learning_rate": 9.954434224335649e-05, | |
| "loss": 2.7268, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.09244261026566933, | |
| "grad_norm": 1.1313235759735107, | |
| "learning_rate": 9.952577531618574e-05, | |
| "loss": 2.7417, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.09326799071446995, | |
| "grad_norm": 0.7514833211898804, | |
| "learning_rate": 9.950683942831328e-05, | |
| "loss": 2.6898, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.09409337116327057, | |
| "grad_norm": 0.9731917381286621, | |
| "learning_rate": 9.948753472080907e-05, | |
| "loss": 2.686, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.0949187516120712, | |
| "grad_norm": 0.8640966415405273, | |
| "learning_rate": 9.946786133749071e-05, | |
| "loss": 2.7168, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.0957441320608718, | |
| "grad_norm": 0.9116567969322205, | |
| "learning_rate": 9.944781942492242e-05, | |
| "loss": 2.7123, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.09656951250967243, | |
| "grad_norm": 1.0034291744232178, | |
| "learning_rate": 9.942740913241386e-05, | |
| "loss": 2.7146, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.09739489295847305, | |
| "grad_norm": 0.8208848237991333, | |
| "learning_rate": 9.94066306120191e-05, | |
| "loss": 2.6773, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.09822027340727367, | |
| "grad_norm": 0.8781367540359497, | |
| "learning_rate": 9.938548401853547e-05, | |
| "loss": 2.719, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.09904565385607428, | |
| "grad_norm": 0.7302896976470947, | |
| "learning_rate": 9.93639695095024e-05, | |
| "loss": 2.7011, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.0998710343048749, | |
| "grad_norm": 0.705086350440979, | |
| "learning_rate": 9.934208724520024e-05, | |
| "loss": 2.6648, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.10069641475367552, | |
| "grad_norm": 0.8350553512573242, | |
| "learning_rate": 9.931983738864904e-05, | |
| "loss": 2.687, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.10152179520247614, | |
| "grad_norm": 0.6524394154548645, | |
| "learning_rate": 9.92972201056074e-05, | |
| "loss": 2.7015, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.10234717565127677, | |
| "grad_norm": 0.6503209471702576, | |
| "learning_rate": 9.927423556457121e-05, | |
| "loss": 2.6148, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.10317255610007738, | |
| "grad_norm": 0.7506954073905945, | |
| "learning_rate": 9.925088393677236e-05, | |
| "loss": 2.6914, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.103997936548878, | |
| "grad_norm": 1.1561987400054932, | |
| "learning_rate": 9.922716539617746e-05, | |
| "loss": 2.6659, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.10482331699767862, | |
| "grad_norm": 1.0000964403152466, | |
| "learning_rate": 9.920308011948665e-05, | |
| "loss": 2.6626, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.10564869744647924, | |
| "grad_norm": 0.8899397850036621, | |
| "learning_rate": 9.917862828613214e-05, | |
| "loss": 2.6666, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.10647407789527985, | |
| "grad_norm": 1.1503660678863525, | |
| "learning_rate": 9.915381007827698e-05, | |
| "loss": 2.6395, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.10729945834408047, | |
| "grad_norm": 0.8070819973945618, | |
| "learning_rate": 9.912862568081364e-05, | |
| "loss": 2.6531, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.1081248387928811, | |
| "grad_norm": 0.8623407483100891, | |
| "learning_rate": 9.910307528136266e-05, | |
| "loss": 2.6588, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.10895021924168172, | |
| "grad_norm": 0.9573660492897034, | |
| "learning_rate": 9.907715907027129e-05, | |
| "loss": 2.6823, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.10977559969048233, | |
| "grad_norm": 1.0500940084457397, | |
| "learning_rate": 9.905087724061195e-05, | |
| "loss": 2.6545, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.11060098013928295, | |
| "grad_norm": 1.0520515441894531, | |
| "learning_rate": 9.902422998818094e-05, | |
| "loss": 2.6371, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.11142636058808357, | |
| "grad_norm": 0.9879215955734253, | |
| "learning_rate": 9.899721751149688e-05, | |
| "loss": 2.6474, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.11225174103688419, | |
| "grad_norm": 0.8972532749176025, | |
| "learning_rate": 9.896984001179925e-05, | |
| "loss": 2.6271, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.11307712148568481, | |
| "grad_norm": 0.6369883418083191, | |
| "learning_rate": 9.894209769304696e-05, | |
| "loss": 2.6054, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.11390250193448542, | |
| "grad_norm": 0.6478956937789917, | |
| "learning_rate": 9.891399076191674e-05, | |
| "loss": 2.6168, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.11472788238328605, | |
| "grad_norm": 0.8620642423629761, | |
| "learning_rate": 9.888551942780162e-05, | |
| "loss": 2.6313, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.11555326283208667, | |
| "grad_norm": 0.740717887878418, | |
| "learning_rate": 9.885668390280941e-05, | |
| "loss": 2.6307, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.11637864328088729, | |
| "grad_norm": 0.7513862252235413, | |
| "learning_rate": 9.882748440176109e-05, | |
| "loss": 2.625, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.1172040237296879, | |
| "grad_norm": 0.8409993052482605, | |
| "learning_rate": 9.879792114218921e-05, | |
| "loss": 2.6034, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.11802940417848852, | |
| "grad_norm": 0.8200739622116089, | |
| "learning_rate": 9.876799434433628e-05, | |
| "loss": 2.599, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.11885478462728914, | |
| "grad_norm": 0.9191763401031494, | |
| "learning_rate": 9.873770423115314e-05, | |
| "loss": 2.6168, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.11968016507608976, | |
| "grad_norm": 0.7739763855934143, | |
| "learning_rate": 9.870705102829723e-05, | |
| "loss": 2.6279, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.12050554552489037, | |
| "grad_norm": 0.6580247282981873, | |
| "learning_rate": 9.867603496413103e-05, | |
| "loss": 2.599, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.121330925973691, | |
| "grad_norm": 0.7197789549827576, | |
| "learning_rate": 9.864465626972023e-05, | |
| "loss": 2.5948, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.12215630642249162, | |
| "grad_norm": 0.9027787446975708, | |
| "learning_rate": 9.861291517883213e-05, | |
| "loss": 2.6058, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.12298168687129224, | |
| "grad_norm": 1.048640489578247, | |
| "learning_rate": 9.858081192793378e-05, | |
| "loss": 2.6128, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.12380706732009286, | |
| "grad_norm": 0.827551543712616, | |
| "learning_rate": 9.85483467561903e-05, | |
| "loss": 2.6058, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.12463244776889347, | |
| "grad_norm": 0.9751214981079102, | |
| "learning_rate": 9.851551990546306e-05, | |
| "loss": 2.57, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.1254578282176941, | |
| "grad_norm": 1.0783475637435913, | |
| "learning_rate": 9.848233162030794e-05, | |
| "loss": 2.6116, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.1262832086664947, | |
| "grad_norm": 0.8441977500915527, | |
| "learning_rate": 9.84487821479734e-05, | |
| "loss": 2.59, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.12710858911529532, | |
| "grad_norm": 1.0184507369995117, | |
| "learning_rate": 9.841487173839873e-05, | |
| "loss": 2.579, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.12793396956409595, | |
| "grad_norm": 0.9782202243804932, | |
| "learning_rate": 9.838060064421217e-05, | |
| "loss": 2.5996, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.12875935001289657, | |
| "grad_norm": 0.8052064180374146, | |
| "learning_rate": 9.834596912072897e-05, | |
| "loss": 2.594, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.1295847304616972, | |
| "grad_norm": 0.765544056892395, | |
| "learning_rate": 9.831097742594958e-05, | |
| "loss": 2.581, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.1304101109104978, | |
| "grad_norm": 0.8481123447418213, | |
| "learning_rate": 9.827562582055765e-05, | |
| "loss": 2.6006, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.13123549135929843, | |
| "grad_norm": 0.8604638576507568, | |
| "learning_rate": 9.823991456791811e-05, | |
| "loss": 2.5875, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.13206087180809906, | |
| "grad_norm": 0.6848525404930115, | |
| "learning_rate": 9.820384393407525e-05, | |
| "loss": 2.5788, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.13288625225689968, | |
| "grad_norm": 0.7860177159309387, | |
| "learning_rate": 9.816741418775066e-05, | |
| "loss": 2.5961, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.13371163270570027, | |
| "grad_norm": 0.7415710091590881, | |
| "learning_rate": 9.813062560034134e-05, | |
| "loss": 2.5805, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.1345370131545009, | |
| "grad_norm": 0.8323041200637817, | |
| "learning_rate": 9.809347844591753e-05, | |
| "loss": 2.5799, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.13536239360330152, | |
| "grad_norm": 0.9540163278579712, | |
| "learning_rate": 9.805597300122081e-05, | |
| "loss": 2.5704, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.13618777405210214, | |
| "grad_norm": 0.7344382405281067, | |
| "learning_rate": 9.801810954566195e-05, | |
| "loss": 2.5649, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.13701315450090276, | |
| "grad_norm": 0.7706190347671509, | |
| "learning_rate": 9.797988836131884e-05, | |
| "loss": 2.5583, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.13783853494970338, | |
| "grad_norm": 0.7073199152946472, | |
| "learning_rate": 9.794130973293445e-05, | |
| "loss": 2.5523, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.138663915398504, | |
| "grad_norm": 0.6726153492927551, | |
| "learning_rate": 9.790237394791461e-05, | |
| "loss": 2.5673, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.13948929584730463, | |
| "grad_norm": 0.6806092262268066, | |
| "learning_rate": 9.786308129632598e-05, | |
| "loss": 2.5748, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.14031467629610522, | |
| "grad_norm": 0.7212201952934265, | |
| "learning_rate": 9.782343207089377e-05, | |
| "loss": 2.5615, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.14114005674490585, | |
| "grad_norm": 0.7233206629753113, | |
| "learning_rate": 9.778342656699964e-05, | |
| "loss": 2.5509, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.14196543719370647, | |
| "grad_norm": 0.6308603882789612, | |
| "learning_rate": 9.77430650826795e-05, | |
| "loss": 2.5133, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.1427908176425071, | |
| "grad_norm": 0.676368772983551, | |
| "learning_rate": 9.770234791862125e-05, | |
| "loss": 2.5293, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.1436161980913077, | |
| "grad_norm": 0.687326192855835, | |
| "learning_rate": 9.766127537816256e-05, | |
| "loss": 2.531, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.14444157854010833, | |
| "grad_norm": 0.7347912788391113, | |
| "learning_rate": 9.761984776728864e-05, | |
| "loss": 2.5468, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.14526695898890896, | |
| "grad_norm": 0.5843812227249146, | |
| "learning_rate": 9.757806539462985e-05, | |
| "loss": 2.539, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.14609233943770958, | |
| "grad_norm": 0.7662450671195984, | |
| "learning_rate": 9.753592857145957e-05, | |
| "loss": 2.5297, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.1469177198865102, | |
| "grad_norm": 0.6813721656799316, | |
| "learning_rate": 9.749343761169171e-05, | |
| "loss": 2.5519, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.1477431003353108, | |
| "grad_norm": 0.7090803384780884, | |
| "learning_rate": 9.745059283187857e-05, | |
| "loss": 2.515, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.14856848078411142, | |
| "grad_norm": 0.5888863801956177, | |
| "learning_rate": 9.74073945512082e-05, | |
| "loss": 2.5622, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.14939386123291204, | |
| "grad_norm": 0.7369230389595032, | |
| "learning_rate": 9.736384309150233e-05, | |
| "loss": 2.5482, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.15021924168171266, | |
| "grad_norm": 0.6921555995941162, | |
| "learning_rate": 9.731993877721377e-05, | |
| "loss": 2.5304, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.15104462213051328, | |
| "grad_norm": 0.755409300327301, | |
| "learning_rate": 9.727568193542403e-05, | |
| "loss": 2.5214, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.1518700025793139, | |
| "grad_norm": 0.9340344071388245, | |
| "learning_rate": 9.723107289584095e-05, | |
| "loss": 2.5248, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.15269538302811453, | |
| "grad_norm": 0.9866952300071716, | |
| "learning_rate": 9.718611199079617e-05, | |
| "loss": 2.5281, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.15352076347691515, | |
| "grad_norm": 0.8538560271263123, | |
| "learning_rate": 9.714079955524269e-05, | |
| "loss": 2.5436, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.15434614392571577, | |
| "grad_norm": 0.6923696398735046, | |
| "learning_rate": 9.709513592675236e-05, | |
| "loss": 2.5431, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.15517152437451637, | |
| "grad_norm": 0.6311334371566772, | |
| "learning_rate": 9.704912144551341e-05, | |
| "loss": 2.5473, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.155996904823317, | |
| "grad_norm": 0.7558380961418152, | |
| "learning_rate": 9.700275645432784e-05, | |
| "loss": 2.4998, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.1568222852721176, | |
| "grad_norm": 0.8375957608222961, | |
| "learning_rate": 9.695604129860889e-05, | |
| "loss": 2.524, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.15764766572091823, | |
| "grad_norm": 0.9554900527000427, | |
| "learning_rate": 9.690897632637852e-05, | |
| "loss": 2.5184, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.15847304616971886, | |
| "grad_norm": 0.7582331299781799, | |
| "learning_rate": 9.686156188826478e-05, | |
| "loss": 2.5177, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.15929842661851948, | |
| "grad_norm": 0.8506788611412048, | |
| "learning_rate": 9.681379833749915e-05, | |
| "loss": 2.4866, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.1601238070673201, | |
| "grad_norm": 0.738979697227478, | |
| "learning_rate": 9.676568602991399e-05, | |
| "loss": 2.5286, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.16094918751612072, | |
| "grad_norm": 0.6935485005378723, | |
| "learning_rate": 9.671722532393985e-05, | |
| "loss": 2.532, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.16177456796492132, | |
| "grad_norm": 0.7109572887420654, | |
| "learning_rate": 9.666841658060282e-05, | |
| "loss": 2.5126, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.16259994841372194, | |
| "grad_norm": 0.8224465847015381, | |
| "learning_rate": 9.661926016352178e-05, | |
| "loss": 2.5062, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.16342532886252256, | |
| "grad_norm": 0.7705041766166687, | |
| "learning_rate": 9.656975643890578e-05, | |
| "loss": 2.5173, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.16425070931132318, | |
| "grad_norm": 0.6699206829071045, | |
| "learning_rate": 9.651990577555122e-05, | |
| "loss": 2.5047, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.1650760897601238, | |
| "grad_norm": 0.6773229241371155, | |
| "learning_rate": 9.64697085448392e-05, | |
| "loss": 2.5437, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.16590147020892443, | |
| "grad_norm": 0.6482560634613037, | |
| "learning_rate": 9.641916512073268e-05, | |
| "loss": 2.4866, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.16672685065772505, | |
| "grad_norm": 0.5687413811683655, | |
| "learning_rate": 9.636827587977368e-05, | |
| "loss": 2.5004, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.16755223110652567, | |
| "grad_norm": 0.591502845287323, | |
| "learning_rate": 9.63170412010806e-05, | |
| "loss": 2.4925, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.1683776115553263, | |
| "grad_norm": 0.7202515602111816, | |
| "learning_rate": 9.626546146634523e-05, | |
| "loss": 2.4834, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.1692029920041269, | |
| "grad_norm": 0.5923997163772583, | |
| "learning_rate": 9.621353705982998e-05, | |
| "loss": 2.4832, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.1700283724529275, | |
| "grad_norm": 0.519095242023468, | |
| "learning_rate": 9.616126836836508e-05, | |
| "loss": 2.4909, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.17085375290172813, | |
| "grad_norm": 0.6338533759117126, | |
| "learning_rate": 9.61086557813456e-05, | |
| "loss": 2.5027, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.17167913335052876, | |
| "grad_norm": 0.5483947992324829, | |
| "learning_rate": 9.60556996907286e-05, | |
| "loss": 2.4864, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.17250451379932938, | |
| "grad_norm": 0.6210249662399292, | |
| "learning_rate": 9.600240049103017e-05, | |
| "loss": 2.4987, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.17332989424813, | |
| "grad_norm": 0.6927972435951233, | |
| "learning_rate": 9.594875857932258e-05, | |
| "loss": 2.4987, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.17415527469693062, | |
| "grad_norm": 0.6685944199562073, | |
| "learning_rate": 9.589477435523118e-05, | |
| "loss": 2.4794, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.17498065514573125, | |
| "grad_norm": 0.89150470495224, | |
| "learning_rate": 9.584044822093157e-05, | |
| "loss": 2.467, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.17580603559453187, | |
| "grad_norm": 0.8726872205734253, | |
| "learning_rate": 9.57857805811465e-05, | |
| "loss": 2.4917, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.17663141604333246, | |
| "grad_norm": 0.8355669379234314, | |
| "learning_rate": 9.573077184314294e-05, | |
| "loss": 2.5095, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.17745679649213308, | |
| "grad_norm": 0.7523061037063599, | |
| "learning_rate": 9.567542241672891e-05, | |
| "loss": 2.4695, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.1782821769409337, | |
| "grad_norm": 0.5890743732452393, | |
| "learning_rate": 9.561973271425061e-05, | |
| "loss": 2.4748, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.17910755738973433, | |
| "grad_norm": 0.6020349860191345, | |
| "learning_rate": 9.55637031505892e-05, | |
| "loss": 2.4746, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.17993293783853495, | |
| "grad_norm": 0.6291228532791138, | |
| "learning_rate": 9.550733414315776e-05, | |
| "loss": 2.4537, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.18075831828733557, | |
| "grad_norm": 0.6267942786216736, | |
| "learning_rate": 9.545062611189821e-05, | |
| "loss": 2.4663, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.1815836987361362, | |
| "grad_norm": 0.7870015501976013, | |
| "learning_rate": 9.539357947927815e-05, | |
| "loss": 2.4794, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.18240907918493682, | |
| "grad_norm": 0.818481981754303, | |
| "learning_rate": 9.53361946702877e-05, | |
| "loss": 2.4538, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.1832344596337374, | |
| "grad_norm": 0.6101433634757996, | |
| "learning_rate": 9.527847211243635e-05, | |
| "loss": 2.5041, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.18405984008253803, | |
| "grad_norm": 0.7772427201271057, | |
| "learning_rate": 9.52204122357498e-05, | |
| "loss": 2.4497, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.18488522053133866, | |
| "grad_norm": 0.6459339261054993, | |
| "learning_rate": 9.516201547276668e-05, | |
| "loss": 2.4636, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.18571060098013928, | |
| "grad_norm": 0.6417956948280334, | |
| "learning_rate": 9.510328225853549e-05, | |
| "loss": 2.4489, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1865359814289399, | |
| "grad_norm": 0.5950794219970703, | |
| "learning_rate": 9.50442130306111e-05, | |
| "loss": 2.4418, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.18736136187774052, | |
| "grad_norm": 0.4874417185783386, | |
| "learning_rate": 9.498480822905176e-05, | |
| "loss": 2.4458, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.18818674232654115, | |
| "grad_norm": 0.6025642156600952, | |
| "learning_rate": 9.492506829641566e-05, | |
| "loss": 2.4582, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.18901212277534177, | |
| "grad_norm": 0.5527782440185547, | |
| "learning_rate": 9.486499367775764e-05, | |
| "loss": 2.4732, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.1898375032241424, | |
| "grad_norm": 0.5585253238677979, | |
| "learning_rate": 9.480458482062594e-05, | |
| "loss": 2.4494, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.19066288367294298, | |
| "grad_norm": 0.7675066590309143, | |
| "learning_rate": 9.474384217505883e-05, | |
| "loss": 2.4404, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.1914882641217436, | |
| "grad_norm": 0.6983161568641663, | |
| "learning_rate": 9.468276619358129e-05, | |
| "loss": 2.4409, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.19231364457054423, | |
| "grad_norm": 0.6638076305389404, | |
| "learning_rate": 9.462135733120156e-05, | |
| "loss": 2.4746, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.19313902501934485, | |
| "grad_norm": 0.7150386571884155, | |
| "learning_rate": 9.455961604540784e-05, | |
| "loss": 2.4841, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.19396440546814547, | |
| "grad_norm": 0.6076739430427551, | |
| "learning_rate": 9.449754279616481e-05, | |
| "loss": 2.4154, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.1947897859169461, | |
| "grad_norm": 0.6565660238265991, | |
| "learning_rate": 9.443513804591026e-05, | |
| "loss": 2.4033, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.19561516636574672, | |
| "grad_norm": 0.5531965494155884, | |
| "learning_rate": 9.43724022595516e-05, | |
| "loss": 2.4344, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.19644054681454734, | |
| "grad_norm": 0.6353370547294617, | |
| "learning_rate": 9.430933590446244e-05, | |
| "loss": 2.4631, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.19726592726334796, | |
| "grad_norm": 0.6833090782165527, | |
| "learning_rate": 9.424593945047906e-05, | |
| "loss": 2.4402, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.19809130771214856, | |
| "grad_norm": 0.5862318277359009, | |
| "learning_rate": 9.418221336989695e-05, | |
| "loss": 2.4308, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.19891668816094918, | |
| "grad_norm": 0.6512525081634521, | |
| "learning_rate": 9.411815813746726e-05, | |
| "loss": 2.4239, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.1997420686097498, | |
| "grad_norm": 0.6480604410171509, | |
| "learning_rate": 9.405377423039331e-05, | |
| "loss": 2.4211, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.20056744905855042, | |
| "grad_norm": 0.5605840682983398, | |
| "learning_rate": 9.398906212832699e-05, | |
| "loss": 2.4346, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.20139282950735105, | |
| "grad_norm": 0.6704816818237305, | |
| "learning_rate": 9.392402231336518e-05, | |
| "loss": 2.4309, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.20221820995615167, | |
| "grad_norm": 1.019185185432434, | |
| "learning_rate": 9.38586552700462e-05, | |
| "loss": 2.4341, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.2030435904049523, | |
| "grad_norm": 0.816562294960022, | |
| "learning_rate": 9.379296148534619e-05, | |
| "loss": 2.4389, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.2038689708537529, | |
| "grad_norm": 0.8160488605499268, | |
| "learning_rate": 9.372694144867544e-05, | |
| "loss": 2.4259, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.20469435130255353, | |
| "grad_norm": 0.6071799993515015, | |
| "learning_rate": 9.36605956518748e-05, | |
| "loss": 2.4479, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.20551973175135413, | |
| "grad_norm": 0.5928800106048584, | |
| "learning_rate": 9.359392458921198e-05, | |
| "loss": 2.4467, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.20634511220015475, | |
| "grad_norm": 0.6091005802154541, | |
| "learning_rate": 9.352692875737787e-05, | |
| "loss": 2.4204, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.20717049264895537, | |
| "grad_norm": 0.6261982917785645, | |
| "learning_rate": 9.34596086554829e-05, | |
| "loss": 2.3813, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.207995873097756, | |
| "grad_norm": 0.597626805305481, | |
| "learning_rate": 9.339196478505321e-05, | |
| "loss": 2.4288, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.20882125354655662, | |
| "grad_norm": 0.6542629599571228, | |
| "learning_rate": 9.332399765002698e-05, | |
| "loss": 2.4192, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.20964663399535724, | |
| "grad_norm": 0.5791048407554626, | |
| "learning_rate": 9.32557077567507e-05, | |
| "loss": 2.3969, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.21047201444415786, | |
| "grad_norm": 0.7887916564941406, | |
| "learning_rate": 9.318709561397537e-05, | |
| "loss": 2.4048, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.21129739489295848, | |
| "grad_norm": 0.5982603430747986, | |
| "learning_rate": 9.311816173285268e-05, | |
| "loss": 2.4113, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.21212277534175908, | |
| "grad_norm": 0.6093983054161072, | |
| "learning_rate": 9.304890662693123e-05, | |
| "loss": 2.4111, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.2129481557905597, | |
| "grad_norm": 0.6429126858711243, | |
| "learning_rate": 9.297933081215273e-05, | |
| "loss": 2.4279, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.21377353623936032, | |
| "grad_norm": 0.5441975593566895, | |
| "learning_rate": 9.29094348068481e-05, | |
| "loss": 2.4287, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.21459891668816095, | |
| "grad_norm": 0.635387659072876, | |
| "learning_rate": 9.283921913173368e-05, | |
| "loss": 2.43, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.21542429713696157, | |
| "grad_norm": 0.5462561845779419, | |
| "learning_rate": 9.276868430990726e-05, | |
| "loss": 2.4366, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.2162496775857622, | |
| "grad_norm": 0.53558748960495, | |
| "learning_rate": 9.269783086684428e-05, | |
| "loss": 2.4123, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.2170750580345628, | |
| "grad_norm": 0.6640864014625549, | |
| "learning_rate": 9.262665933039381e-05, | |
| "loss": 2.4034, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.21790043848336343, | |
| "grad_norm": 0.6774006485939026, | |
| "learning_rate": 9.255517023077472e-05, | |
| "loss": 2.4487, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.21872581893216406, | |
| "grad_norm": 0.7681392431259155, | |
| "learning_rate": 9.248336410057168e-05, | |
| "loss": 2.4377, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.21955119938096465, | |
| "grad_norm": 0.6729565262794495, | |
| "learning_rate": 9.24112414747312e-05, | |
| "loss": 2.4305, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.22037657982976527, | |
| "grad_norm": 0.5404065847396851, | |
| "learning_rate": 9.233880289055761e-05, | |
| "loss": 2.3878, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.2212019602785659, | |
| "grad_norm": 0.5672969818115234, | |
| "learning_rate": 9.226604888770911e-05, | |
| "loss": 2.4138, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.22202734072736652, | |
| "grad_norm": 0.5919613838195801, | |
| "learning_rate": 9.219298000819376e-05, | |
| "loss": 2.4219, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.22285272117616714, | |
| "grad_norm": 0.6129051446914673, | |
| "learning_rate": 9.211959679636535e-05, | |
| "loss": 2.3922, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.22367810162496776, | |
| "grad_norm": 0.5665661096572876, | |
| "learning_rate": 9.204589979891946e-05, | |
| "loss": 2.4229, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.22450348207376838, | |
| "grad_norm": 0.6748535633087158, | |
| "learning_rate": 9.197188956488931e-05, | |
| "loss": 2.3801, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.225328862522569, | |
| "grad_norm": 0.666181743144989, | |
| "learning_rate": 9.189756664564167e-05, | |
| "loss": 2.3888, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.22615424297136963, | |
| "grad_norm": 0.6297335028648376, | |
| "learning_rate": 9.182293159487281e-05, | |
| "loss": 2.3812, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.22697962342017022, | |
| "grad_norm": 0.6825816631317139, | |
| "learning_rate": 9.174798496860433e-05, | |
| "loss": 2.3837, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.22780500386897085, | |
| "grad_norm": 0.7759227156639099, | |
| "learning_rate": 9.167272732517903e-05, | |
| "loss": 2.3769, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.22863038431777147, | |
| "grad_norm": 0.6477057933807373, | |
| "learning_rate": 9.159715922525673e-05, | |
| "loss": 2.3852, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.2294557647665721, | |
| "grad_norm": 0.8262442946434021, | |
| "learning_rate": 9.152128123181013e-05, | |
| "loss": 2.3868, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.2302811452153727, | |
| "grad_norm": 0.6992378830909729, | |
| "learning_rate": 9.14450939101206e-05, | |
| "loss": 2.3896, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.23110652566417333, | |
| "grad_norm": 0.6080542206764221, | |
| "learning_rate": 9.136859782777394e-05, | |
| "loss": 2.3806, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.23193190611297396, | |
| "grad_norm": 0.7257338166236877, | |
| "learning_rate": 9.129179355465621e-05, | |
| "loss": 2.4114, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.23275728656177458, | |
| "grad_norm": 0.6741151213645935, | |
| "learning_rate": 9.121468166294945e-05, | |
| "loss": 2.3753, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.23358266701057517, | |
| "grad_norm": 0.6519246101379395, | |
| "learning_rate": 9.113726272712734e-05, | |
| "loss": 2.3937, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.2344080474593758, | |
| "grad_norm": 0.7125470042228699, | |
| "learning_rate": 9.105953732395116e-05, | |
| "loss": 2.4047, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.23523342790817642, | |
| "grad_norm": 0.5559272170066833, | |
| "learning_rate": 9.098150603246517e-05, | |
| "loss": 2.3927, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.23605880835697704, | |
| "grad_norm": 0.5721775889396667, | |
| "learning_rate": 9.090316943399255e-05, | |
| "loss": 2.3992, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.23688418880577766, | |
| "grad_norm": 0.5262630581855774, | |
| "learning_rate": 9.082452811213095e-05, | |
| "loss": 2.3898, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.23770956925457828, | |
| "grad_norm": 0.5832549333572388, | |
| "learning_rate": 9.074558265274819e-05, | |
| "loss": 2.3943, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.2385349497033789, | |
| "grad_norm": 0.5760109424591064, | |
| "learning_rate": 9.066633364397786e-05, | |
| "loss": 2.4223, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.23936033015217953, | |
| "grad_norm": 0.5974195003509521, | |
| "learning_rate": 9.058678167621493e-05, | |
| "loss": 2.402, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.24018571060098015, | |
| "grad_norm": 0.5613354444503784, | |
| "learning_rate": 9.050692734211142e-05, | |
| "loss": 2.3602, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.24101109104978075, | |
| "grad_norm": 0.6307066082954407, | |
| "learning_rate": 9.042677123657191e-05, | |
| "loss": 2.4034, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.24183647149858137, | |
| "grad_norm": 0.7120583057403564, | |
| "learning_rate": 9.034631395674917e-05, | |
| "loss": 2.3909, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.242661851947382, | |
| "grad_norm": 0.7483564615249634, | |
| "learning_rate": 9.026555610203964e-05, | |
| "loss": 2.4026, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.2434872323961826, | |
| "grad_norm": 0.5825770497322083, | |
| "learning_rate": 9.018449827407905e-05, | |
| "loss": 2.3736, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.24431261284498323, | |
| "grad_norm": 0.5669400095939636, | |
| "learning_rate": 9.010314107673783e-05, | |
| "loss": 2.376, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.24513799329378386, | |
| "grad_norm": 0.6108945608139038, | |
| "learning_rate": 9.002148511611675e-05, | |
| "loss": 2.3765, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.24596337374258448, | |
| "grad_norm": 0.499024361371994, | |
| "learning_rate": 8.993953100054224e-05, | |
| "loss": 2.4033, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.2467887541913851, | |
| "grad_norm": 0.4951154291629791, | |
| "learning_rate": 8.985727934056207e-05, | |
| "loss": 2.3808, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.24761413464018572, | |
| "grad_norm": 0.4967551529407501, | |
| "learning_rate": 8.977473074894052e-05, | |
| "loss": 2.3749, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.24843951508898632, | |
| "grad_norm": 0.6024683117866516, | |
| "learning_rate": 8.969188584065412e-05, | |
| "loss": 2.3745, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.24926489553778694, | |
| "grad_norm": 0.5306781530380249, | |
| "learning_rate": 8.960874523288683e-05, | |
| "loss": 2.3824, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.25009027598658756, | |
| "grad_norm": 0.5470788478851318, | |
| "learning_rate": 8.952530954502557e-05, | |
| "loss": 2.3828, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.2509156564353882, | |
| "grad_norm": 0.5227565765380859, | |
| "learning_rate": 8.944157939865556e-05, | |
| "loss": 2.3512, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.2517410368841888, | |
| "grad_norm": 0.5495042204856873, | |
| "learning_rate": 8.935755541755569e-05, | |
| "loss": 2.3553, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.2525664173329894, | |
| "grad_norm": 0.554063081741333, | |
| "learning_rate": 8.927323822769386e-05, | |
| "loss": 2.3492, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.25339179778179005, | |
| "grad_norm": 0.596449077129364, | |
| "learning_rate": 8.918862845722243e-05, | |
| "loss": 2.3708, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.25421717823059065, | |
| "grad_norm": 0.5647444128990173, | |
| "learning_rate": 8.910372673647336e-05, | |
| "loss": 2.3716, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.2550425586793913, | |
| "grad_norm": 0.6036911606788635, | |
| "learning_rate": 8.901853369795361e-05, | |
| "loss": 2.3728, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.2558679391281919, | |
| "grad_norm": 0.6191892027854919, | |
| "learning_rate": 8.893304997634045e-05, | |
| "loss": 2.36, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.25669331957699254, | |
| "grad_norm": 0.5799134373664856, | |
| "learning_rate": 8.884727620847669e-05, | |
| "loss": 2.3686, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.25751870002579313, | |
| "grad_norm": 0.6006870865821838, | |
| "learning_rate": 8.876121303336596e-05, | |
| "loss": 2.3705, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.2583440804745938, | |
| "grad_norm": 0.6628080606460571, | |
| "learning_rate": 8.867486109216789e-05, | |
| "loss": 2.3723, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.2591694609233944, | |
| "grad_norm": 0.6496407389640808, | |
| "learning_rate": 8.858822102819347e-05, | |
| "loss": 2.3438, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.259994841372195, | |
| "grad_norm": 0.5622187256813049, | |
| "learning_rate": 8.850129348690004e-05, | |
| "loss": 2.3729, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.2608202218209956, | |
| "grad_norm": 0.6995213031768799, | |
| "learning_rate": 8.841407911588675e-05, | |
| "loss": 2.3889, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.2616456022697962, | |
| "grad_norm": 0.6655398607254028, | |
| "learning_rate": 8.832657856488949e-05, | |
| "loss": 2.3803, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.26247098271859687, | |
| "grad_norm": 0.7044237852096558, | |
| "learning_rate": 8.82387924857762e-05, | |
| "loss": 2.3709, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.26329636316739746, | |
| "grad_norm": 0.6213721036911011, | |
| "learning_rate": 8.815072153254195e-05, | |
| "loss": 2.3698, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.2641217436161981, | |
| "grad_norm": 0.6705930233001709, | |
| "learning_rate": 8.806236636130411e-05, | |
| "loss": 2.329, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.2649471240649987, | |
| "grad_norm": 0.4926806688308716, | |
| "learning_rate": 8.797372763029742e-05, | |
| "loss": 2.3431, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.26577250451379936, | |
| "grad_norm": 0.5596938729286194, | |
| "learning_rate": 8.78848059998691e-05, | |
| "loss": 2.3888, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.26659788496259995, | |
| "grad_norm": 0.4732595980167389, | |
| "learning_rate": 8.779560213247395e-05, | |
| "loss": 2.3267, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.26742326541140055, | |
| "grad_norm": 0.5391475558280945, | |
| "learning_rate": 8.770611669266938e-05, | |
| "loss": 2.3869, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.2682486458602012, | |
| "grad_norm": 0.5030598640441895, | |
| "learning_rate": 8.76163503471105e-05, | |
| "loss": 2.3535, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.2690740263090018, | |
| "grad_norm": 0.6067845225334167, | |
| "learning_rate": 8.752630376454511e-05, | |
| "loss": 2.3711, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.26989940675780244, | |
| "grad_norm": 0.5053196549415588, | |
| "learning_rate": 8.743597761580877e-05, | |
| "loss": 2.3643, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.27072478720660303, | |
| "grad_norm": 0.5403187274932861, | |
| "learning_rate": 8.734537257381973e-05, | |
| "loss": 2.3531, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.2715501676554037, | |
| "grad_norm": 0.498923659324646, | |
| "learning_rate": 8.7254489313574e-05, | |
| "loss": 2.3423, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.2723755481042043, | |
| "grad_norm": 0.5593265295028687, | |
| "learning_rate": 8.716332851214024e-05, | |
| "loss": 2.3548, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.2732009285530049, | |
| "grad_norm": 0.5866305232048035, | |
| "learning_rate": 8.707189084865481e-05, | |
| "loss": 2.3677, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.2740263090018055, | |
| "grad_norm": 0.5718086957931519, | |
| "learning_rate": 8.698017700431662e-05, | |
| "loss": 2.3416, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.2748516894506061, | |
| "grad_norm": 0.6683026552200317, | |
| "learning_rate": 8.688818766238208e-05, | |
| "loss": 2.3658, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.27567706989940677, | |
| "grad_norm": 0.682115912437439, | |
| "learning_rate": 8.679592350816007e-05, | |
| "loss": 2.3287, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.27650245034820736, | |
| "grad_norm": 0.6516245007514954, | |
| "learning_rate": 8.670338522900672e-05, | |
| "loss": 2.3095, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.277327830797008, | |
| "grad_norm": 0.5818968415260315, | |
| "learning_rate": 8.66105735143204e-05, | |
| "loss": 2.3354, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.2781532112458086, | |
| "grad_norm": 0.5822069644927979, | |
| "learning_rate": 8.651748905553656e-05, | |
| "loss": 2.3402, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.27897859169460926, | |
| "grad_norm": 0.5988590717315674, | |
| "learning_rate": 8.642413254612251e-05, | |
| "loss": 2.3488, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.27980397214340985, | |
| "grad_norm": 0.5264620184898376, | |
| "learning_rate": 8.633050468157234e-05, | |
| "loss": 2.3336, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.28062935259221045, | |
| "grad_norm": 0.5173693895339966, | |
| "learning_rate": 8.623660615940165e-05, | |
| "loss": 2.3201, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.2814547330410111, | |
| "grad_norm": 0.5370919108390808, | |
| "learning_rate": 8.61424376791425e-05, | |
| "loss": 2.353, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.2822801134898117, | |
| "grad_norm": 0.5648570656776428, | |
| "learning_rate": 8.604799994233798e-05, | |
| "loss": 2.3517, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.28310549393861234, | |
| "grad_norm": 0.5215715169906616, | |
| "learning_rate": 8.595329365253719e-05, | |
| "loss": 2.3333, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.28393087438741293, | |
| "grad_norm": 0.6153488755226135, | |
| "learning_rate": 8.585831951528991e-05, | |
| "loss": 2.3617, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.2847562548362136, | |
| "grad_norm": 0.6054225564002991, | |
| "learning_rate": 8.576307823814132e-05, | |
| "loss": 2.3149, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.2855816352850142, | |
| "grad_norm": 0.5253807902336121, | |
| "learning_rate": 8.566757053062678e-05, | |
| "loss": 2.3114, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.28640701573381483, | |
| "grad_norm": 0.5196321606636047, | |
| "learning_rate": 8.55717971042665e-05, | |
| "loss": 2.3275, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.2872323961826154, | |
| "grad_norm": 0.5359232425689697, | |
| "learning_rate": 8.54757586725603e-05, | |
| "loss": 2.3108, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.288057776631416, | |
| "grad_norm": 0.5835343599319458, | |
| "learning_rate": 8.537945595098222e-05, | |
| "loss": 2.3423, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.28888315708021667, | |
| "grad_norm": 0.49723029136657715, | |
| "learning_rate": 8.528288965697523e-05, | |
| "loss": 2.2839, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.28970853752901726, | |
| "grad_norm": 0.5888681411743164, | |
| "learning_rate": 8.518606050994591e-05, | |
| "loss": 2.3485, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.2905339179778179, | |
| "grad_norm": 0.5658313632011414, | |
| "learning_rate": 8.508896923125901e-05, | |
| "loss": 2.359, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.2913592984266185, | |
| "grad_norm": 0.5612741112709045, | |
| "learning_rate": 8.499161654423219e-05, | |
| "loss": 2.3482, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.29218467887541916, | |
| "grad_norm": 0.51905757188797, | |
| "learning_rate": 8.48940031741305e-05, | |
| "loss": 2.3349, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.29301005932421975, | |
| "grad_norm": 0.5525624752044678, | |
| "learning_rate": 8.479612984816112e-05, | |
| "loss": 2.3318, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.2938354397730204, | |
| "grad_norm": 0.6043487191200256, | |
| "learning_rate": 8.469799729546781e-05, | |
| "loss": 2.3399, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.294660820221821, | |
| "grad_norm": 0.5291466116905212, | |
| "learning_rate": 8.459960624712556e-05, | |
| "loss": 2.3175, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.2954862006706216, | |
| "grad_norm": 0.5733122229576111, | |
| "learning_rate": 8.450095743613512e-05, | |
| "loss": 2.3366, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.29631158111942224, | |
| "grad_norm": 0.5417082905769348, | |
| "learning_rate": 8.440205159741752e-05, | |
| "loss": 2.3335, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.29713696156822283, | |
| "grad_norm": 0.565000593662262, | |
| "learning_rate": 8.430288946780865e-05, | |
| "loss": 2.3384, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.2979623420170235, | |
| "grad_norm": 0.6104756593704224, | |
| "learning_rate": 8.420347178605367e-05, | |
| "loss": 2.3587, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.2987877224658241, | |
| "grad_norm": 0.5492483973503113, | |
| "learning_rate": 8.410379929280168e-05, | |
| "loss": 2.3169, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.29961310291462473, | |
| "grad_norm": 0.5351945161819458, | |
| "learning_rate": 8.400387273059998e-05, | |
| "loss": 2.3195, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.3004384833634253, | |
| "grad_norm": 0.5342651605606079, | |
| "learning_rate": 8.39036928438887e-05, | |
| "loss": 2.2912, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.301263863812226, | |
| "grad_norm": 0.5106682181358337, | |
| "learning_rate": 8.380326037899522e-05, | |
| "loss": 2.3283, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.30208924426102657, | |
| "grad_norm": 0.5767691731452942, | |
| "learning_rate": 8.370257608412857e-05, | |
| "loss": 2.3286, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.30291462470982716, | |
| "grad_norm": 0.54031902551651, | |
| "learning_rate": 8.360164070937389e-05, | |
| "loss": 2.3243, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.3037400051586278, | |
| "grad_norm": 0.6625233888626099, | |
| "learning_rate": 8.350045500668681e-05, | |
| "loss": 2.3232, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.3045653856074284, | |
| "grad_norm": 0.6196743845939636, | |
| "learning_rate": 8.339901972988795e-05, | |
| "loss": 2.3216, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.30539076605622906, | |
| "grad_norm": 0.5990370512008667, | |
| "learning_rate": 8.329733563465711e-05, | |
| "loss": 2.3401, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.30621614650502965, | |
| "grad_norm": 0.5713345408439636, | |
| "learning_rate": 8.319540347852786e-05, | |
| "loss": 2.3416, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.3070415269538303, | |
| "grad_norm": 0.5813875198364258, | |
| "learning_rate": 8.30932240208817e-05, | |
| "loss": 2.2979, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.3078669074026309, | |
| "grad_norm": 0.6877428889274597, | |
| "learning_rate": 8.299079802294258e-05, | |
| "loss": 2.317, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.30869228785143155, | |
| "grad_norm": 0.5318434834480286, | |
| "learning_rate": 8.288812624777109e-05, | |
| "loss": 2.2947, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.30951766830023214, | |
| "grad_norm": 0.5925495624542236, | |
| "learning_rate": 8.278520946025884e-05, | |
| "loss": 2.3114, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.31034304874903273, | |
| "grad_norm": 0.5941009521484375, | |
| "learning_rate": 8.268204842712278e-05, | |
| "loss": 2.3463, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.3111684291978334, | |
| "grad_norm": 0.5976232886314392, | |
| "learning_rate": 8.25786439168994e-05, | |
| "loss": 2.2996, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.311993809646634, | |
| "grad_norm": 0.5421935319900513, | |
| "learning_rate": 8.24749966999391e-05, | |
| "loss": 2.3143, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.31281919009543463, | |
| "grad_norm": 0.5300918817520142, | |
| "learning_rate": 8.237110754840043e-05, | |
| "loss": 2.2946, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.3136445705442352, | |
| "grad_norm": 0.5926389098167419, | |
| "learning_rate": 8.22669772362443e-05, | |
| "loss": 2.3312, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.3144699509930359, | |
| "grad_norm": 0.5365331768989563, | |
| "learning_rate": 8.216260653922823e-05, | |
| "loss": 2.3165, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.31529533144183647, | |
| "grad_norm": 0.5278505086898804, | |
| "learning_rate": 8.205799623490055e-05, | |
| "loss": 2.2811, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.31612071189063706, | |
| "grad_norm": 0.5568265318870544, | |
| "learning_rate": 8.195314710259475e-05, | |
| "loss": 2.2899, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.3169460923394377, | |
| "grad_norm": 0.6386959552764893, | |
| "learning_rate": 8.184805992342342e-05, | |
| "loss": 2.3193, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.3177714727882383, | |
| "grad_norm": 0.5363957285881042, | |
| "learning_rate": 8.174273548027262e-05, | |
| "loss": 2.3255, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.31859685323703896, | |
| "grad_norm": 0.6581987142562866, | |
| "learning_rate": 8.163717455779602e-05, | |
| "loss": 2.3159, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.31942223368583955, | |
| "grad_norm": 0.7844798564910889, | |
| "learning_rate": 8.153137794240903e-05, | |
| "loss": 2.3241, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.3202476141346402, | |
| "grad_norm": 0.5486951470375061, | |
| "learning_rate": 8.142534642228288e-05, | |
| "loss": 2.2834, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.3210729945834408, | |
| "grad_norm": 0.6187033653259277, | |
| "learning_rate": 8.13190807873389e-05, | |
| "loss": 2.31, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.32189837503224145, | |
| "grad_norm": 0.6599840521812439, | |
| "learning_rate": 8.121258182924247e-05, | |
| "loss": 2.3075, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.32272375548104204, | |
| "grad_norm": 0.6156419515609741, | |
| "learning_rate": 8.110585034139723e-05, | |
| "loss": 2.3171, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.32354913592984264, | |
| "grad_norm": 0.652897834777832, | |
| "learning_rate": 8.099888711893917e-05, | |
| "loss": 2.2963, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.3243745163786433, | |
| "grad_norm": 0.5405826568603516, | |
| "learning_rate": 8.089169295873058e-05, | |
| "loss": 2.299, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.3251998968274439, | |
| "grad_norm": 0.5489581823348999, | |
| "learning_rate": 8.078426865935432e-05, | |
| "loss": 2.3051, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.32602527727624453, | |
| "grad_norm": 0.5330743789672852, | |
| "learning_rate": 8.067661502110768e-05, | |
| "loss": 2.306, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.3268506577250451, | |
| "grad_norm": 0.5833327174186707, | |
| "learning_rate": 8.056873284599648e-05, | |
| "loss": 2.3197, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.3276760381738458, | |
| "grad_norm": 0.638880729675293, | |
| "learning_rate": 8.046062293772922e-05, | |
| "loss": 2.309, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 0.32850141862264637, | |
| "grad_norm": 0.6887226104736328, | |
| "learning_rate": 8.035228610171085e-05, | |
| "loss": 2.3192, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.329326799071447, | |
| "grad_norm": 0.5981518030166626, | |
| "learning_rate": 8.024372314503701e-05, | |
| "loss": 2.2958, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 0.3301521795202476, | |
| "grad_norm": 0.5953544974327087, | |
| "learning_rate": 8.013493487648782e-05, | |
| "loss": 2.3161, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.3309775599690482, | |
| "grad_norm": 0.576503574848175, | |
| "learning_rate": 8.002592210652202e-05, | |
| "loss": 2.2901, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 0.33180294041784886, | |
| "grad_norm": 0.6154365539550781, | |
| "learning_rate": 7.991668564727082e-05, | |
| "loss": 2.3093, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.33262832086664945, | |
| "grad_norm": 0.5416200757026672, | |
| "learning_rate": 7.98072263125319e-05, | |
| "loss": 2.3004, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.3334537013154501, | |
| "grad_norm": 0.5973731875419617, | |
| "learning_rate": 7.969754491776329e-05, | |
| "loss": 2.3224, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.3342790817642507, | |
| "grad_norm": 0.5144022107124329, | |
| "learning_rate": 7.958764228007741e-05, | |
| "loss": 2.2899, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.33510446221305135, | |
| "grad_norm": 0.627142608165741, | |
| "learning_rate": 7.947751921823488e-05, | |
| "loss": 2.3196, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.33592984266185194, | |
| "grad_norm": 0.5556638240814209, | |
| "learning_rate": 7.936717655263841e-05, | |
| "loss": 2.2991, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 0.3367552231106526, | |
| "grad_norm": 0.5637221336364746, | |
| "learning_rate": 7.925661510532681e-05, | |
| "loss": 2.3093, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.3375806035594532, | |
| "grad_norm": 0.6272327899932861, | |
| "learning_rate": 7.91458356999687e-05, | |
| "loss": 2.3286, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 0.3384059840082538, | |
| "grad_norm": 0.5124315023422241, | |
| "learning_rate": 7.903483916185654e-05, | |
| "loss": 2.2879, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.33923136445705443, | |
| "grad_norm": 0.5467550158500671, | |
| "learning_rate": 7.892362631790035e-05, | |
| "loss": 2.3107, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 0.340056744905855, | |
| "grad_norm": 0.5180369019508362, | |
| "learning_rate": 7.881219799662164e-05, | |
| "loss": 2.2784, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 0.3408821253546557, | |
| "grad_norm": 0.5402101874351501, | |
| "learning_rate": 7.870055502814714e-05, | |
| "loss": 2.2704, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 0.34170750580345627, | |
| "grad_norm": 0.4879278242588043, | |
| "learning_rate": 7.858869824420272e-05, | |
| "loss": 2.2913, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.3425328862522569, | |
| "grad_norm": 0.5525892972946167, | |
| "learning_rate": 7.847662847810713e-05, | |
| "loss": 2.3071, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.3433582667010575, | |
| "grad_norm": 0.517867922782898, | |
| "learning_rate": 7.836434656476583e-05, | |
| "loss": 2.2748, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.34418364714985816, | |
| "grad_norm": 0.5947239398956299, | |
| "learning_rate": 7.825185334066475e-05, | |
| "loss": 2.2687, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 0.34500902759865876, | |
| "grad_norm": 0.6157646775245667, | |
| "learning_rate": 7.813914964386401e-05, | |
| "loss": 2.2769, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 0.34583440804745935, | |
| "grad_norm": 0.5485296845436096, | |
| "learning_rate": 7.802623631399176e-05, | |
| "loss": 2.2663, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 0.34665978849626, | |
| "grad_norm": 0.4869697093963623, | |
| "learning_rate": 7.791311419223791e-05, | |
| "loss": 2.2908, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.3474851689450606, | |
| "grad_norm": 0.5534776449203491, | |
| "learning_rate": 7.779978412134783e-05, | |
| "loss": 2.2998, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 0.34831054939386125, | |
| "grad_norm": 0.5947521328926086, | |
| "learning_rate": 7.768624694561604e-05, | |
| "loss": 2.2907, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 0.34913592984266184, | |
| "grad_norm": 0.5616730451583862, | |
| "learning_rate": 7.757250351088004e-05, | |
| "loss": 2.3003, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 0.3499613102914625, | |
| "grad_norm": 0.5840707421302795, | |
| "learning_rate": 7.745855466451385e-05, | |
| "loss": 2.2683, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 0.3507866907402631, | |
| "grad_norm": 0.5235263705253601, | |
| "learning_rate": 7.734440125542186e-05, | |
| "loss": 2.2976, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.35161207118906374, | |
| "grad_norm": 0.5673782229423523, | |
| "learning_rate": 7.723004413403238e-05, | |
| "loss": 2.2936, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 0.35243745163786433, | |
| "grad_norm": 0.5472978353500366, | |
| "learning_rate": 7.711548415229135e-05, | |
| "loss": 2.2896, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 0.3532628320866649, | |
| "grad_norm": 0.5697288513183594, | |
| "learning_rate": 7.700072216365602e-05, | |
| "loss": 2.3239, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 0.3540882125354656, | |
| "grad_norm": 0.6162300705909729, | |
| "learning_rate": 7.688575902308854e-05, | |
| "loss": 2.2973, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 0.35491359298426617, | |
| "grad_norm": 0.4970763027667999, | |
| "learning_rate": 7.677059558704965e-05, | |
| "loss": 2.3018, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.3557389734330668, | |
| "grad_norm": 0.486848920583725, | |
| "learning_rate": 7.665523271349221e-05, | |
| "loss": 2.3116, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 0.3565643538818674, | |
| "grad_norm": 0.5172179341316223, | |
| "learning_rate": 7.65396712618549e-05, | |
| "loss": 2.2898, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 0.35738973433066806, | |
| "grad_norm": 0.5573694705963135, | |
| "learning_rate": 7.642391209305581e-05, | |
| "loss": 2.2866, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 0.35821511477946866, | |
| "grad_norm": 0.540338397026062, | |
| "learning_rate": 7.630795606948592e-05, | |
| "loss": 2.3066, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 0.3590404952282693, | |
| "grad_norm": 0.5216573476791382, | |
| "learning_rate": 7.619180405500284e-05, | |
| "loss": 2.2489, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.3598658756770699, | |
| "grad_norm": 0.5518911480903625, | |
| "learning_rate": 7.607545691492421e-05, | |
| "loss": 2.2621, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 0.3606912561258705, | |
| "grad_norm": 0.5395997166633606, | |
| "learning_rate": 7.595891551602139e-05, | |
| "loss": 2.2828, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 0.36151663657467115, | |
| "grad_norm": 0.5570027828216553, | |
| "learning_rate": 7.584218072651291e-05, | |
| "loss": 2.2721, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 0.36234201702347174, | |
| "grad_norm": 0.530053436756134, | |
| "learning_rate": 7.572525341605805e-05, | |
| "loss": 2.3146, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 0.3631673974722724, | |
| "grad_norm": 0.5206712484359741, | |
| "learning_rate": 7.560813445575032e-05, | |
| "loss": 2.2633, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.363992777921073, | |
| "grad_norm": 0.5709933638572693, | |
| "learning_rate": 7.549082471811105e-05, | |
| "loss": 2.2524, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 0.36481815836987364, | |
| "grad_norm": 0.5382503271102905, | |
| "learning_rate": 7.53733250770828e-05, | |
| "loss": 2.2871, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 0.36564353881867423, | |
| "grad_norm": 0.49123620986938477, | |
| "learning_rate": 7.525563640802286e-05, | |
| "loss": 2.2906, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 0.3664689192674748, | |
| "grad_norm": 0.48953843116760254, | |
| "learning_rate": 7.513775958769683e-05, | |
| "loss": 2.2912, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 0.3672942997162755, | |
| "grad_norm": 0.5467056632041931, | |
| "learning_rate": 7.501969549427195e-05, | |
| "loss": 2.2961, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.36811968016507607, | |
| "grad_norm": 0.5301052331924438, | |
| "learning_rate": 7.49014450073106e-05, | |
| "loss": 2.2588, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 0.3689450606138767, | |
| "grad_norm": 0.5778504014015198, | |
| "learning_rate": 7.478300900776387e-05, | |
| "loss": 2.2627, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 0.3697704410626773, | |
| "grad_norm": 0.5600552558898926, | |
| "learning_rate": 7.46643883779648e-05, | |
| "loss": 2.2874, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.37059582151147796, | |
| "grad_norm": 0.5173321962356567, | |
| "learning_rate": 7.454558400162195e-05, | |
| "loss": 2.2787, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 0.37142120196027856, | |
| "grad_norm": 0.534444272518158, | |
| "learning_rate": 7.442659676381275e-05, | |
| "loss": 2.2883, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.3722465824090792, | |
| "grad_norm": 0.5139839053153992, | |
| "learning_rate": 7.430742755097689e-05, | |
| "loss": 2.2999, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 0.3730719628578798, | |
| "grad_norm": 0.4908299744129181, | |
| "learning_rate": 7.418807725090983e-05, | |
| "loss": 2.285, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 0.3738973433066804, | |
| "grad_norm": 0.49594688415527344, | |
| "learning_rate": 7.406854675275605e-05, | |
| "loss": 2.2884, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 0.37472272375548105, | |
| "grad_norm": 0.49902480840682983, | |
| "learning_rate": 7.39488369470025e-05, | |
| "loss": 2.2778, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 0.37554810420428164, | |
| "grad_norm": 0.46000784635543823, | |
| "learning_rate": 7.382894872547195e-05, | |
| "loss": 2.2797, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.3763734846530823, | |
| "grad_norm": 0.4911547899246216, | |
| "learning_rate": 7.370888298131633e-05, | |
| "loss": 2.2611, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 0.3771988651018829, | |
| "grad_norm": 0.5099422335624695, | |
| "learning_rate": 7.35886406090101e-05, | |
| "loss": 2.2536, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 0.37802424555068354, | |
| "grad_norm": 0.42195364832878113, | |
| "learning_rate": 7.34682225043436e-05, | |
| "loss": 2.2597, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 0.37884962599948413, | |
| "grad_norm": 0.5352451205253601, | |
| "learning_rate": 7.334762956441632e-05, | |
| "loss": 2.2866, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 0.3796750064482848, | |
| "grad_norm": 0.4699161946773529, | |
| "learning_rate": 7.322686268763026e-05, | |
| "loss": 2.2801, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.3805003868970854, | |
| "grad_norm": 0.5075897574424744, | |
| "learning_rate": 7.310592277368322e-05, | |
| "loss": 2.2814, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 0.38132576734588597, | |
| "grad_norm": 0.4679708480834961, | |
| "learning_rate": 7.298481072356214e-05, | |
| "loss": 2.2524, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 0.3821511477946866, | |
| "grad_norm": 0.5117970705032349, | |
| "learning_rate": 7.28635274395363e-05, | |
| "loss": 2.2605, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 0.3829765282434872, | |
| "grad_norm": 0.518598735332489, | |
| "learning_rate": 7.274207382515071e-05, | |
| "loss": 2.2685, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 0.38380190869228786, | |
| "grad_norm": 0.5765763521194458, | |
| "learning_rate": 7.262045078521924e-05, | |
| "loss": 2.2894, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.38462728914108846, | |
| "grad_norm": 0.48513707518577576, | |
| "learning_rate": 7.249865922581807e-05, | |
| "loss": 2.2658, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 0.3854526695898891, | |
| "grad_norm": 0.5026444792747498, | |
| "learning_rate": 7.237670005427872e-05, | |
| "loss": 2.283, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 0.3862780500386897, | |
| "grad_norm": 0.5490180253982544, | |
| "learning_rate": 7.225457417918144e-05, | |
| "loss": 2.241, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 0.38710343048749035, | |
| "grad_norm": 0.54227215051651, | |
| "learning_rate": 7.213228251034844e-05, | |
| "loss": 2.2338, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 0.38792881093629095, | |
| "grad_norm": 0.5452665686607361, | |
| "learning_rate": 7.2009825958837e-05, | |
| "loss": 2.2478, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.38875419138509154, | |
| "grad_norm": 0.5434293150901794, | |
| "learning_rate": 7.188720543693283e-05, | |
| "loss": 2.2489, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 0.3895795718338922, | |
| "grad_norm": 0.5174756050109863, | |
| "learning_rate": 7.176442185814312e-05, | |
| "loss": 2.2562, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 0.3904049522826928, | |
| "grad_norm": 0.5360869765281677, | |
| "learning_rate": 7.164147613718986e-05, | |
| "loss": 2.254, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 0.39123033273149344, | |
| "grad_norm": 0.5211062431335449, | |
| "learning_rate": 7.151836919000299e-05, | |
| "loss": 2.2501, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 0.39205571318029403, | |
| "grad_norm": 0.5354079604148865, | |
| "learning_rate": 7.139510193371352e-05, | |
| "loss": 2.2585, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.3928810936290947, | |
| "grad_norm": 0.5471782684326172, | |
| "learning_rate": 7.127167528664682e-05, | |
| "loss": 2.2502, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 0.3937064740778953, | |
| "grad_norm": 0.5891269445419312, | |
| "learning_rate": 7.114809016831558e-05, | |
| "loss": 2.2916, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 0.3945318545266959, | |
| "grad_norm": 0.6164124608039856, | |
| "learning_rate": 7.10243474994132e-05, | |
| "loss": 2.2526, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 0.3953572349754965, | |
| "grad_norm": 0.5750834345817566, | |
| "learning_rate": 7.090044820180673e-05, | |
| "loss": 2.25, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 0.3961826154242971, | |
| "grad_norm": 0.5516604781150818, | |
| "learning_rate": 7.077639319853013e-05, | |
| "loss": 2.2624, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.39700799587309776, | |
| "grad_norm": 0.5377829670906067, | |
| "learning_rate": 7.065218341377734e-05, | |
| "loss": 2.2542, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 0.39783337632189836, | |
| "grad_norm": 0.5297653079032898, | |
| "learning_rate": 7.05278197728954e-05, | |
| "loss": 2.2784, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 0.398658756770699, | |
| "grad_norm": 0.5507510304450989, | |
| "learning_rate": 7.040330320237752e-05, | |
| "loss": 2.2848, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 0.3994841372194996, | |
| "grad_norm": 0.571143388748169, | |
| "learning_rate": 7.027863462985628e-05, | |
| "loss": 2.2697, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 0.40030951766830025, | |
| "grad_norm": 0.538505494594574, | |
| "learning_rate": 7.015381498409661e-05, | |
| "loss": 2.2532, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.40113489811710085, | |
| "grad_norm": 0.53383868932724, | |
| "learning_rate": 7.002884519498895e-05, | |
| "loss": 2.2399, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 0.4019602785659015, | |
| "grad_norm": 0.46435126662254333, | |
| "learning_rate": 6.990372619354224e-05, | |
| "loss": 2.2689, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 0.4027856590147021, | |
| "grad_norm": 0.49260959029197693, | |
| "learning_rate": 6.977845891187708e-05, | |
| "loss": 2.2653, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 0.4036110394635027, | |
| "grad_norm": 0.4841521084308624, | |
| "learning_rate": 6.965304428321874e-05, | |
| "loss": 2.2743, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 0.40443641991230334, | |
| "grad_norm": 0.45740070939064026, | |
| "learning_rate": 6.952748324189016e-05, | |
| "loss": 2.2406, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.40526180036110393, | |
| "grad_norm": 0.4746413826942444, | |
| "learning_rate": 6.940177672330508e-05, | |
| "loss": 2.2461, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 0.4060871808099046, | |
| "grad_norm": 0.4509133994579315, | |
| "learning_rate": 6.9275925663961e-05, | |
| "loss": 2.2287, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 0.4069125612587052, | |
| "grad_norm": 0.44701477885246277, | |
| "learning_rate": 6.914993100143224e-05, | |
| "loss": 2.2771, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 0.4077379417075058, | |
| "grad_norm": 0.49677759408950806, | |
| "learning_rate": 6.902379367436296e-05, | |
| "loss": 2.2434, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 0.4085633221563064, | |
| "grad_norm": 0.5722448229789734, | |
| "learning_rate": 6.889751462246013e-05, | |
| "loss": 2.228, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.40938870260510707, | |
| "grad_norm": 0.4709642827510834, | |
| "learning_rate": 6.877109478648656e-05, | |
| "loss": 2.2444, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 0.41021408305390766, | |
| "grad_norm": 0.6394116282463074, | |
| "learning_rate": 6.864453510825388e-05, | |
| "loss": 2.2317, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 0.41103946350270826, | |
| "grad_norm": 0.5443106889724731, | |
| "learning_rate": 6.851783653061555e-05, | |
| "loss": 2.2135, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 0.4118648439515089, | |
| "grad_norm": 0.6894340515136719, | |
| "learning_rate": 6.83909999974598e-05, | |
| "loss": 2.2545, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 0.4126902244003095, | |
| "grad_norm": 0.5471647381782532, | |
| "learning_rate": 6.826402645370256e-05, | |
| "loss": 2.2036, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.41351560484911015, | |
| "grad_norm": 0.5389876365661621, | |
| "learning_rate": 6.813691684528054e-05, | |
| "loss": 2.2575, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 0.41434098529791075, | |
| "grad_norm": 0.5603637099266052, | |
| "learning_rate": 6.800967211914409e-05, | |
| "loss": 2.2577, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 0.4151663657467114, | |
| "grad_norm": 0.5469692349433899, | |
| "learning_rate": 6.788229322325022e-05, | |
| "loss": 2.2472, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 0.415991746195512, | |
| "grad_norm": 0.5489314794540405, | |
| "learning_rate": 6.775478110655535e-05, | |
| "loss": 2.2818, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 0.4168171266443126, | |
| "grad_norm": 0.4870460331439972, | |
| "learning_rate": 6.762713671900853e-05, | |
| "loss": 2.2747, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.41764250709311324, | |
| "grad_norm": 0.5328664183616638, | |
| "learning_rate": 6.74993610115441e-05, | |
| "loss": 2.2826, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 0.41846788754191383, | |
| "grad_norm": 0.5465196371078491, | |
| "learning_rate": 6.737145493607482e-05, | |
| "loss": 2.2562, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 0.4192932679907145, | |
| "grad_norm": 0.4999396502971649, | |
| "learning_rate": 6.724341944548459e-05, | |
| "loss": 2.2545, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 0.4201186484395151, | |
| "grad_norm": 0.49227967858314514, | |
| "learning_rate": 6.711525549362144e-05, | |
| "loss": 2.2322, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 0.4209440288883157, | |
| "grad_norm": 0.4593503773212433, | |
| "learning_rate": 6.698696403529049e-05, | |
| "loss": 2.2817, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.4217694093371163, | |
| "grad_norm": 0.477332204580307, | |
| "learning_rate": 6.685854602624668e-05, | |
| "loss": 2.25, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 0.42259478978591697, | |
| "grad_norm": 0.4970756471157074, | |
| "learning_rate": 6.673000242318782e-05, | |
| "loss": 2.2937, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.42342017023471756, | |
| "grad_norm": 0.4330197274684906, | |
| "learning_rate": 6.660133418374732e-05, | |
| "loss": 2.2231, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 0.42424555068351816, | |
| "grad_norm": 0.45938703417778015, | |
| "learning_rate": 6.647254226648711e-05, | |
| "loss": 2.2166, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 0.4250709311323188, | |
| "grad_norm": 0.45608118176460266, | |
| "learning_rate": 6.634362763089056e-05, | |
| "loss": 2.2346, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.4258963115811194, | |
| "grad_norm": 0.5052918195724487, | |
| "learning_rate": 6.621459123735522e-05, | |
| "loss": 2.2608, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 0.42672169202992005, | |
| "grad_norm": 0.4969039559364319, | |
| "learning_rate": 6.608543404718578e-05, | |
| "loss": 2.2349, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 0.42754707247872065, | |
| "grad_norm": 0.5392360091209412, | |
| "learning_rate": 6.595615702258676e-05, | |
| "loss": 2.2463, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 0.4283724529275213, | |
| "grad_norm": 0.5267171859741211, | |
| "learning_rate": 6.58267611266555e-05, | |
| "loss": 2.2157, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 0.4291978333763219, | |
| "grad_norm": 0.4976198673248291, | |
| "learning_rate": 6.569724732337495e-05, | |
| "loss": 2.214, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.43002321382512254, | |
| "grad_norm": 0.46527931094169617, | |
| "learning_rate": 6.556761657760635e-05, | |
| "loss": 2.2162, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 0.43084859427392314, | |
| "grad_norm": 0.4234091341495514, | |
| "learning_rate": 6.543786985508223e-05, | |
| "loss": 2.2026, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 0.43167397472272373, | |
| "grad_norm": 0.41595569252967834, | |
| "learning_rate": 6.53080081223991e-05, | |
| "loss": 2.2413, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 0.4324993551715244, | |
| "grad_norm": 0.43087056279182434, | |
| "learning_rate": 6.517803234701025e-05, | |
| "loss": 2.2465, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 0.433324735620325, | |
| "grad_norm": 0.4584639072418213, | |
| "learning_rate": 6.504794349721866e-05, | |
| "loss": 2.2353, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.4341501160691256, | |
| "grad_norm": 0.49019527435302734, | |
| "learning_rate": 6.491774254216963e-05, | |
| "loss": 2.218, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 0.4349754965179262, | |
| "grad_norm": 0.46099725365638733, | |
| "learning_rate": 6.478743045184365e-05, | |
| "loss": 2.2402, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 0.43580087696672687, | |
| "grad_norm": 0.48430031538009644, | |
| "learning_rate": 6.465700819704913e-05, | |
| "loss": 2.2382, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 0.43662625741552746, | |
| "grad_norm": 0.48902517557144165, | |
| "learning_rate": 6.452647674941524e-05, | |
| "loss": 2.2328, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 0.4374516378643281, | |
| "grad_norm": 0.46378427743911743, | |
| "learning_rate": 6.439583708138459e-05, | |
| "loss": 2.2587, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.4382770183131287, | |
| "grad_norm": 0.4577491879463196, | |
| "learning_rate": 6.426509016620603e-05, | |
| "loss": 2.2379, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 0.4391023987619293, | |
| "grad_norm": 0.44876828789711, | |
| "learning_rate": 6.413423697792737e-05, | |
| "loss": 2.2537, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 0.43992777921072995, | |
| "grad_norm": 0.501110851764679, | |
| "learning_rate": 6.400327849138814e-05, | |
| "loss": 2.2101, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 0.44075315965953055, | |
| "grad_norm": 0.43247148394584656, | |
| "learning_rate": 6.387221568221238e-05, | |
| "loss": 2.2493, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 0.4415785401083312, | |
| "grad_norm": 0.5204429030418396, | |
| "learning_rate": 6.374104952680125e-05, | |
| "loss": 2.2396, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.4424039205571318, | |
| "grad_norm": 0.4856724143028259, | |
| "learning_rate": 6.360978100232587e-05, | |
| "loss": 2.2443, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 0.44322930100593244, | |
| "grad_norm": 0.5140888690948486, | |
| "learning_rate": 6.347841108672e-05, | |
| "loss": 2.2539, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 0.44405468145473304, | |
| "grad_norm": 0.4858645796775818, | |
| "learning_rate": 6.334694075867269e-05, | |
| "loss": 2.2525, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 0.4448800619035337, | |
| "grad_norm": 0.4742701053619385, | |
| "learning_rate": 6.321537099762114e-05, | |
| "loss": 2.244, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 0.4457054423523343, | |
| "grad_norm": 0.4515312910079956, | |
| "learning_rate": 6.308370278374325e-05, | |
| "loss": 2.225, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.4465308228011349, | |
| "grad_norm": 0.46387842297554016, | |
| "learning_rate": 6.295193709795037e-05, | |
| "loss": 2.2135, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 0.4473562032499355, | |
| "grad_norm": 0.44970181584358215, | |
| "learning_rate": 6.282007492188011e-05, | |
| "loss": 2.248, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 0.4481815836987361, | |
| "grad_norm": 0.46752220392227173, | |
| "learning_rate": 6.268811723788877e-05, | |
| "loss": 2.2795, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 0.44900696414753677, | |
| "grad_norm": 0.48435142636299133, | |
| "learning_rate": 6.255606502904429e-05, | |
| "loss": 2.2396, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 0.44983234459633736, | |
| "grad_norm": 0.4091418385505676, | |
| "learning_rate": 6.242391927911872e-05, | |
| "loss": 2.2252, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.450657725045138, | |
| "grad_norm": 0.4950377345085144, | |
| "learning_rate": 6.229168097258106e-05, | |
| "loss": 2.2474, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 0.4514831054939386, | |
| "grad_norm": 0.4584418535232544, | |
| "learning_rate": 6.21593510945898e-05, | |
| "loss": 2.245, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 0.45230848594273926, | |
| "grad_norm": 0.4853164851665497, | |
| "learning_rate": 6.202693063098561e-05, | |
| "loss": 2.2587, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 0.45313386639153985, | |
| "grad_norm": 0.48432743549346924, | |
| "learning_rate": 6.189442056828407e-05, | |
| "loss": 2.2225, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 0.45395924684034045, | |
| "grad_norm": 0.4741179943084717, | |
| "learning_rate": 6.176182189366819e-05, | |
| "loss": 2.2461, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.4547846272891411, | |
| "grad_norm": 0.48972228169441223, | |
| "learning_rate": 6.16291355949812e-05, | |
| "loss": 2.2534, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 0.4556100077379417, | |
| "grad_norm": 0.501619279384613, | |
| "learning_rate": 6.149636266071904e-05, | |
| "loss": 2.2196, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 0.45643538818674234, | |
| "grad_norm": 0.4932022988796234, | |
| "learning_rate": 6.136350408002314e-05, | |
| "loss": 2.1943, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 0.45726076863554294, | |
| "grad_norm": 0.4711919128894806, | |
| "learning_rate": 6.123056084267296e-05, | |
| "loss": 2.2274, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 0.4580861490843436, | |
| "grad_norm": 0.4498264491558075, | |
| "learning_rate": 6.109753393907862e-05, | |
| "loss": 2.2332, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.4589115295331442, | |
| "grad_norm": 0.4392760694026947, | |
| "learning_rate": 6.09644243602736e-05, | |
| "loss": 2.2566, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 0.45973690998194483, | |
| "grad_norm": 0.48753243684768677, | |
| "learning_rate": 6.0831233097907236e-05, | |
| "loss": 2.2242, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 0.4605622904307454, | |
| "grad_norm": 0.4300667643547058, | |
| "learning_rate": 6.069796114423743e-05, | |
| "loss": 2.2305, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 0.461387670879546, | |
| "grad_norm": 0.4927161633968353, | |
| "learning_rate": 6.056460949212324e-05, | |
| "loss": 2.2336, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 0.46221305132834667, | |
| "grad_norm": 0.4289657175540924, | |
| "learning_rate": 6.043117913501741e-05, | |
| "loss": 2.2242, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.46303843177714726, | |
| "grad_norm": 0.44931545853614807, | |
| "learning_rate": 6.029767106695909e-05, | |
| "loss": 2.222, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 0.4638638122259479, | |
| "grad_norm": 0.5036013126373291, | |
| "learning_rate": 6.0164086282566326e-05, | |
| "loss": 2.2366, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 0.4646891926747485, | |
| "grad_norm": 0.536802351474762, | |
| "learning_rate": 6.0030425777028685e-05, | |
| "loss": 2.2069, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 0.46551457312354916, | |
| "grad_norm": 0.46372362971305847, | |
| "learning_rate": 5.9896690546099906e-05, | |
| "loss": 2.2497, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 0.46633995357234975, | |
| "grad_norm": 0.48737701773643494, | |
| "learning_rate": 5.9762881586090344e-05, | |
| "loss": 2.2389, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.46716533402115035, | |
| "grad_norm": 0.4532935321331024, | |
| "learning_rate": 5.962899989385969e-05, | |
| "loss": 2.1895, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 0.467990714469951, | |
| "grad_norm": 0.45759317278862, | |
| "learning_rate": 5.9495046466809444e-05, | |
| "loss": 2.2228, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 0.4688160949187516, | |
| "grad_norm": 0.41080954670906067, | |
| "learning_rate": 5.936102230287553e-05, | |
| "loss": 2.2269, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 0.46964147536755224, | |
| "grad_norm": 0.42970025539398193, | |
| "learning_rate": 5.9226928400520854e-05, | |
| "loss": 2.2291, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 0.47046685581635284, | |
| "grad_norm": 0.4297996759414673, | |
| "learning_rate": 5.9092765758727854e-05, | |
| "loss": 2.2319, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.4712922362651535, | |
| "grad_norm": 0.42044776678085327, | |
| "learning_rate": 5.8958535376991106e-05, | |
| "loss": 2.1991, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 0.4721176167139541, | |
| "grad_norm": 0.4343169927597046, | |
| "learning_rate": 5.882423825530981e-05, | |
| "loss": 2.224, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 0.47294299716275473, | |
| "grad_norm": 0.41891202330589294, | |
| "learning_rate": 5.8689875394180335e-05, | |
| "loss": 2.2232, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 0.4737683776115553, | |
| "grad_norm": 0.4097048342227936, | |
| "learning_rate": 5.855544779458887e-05, | |
| "loss": 2.2341, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 0.4745937580603559, | |
| "grad_norm": 0.5282487273216248, | |
| "learning_rate": 5.8420956458003876e-05, | |
| "loss": 2.2616, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.47541913850915657, | |
| "grad_norm": 0.441388875246048, | |
| "learning_rate": 5.828640238636861e-05, | |
| "loss": 2.1994, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 0.47624451895795716, | |
| "grad_norm": 0.4527244567871094, | |
| "learning_rate": 5.815178658209372e-05, | |
| "loss": 2.2287, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 0.4770698994067578, | |
| "grad_norm": 0.4659084379673004, | |
| "learning_rate": 5.801711004804979e-05, | |
| "loss": 2.2059, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 0.4778952798555584, | |
| "grad_norm": 0.46441391110420227, | |
| "learning_rate": 5.7882373787559775e-05, | |
| "loss": 2.2184, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 0.47872066030435906, | |
| "grad_norm": 0.47255754470825195, | |
| "learning_rate": 5.7747578804391624e-05, | |
| "loss": 2.2101, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.47954604075315965, | |
| "grad_norm": 0.4485050439834595, | |
| "learning_rate": 5.761272610275074e-05, | |
| "loss": 2.2034, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 0.4803714212019603, | |
| "grad_norm": 0.402884840965271, | |
| "learning_rate": 5.747781668727251e-05, | |
| "loss": 2.2278, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 0.4811968016507609, | |
| "grad_norm": 0.45074397325515747, | |
| "learning_rate": 5.73428515630149e-05, | |
| "loss": 2.2272, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 0.4820221820995615, | |
| "grad_norm": 0.48101773858070374, | |
| "learning_rate": 5.72078317354508e-05, | |
| "loss": 2.2207, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 0.48284756254836214, | |
| "grad_norm": 0.5213938355445862, | |
| "learning_rate": 5.7072758210460716e-05, | |
| "loss": 2.2025, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.48367294299716274, | |
| "grad_norm": 0.5374065041542053, | |
| "learning_rate": 5.693763199432516e-05, | |
| "loss": 2.2319, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 0.4844983234459634, | |
| "grad_norm": 0.5042341351509094, | |
| "learning_rate": 5.680245409371716e-05, | |
| "loss": 2.2216, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 0.485323703894764, | |
| "grad_norm": 0.4710129499435425, | |
| "learning_rate": 5.666722551569484e-05, | |
| "loss": 2.2126, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 0.48614908434356463, | |
| "grad_norm": 0.46355435252189636, | |
| "learning_rate": 5.653194726769382e-05, | |
| "loss": 2.2465, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 0.4869744647923652, | |
| "grad_norm": 0.5684232711791992, | |
| "learning_rate": 5.63966203575198e-05, | |
| "loss": 2.2307, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.4877998452411659, | |
| "grad_norm": 0.5232070684432983, | |
| "learning_rate": 5.6261245793340944e-05, | |
| "loss": 2.204, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 0.48862522568996647, | |
| "grad_norm": 0.47950464487075806, | |
| "learning_rate": 5.612582458368047e-05, | |
| "loss": 2.2277, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 0.48945060613876706, | |
| "grad_norm": 0.551558792591095, | |
| "learning_rate": 5.599035773740915e-05, | |
| "loss": 2.2111, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 0.4902759865875677, | |
| "grad_norm": 0.5343023538589478, | |
| "learning_rate": 5.5854846263737625e-05, | |
| "loss": 2.2322, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 0.4911013670363683, | |
| "grad_norm": 0.4963313639163971, | |
| "learning_rate": 5.571929117220911e-05, | |
| "loss": 2.2048, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.49192674748516896, | |
| "grad_norm": 0.564353883266449, | |
| "learning_rate": 5.558369347269169e-05, | |
| "loss": 2.2381, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 0.49275212793396955, | |
| "grad_norm": 0.5161218047142029, | |
| "learning_rate": 5.544805417537096e-05, | |
| "loss": 2.2253, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 0.4935775083827702, | |
| "grad_norm": 0.5393477082252502, | |
| "learning_rate": 5.531237429074231e-05, | |
| "loss": 2.2268, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 0.4944028888315708, | |
| "grad_norm": 0.5184178948402405, | |
| "learning_rate": 5.517665482960359e-05, | |
| "loss": 2.2219, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 0.49522826928037145, | |
| "grad_norm": 0.45648929476737976, | |
| "learning_rate": 5.504089680304745e-05, | |
| "loss": 2.207, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.49605364972917204, | |
| "grad_norm": 0.4045144021511078, | |
| "learning_rate": 5.490510122245384e-05, | |
| "loss": 2.2117, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 0.49687903017797264, | |
| "grad_norm": 0.4706050157546997, | |
| "learning_rate": 5.47692690994825e-05, | |
| "loss": 2.1831, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 0.4977044106267733, | |
| "grad_norm": 0.428549587726593, | |
| "learning_rate": 5.463340144606541e-05, | |
| "loss": 2.207, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 0.4985297910755739, | |
| "grad_norm": 0.41387081146240234, | |
| "learning_rate": 5.449749927439922e-05, | |
| "loss": 2.1875, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 0.49935517152437453, | |
| "grad_norm": 0.4002302587032318, | |
| "learning_rate": 5.436156359693777e-05, | |
| "loss": 2.2019, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.5001805519731751, | |
| "grad_norm": 0.4339626133441925, | |
| "learning_rate": 5.422559542638448e-05, | |
| "loss": 2.1963, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 0.5010059324219758, | |
| "grad_norm": 0.47335943579673767, | |
| "learning_rate": 5.4089595775684886e-05, | |
| "loss": 2.2179, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 0.5018313128707764, | |
| "grad_norm": 0.5027151107788086, | |
| "learning_rate": 5.395356565801899e-05, | |
| "loss": 2.1787, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 0.502656693319577, | |
| "grad_norm": 0.6695459485054016, | |
| "learning_rate": 5.3817506086793813e-05, | |
| "loss": 2.1898, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 0.5034820737683776, | |
| "grad_norm": 0.46689918637275696, | |
| "learning_rate": 5.368141807563578e-05, | |
| "loss": 2.198, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.5043074542171783, | |
| "grad_norm": 0.42445775866508484, | |
| "learning_rate": 5.35453026383832e-05, | |
| "loss": 2.1951, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 0.5051328346659788, | |
| "grad_norm": 0.4324966371059418, | |
| "learning_rate": 5.34091607890787e-05, | |
| "loss": 2.2097, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 0.5059582151147795, | |
| "grad_norm": 0.5108333826065063, | |
| "learning_rate": 5.327299354196167e-05, | |
| "loss": 2.219, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 0.5067835955635801, | |
| "grad_norm": 0.4502308964729309, | |
| "learning_rate": 5.313680191146071e-05, | |
| "loss": 2.1852, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 0.5076089760123808, | |
| "grad_norm": 0.4594053328037262, | |
| "learning_rate": 5.300058691218607e-05, | |
| "loss": 2.2387, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.5084343564611813, | |
| "grad_norm": 0.4119742512702942, | |
| "learning_rate": 5.28643495589221e-05, | |
| "loss": 2.1933, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 0.5092597369099819, | |
| "grad_norm": 0.48443832993507385, | |
| "learning_rate": 5.272809086661972e-05, | |
| "loss": 2.1906, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 0.5100851173587826, | |
| "grad_norm": 0.5221107602119446, | |
| "learning_rate": 5.2591811850388726e-05, | |
| "loss": 2.2195, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 0.5109104978075831, | |
| "grad_norm": 0.47215166687965393, | |
| "learning_rate": 5.245551352549044e-05, | |
| "loss": 2.192, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 0.5117358782563838, | |
| "grad_norm": 0.49213218688964844, | |
| "learning_rate": 5.231919690732995e-05, | |
| "loss": 2.2189, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.5125612587051844, | |
| "grad_norm": 0.4713018536567688, | |
| "learning_rate": 5.218286301144867e-05, | |
| "loss": 2.1888, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 0.5133866391539851, | |
| "grad_norm": 0.47867047786712646, | |
| "learning_rate": 5.2046512853516696e-05, | |
| "loss": 2.2104, | |
| "step": 1244 | |
| }, | |
| { | |
| "epoch": 0.5142120196027856, | |
| "grad_norm": 0.4218139946460724, | |
| "learning_rate": 5.1910147449325295e-05, | |
| "loss": 2.1981, | |
| "step": 1246 | |
| }, | |
| { | |
| "epoch": 0.5150374000515863, | |
| "grad_norm": 0.3904942572116852, | |
| "learning_rate": 5.177376781477933e-05, | |
| "loss": 2.1932, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 0.5158627805003869, | |
| "grad_norm": 0.4622825086116791, | |
| "learning_rate": 5.163737496588964e-05, | |
| "loss": 2.2195, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.5166881609491876, | |
| "grad_norm": 0.49373888969421387, | |
| "learning_rate": 5.150096991876556e-05, | |
| "loss": 2.2207, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 0.5175135413979881, | |
| "grad_norm": 0.4939506947994232, | |
| "learning_rate": 5.1364553689607264e-05, | |
| "loss": 2.214, | |
| "step": 1254 | |
| }, | |
| { | |
| "epoch": 0.5183389218467888, | |
| "grad_norm": 0.4570372402667999, | |
| "learning_rate": 5.1228127294698225e-05, | |
| "loss": 2.173, | |
| "step": 1256 | |
| }, | |
| { | |
| "epoch": 0.5191643022955894, | |
| "grad_norm": 0.4791816174983978, | |
| "learning_rate": 5.1091691750397675e-05, | |
| "loss": 2.2091, | |
| "step": 1258 | |
| }, | |
| { | |
| "epoch": 0.51998968274439, | |
| "grad_norm": 0.4587259292602539, | |
| "learning_rate": 5.0955248073132975e-05, | |
| "loss": 2.2092, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.5208150631931906, | |
| "grad_norm": 0.4694560170173645, | |
| "learning_rate": 5.081879727939214e-05, | |
| "loss": 2.188, | |
| "step": 1262 | |
| }, | |
| { | |
| "epoch": 0.5216404436419912, | |
| "grad_norm": 0.4412589371204376, | |
| "learning_rate": 5.068234038571612e-05, | |
| "loss": 2.2085, | |
| "step": 1264 | |
| }, | |
| { | |
| "epoch": 0.5224658240907919, | |
| "grad_norm": 0.4659062325954437, | |
| "learning_rate": 5.054587840869136e-05, | |
| "loss": 2.206, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 0.5232912045395924, | |
| "grad_norm": 0.45120978355407715, | |
| "learning_rate": 5.0409412364942165e-05, | |
| "loss": 2.2031, | |
| "step": 1268 | |
| }, | |
| { | |
| "epoch": 0.5241165849883931, | |
| "grad_norm": 0.5263309478759766, | |
| "learning_rate": 5.027294327112314e-05, | |
| "loss": 2.2222, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.5249419654371937, | |
| "grad_norm": 0.452806681394577, | |
| "learning_rate": 5.01364721439116e-05, | |
| "loss": 2.2038, | |
| "step": 1272 | |
| }, | |
| { | |
| "epoch": 0.5257673458859943, | |
| "grad_norm": 0.4813989996910095, | |
| "learning_rate": 5e-05, | |
| "loss": 2.2207, | |
| "step": 1274 | |
| }, | |
| { | |
| "epoch": 0.5265927263347949, | |
| "grad_norm": 0.5097128748893738, | |
| "learning_rate": 4.986352785608842e-05, | |
| "loss": 2.1874, | |
| "step": 1276 | |
| }, | |
| { | |
| "epoch": 0.5274181067835956, | |
| "grad_norm": 0.4897962510585785, | |
| "learning_rate": 4.9727056728876865e-05, | |
| "loss": 2.2289, | |
| "step": 1278 | |
| }, | |
| { | |
| "epoch": 0.5282434872323962, | |
| "grad_norm": 0.4591223895549774, | |
| "learning_rate": 4.959058763505784e-05, | |
| "loss": 2.2128, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.5290688676811968, | |
| "grad_norm": 0.4303410053253174, | |
| "learning_rate": 4.945412159130864e-05, | |
| "loss": 2.2377, | |
| "step": 1282 | |
| }, | |
| { | |
| "epoch": 0.5298942481299974, | |
| "grad_norm": 0.4496016800403595, | |
| "learning_rate": 4.931765961428389e-05, | |
| "loss": 2.2125, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 0.5307196285787981, | |
| "grad_norm": 0.49691838026046753, | |
| "learning_rate": 4.9181202720607874e-05, | |
| "loss": 2.1943, | |
| "step": 1286 | |
| }, | |
| { | |
| "epoch": 0.5315450090275987, | |
| "grad_norm": 0.4995412230491638, | |
| "learning_rate": 4.904475192686702e-05, | |
| "loss": 2.1847, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 0.5323703894763993, | |
| "grad_norm": 0.47865188121795654, | |
| "learning_rate": 4.890830824960234e-05, | |
| "loss": 2.2069, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.5331957699251999, | |
| "grad_norm": 0.40386030077934265, | |
| "learning_rate": 4.877187270530178e-05, | |
| "loss": 2.1561, | |
| "step": 1292 | |
| }, | |
| { | |
| "epoch": 0.5340211503740006, | |
| "grad_norm": 0.42778220772743225, | |
| "learning_rate": 4.863544631039275e-05, | |
| "loss": 2.2129, | |
| "step": 1294 | |
| }, | |
| { | |
| "epoch": 0.5348465308228011, | |
| "grad_norm": 0.4645983874797821, | |
| "learning_rate": 4.8499030081234444e-05, | |
| "loss": 2.2059, | |
| "step": 1296 | |
| }, | |
| { | |
| "epoch": 0.5356719112716017, | |
| "grad_norm": 0.4080425202846527, | |
| "learning_rate": 4.8362625034110354e-05, | |
| "loss": 2.1921, | |
| "step": 1298 | |
| }, | |
| { | |
| "epoch": 0.5364972917204024, | |
| "grad_norm": 0.49392467737197876, | |
| "learning_rate": 4.8226232185220684e-05, | |
| "loss": 2.2011, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.537322672169203, | |
| "grad_norm": 0.46411821246147156, | |
| "learning_rate": 4.808985255067473e-05, | |
| "loss": 2.1962, | |
| "step": 1302 | |
| }, | |
| { | |
| "epoch": 0.5381480526180036, | |
| "grad_norm": 0.4228093922138214, | |
| "learning_rate": 4.7953487146483315e-05, | |
| "loss": 2.1824, | |
| "step": 1304 | |
| }, | |
| { | |
| "epoch": 0.5389734330668042, | |
| "grad_norm": 0.44242537021636963, | |
| "learning_rate": 4.781713698855135e-05, | |
| "loss": 2.1818, | |
| "step": 1306 | |
| }, | |
| { | |
| "epoch": 0.5397988135156049, | |
| "grad_norm": 0.4092284142971039, | |
| "learning_rate": 4.7680803092670054e-05, | |
| "loss": 2.2122, | |
| "step": 1308 | |
| }, | |
| { | |
| "epoch": 0.5406241939644054, | |
| "grad_norm": 0.4461856484413147, | |
| "learning_rate": 4.754448647450957e-05, | |
| "loss": 2.2202, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.5414495744132061, | |
| "grad_norm": 0.42323821783065796, | |
| "learning_rate": 4.740818814961129e-05, | |
| "loss": 2.1846, | |
| "step": 1312 | |
| }, | |
| { | |
| "epoch": 0.5422749548620067, | |
| "grad_norm": 0.4296724498271942, | |
| "learning_rate": 4.72719091333803e-05, | |
| "loss": 2.1705, | |
| "step": 1314 | |
| }, | |
| { | |
| "epoch": 0.5431003353108074, | |
| "grad_norm": 0.44609442353248596, | |
| "learning_rate": 4.713565044107792e-05, | |
| "loss": 2.2015, | |
| "step": 1316 | |
| }, | |
| { | |
| "epoch": 0.5439257157596079, | |
| "grad_norm": 0.44201433658599854, | |
| "learning_rate": 4.699941308781394e-05, | |
| "loss": 2.1932, | |
| "step": 1318 | |
| }, | |
| { | |
| "epoch": 0.5447510962084086, | |
| "grad_norm": 0.4727790653705597, | |
| "learning_rate": 4.686319808853931e-05, | |
| "loss": 2.1913, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.5455764766572092, | |
| "grad_norm": 0.4268248677253723, | |
| "learning_rate": 4.6727006458038357e-05, | |
| "loss": 2.1954, | |
| "step": 1322 | |
| }, | |
| { | |
| "epoch": 0.5464018571060097, | |
| "grad_norm": 0.4677101969718933, | |
| "learning_rate": 4.659083921092131e-05, | |
| "loss": 2.2065, | |
| "step": 1324 | |
| }, | |
| { | |
| "epoch": 0.5472272375548104, | |
| "grad_norm": 0.44317469000816345, | |
| "learning_rate": 4.6454697361616816e-05, | |
| "loss": 2.203, | |
| "step": 1326 | |
| }, | |
| { | |
| "epoch": 0.548052618003611, | |
| "grad_norm": 0.40845441818237305, | |
| "learning_rate": 4.631858192436422e-05, | |
| "loss": 2.1871, | |
| "step": 1328 | |
| }, | |
| { | |
| "epoch": 0.5488779984524117, | |
| "grad_norm": 0.42979684472084045, | |
| "learning_rate": 4.6182493913206205e-05, | |
| "loss": 2.1843, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.5497033789012122, | |
| "grad_norm": 0.43987905979156494, | |
| "learning_rate": 4.6046434341981034e-05, | |
| "loss": 2.1977, | |
| "step": 1332 | |
| }, | |
| { | |
| "epoch": 0.5505287593500129, | |
| "grad_norm": 0.4020345211029053, | |
| "learning_rate": 4.591040422431512e-05, | |
| "loss": 2.1815, | |
| "step": 1334 | |
| }, | |
| { | |
| "epoch": 0.5513541397988135, | |
| "grad_norm": 0.42069509625434875, | |
| "learning_rate": 4.5774404573615534e-05, | |
| "loss": 2.1759, | |
| "step": 1336 | |
| }, | |
| { | |
| "epoch": 0.5521795202476142, | |
| "grad_norm": 0.45546865463256836, | |
| "learning_rate": 4.563843640306225e-05, | |
| "loss": 2.2009, | |
| "step": 1338 | |
| }, | |
| { | |
| "epoch": 0.5530049006964147, | |
| "grad_norm": 0.42900213599205017, | |
| "learning_rate": 4.550250072560079e-05, | |
| "loss": 2.224, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.5538302811452154, | |
| "grad_norm": 0.47856220602989197, | |
| "learning_rate": 4.5366598553934606e-05, | |
| "loss": 2.1765, | |
| "step": 1342 | |
| }, | |
| { | |
| "epoch": 0.554655661594016, | |
| "grad_norm": 0.44292357563972473, | |
| "learning_rate": 4.523073090051751e-05, | |
| "loss": 2.2077, | |
| "step": 1344 | |
| }, | |
| { | |
| "epoch": 0.5554810420428166, | |
| "grad_norm": 0.4365878403186798, | |
| "learning_rate": 4.5094898777546175e-05, | |
| "loss": 2.1926, | |
| "step": 1346 | |
| }, | |
| { | |
| "epoch": 0.5563064224916172, | |
| "grad_norm": 0.46365126967430115, | |
| "learning_rate": 4.495910319695257e-05, | |
| "loss": 2.1937, | |
| "step": 1348 | |
| }, | |
| { | |
| "epoch": 0.5571318029404179, | |
| "grad_norm": 0.444984495639801, | |
| "learning_rate": 4.4823345170396417e-05, | |
| "loss": 2.1887, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.5579571833892185, | |
| "grad_norm": 0.45117077231407166, | |
| "learning_rate": 4.468762570925771e-05, | |
| "loss": 2.173, | |
| "step": 1352 | |
| }, | |
| { | |
| "epoch": 0.558782563838019, | |
| "grad_norm": 0.42752307653427124, | |
| "learning_rate": 4.455194582462906e-05, | |
| "loss": 2.1704, | |
| "step": 1354 | |
| }, | |
| { | |
| "epoch": 0.5596079442868197, | |
| "grad_norm": 0.42912495136260986, | |
| "learning_rate": 4.441630652730831e-05, | |
| "loss": 2.1733, | |
| "step": 1356 | |
| }, | |
| { | |
| "epoch": 0.5604333247356204, | |
| "grad_norm": 0.43053561449050903, | |
| "learning_rate": 4.428070882779091e-05, | |
| "loss": 2.1811, | |
| "step": 1358 | |
| }, | |
| { | |
| "epoch": 0.5612587051844209, | |
| "grad_norm": 0.43450236320495605, | |
| "learning_rate": 4.4145153736262387e-05, | |
| "loss": 2.1957, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.5620840856332215, | |
| "grad_norm": 0.4460732936859131, | |
| "learning_rate": 4.400964226259087e-05, | |
| "loss": 2.2151, | |
| "step": 1362 | |
| }, | |
| { | |
| "epoch": 0.5629094660820222, | |
| "grad_norm": 0.4485209584236145, | |
| "learning_rate": 4.387417541631952e-05, | |
| "loss": 2.208, | |
| "step": 1364 | |
| }, | |
| { | |
| "epoch": 0.5637348465308228, | |
| "grad_norm": 0.431249737739563, | |
| "learning_rate": 4.373875420665907e-05, | |
| "loss": 2.1702, | |
| "step": 1366 | |
| }, | |
| { | |
| "epoch": 0.5645602269796234, | |
| "grad_norm": 0.42122384905815125, | |
| "learning_rate": 4.3603379642480216e-05, | |
| "loss": 2.1714, | |
| "step": 1368 | |
| }, | |
| { | |
| "epoch": 0.565385607428424, | |
| "grad_norm": 0.4479855000972748, | |
| "learning_rate": 4.3468052732306184e-05, | |
| "loss": 2.1934, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.5662109878772247, | |
| "grad_norm": 0.438849538564682, | |
| "learning_rate": 4.333277448430517e-05, | |
| "loss": 2.2139, | |
| "step": 1372 | |
| }, | |
| { | |
| "epoch": 0.5670363683260253, | |
| "grad_norm": 0.4208733141422272, | |
| "learning_rate": 4.3197545906282845e-05, | |
| "loss": 2.2021, | |
| "step": 1374 | |
| }, | |
| { | |
| "epoch": 0.5678617487748259, | |
| "grad_norm": 0.45249640941619873, | |
| "learning_rate": 4.306236800567485e-05, | |
| "loss": 2.2137, | |
| "step": 1376 | |
| }, | |
| { | |
| "epoch": 0.5686871292236265, | |
| "grad_norm": 0.3885250985622406, | |
| "learning_rate": 4.292724178953929e-05, | |
| "loss": 2.1872, | |
| "step": 1378 | |
| }, | |
| { | |
| "epoch": 0.5695125096724272, | |
| "grad_norm": 0.45552772283554077, | |
| "learning_rate": 4.27921682645492e-05, | |
| "loss": 2.1896, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.5703378901212277, | |
| "grad_norm": 0.4536789357662201, | |
| "learning_rate": 4.265714843698511e-05, | |
| "loss": 2.1669, | |
| "step": 1382 | |
| }, | |
| { | |
| "epoch": 0.5711632705700284, | |
| "grad_norm": 0.4505348801612854, | |
| "learning_rate": 4.2522183312727496e-05, | |
| "loss": 2.1951, | |
| "step": 1384 | |
| }, | |
| { | |
| "epoch": 0.571988651018829, | |
| "grad_norm": 0.4032357931137085, | |
| "learning_rate": 4.238727389724927e-05, | |
| "loss": 2.1917, | |
| "step": 1386 | |
| }, | |
| { | |
| "epoch": 0.5728140314676297, | |
| "grad_norm": 0.4129430651664734, | |
| "learning_rate": 4.225242119560839e-05, | |
| "loss": 2.2047, | |
| "step": 1388 | |
| }, | |
| { | |
| "epoch": 0.5736394119164302, | |
| "grad_norm": 0.41739246249198914, | |
| "learning_rate": 4.211762621244022e-05, | |
| "loss": 2.1966, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.5744647923652308, | |
| "grad_norm": 0.4234218895435333, | |
| "learning_rate": 4.198288995195021e-05, | |
| "loss": 2.1865, | |
| "step": 1392 | |
| }, | |
| { | |
| "epoch": 0.5752901728140315, | |
| "grad_norm": 0.4507383704185486, | |
| "learning_rate": 4.184821341790629e-05, | |
| "loss": 2.1878, | |
| "step": 1394 | |
| }, | |
| { | |
| "epoch": 0.576115553262832, | |
| "grad_norm": 0.4859677851200104, | |
| "learning_rate": 4.17135976136314e-05, | |
| "loss": 2.1886, | |
| "step": 1396 | |
| }, | |
| { | |
| "epoch": 0.5769409337116327, | |
| "grad_norm": 0.48982110619544983, | |
| "learning_rate": 4.1579043541996136e-05, | |
| "loss": 2.1965, | |
| "step": 1398 | |
| }, | |
| { | |
| "epoch": 0.5777663141604333, | |
| "grad_norm": 0.4520018696784973, | |
| "learning_rate": 4.144455220541112e-05, | |
| "loss": 2.1914, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.578591694609234, | |
| "grad_norm": 0.46317774057388306, | |
| "learning_rate": 4.131012460581967e-05, | |
| "loss": 2.1601, | |
| "step": 1402 | |
| }, | |
| { | |
| "epoch": 0.5794170750580345, | |
| "grad_norm": 0.4065730571746826, | |
| "learning_rate": 4.1175761744690225e-05, | |
| "loss": 2.1799, | |
| "step": 1404 | |
| }, | |
| { | |
| "epoch": 0.5802424555068352, | |
| "grad_norm": 0.42802825570106506, | |
| "learning_rate": 4.10414646230089e-05, | |
| "loss": 2.1827, | |
| "step": 1406 | |
| }, | |
| { | |
| "epoch": 0.5810678359556358, | |
| "grad_norm": 0.39025866985321045, | |
| "learning_rate": 4.090723424127216e-05, | |
| "loss": 2.1761, | |
| "step": 1408 | |
| }, | |
| { | |
| "epoch": 0.5818932164044365, | |
| "grad_norm": 0.410810649394989, | |
| "learning_rate": 4.077307159947915e-05, | |
| "loss": 2.1966, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.582718596853237, | |
| "grad_norm": 0.42078617215156555, | |
| "learning_rate": 4.063897769712448e-05, | |
| "loss": 2.1785, | |
| "step": 1412 | |
| }, | |
| { | |
| "epoch": 0.5835439773020377, | |
| "grad_norm": 0.3961106538772583, | |
| "learning_rate": 4.0504953533190575e-05, | |
| "loss": 2.199, | |
| "step": 1414 | |
| }, | |
| { | |
| "epoch": 0.5843693577508383, | |
| "grad_norm": 0.4196956753730774, | |
| "learning_rate": 4.037100010614031e-05, | |
| "loss": 2.1855, | |
| "step": 1416 | |
| }, | |
| { | |
| "epoch": 0.5851947381996389, | |
| "grad_norm": 0.4017656445503235, | |
| "learning_rate": 4.023711841390966e-05, | |
| "loss": 2.1688, | |
| "step": 1418 | |
| }, | |
| { | |
| "epoch": 0.5860201186484395, | |
| "grad_norm": 0.39641791582107544, | |
| "learning_rate": 4.010330945390012e-05, | |
| "loss": 2.1595, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.5868454990972402, | |
| "grad_norm": 0.4096054434776306, | |
| "learning_rate": 3.996957422297132e-05, | |
| "loss": 2.1538, | |
| "step": 1422 | |
| }, | |
| { | |
| "epoch": 0.5876708795460408, | |
| "grad_norm": 0.4470682740211487, | |
| "learning_rate": 3.98359137174337e-05, | |
| "loss": 2.1905, | |
| "step": 1424 | |
| }, | |
| { | |
| "epoch": 0.5884962599948413, | |
| "grad_norm": 0.45044365525245667, | |
| "learning_rate": 3.9702328933040914e-05, | |
| "loss": 2.1748, | |
| "step": 1426 | |
| }, | |
| { | |
| "epoch": 0.589321640443642, | |
| "grad_norm": 0.44126656651496887, | |
| "learning_rate": 3.9568820864982606e-05, | |
| "loss": 2.1731, | |
| "step": 1428 | |
| }, | |
| { | |
| "epoch": 0.5901470208924426, | |
| "grad_norm": 0.4331135153770447, | |
| "learning_rate": 3.943539050787678e-05, | |
| "loss": 2.1852, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.5909724013412432, | |
| "grad_norm": 0.44461002945899963, | |
| "learning_rate": 3.930203885576257e-05, | |
| "loss": 2.1626, | |
| "step": 1432 | |
| }, | |
| { | |
| "epoch": 0.5917977817900438, | |
| "grad_norm": 0.3985869288444519, | |
| "learning_rate": 3.9168766902092776e-05, | |
| "loss": 2.1839, | |
| "step": 1434 | |
| }, | |
| { | |
| "epoch": 0.5926231622388445, | |
| "grad_norm": 0.42710578441619873, | |
| "learning_rate": 3.903557563972641e-05, | |
| "loss": 2.1772, | |
| "step": 1436 | |
| }, | |
| { | |
| "epoch": 0.5934485426876451, | |
| "grad_norm": 0.4027608036994934, | |
| "learning_rate": 3.890246606092139e-05, | |
| "loss": 2.1583, | |
| "step": 1438 | |
| }, | |
| { | |
| "epoch": 0.5942739231364457, | |
| "grad_norm": 0.4465287923812866, | |
| "learning_rate": 3.876943915732706e-05, | |
| "loss": 2.1724, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.5950993035852463, | |
| "grad_norm": 0.4476611912250519, | |
| "learning_rate": 3.863649591997688e-05, | |
| "loss": 2.1655, | |
| "step": 1442 | |
| }, | |
| { | |
| "epoch": 0.595924684034047, | |
| "grad_norm": 0.4644310474395752, | |
| "learning_rate": 3.850363733928098e-05, | |
| "loss": 2.1907, | |
| "step": 1444 | |
| }, | |
| { | |
| "epoch": 0.5967500644828475, | |
| "grad_norm": 0.4769267141819, | |
| "learning_rate": 3.8370864405018816e-05, | |
| "loss": 2.156, | |
| "step": 1446 | |
| }, | |
| { | |
| "epoch": 0.5975754449316482, | |
| "grad_norm": 0.4150826632976532, | |
| "learning_rate": 3.823817810633181e-05, | |
| "loss": 2.1603, | |
| "step": 1448 | |
| }, | |
| { | |
| "epoch": 0.5984008253804488, | |
| "grad_norm": 0.4119608998298645, | |
| "learning_rate": 3.810557943171594e-05, | |
| "loss": 2.1631, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.5992262058292495, | |
| "grad_norm": 0.415431946516037, | |
| "learning_rate": 3.797306936901439e-05, | |
| "loss": 2.1926, | |
| "step": 1452 | |
| }, | |
| { | |
| "epoch": 0.60005158627805, | |
| "grad_norm": 0.38659125566482544, | |
| "learning_rate": 3.784064890541022e-05, | |
| "loss": 2.1918, | |
| "step": 1454 | |
| }, | |
| { | |
| "epoch": 0.6008769667268506, | |
| "grad_norm": 0.37595850229263306, | |
| "learning_rate": 3.770831902741895e-05, | |
| "loss": 2.1718, | |
| "step": 1456 | |
| }, | |
| { | |
| "epoch": 0.6017023471756513, | |
| "grad_norm": 0.39984071254730225, | |
| "learning_rate": 3.757608072088129e-05, | |
| "loss": 2.1446, | |
| "step": 1458 | |
| }, | |
| { | |
| "epoch": 0.602527727624452, | |
| "grad_norm": 0.4029097557067871, | |
| "learning_rate": 3.744393497095573e-05, | |
| "loss": 2.1848, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.6033531080732525, | |
| "grad_norm": 0.42148053646087646, | |
| "learning_rate": 3.7311882762111235e-05, | |
| "loss": 2.1741, | |
| "step": 1462 | |
| }, | |
| { | |
| "epoch": 0.6041784885220531, | |
| "grad_norm": 0.45567959547042847, | |
| "learning_rate": 3.7179925078119905e-05, | |
| "loss": 2.1867, | |
| "step": 1464 | |
| }, | |
| { | |
| "epoch": 0.6050038689708538, | |
| "grad_norm": 0.4234828054904938, | |
| "learning_rate": 3.704806290204963e-05, | |
| "loss": 2.1619, | |
| "step": 1466 | |
| }, | |
| { | |
| "epoch": 0.6058292494196543, | |
| "grad_norm": 0.4149649143218994, | |
| "learning_rate": 3.691629721625677e-05, | |
| "loss": 2.1913, | |
| "step": 1468 | |
| }, | |
| { | |
| "epoch": 0.606654629868455, | |
| "grad_norm": 0.4161490797996521, | |
| "learning_rate": 3.678462900237888e-05, | |
| "loss": 2.1616, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.6074800103172556, | |
| "grad_norm": 0.394819974899292, | |
| "learning_rate": 3.6653059241327314e-05, | |
| "loss": 2.1712, | |
| "step": 1472 | |
| }, | |
| { | |
| "epoch": 0.6083053907660563, | |
| "grad_norm": 0.37029707431793213, | |
| "learning_rate": 3.6521588913280024e-05, | |
| "loss": 2.1795, | |
| "step": 1474 | |
| }, | |
| { | |
| "epoch": 0.6091307712148568, | |
| "grad_norm": 0.3810674846172333, | |
| "learning_rate": 3.6390218997674144e-05, | |
| "loss": 2.1665, | |
| "step": 1476 | |
| }, | |
| { | |
| "epoch": 0.6099561516636575, | |
| "grad_norm": 0.38873404264450073, | |
| "learning_rate": 3.625895047319875e-05, | |
| "loss": 2.1714, | |
| "step": 1478 | |
| }, | |
| { | |
| "epoch": 0.6107815321124581, | |
| "grad_norm": 0.37483614683151245, | |
| "learning_rate": 3.6127784317787625e-05, | |
| "loss": 2.1974, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.6116069125612587, | |
| "grad_norm": 0.3878437876701355, | |
| "learning_rate": 3.5996721508611854e-05, | |
| "loss": 2.1547, | |
| "step": 1482 | |
| }, | |
| { | |
| "epoch": 0.6124322930100593, | |
| "grad_norm": 0.4019594192504883, | |
| "learning_rate": 3.586576302207264e-05, | |
| "loss": 2.185, | |
| "step": 1484 | |
| }, | |
| { | |
| "epoch": 0.61325767345886, | |
| "grad_norm": 0.39813095331192017, | |
| "learning_rate": 3.573490983379399e-05, | |
| "loss": 2.1578, | |
| "step": 1486 | |
| }, | |
| { | |
| "epoch": 0.6140830539076606, | |
| "grad_norm": 0.3950514495372772, | |
| "learning_rate": 3.5604162918615413e-05, | |
| "loss": 2.2005, | |
| "step": 1488 | |
| }, | |
| { | |
| "epoch": 0.6149084343564611, | |
| "grad_norm": 0.4188961088657379, | |
| "learning_rate": 3.5473523250584765e-05, | |
| "loss": 2.1929, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.6157338148052618, | |
| "grad_norm": 0.3965320885181427, | |
| "learning_rate": 3.534299180295087e-05, | |
| "loss": 2.1793, | |
| "step": 1492 | |
| }, | |
| { | |
| "epoch": 0.6165591952540624, | |
| "grad_norm": 0.40151247382164, | |
| "learning_rate": 3.5212569548156366e-05, | |
| "loss": 2.1846, | |
| "step": 1494 | |
| }, | |
| { | |
| "epoch": 0.6173845757028631, | |
| "grad_norm": 0.38915884494781494, | |
| "learning_rate": 3.508225745783039e-05, | |
| "loss": 2.162, | |
| "step": 1496 | |
| }, | |
| { | |
| "epoch": 0.6182099561516636, | |
| "grad_norm": 0.39938920736312866, | |
| "learning_rate": 3.4952056502781336e-05, | |
| "loss": 2.193, | |
| "step": 1498 | |
| }, | |
| { | |
| "epoch": 0.6190353366004643, | |
| "grad_norm": 0.40345609188079834, | |
| "learning_rate": 3.4821967652989764e-05, | |
| "loss": 2.1492, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.6198607170492649, | |
| "grad_norm": 0.4229806363582611, | |
| "learning_rate": 3.469199187760094e-05, | |
| "loss": 2.1865, | |
| "step": 1502 | |
| }, | |
| { | |
| "epoch": 0.6206860974980655, | |
| "grad_norm": 0.39548221230506897, | |
| "learning_rate": 3.456213014491778e-05, | |
| "loss": 2.1927, | |
| "step": 1504 | |
| }, | |
| { | |
| "epoch": 0.6215114779468661, | |
| "grad_norm": 0.38838130235671997, | |
| "learning_rate": 3.4432383422393666e-05, | |
| "loss": 2.1613, | |
| "step": 1506 | |
| }, | |
| { | |
| "epoch": 0.6223368583956668, | |
| "grad_norm": 0.371660441160202, | |
| "learning_rate": 3.430275267662505e-05, | |
| "loss": 2.1561, | |
| "step": 1508 | |
| }, | |
| { | |
| "epoch": 0.6231622388444674, | |
| "grad_norm": 0.4182002544403076, | |
| "learning_rate": 3.41732388733445e-05, | |
| "loss": 2.1722, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.623987619293268, | |
| "grad_norm": 0.402644544839859, | |
| "learning_rate": 3.404384297741326e-05, | |
| "loss": 2.1692, | |
| "step": 1512 | |
| }, | |
| { | |
| "epoch": 0.6248129997420686, | |
| "grad_norm": 0.4064319133758545, | |
| "learning_rate": 3.3914565952814237e-05, | |
| "loss": 2.1838, | |
| "step": 1514 | |
| }, | |
| { | |
| "epoch": 0.6256383801908693, | |
| "grad_norm": 0.4280416667461395, | |
| "learning_rate": 3.378540876264479e-05, | |
| "loss": 2.2136, | |
| "step": 1516 | |
| }, | |
| { | |
| "epoch": 0.6264637606396698, | |
| "grad_norm": 0.4381312429904938, | |
| "learning_rate": 3.365637236910944e-05, | |
| "loss": 2.1611, | |
| "step": 1518 | |
| }, | |
| { | |
| "epoch": 0.6272891410884704, | |
| "grad_norm": 0.43255355954170227, | |
| "learning_rate": 3.35274577335129e-05, | |
| "loss": 2.1719, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.6281145215372711, | |
| "grad_norm": 0.42183390259742737, | |
| "learning_rate": 3.3398665816252705e-05, | |
| "loss": 2.1929, | |
| "step": 1522 | |
| }, | |
| { | |
| "epoch": 0.6289399019860717, | |
| "grad_norm": 0.39848631620407104, | |
| "learning_rate": 3.326999757681218e-05, | |
| "loss": 2.1474, | |
| "step": 1524 | |
| }, | |
| { | |
| "epoch": 0.6297652824348723, | |
| "grad_norm": 0.4457964599132538, | |
| "learning_rate": 3.3141453973753324e-05, | |
| "loss": 2.1594, | |
| "step": 1526 | |
| }, | |
| { | |
| "epoch": 0.6305906628836729, | |
| "grad_norm": 0.4077092409133911, | |
| "learning_rate": 3.301303596470951e-05, | |
| "loss": 2.1596, | |
| "step": 1528 | |
| }, | |
| { | |
| "epoch": 0.6314160433324736, | |
| "grad_norm": 0.3941916227340698, | |
| "learning_rate": 3.288474450637857e-05, | |
| "loss": 2.1848, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.6322414237812741, | |
| "grad_norm": 0.402087539434433, | |
| "learning_rate": 3.2756580554515434e-05, | |
| "loss": 2.1598, | |
| "step": 1532 | |
| }, | |
| { | |
| "epoch": 0.6330668042300748, | |
| "grad_norm": 0.4281771183013916, | |
| "learning_rate": 3.262854506392519e-05, | |
| "loss": 2.1842, | |
| "step": 1534 | |
| }, | |
| { | |
| "epoch": 0.6338921846788754, | |
| "grad_norm": 0.40748128294944763, | |
| "learning_rate": 3.25006389884559e-05, | |
| "loss": 2.1477, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 0.6347175651276761, | |
| "grad_norm": 0.4175991415977478, | |
| "learning_rate": 3.237286328099149e-05, | |
| "loss": 2.1535, | |
| "step": 1538 | |
| }, | |
| { | |
| "epoch": 0.6355429455764766, | |
| "grad_norm": 0.4628128111362457, | |
| "learning_rate": 3.2245218893444664e-05, | |
| "loss": 2.1645, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.6363683260252773, | |
| "grad_norm": 0.45409590005874634, | |
| "learning_rate": 3.2117706776749815e-05, | |
| "loss": 2.1625, | |
| "step": 1542 | |
| }, | |
| { | |
| "epoch": 0.6371937064740779, | |
| "grad_norm": 0.39506542682647705, | |
| "learning_rate": 3.19903278808559e-05, | |
| "loss": 2.1765, | |
| "step": 1544 | |
| }, | |
| { | |
| "epoch": 0.6380190869228786, | |
| "grad_norm": 0.3923264145851135, | |
| "learning_rate": 3.186308315471947e-05, | |
| "loss": 2.16, | |
| "step": 1546 | |
| }, | |
| { | |
| "epoch": 0.6388444673716791, | |
| "grad_norm": 0.35540321469306946, | |
| "learning_rate": 3.173597354629746e-05, | |
| "loss": 2.1815, | |
| "step": 1548 | |
| }, | |
| { | |
| "epoch": 0.6396698478204798, | |
| "grad_norm": 0.37577253580093384, | |
| "learning_rate": 3.1609000002540226e-05, | |
| "loss": 2.1604, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.6404952282692804, | |
| "grad_norm": 0.38430941104888916, | |
| "learning_rate": 3.148216346938446e-05, | |
| "loss": 2.1789, | |
| "step": 1552 | |
| }, | |
| { | |
| "epoch": 0.6413206087180809, | |
| "grad_norm": 0.3682323694229126, | |
| "learning_rate": 3.135546489174612e-05, | |
| "loss": 2.1718, | |
| "step": 1554 | |
| }, | |
| { | |
| "epoch": 0.6421459891668816, | |
| "grad_norm": 0.38881126046180725, | |
| "learning_rate": 3.122890521351345e-05, | |
| "loss": 2.1798, | |
| "step": 1556 | |
| }, | |
| { | |
| "epoch": 0.6429713696156822, | |
| "grad_norm": 0.41004738211631775, | |
| "learning_rate": 3.1102485377539886e-05, | |
| "loss": 2.1297, | |
| "step": 1558 | |
| }, | |
| { | |
| "epoch": 0.6437967500644829, | |
| "grad_norm": 0.39379170536994934, | |
| "learning_rate": 3.0976206325637046e-05, | |
| "loss": 2.1507, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.6446221305132834, | |
| "grad_norm": 0.40458956360816956, | |
| "learning_rate": 3.085006899856777e-05, | |
| "loss": 2.1678, | |
| "step": 1562 | |
| }, | |
| { | |
| "epoch": 0.6454475109620841, | |
| "grad_norm": 0.3773816227912903, | |
| "learning_rate": 3.072407433603901e-05, | |
| "loss": 2.1558, | |
| "step": 1564 | |
| }, | |
| { | |
| "epoch": 0.6462728914108847, | |
| "grad_norm": 0.3630047142505646, | |
| "learning_rate": 3.059822327669494e-05, | |
| "loss": 2.165, | |
| "step": 1566 | |
| }, | |
| { | |
| "epoch": 0.6470982718596853, | |
| "grad_norm": 0.3747231066226959, | |
| "learning_rate": 3.0472516758109847e-05, | |
| "loss": 2.1576, | |
| "step": 1568 | |
| }, | |
| { | |
| "epoch": 0.6479236523084859, | |
| "grad_norm": 0.41273626685142517, | |
| "learning_rate": 3.034695571678127e-05, | |
| "loss": 2.1395, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.6487490327572866, | |
| "grad_norm": 0.4430343508720398, | |
| "learning_rate": 3.0221541088122917e-05, | |
| "loss": 2.1734, | |
| "step": 1572 | |
| }, | |
| { | |
| "epoch": 0.6495744132060872, | |
| "grad_norm": 0.40412119030952454, | |
| "learning_rate": 3.0096273806457763e-05, | |
| "loss": 2.1532, | |
| "step": 1574 | |
| }, | |
| { | |
| "epoch": 0.6503997936548878, | |
| "grad_norm": 0.37069565057754517, | |
| "learning_rate": 2.997115480501106e-05, | |
| "loss": 2.1957, | |
| "step": 1576 | |
| }, | |
| { | |
| "epoch": 0.6512251741036884, | |
| "grad_norm": 0.38193821907043457, | |
| "learning_rate": 2.9846185015903394e-05, | |
| "loss": 2.1536, | |
| "step": 1578 | |
| }, | |
| { | |
| "epoch": 0.6520505545524891, | |
| "grad_norm": 0.3995210826396942, | |
| "learning_rate": 2.9721365370143722e-05, | |
| "loss": 2.181, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.6528759350012897, | |
| "grad_norm": 0.3900876045227051, | |
| "learning_rate": 2.959669679762248e-05, | |
| "loss": 2.1457, | |
| "step": 1582 | |
| }, | |
| { | |
| "epoch": 0.6537013154500902, | |
| "grad_norm": 0.3811575174331665, | |
| "learning_rate": 2.9472180227104628e-05, | |
| "loss": 2.1654, | |
| "step": 1584 | |
| }, | |
| { | |
| "epoch": 0.6545266958988909, | |
| "grad_norm": 0.420245885848999, | |
| "learning_rate": 2.9347816586222654e-05, | |
| "loss": 2.1728, | |
| "step": 1586 | |
| }, | |
| { | |
| "epoch": 0.6553520763476915, | |
| "grad_norm": 0.40868568420410156, | |
| "learning_rate": 2.9223606801469882e-05, | |
| "loss": 2.1794, | |
| "step": 1588 | |
| }, | |
| { | |
| "epoch": 0.6561774567964921, | |
| "grad_norm": 0.42787596583366394, | |
| "learning_rate": 2.909955179819328e-05, | |
| "loss": 2.1563, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.6570028372452927, | |
| "grad_norm": 0.4084904193878174, | |
| "learning_rate": 2.8975652500586826e-05, | |
| "loss": 2.1449, | |
| "step": 1592 | |
| }, | |
| { | |
| "epoch": 0.6578282176940934, | |
| "grad_norm": 0.4251170754432678, | |
| "learning_rate": 2.885190983168444e-05, | |
| "loss": 2.1528, | |
| "step": 1594 | |
| }, | |
| { | |
| "epoch": 0.658653598142894, | |
| "grad_norm": 0.3818851709365845, | |
| "learning_rate": 2.8728324713353194e-05, | |
| "loss": 2.1351, | |
| "step": 1596 | |
| }, | |
| { | |
| "epoch": 0.6594789785916946, | |
| "grad_norm": 0.4211563467979431, | |
| "learning_rate": 2.860489806628648e-05, | |
| "loss": 2.1689, | |
| "step": 1598 | |
| }, | |
| { | |
| "epoch": 0.6603043590404952, | |
| "grad_norm": 0.3683875799179077, | |
| "learning_rate": 2.8481630809997022e-05, | |
| "loss": 2.1598, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.6611297394892959, | |
| "grad_norm": 0.375102698802948, | |
| "learning_rate": 2.8358523862810155e-05, | |
| "loss": 2.1573, | |
| "step": 1602 | |
| }, | |
| { | |
| "epoch": 0.6619551199380964, | |
| "grad_norm": 0.3758949637413025, | |
| "learning_rate": 2.823557814185691e-05, | |
| "loss": 2.1531, | |
| "step": 1604 | |
| }, | |
| { | |
| "epoch": 0.6627805003868971, | |
| "grad_norm": 0.43883299827575684, | |
| "learning_rate": 2.811279456306718e-05, | |
| "loss": 2.1656, | |
| "step": 1606 | |
| }, | |
| { | |
| "epoch": 0.6636058808356977, | |
| "grad_norm": 0.45055341720581055, | |
| "learning_rate": 2.7990174041162997e-05, | |
| "loss": 2.1664, | |
| "step": 1608 | |
| }, | |
| { | |
| "epoch": 0.6644312612844984, | |
| "grad_norm": 0.3908529281616211, | |
| "learning_rate": 2.7867717489651567e-05, | |
| "loss": 2.1549, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.6652566417332989, | |
| "grad_norm": 0.3923724889755249, | |
| "learning_rate": 2.7745425820818567e-05, | |
| "loss": 2.1702, | |
| "step": 1612 | |
| }, | |
| { | |
| "epoch": 0.6660820221820996, | |
| "grad_norm": 0.39021268486976624, | |
| "learning_rate": 2.7623299945721302e-05, | |
| "loss": 2.1695, | |
| "step": 1614 | |
| }, | |
| { | |
| "epoch": 0.6669074026309002, | |
| "grad_norm": 0.3864549398422241, | |
| "learning_rate": 2.7501340774181934e-05, | |
| "loss": 2.1723, | |
| "step": 1616 | |
| }, | |
| { | |
| "epoch": 0.6677327830797009, | |
| "grad_norm": 0.39826998114585876, | |
| "learning_rate": 2.7379549214780754e-05, | |
| "loss": 2.1585, | |
| "step": 1618 | |
| }, | |
| { | |
| "epoch": 0.6685581635285014, | |
| "grad_norm": 0.41055572032928467, | |
| "learning_rate": 2.7257926174849328e-05, | |
| "loss": 2.1842, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.669383543977302, | |
| "grad_norm": 0.39198827743530273, | |
| "learning_rate": 2.713647256046371e-05, | |
| "loss": 2.1697, | |
| "step": 1622 | |
| }, | |
| { | |
| "epoch": 0.6702089244261027, | |
| "grad_norm": 0.3906678855419159, | |
| "learning_rate": 2.7015189276437884e-05, | |
| "loss": 2.1513, | |
| "step": 1624 | |
| }, | |
| { | |
| "epoch": 0.6710343048749032, | |
| "grad_norm": 0.39475318789482117, | |
| "learning_rate": 2.6894077226316783e-05, | |
| "loss": 2.1347, | |
| "step": 1626 | |
| }, | |
| { | |
| "epoch": 0.6718596853237039, | |
| "grad_norm": 0.3569572865962982, | |
| "learning_rate": 2.6773137312369756e-05, | |
| "loss": 2.1333, | |
| "step": 1628 | |
| }, | |
| { | |
| "epoch": 0.6726850657725045, | |
| "grad_norm": 0.390018105506897, | |
| "learning_rate": 2.6652370435583712e-05, | |
| "loss": 2.1843, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.6735104462213052, | |
| "grad_norm": 0.36904361844062805, | |
| "learning_rate": 2.6531777495656406e-05, | |
| "loss": 2.1879, | |
| "step": 1632 | |
| }, | |
| { | |
| "epoch": 0.6743358266701057, | |
| "grad_norm": 0.4072750210762024, | |
| "learning_rate": 2.641135939098991e-05, | |
| "loss": 2.1533, | |
| "step": 1634 | |
| }, | |
| { | |
| "epoch": 0.6751612071189064, | |
| "grad_norm": 0.40311920642852783, | |
| "learning_rate": 2.6291117018683676e-05, | |
| "loss": 2.1724, | |
| "step": 1636 | |
| }, | |
| { | |
| "epoch": 0.675986587567707, | |
| "grad_norm": 0.39063480496406555, | |
| "learning_rate": 2.6171051274528057e-05, | |
| "loss": 2.1462, | |
| "step": 1638 | |
| }, | |
| { | |
| "epoch": 0.6768119680165076, | |
| "grad_norm": 0.38110098242759705, | |
| "learning_rate": 2.6051163052997528e-05, | |
| "loss": 2.1527, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.6776373484653082, | |
| "grad_norm": 0.37648850679397583, | |
| "learning_rate": 2.5931453247243963e-05, | |
| "loss": 2.148, | |
| "step": 1642 | |
| }, | |
| { | |
| "epoch": 0.6784627289141089, | |
| "grad_norm": 0.3929273188114166, | |
| "learning_rate": 2.5811922749090188e-05, | |
| "loss": 2.1445, | |
| "step": 1644 | |
| }, | |
| { | |
| "epoch": 0.6792881093629095, | |
| "grad_norm": 0.3670939803123474, | |
| "learning_rate": 2.569257244902311e-05, | |
| "loss": 2.1263, | |
| "step": 1646 | |
| }, | |
| { | |
| "epoch": 0.68011348981171, | |
| "grad_norm": 0.36248132586479187, | |
| "learning_rate": 2.5573403236187287e-05, | |
| "loss": 2.1398, | |
| "step": 1648 | |
| }, | |
| { | |
| "epoch": 0.6809388702605107, | |
| "grad_norm": 0.37559229135513306, | |
| "learning_rate": 2.5454415998378073e-05, | |
| "loss": 2.1421, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.6817642507093113, | |
| "grad_norm": 0.3896510899066925, | |
| "learning_rate": 2.5335611622035198e-05, | |
| "loss": 2.1469, | |
| "step": 1652 | |
| }, | |
| { | |
| "epoch": 0.6825896311581119, | |
| "grad_norm": 0.3731015920639038, | |
| "learning_rate": 2.5216990992236135e-05, | |
| "loss": 2.1438, | |
| "step": 1654 | |
| }, | |
| { | |
| "epoch": 0.6834150116069125, | |
| "grad_norm": 0.40263113379478455, | |
| "learning_rate": 2.509855499268938e-05, | |
| "loss": 2.1549, | |
| "step": 1656 | |
| }, | |
| { | |
| "epoch": 0.6842403920557132, | |
| "grad_norm": 0.3805602192878723, | |
| "learning_rate": 2.498030450572808e-05, | |
| "loss": 2.1888, | |
| "step": 1658 | |
| }, | |
| { | |
| "epoch": 0.6850657725045138, | |
| "grad_norm": 0.3967129588127136, | |
| "learning_rate": 2.486224041230319e-05, | |
| "loss": 2.1246, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.6858911529533144, | |
| "grad_norm": 0.39894452691078186, | |
| "learning_rate": 2.474436359197714e-05, | |
| "loss": 2.1717, | |
| "step": 1662 | |
| }, | |
| { | |
| "epoch": 0.686716533402115, | |
| "grad_norm": 0.38463294506073, | |
| "learning_rate": 2.4626674922917207e-05, | |
| "loss": 2.1304, | |
| "step": 1664 | |
| }, | |
| { | |
| "epoch": 0.6875419138509157, | |
| "grad_norm": 0.38346001505851746, | |
| "learning_rate": 2.4509175281888957e-05, | |
| "loss": 2.1603, | |
| "step": 1666 | |
| }, | |
| { | |
| "epoch": 0.6883672942997163, | |
| "grad_norm": 0.3817712664604187, | |
| "learning_rate": 2.4391865544249687e-05, | |
| "loss": 2.1568, | |
| "step": 1668 | |
| }, | |
| { | |
| "epoch": 0.6891926747485169, | |
| "grad_norm": 0.41481316089630127, | |
| "learning_rate": 2.4274746583941975e-05, | |
| "loss": 2.1913, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.6900180551973175, | |
| "grad_norm": 0.3687182366847992, | |
| "learning_rate": 2.415781927348709e-05, | |
| "loss": 2.1775, | |
| "step": 1672 | |
| }, | |
| { | |
| "epoch": 0.6908434356461182, | |
| "grad_norm": 0.39612531661987305, | |
| "learning_rate": 2.4041084483978616e-05, | |
| "loss": 2.1445, | |
| "step": 1674 | |
| }, | |
| { | |
| "epoch": 0.6916688160949187, | |
| "grad_norm": 0.405380517244339, | |
| "learning_rate": 2.39245430850758e-05, | |
| "loss": 2.1408, | |
| "step": 1676 | |
| }, | |
| { | |
| "epoch": 0.6924941965437194, | |
| "grad_norm": 0.40155723690986633, | |
| "learning_rate": 2.380819594499718e-05, | |
| "loss": 2.1568, | |
| "step": 1678 | |
| }, | |
| { | |
| "epoch": 0.69331957699252, | |
| "grad_norm": 0.3637157082557678, | |
| "learning_rate": 2.369204393051409e-05, | |
| "loss": 2.1374, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.6941449574413207, | |
| "grad_norm": 0.3784123659133911, | |
| "learning_rate": 2.3576087906944195e-05, | |
| "loss": 2.1413, | |
| "step": 1682 | |
| }, | |
| { | |
| "epoch": 0.6949703378901212, | |
| "grad_norm": 0.3910234570503235, | |
| "learning_rate": 2.34603287381451e-05, | |
| "loss": 2.1441, | |
| "step": 1684 | |
| }, | |
| { | |
| "epoch": 0.6957957183389218, | |
| "grad_norm": 0.38652199506759644, | |
| "learning_rate": 2.33447672865078e-05, | |
| "loss": 2.1682, | |
| "step": 1686 | |
| }, | |
| { | |
| "epoch": 0.6966210987877225, | |
| "grad_norm": 0.3730972409248352, | |
| "learning_rate": 2.322940441295036e-05, | |
| "loss": 2.0956, | |
| "step": 1688 | |
| }, | |
| { | |
| "epoch": 0.697446479236523, | |
| "grad_norm": 0.38551416993141174, | |
| "learning_rate": 2.3114240976911466e-05, | |
| "loss": 2.1744, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.6982718596853237, | |
| "grad_norm": 0.38724467158317566, | |
| "learning_rate": 2.2999277836343973e-05, | |
| "loss": 2.1704, | |
| "step": 1692 | |
| }, | |
| { | |
| "epoch": 0.6990972401341243, | |
| "grad_norm": 0.4469600319862366, | |
| "learning_rate": 2.2884515847708648e-05, | |
| "loss": 2.1334, | |
| "step": 1694 | |
| }, | |
| { | |
| "epoch": 0.699922620582925, | |
| "grad_norm": 0.38431516289711, | |
| "learning_rate": 2.276995586596763e-05, | |
| "loss": 2.1727, | |
| "step": 1696 | |
| }, | |
| { | |
| "epoch": 0.7007480010317255, | |
| "grad_norm": 0.39912179112434387, | |
| "learning_rate": 2.265559874457815e-05, | |
| "loss": 2.1739, | |
| "step": 1698 | |
| }, | |
| { | |
| "epoch": 0.7015733814805262, | |
| "grad_norm": 0.3802652955055237, | |
| "learning_rate": 2.254144533548616e-05, | |
| "loss": 2.1575, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.7023987619293268, | |
| "grad_norm": 0.40614211559295654, | |
| "learning_rate": 2.2427496489119986e-05, | |
| "loss": 2.1458, | |
| "step": 1702 | |
| }, | |
| { | |
| "epoch": 0.7032241423781275, | |
| "grad_norm": 0.39186087250709534, | |
| "learning_rate": 2.2313753054383958e-05, | |
| "loss": 2.1891, | |
| "step": 1704 | |
| }, | |
| { | |
| "epoch": 0.704049522826928, | |
| "grad_norm": 0.4107969105243683, | |
| "learning_rate": 2.220021587865218e-05, | |
| "loss": 2.134, | |
| "step": 1706 | |
| }, | |
| { | |
| "epoch": 0.7048749032757287, | |
| "grad_norm": 0.4244343936443329, | |
| "learning_rate": 2.2086885807762093e-05, | |
| "loss": 2.1632, | |
| "step": 1708 | |
| }, | |
| { | |
| "epoch": 0.7057002837245293, | |
| "grad_norm": 0.3787813186645508, | |
| "learning_rate": 2.197376368600825e-05, | |
| "loss": 2.1424, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.7065256641733298, | |
| "grad_norm": 0.41439372301101685, | |
| "learning_rate": 2.1860850356136015e-05, | |
| "loss": 2.1364, | |
| "step": 1712 | |
| }, | |
| { | |
| "epoch": 0.7073510446221305, | |
| "grad_norm": 0.4174649715423584, | |
| "learning_rate": 2.1748146659335256e-05, | |
| "loss": 2.1588, | |
| "step": 1714 | |
| }, | |
| { | |
| "epoch": 0.7081764250709311, | |
| "grad_norm": 0.4233049750328064, | |
| "learning_rate": 2.163565343523416e-05, | |
| "loss": 2.1649, | |
| "step": 1716 | |
| }, | |
| { | |
| "epoch": 0.7090018055197318, | |
| "grad_norm": 0.4185529947280884, | |
| "learning_rate": 2.152337152189287e-05, | |
| "loss": 2.1661, | |
| "step": 1718 | |
| }, | |
| { | |
| "epoch": 0.7098271859685323, | |
| "grad_norm": 0.374424010515213, | |
| "learning_rate": 2.1411301755797293e-05, | |
| "loss": 2.1505, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.710652566417333, | |
| "grad_norm": 0.3833918869495392, | |
| "learning_rate": 2.1299444971852876e-05, | |
| "loss": 2.1284, | |
| "step": 1722 | |
| }, | |
| { | |
| "epoch": 0.7114779468661336, | |
| "grad_norm": 0.3803061544895172, | |
| "learning_rate": 2.118780200337836e-05, | |
| "loss": 2.1704, | |
| "step": 1724 | |
| }, | |
| { | |
| "epoch": 0.7123033273149342, | |
| "grad_norm": 0.36754024028778076, | |
| "learning_rate": 2.107637368209966e-05, | |
| "loss": 2.1817, | |
| "step": 1726 | |
| }, | |
| { | |
| "epoch": 0.7131287077637348, | |
| "grad_norm": 0.38716909289360046, | |
| "learning_rate": 2.096516083814346e-05, | |
| "loss": 2.1664, | |
| "step": 1728 | |
| }, | |
| { | |
| "epoch": 0.7139540882125355, | |
| "grad_norm": 0.373674601316452, | |
| "learning_rate": 2.085416430003131e-05, | |
| "loss": 2.141, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.7147794686613361, | |
| "grad_norm": 0.40231001377105713, | |
| "learning_rate": 2.074338489467322e-05, | |
| "loss": 2.1501, | |
| "step": 1732 | |
| }, | |
| { | |
| "epoch": 0.7156048491101367, | |
| "grad_norm": 0.3659866452217102, | |
| "learning_rate": 2.0632823447361593e-05, | |
| "loss": 2.15, | |
| "step": 1734 | |
| }, | |
| { | |
| "epoch": 0.7164302295589373, | |
| "grad_norm": 0.3718184232711792, | |
| "learning_rate": 2.0522480781765153e-05, | |
| "loss": 2.1515, | |
| "step": 1736 | |
| }, | |
| { | |
| "epoch": 0.717255610007738, | |
| "grad_norm": 0.3444000780582428, | |
| "learning_rate": 2.0412357719922593e-05, | |
| "loss": 2.1162, | |
| "step": 1738 | |
| }, | |
| { | |
| "epoch": 0.7180809904565386, | |
| "grad_norm": 0.37671560049057007, | |
| "learning_rate": 2.0302455082236716e-05, | |
| "loss": 2.1391, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.7189063709053392, | |
| "grad_norm": 0.3651193380355835, | |
| "learning_rate": 2.019277368746812e-05, | |
| "loss": 2.1568, | |
| "step": 1742 | |
| }, | |
| { | |
| "epoch": 0.7197317513541398, | |
| "grad_norm": 0.3553023934364319, | |
| "learning_rate": 2.008331435272917e-05, | |
| "loss": 2.1679, | |
| "step": 1744 | |
| }, | |
| { | |
| "epoch": 0.7205571318029405, | |
| "grad_norm": 0.36929070949554443, | |
| "learning_rate": 1.997407789347799e-05, | |
| "loss": 2.1376, | |
| "step": 1746 | |
| }, | |
| { | |
| "epoch": 0.721382512251741, | |
| "grad_norm": 0.355954110622406, | |
| "learning_rate": 1.9865065123512194e-05, | |
| "loss": 2.1242, | |
| "step": 1748 | |
| }, | |
| { | |
| "epoch": 0.7222078927005416, | |
| "grad_norm": 0.3636607527732849, | |
| "learning_rate": 1.9756276854963002e-05, | |
| "loss": 2.1559, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.7230332731493423, | |
| "grad_norm": 0.37725040316581726, | |
| "learning_rate": 1.9647713898289154e-05, | |
| "loss": 2.131, | |
| "step": 1752 | |
| }, | |
| { | |
| "epoch": 0.7238586535981429, | |
| "grad_norm": 0.3439856171607971, | |
| "learning_rate": 1.953937706227078e-05, | |
| "loss": 2.1499, | |
| "step": 1754 | |
| }, | |
| { | |
| "epoch": 0.7246840340469435, | |
| "grad_norm": 0.3753752112388611, | |
| "learning_rate": 1.943126715400353e-05, | |
| "loss": 2.1567, | |
| "step": 1756 | |
| }, | |
| { | |
| "epoch": 0.7255094144957441, | |
| "grad_norm": 0.3847252428531647, | |
| "learning_rate": 1.9323384978892357e-05, | |
| "loss": 2.149, | |
| "step": 1758 | |
| }, | |
| { | |
| "epoch": 0.7263347949445448, | |
| "grad_norm": 0.37161755561828613, | |
| "learning_rate": 1.921573134064569e-05, | |
| "loss": 2.1394, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.7271601753933453, | |
| "grad_norm": 0.3596508204936981, | |
| "learning_rate": 1.9108307041269418e-05, | |
| "loss": 2.1586, | |
| "step": 1762 | |
| }, | |
| { | |
| "epoch": 0.727985555842146, | |
| "grad_norm": 0.37054118514060974, | |
| "learning_rate": 1.9001112881060845e-05, | |
| "loss": 2.1511, | |
| "step": 1764 | |
| }, | |
| { | |
| "epoch": 0.7288109362909466, | |
| "grad_norm": 0.3763185739517212, | |
| "learning_rate": 1.8894149658602767e-05, | |
| "loss": 2.1515, | |
| "step": 1766 | |
| }, | |
| { | |
| "epoch": 0.7296363167397473, | |
| "grad_norm": 0.3773542642593384, | |
| "learning_rate": 1.878741817075754e-05, | |
| "loss": 2.1281, | |
| "step": 1768 | |
| }, | |
| { | |
| "epoch": 0.7304616971885478, | |
| "grad_norm": 0.35355421900749207, | |
| "learning_rate": 1.8680919212661097e-05, | |
| "loss": 2.1556, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.7312870776373485, | |
| "grad_norm": 0.36710578203201294, | |
| "learning_rate": 1.8574653577717116e-05, | |
| "loss": 2.1278, | |
| "step": 1772 | |
| }, | |
| { | |
| "epoch": 0.7321124580861491, | |
| "grad_norm": 0.36785537004470825, | |
| "learning_rate": 1.8468622057590978e-05, | |
| "loss": 2.1674, | |
| "step": 1774 | |
| }, | |
| { | |
| "epoch": 0.7329378385349496, | |
| "grad_norm": 0.3685971796512604, | |
| "learning_rate": 1.836282544220398e-05, | |
| "loss": 2.1231, | |
| "step": 1776 | |
| }, | |
| { | |
| "epoch": 0.7337632189837503, | |
| "grad_norm": 0.3786368668079376, | |
| "learning_rate": 1.825726451972739e-05, | |
| "loss": 2.1318, | |
| "step": 1778 | |
| }, | |
| { | |
| "epoch": 0.734588599432551, | |
| "grad_norm": 0.3615210950374603, | |
| "learning_rate": 1.815194007657659e-05, | |
| "loss": 2.1545, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.7354139798813516, | |
| "grad_norm": 0.3552665412425995, | |
| "learning_rate": 1.804685289740526e-05, | |
| "loss": 2.1302, | |
| "step": 1782 | |
| }, | |
| { | |
| "epoch": 0.7362393603301521, | |
| "grad_norm": 0.348206490278244, | |
| "learning_rate": 1.794200376509944e-05, | |
| "loss": 2.1812, | |
| "step": 1784 | |
| }, | |
| { | |
| "epoch": 0.7370647407789528, | |
| "grad_norm": 0.3670431971549988, | |
| "learning_rate": 1.7837393460771795e-05, | |
| "loss": 2.149, | |
| "step": 1786 | |
| }, | |
| { | |
| "epoch": 0.7378901212277534, | |
| "grad_norm": 0.33708223700523376, | |
| "learning_rate": 1.7733022763755725e-05, | |
| "loss": 2.1302, | |
| "step": 1788 | |
| }, | |
| { | |
| "epoch": 0.7387155016765541, | |
| "grad_norm": 0.3709441125392914, | |
| "learning_rate": 1.762889245159957e-05, | |
| "loss": 2.1327, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.7395408821253546, | |
| "grad_norm": 0.3586266338825226, | |
| "learning_rate": 1.7525003300060904e-05, | |
| "loss": 2.179, | |
| "step": 1792 | |
| }, | |
| { | |
| "epoch": 0.7403662625741553, | |
| "grad_norm": 0.3644587993621826, | |
| "learning_rate": 1.7421356083100615e-05, | |
| "loss": 2.1404, | |
| "step": 1794 | |
| }, | |
| { | |
| "epoch": 0.7411916430229559, | |
| "grad_norm": 0.35930705070495605, | |
| "learning_rate": 1.7317951572877237e-05, | |
| "loss": 2.1395, | |
| "step": 1796 | |
| }, | |
| { | |
| "epoch": 0.7420170234717565, | |
| "grad_norm": 0.34341469407081604, | |
| "learning_rate": 1.7214790539741167e-05, | |
| "loss": 2.1475, | |
| "step": 1798 | |
| }, | |
| { | |
| "epoch": 0.7428424039205571, | |
| "grad_norm": 0.3495519757270813, | |
| "learning_rate": 1.7111873752228907e-05, | |
| "loss": 2.182, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.7436677843693578, | |
| "grad_norm": 0.37678736448287964, | |
| "learning_rate": 1.700920197705742e-05, | |
| "loss": 2.1497, | |
| "step": 1802 | |
| }, | |
| { | |
| "epoch": 0.7444931648181584, | |
| "grad_norm": 0.34368839859962463, | |
| "learning_rate": 1.6906775979118307e-05, | |
| "loss": 2.1353, | |
| "step": 1804 | |
| }, | |
| { | |
| "epoch": 0.745318545266959, | |
| "grad_norm": 0.354757696390152, | |
| "learning_rate": 1.680459652147216e-05, | |
| "loss": 2.1045, | |
| "step": 1806 | |
| }, | |
| { | |
| "epoch": 0.7461439257157596, | |
| "grad_norm": 0.3509856164455414, | |
| "learning_rate": 1.67026643653429e-05, | |
| "loss": 2.1332, | |
| "step": 1808 | |
| }, | |
| { | |
| "epoch": 0.7469693061645603, | |
| "grad_norm": 0.3583715856075287, | |
| "learning_rate": 1.6600980270112055e-05, | |
| "loss": 2.1543, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.7477946866133608, | |
| "grad_norm": 0.3545357882976532, | |
| "learning_rate": 1.6499544993313183e-05, | |
| "loss": 2.1412, | |
| "step": 1812 | |
| }, | |
| { | |
| "epoch": 0.7486200670621614, | |
| "grad_norm": 0.35939571261405945, | |
| "learning_rate": 1.6398359290626135e-05, | |
| "loss": 2.1579, | |
| "step": 1814 | |
| }, | |
| { | |
| "epoch": 0.7494454475109621, | |
| "grad_norm": 0.3477668762207031, | |
| "learning_rate": 1.629742391587144e-05, | |
| "loss": 2.1298, | |
| "step": 1816 | |
| }, | |
| { | |
| "epoch": 0.7502708279597627, | |
| "grad_norm": 0.36954382061958313, | |
| "learning_rate": 1.619673962100479e-05, | |
| "loss": 2.1591, | |
| "step": 1818 | |
| }, | |
| { | |
| "epoch": 0.7510962084085633, | |
| "grad_norm": 0.37065422534942627, | |
| "learning_rate": 1.6096307156111312e-05, | |
| "loss": 2.136, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.7519215888573639, | |
| "grad_norm": 0.37768426537513733, | |
| "learning_rate": 1.5996127269400023e-05, | |
| "loss": 2.1142, | |
| "step": 1822 | |
| }, | |
| { | |
| "epoch": 0.7527469693061646, | |
| "grad_norm": 0.36868610978126526, | |
| "learning_rate": 1.589620070719834e-05, | |
| "loss": 2.1538, | |
| "step": 1824 | |
| }, | |
| { | |
| "epoch": 0.7535723497549652, | |
| "grad_norm": 0.3411722183227539, | |
| "learning_rate": 1.579652821394632e-05, | |
| "loss": 2.1304, | |
| "step": 1826 | |
| }, | |
| { | |
| "epoch": 0.7543977302037658, | |
| "grad_norm": 0.33972230553627014, | |
| "learning_rate": 1.5697110532191366e-05, | |
| "loss": 2.1262, | |
| "step": 1828 | |
| }, | |
| { | |
| "epoch": 0.7552231106525664, | |
| "grad_norm": 0.3463905453681946, | |
| "learning_rate": 1.559794840258249e-05, | |
| "loss": 2.1427, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.7560484911013671, | |
| "grad_norm": 0.40490660071372986, | |
| "learning_rate": 1.549904256386488e-05, | |
| "loss": 2.1464, | |
| "step": 1832 | |
| }, | |
| { | |
| "epoch": 0.7568738715501676, | |
| "grad_norm": 0.3470916450023651, | |
| "learning_rate": 1.5400393752874454e-05, | |
| "loss": 2.1483, | |
| "step": 1834 | |
| }, | |
| { | |
| "epoch": 0.7576992519989683, | |
| "grad_norm": 0.36317574977874756, | |
| "learning_rate": 1.5302002704532192e-05, | |
| "loss": 2.1393, | |
| "step": 1836 | |
| }, | |
| { | |
| "epoch": 0.7585246324477689, | |
| "grad_norm": 0.34546831250190735, | |
| "learning_rate": 1.5203870151838884e-05, | |
| "loss": 2.1396, | |
| "step": 1838 | |
| }, | |
| { | |
| "epoch": 0.7593500128965696, | |
| "grad_norm": 0.3429735600948334, | |
| "learning_rate": 1.51059968258695e-05, | |
| "loss": 2.168, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.7601753933453701, | |
| "grad_norm": 0.3513084053993225, | |
| "learning_rate": 1.5008383455767828e-05, | |
| "loss": 2.1301, | |
| "step": 1842 | |
| }, | |
| { | |
| "epoch": 0.7610007737941707, | |
| "grad_norm": 0.37402522563934326, | |
| "learning_rate": 1.4911030768741003e-05, | |
| "loss": 2.1552, | |
| "step": 1844 | |
| }, | |
| { | |
| "epoch": 0.7618261542429714, | |
| "grad_norm": 0.34295645356178284, | |
| "learning_rate": 1.4813939490054095e-05, | |
| "loss": 2.1551, | |
| "step": 1846 | |
| }, | |
| { | |
| "epoch": 0.7626515346917719, | |
| "grad_norm": 0.3719686269760132, | |
| "learning_rate": 1.471711034302477e-05, | |
| "loss": 2.1525, | |
| "step": 1848 | |
| }, | |
| { | |
| "epoch": 0.7634769151405726, | |
| "grad_norm": 0.3539029657840729, | |
| "learning_rate": 1.4620544049017787e-05, | |
| "loss": 2.1572, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.7643022955893732, | |
| "grad_norm": 0.35728850960731506, | |
| "learning_rate": 1.4524241327439708e-05, | |
| "loss": 2.1202, | |
| "step": 1852 | |
| }, | |
| { | |
| "epoch": 0.7651276760381739, | |
| "grad_norm": 0.36121803522109985, | |
| "learning_rate": 1.4428202895733505e-05, | |
| "loss": 2.1547, | |
| "step": 1854 | |
| }, | |
| { | |
| "epoch": 0.7659530564869744, | |
| "grad_norm": 0.32610949873924255, | |
| "learning_rate": 1.4332429469373226e-05, | |
| "loss": 2.1201, | |
| "step": 1856 | |
| }, | |
| { | |
| "epoch": 0.7667784369357751, | |
| "grad_norm": 0.34636059403419495, | |
| "learning_rate": 1.4236921761858685e-05, | |
| "loss": 2.1117, | |
| "step": 1858 | |
| }, | |
| { | |
| "epoch": 0.7676038173845757, | |
| "grad_norm": 0.34160706400871277, | |
| "learning_rate": 1.4141680484710095e-05, | |
| "loss": 2.1156, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.7684291978333764, | |
| "grad_norm": 0.343398779630661, | |
| "learning_rate": 1.4046706347462819e-05, | |
| "loss": 2.1488, | |
| "step": 1862 | |
| }, | |
| { | |
| "epoch": 0.7692545782821769, | |
| "grad_norm": 0.3486791253089905, | |
| "learning_rate": 1.395200005766204e-05, | |
| "loss": 2.1227, | |
| "step": 1864 | |
| }, | |
| { | |
| "epoch": 0.7700799587309776, | |
| "grad_norm": 0.3483271598815918, | |
| "learning_rate": 1.3857562320857526e-05, | |
| "loss": 2.1152, | |
| "step": 1866 | |
| }, | |
| { | |
| "epoch": 0.7709053391797782, | |
| "grad_norm": 0.341259241104126, | |
| "learning_rate": 1.3763393840598338e-05, | |
| "loss": 2.1577, | |
| "step": 1868 | |
| }, | |
| { | |
| "epoch": 0.7717307196285788, | |
| "grad_norm": 0.3476760983467102, | |
| "learning_rate": 1.3669495318427666e-05, | |
| "loss": 2.1201, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.7725561000773794, | |
| "grad_norm": 0.34470468759536743, | |
| "learning_rate": 1.3575867453877488e-05, | |
| "loss": 2.1348, | |
| "step": 1872 | |
| }, | |
| { | |
| "epoch": 0.77338148052618, | |
| "grad_norm": 0.3454931974411011, | |
| "learning_rate": 1.3482510944463445e-05, | |
| "loss": 2.1165, | |
| "step": 1874 | |
| }, | |
| { | |
| "epoch": 0.7742068609749807, | |
| "grad_norm": 0.3486715257167816, | |
| "learning_rate": 1.3389426485679607e-05, | |
| "loss": 2.1365, | |
| "step": 1876 | |
| }, | |
| { | |
| "epoch": 0.7750322414237812, | |
| "grad_norm": 0.35253047943115234, | |
| "learning_rate": 1.3296614770993293e-05, | |
| "loss": 2.1237, | |
| "step": 1878 | |
| }, | |
| { | |
| "epoch": 0.7758576218725819, | |
| "grad_norm": 0.33677321672439575, | |
| "learning_rate": 1.320407649183995e-05, | |
| "loss": 2.1594, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.7766830023213825, | |
| "grad_norm": 0.34186917543411255, | |
| "learning_rate": 1.3111812337617924e-05, | |
| "loss": 2.1212, | |
| "step": 1882 | |
| }, | |
| { | |
| "epoch": 0.7775083827701831, | |
| "grad_norm": 0.33094942569732666, | |
| "learning_rate": 1.3019822995683395e-05, | |
| "loss": 2.1489, | |
| "step": 1884 | |
| }, | |
| { | |
| "epoch": 0.7783337632189837, | |
| "grad_norm": 0.3544219434261322, | |
| "learning_rate": 1.2928109151345196e-05, | |
| "loss": 2.1515, | |
| "step": 1886 | |
| }, | |
| { | |
| "epoch": 0.7791591436677844, | |
| "grad_norm": 0.34201499819755554, | |
| "learning_rate": 1.2836671487859754e-05, | |
| "loss": 2.1533, | |
| "step": 1888 | |
| }, | |
| { | |
| "epoch": 0.779984524116585, | |
| "grad_norm": 0.34465253353118896, | |
| "learning_rate": 1.274551068642601e-05, | |
| "loss": 2.1618, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.7808099045653856, | |
| "grad_norm": 0.3615592420101166, | |
| "learning_rate": 1.2654627426180277e-05, | |
| "loss": 2.1255, | |
| "step": 1892 | |
| }, | |
| { | |
| "epoch": 0.7816352850141862, | |
| "grad_norm": 0.3284577429294586, | |
| "learning_rate": 1.2564022384191243e-05, | |
| "loss": 2.1225, | |
| "step": 1894 | |
| }, | |
| { | |
| "epoch": 0.7824606654629869, | |
| "grad_norm": 0.34497541189193726, | |
| "learning_rate": 1.2473696235454896e-05, | |
| "loss": 2.138, | |
| "step": 1896 | |
| }, | |
| { | |
| "epoch": 0.7832860459117874, | |
| "grad_norm": 0.32846176624298096, | |
| "learning_rate": 1.2383649652889501e-05, | |
| "loss": 2.1531, | |
| "step": 1898 | |
| }, | |
| { | |
| "epoch": 0.7841114263605881, | |
| "grad_norm": 0.32436785101890564, | |
| "learning_rate": 1.2293883307330622e-05, | |
| "loss": 2.1174, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.7849368068093887, | |
| "grad_norm": 0.358806848526001, | |
| "learning_rate": 1.2204397867526069e-05, | |
| "loss": 2.1205, | |
| "step": 1902 | |
| }, | |
| { | |
| "epoch": 0.7857621872581894, | |
| "grad_norm": 0.3195858597755432, | |
| "learning_rate": 1.2115194000130903e-05, | |
| "loss": 2.1283, | |
| "step": 1904 | |
| }, | |
| { | |
| "epoch": 0.7865875677069899, | |
| "grad_norm": 0.33586594462394714, | |
| "learning_rate": 1.202627236970259e-05, | |
| "loss": 2.1421, | |
| "step": 1906 | |
| }, | |
| { | |
| "epoch": 0.7874129481557905, | |
| "grad_norm": 0.3359866440296173, | |
| "learning_rate": 1.1937633638695883e-05, | |
| "loss": 2.1415, | |
| "step": 1908 | |
| }, | |
| { | |
| "epoch": 0.7882383286045912, | |
| "grad_norm": 0.33562368154525757, | |
| "learning_rate": 1.1849278467458048e-05, | |
| "loss": 2.1375, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.7890637090533918, | |
| "grad_norm": 0.32866016030311584, | |
| "learning_rate": 1.1761207514223822e-05, | |
| "loss": 2.1432, | |
| "step": 1912 | |
| }, | |
| { | |
| "epoch": 0.7898890895021924, | |
| "grad_norm": 0.3484470248222351, | |
| "learning_rate": 1.1673421435110522e-05, | |
| "loss": 2.1184, | |
| "step": 1914 | |
| }, | |
| { | |
| "epoch": 0.790714469950993, | |
| "grad_norm": 0.31653323769569397, | |
| "learning_rate": 1.1585920884113261e-05, | |
| "loss": 2.1209, | |
| "step": 1916 | |
| }, | |
| { | |
| "epoch": 0.7915398503997937, | |
| "grad_norm": 0.32875025272369385, | |
| "learning_rate": 1.1498706513099949e-05, | |
| "loss": 2.1288, | |
| "step": 1918 | |
| }, | |
| { | |
| "epoch": 0.7923652308485942, | |
| "grad_norm": 0.341203510761261, | |
| "learning_rate": 1.1411778971806558e-05, | |
| "loss": 2.1301, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.7931906112973949, | |
| "grad_norm": 0.32854121923446655, | |
| "learning_rate": 1.1325138907832122e-05, | |
| "loss": 2.1586, | |
| "step": 1922 | |
| }, | |
| { | |
| "epoch": 0.7940159917461955, | |
| "grad_norm": 0.33962199091911316, | |
| "learning_rate": 1.1238786966634052e-05, | |
| "loss": 2.1464, | |
| "step": 1924 | |
| }, | |
| { | |
| "epoch": 0.7948413721949962, | |
| "grad_norm": 0.3306322991847992, | |
| "learning_rate": 1.1152723791523318e-05, | |
| "loss": 2.155, | |
| "step": 1926 | |
| }, | |
| { | |
| "epoch": 0.7956667526437967, | |
| "grad_norm": 0.3397471606731415, | |
| "learning_rate": 1.1066950023659545e-05, | |
| "loss": 2.1381, | |
| "step": 1928 | |
| }, | |
| { | |
| "epoch": 0.7964921330925974, | |
| "grad_norm": 0.3407101035118103, | |
| "learning_rate": 1.0981466302046406e-05, | |
| "loss": 2.1408, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.797317513541398, | |
| "grad_norm": 0.34207984805107117, | |
| "learning_rate": 1.0896273263526663e-05, | |
| "loss": 2.1525, | |
| "step": 1932 | |
| }, | |
| { | |
| "epoch": 0.7981428939901986, | |
| "grad_norm": 0.34368953108787537, | |
| "learning_rate": 1.0811371542777571e-05, | |
| "loss": 2.1234, | |
| "step": 1934 | |
| }, | |
| { | |
| "epoch": 0.7989682744389992, | |
| "grad_norm": 0.3221174478530884, | |
| "learning_rate": 1.0726761772306137e-05, | |
| "loss": 2.1171, | |
| "step": 1936 | |
| }, | |
| { | |
| "epoch": 0.7997936548877999, | |
| "grad_norm": 0.3296349048614502, | |
| "learning_rate": 1.0642444582444322e-05, | |
| "loss": 2.1252, | |
| "step": 1938 | |
| }, | |
| { | |
| "epoch": 0.8006190353366005, | |
| "grad_norm": 0.3344162404537201, | |
| "learning_rate": 1.055842060134446e-05, | |
| "loss": 2.1235, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.801444415785401, | |
| "grad_norm": 0.33650079369544983, | |
| "learning_rate": 1.0474690454974445e-05, | |
| "loss": 2.1216, | |
| "step": 1942 | |
| }, | |
| { | |
| "epoch": 0.8022697962342017, | |
| "grad_norm": 0.34013551473617554, | |
| "learning_rate": 1.0391254767113169e-05, | |
| "loss": 2.1177, | |
| "step": 1944 | |
| }, | |
| { | |
| "epoch": 0.8030951766830023, | |
| "grad_norm": 0.3282875716686249, | |
| "learning_rate": 1.0308114159345883e-05, | |
| "loss": 2.1248, | |
| "step": 1946 | |
| }, | |
| { | |
| "epoch": 0.803920557131803, | |
| "grad_norm": 0.34421661496162415, | |
| "learning_rate": 1.0225269251059483e-05, | |
| "loss": 2.138, | |
| "step": 1948 | |
| }, | |
| { | |
| "epoch": 0.8047459375806035, | |
| "grad_norm": 0.339918851852417, | |
| "learning_rate": 1.0142720659437955e-05, | |
| "loss": 2.1582, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.8055713180294042, | |
| "grad_norm": 0.3220682144165039, | |
| "learning_rate": 1.0060468999457767e-05, | |
| "loss": 2.124, | |
| "step": 1952 | |
| }, | |
| { | |
| "epoch": 0.8063966984782048, | |
| "grad_norm": 0.3417309522628784, | |
| "learning_rate": 9.978514883883266e-06, | |
| "loss": 2.1242, | |
| "step": 1954 | |
| }, | |
| { | |
| "epoch": 0.8072220789270054, | |
| "grad_norm": 0.3351197838783264, | |
| "learning_rate": 9.89685892326218e-06, | |
| "loss": 2.1391, | |
| "step": 1956 | |
| }, | |
| { | |
| "epoch": 0.808047459375806, | |
| "grad_norm": 0.3231659233570099, | |
| "learning_rate": 9.815501725920972e-06, | |
| "loss": 2.1391, | |
| "step": 1958 | |
| }, | |
| { | |
| "epoch": 0.8088728398246067, | |
| "grad_norm": 0.32000261545181274, | |
| "learning_rate": 9.734443897960372e-06, | |
| "loss": 2.1292, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.8096982202734073, | |
| "grad_norm": 0.34334588050842285, | |
| "learning_rate": 9.653686043250848e-06, | |
| "loss": 2.128, | |
| "step": 1962 | |
| }, | |
| { | |
| "epoch": 0.8105236007222079, | |
| "grad_norm": 0.3440937101840973, | |
| "learning_rate": 9.573228763428093e-06, | |
| "loss": 2.1214, | |
| "step": 1964 | |
| }, | |
| { | |
| "epoch": 0.8113489811710085, | |
| "grad_norm": 0.33723366260528564, | |
| "learning_rate": 9.493072657888597e-06, | |
| "loss": 2.143, | |
| "step": 1966 | |
| }, | |
| { | |
| "epoch": 0.8121743616198092, | |
| "grad_norm": 0.34954842925071716, | |
| "learning_rate": 9.413218323785084e-06, | |
| "loss": 2.1214, | |
| "step": 1968 | |
| }, | |
| { | |
| "epoch": 0.8129997420686097, | |
| "grad_norm": 0.3392115533351898, | |
| "learning_rate": 9.333666356022158e-06, | |
| "loss": 2.1349, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.8138251225174103, | |
| "grad_norm": 0.32914984226226807, | |
| "learning_rate": 9.254417347251815e-06, | |
| "loss": 2.1342, | |
| "step": 1972 | |
| }, | |
| { | |
| "epoch": 0.814650502966211, | |
| "grad_norm": 0.3356204628944397, | |
| "learning_rate": 9.175471887869042e-06, | |
| "loss": 2.1477, | |
| "step": 1974 | |
| }, | |
| { | |
| "epoch": 0.8154758834150116, | |
| "grad_norm": 0.3248778283596039, | |
| "learning_rate": 9.096830566007452e-06, | |
| "loss": 2.1274, | |
| "step": 1976 | |
| }, | |
| { | |
| "epoch": 0.8163012638638122, | |
| "grad_norm": 0.3299380838871002, | |
| "learning_rate": 9.018493967534835e-06, | |
| "loss": 2.1791, | |
| "step": 1978 | |
| }, | |
| { | |
| "epoch": 0.8171266443126128, | |
| "grad_norm": 0.3330373466014862, | |
| "learning_rate": 8.940462676048855e-06, | |
| "loss": 2.1213, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.8179520247614135, | |
| "grad_norm": 0.35274261236190796, | |
| "learning_rate": 8.862737272872657e-06, | |
| "loss": 2.1165, | |
| "step": 1982 | |
| }, | |
| { | |
| "epoch": 0.8187774052102141, | |
| "grad_norm": 0.32712242007255554, | |
| "learning_rate": 8.78531833705058e-06, | |
| "loss": 2.1038, | |
| "step": 1984 | |
| }, | |
| { | |
| "epoch": 0.8196027856590147, | |
| "grad_norm": 0.332295298576355, | |
| "learning_rate": 8.708206445343791e-06, | |
| "loss": 2.1441, | |
| "step": 1986 | |
| }, | |
| { | |
| "epoch": 0.8204281661078153, | |
| "grad_norm": 0.33298367261886597, | |
| "learning_rate": 8.631402172226061e-06, | |
| "loss": 2.1153, | |
| "step": 1988 | |
| }, | |
| { | |
| "epoch": 0.821253546556616, | |
| "grad_norm": 0.33189794421195984, | |
| "learning_rate": 8.554906089879411e-06, | |
| "loss": 2.1432, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.8220789270054165, | |
| "grad_norm": 0.3431938886642456, | |
| "learning_rate": 8.478718768189875e-06, | |
| "loss": 2.1066, | |
| "step": 1992 | |
| }, | |
| { | |
| "epoch": 0.8229043074542172, | |
| "grad_norm": 0.33813905715942383, | |
| "learning_rate": 8.402840774743281e-06, | |
| "loss": 2.1481, | |
| "step": 1994 | |
| }, | |
| { | |
| "epoch": 0.8237296879030178, | |
| "grad_norm": 0.3725796341896057, | |
| "learning_rate": 8.327272674820974e-06, | |
| "loss": 2.1325, | |
| "step": 1996 | |
| }, | |
| { | |
| "epoch": 0.8245550683518185, | |
| "grad_norm": 0.3199642300605774, | |
| "learning_rate": 8.252015031395672e-06, | |
| "loss": 2.1408, | |
| "step": 1998 | |
| }, | |
| { | |
| "epoch": 0.825380448800619, | |
| "grad_norm": 0.32736214995384216, | |
| "learning_rate": 8.177068405127198e-06, | |
| "loss": 2.1613, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.8262058292494197, | |
| "grad_norm": 0.3452676236629486, | |
| "learning_rate": 8.10243335435834e-06, | |
| "loss": 2.1339, | |
| "step": 2002 | |
| }, | |
| { | |
| "epoch": 0.8270312096982203, | |
| "grad_norm": 0.3281574547290802, | |
| "learning_rate": 8.028110435110709e-06, | |
| "loss": 2.1142, | |
| "step": 2004 | |
| }, | |
| { | |
| "epoch": 0.8278565901470208, | |
| "grad_norm": 0.32193323969841003, | |
| "learning_rate": 7.954100201080538e-06, | |
| "loss": 2.132, | |
| "step": 2006 | |
| }, | |
| { | |
| "epoch": 0.8286819705958215, | |
| "grad_norm": 0.3320666253566742, | |
| "learning_rate": 7.880403203634657e-06, | |
| "loss": 2.1461, | |
| "step": 2008 | |
| }, | |
| { | |
| "epoch": 0.8295073510446221, | |
| "grad_norm": 0.3370856046676636, | |
| "learning_rate": 7.807019991806247e-06, | |
| "loss": 2.1272, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.8303327314934228, | |
| "grad_norm": 0.320910781621933, | |
| "learning_rate": 7.733951112290894e-06, | |
| "loss": 2.0938, | |
| "step": 2012 | |
| }, | |
| { | |
| "epoch": 0.8311581119422233, | |
| "grad_norm": 0.3256378471851349, | |
| "learning_rate": 7.661197109442409e-06, | |
| "loss": 2.1145, | |
| "step": 2014 | |
| }, | |
| { | |
| "epoch": 0.831983492391024, | |
| "grad_norm": 0.3386472761631012, | |
| "learning_rate": 7.588758525268808e-06, | |
| "loss": 2.1497, | |
| "step": 2016 | |
| }, | |
| { | |
| "epoch": 0.8328088728398246, | |
| "grad_norm": 0.3254534900188446, | |
| "learning_rate": 7.516635899428331e-06, | |
| "loss": 2.1381, | |
| "step": 2018 | |
| }, | |
| { | |
| "epoch": 0.8336342532886252, | |
| "grad_norm": 0.33104851841926575, | |
| "learning_rate": 7.444829769225286e-06, | |
| "loss": 2.1403, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.8344596337374258, | |
| "grad_norm": 0.3282712399959564, | |
| "learning_rate": 7.373340669606205e-06, | |
| "loss": 2.1488, | |
| "step": 2022 | |
| }, | |
| { | |
| "epoch": 0.8352850141862265, | |
| "grad_norm": 0.3317227363586426, | |
| "learning_rate": 7.30216913315574e-06, | |
| "loss": 2.0801, | |
| "step": 2024 | |
| }, | |
| { | |
| "epoch": 0.8361103946350271, | |
| "grad_norm": 0.3185071051120758, | |
| "learning_rate": 7.231315690092733e-06, | |
| "loss": 2.1338, | |
| "step": 2026 | |
| }, | |
| { | |
| "epoch": 0.8369357750838277, | |
| "grad_norm": 0.3287782073020935, | |
| "learning_rate": 7.1607808682663315e-06, | |
| "loss": 2.1071, | |
| "step": 2028 | |
| }, | |
| { | |
| "epoch": 0.8377611555326283, | |
| "grad_norm": 0.33722466230392456, | |
| "learning_rate": 7.090565193151905e-06, | |
| "loss": 2.1764, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.838586535981429, | |
| "grad_norm": 0.32282426953315735, | |
| "learning_rate": 7.020669187847278e-06, | |
| "loss": 2.1369, | |
| "step": 2032 | |
| }, | |
| { | |
| "epoch": 0.8394119164302296, | |
| "grad_norm": 0.3378765881061554, | |
| "learning_rate": 6.951093373068779e-06, | |
| "loss": 2.144, | |
| "step": 2034 | |
| }, | |
| { | |
| "epoch": 0.8402372968790301, | |
| "grad_norm": 0.32491791248321533, | |
| "learning_rate": 6.881838267147334e-06, | |
| "loss": 2.1442, | |
| "step": 2036 | |
| }, | |
| { | |
| "epoch": 0.8410626773278308, | |
| "grad_norm": 0.34893786907196045, | |
| "learning_rate": 6.812904386024644e-06, | |
| "loss": 2.1357, | |
| "step": 2038 | |
| }, | |
| { | |
| "epoch": 0.8418880577766314, | |
| "grad_norm": 0.3235718011856079, | |
| "learning_rate": 6.744292243249306e-06, | |
| "loss": 2.099, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.842713438225432, | |
| "grad_norm": 0.31949764490127563, | |
| "learning_rate": 6.676002349973027e-06, | |
| "loss": 2.1416, | |
| "step": 2042 | |
| }, | |
| { | |
| "epoch": 0.8435388186742326, | |
| "grad_norm": 0.3352109491825104, | |
| "learning_rate": 6.608035214946806e-06, | |
| "loss": 2.1137, | |
| "step": 2044 | |
| }, | |
| { | |
| "epoch": 0.8443641991230333, | |
| "grad_norm": 0.32356804609298706, | |
| "learning_rate": 6.540391344517105e-06, | |
| "loss": 2.0998, | |
| "step": 2046 | |
| }, | |
| { | |
| "epoch": 0.8451895795718339, | |
| "grad_norm": 0.33448925614356995, | |
| "learning_rate": 6.473071242622131e-06, | |
| "loss": 2.129, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 0.8460149600206345, | |
| "grad_norm": 0.32092127203941345, | |
| "learning_rate": 6.406075410788037e-06, | |
| "loss": 2.1348, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.8468403404694351, | |
| "grad_norm": 0.32737207412719727, | |
| "learning_rate": 6.339404348125205e-06, | |
| "loss": 2.1259, | |
| "step": 2052 | |
| }, | |
| { | |
| "epoch": 0.8476657209182358, | |
| "grad_norm": 0.32666078209877014, | |
| "learning_rate": 6.273058551324568e-06, | |
| "loss": 2.1058, | |
| "step": 2054 | |
| }, | |
| { | |
| "epoch": 0.8484911013670363, | |
| "grad_norm": 0.323102205991745, | |
| "learning_rate": 6.207038514653818e-06, | |
| "loss": 2.1286, | |
| "step": 2056 | |
| }, | |
| { | |
| "epoch": 0.849316481815837, | |
| "grad_norm": 0.32299962639808655, | |
| "learning_rate": 6.141344729953802e-06, | |
| "loss": 2.1255, | |
| "step": 2058 | |
| }, | |
| { | |
| "epoch": 0.8501418622646376, | |
| "grad_norm": 0.3298660218715668, | |
| "learning_rate": 6.075977686634831e-06, | |
| "loss": 2.1577, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.8509672427134383, | |
| "grad_norm": 0.32427558302879333, | |
| "learning_rate": 6.010937871673017e-06, | |
| "loss": 2.1234, | |
| "step": 2062 | |
| }, | |
| { | |
| "epoch": 0.8517926231622388, | |
| "grad_norm": 0.33669888973236084, | |
| "learning_rate": 5.94622576960669e-06, | |
| "loss": 2.1322, | |
| "step": 2064 | |
| }, | |
| { | |
| "epoch": 0.8526180036110395, | |
| "grad_norm": 0.3118768036365509, | |
| "learning_rate": 5.88184186253275e-06, | |
| "loss": 2.1201, | |
| "step": 2066 | |
| }, | |
| { | |
| "epoch": 0.8534433840598401, | |
| "grad_norm": 0.3149246871471405, | |
| "learning_rate": 5.817786630103067e-06, | |
| "loss": 2.1169, | |
| "step": 2068 | |
| }, | |
| { | |
| "epoch": 0.8542687645086408, | |
| "grad_norm": 0.310986191034317, | |
| "learning_rate": 5.754060549520956e-06, | |
| "loss": 2.1262, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.8550941449574413, | |
| "grad_norm": 0.315181702375412, | |
| "learning_rate": 5.690664095537568e-06, | |
| "loss": 2.1019, | |
| "step": 2072 | |
| }, | |
| { | |
| "epoch": 0.8559195254062419, | |
| "grad_norm": 0.30971235036849976, | |
| "learning_rate": 5.627597740448398e-06, | |
| "loss": 2.1539, | |
| "step": 2074 | |
| }, | |
| { | |
| "epoch": 0.8567449058550426, | |
| "grad_norm": 0.33076462149620056, | |
| "learning_rate": 5.5648619540897395e-06, | |
| "loss": 2.1236, | |
| "step": 2076 | |
| }, | |
| { | |
| "epoch": 0.8575702863038431, | |
| "grad_norm": 0.3266281485557556, | |
| "learning_rate": 5.502457203835187e-06, | |
| "loss": 2.1454, | |
| "step": 2078 | |
| }, | |
| { | |
| "epoch": 0.8583956667526438, | |
| "grad_norm": 0.329086571931839, | |
| "learning_rate": 5.4403839545921595e-06, | |
| "loss": 2.1057, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.8592210472014444, | |
| "grad_norm": 0.315960168838501, | |
| "learning_rate": 5.378642668798428e-06, | |
| "loss": 2.135, | |
| "step": 2082 | |
| }, | |
| { | |
| "epoch": 0.8600464276502451, | |
| "grad_norm": 0.31883350014686584, | |
| "learning_rate": 5.317233806418708e-06, | |
| "loss": 2.1407, | |
| "step": 2084 | |
| }, | |
| { | |
| "epoch": 0.8608718080990456, | |
| "grad_norm": 0.3129589259624481, | |
| "learning_rate": 5.2561578249411824e-06, | |
| "loss": 2.107, | |
| "step": 2086 | |
| }, | |
| { | |
| "epoch": 0.8616971885478463, | |
| "grad_norm": 0.3180556297302246, | |
| "learning_rate": 5.19541517937408e-06, | |
| "loss": 2.1278, | |
| "step": 2088 | |
| }, | |
| { | |
| "epoch": 0.8625225689966469, | |
| "grad_norm": 0.30829039216041565, | |
| "learning_rate": 5.135006322242386e-06, | |
| "loss": 2.1136, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.8633479494454475, | |
| "grad_norm": 0.3156387209892273, | |
| "learning_rate": 5.074931703584352e-06, | |
| "loss": 2.1406, | |
| "step": 2092 | |
| }, | |
| { | |
| "epoch": 0.8641733298942481, | |
| "grad_norm": 0.31208303570747375, | |
| "learning_rate": 5.015191770948241e-06, | |
| "loss": 2.1258, | |
| "step": 2094 | |
| }, | |
| { | |
| "epoch": 0.8649987103430488, | |
| "grad_norm": 0.31232208013534546, | |
| "learning_rate": 4.955786969388909e-06, | |
| "loss": 2.1439, | |
| "step": 2096 | |
| }, | |
| { | |
| "epoch": 0.8658240907918494, | |
| "grad_norm": 0.3171435594558716, | |
| "learning_rate": 4.896717741464524e-06, | |
| "loss": 2.1412, | |
| "step": 2098 | |
| }, | |
| { | |
| "epoch": 0.86664947124065, | |
| "grad_norm": 0.30812162160873413, | |
| "learning_rate": 4.837984527233314e-06, | |
| "loss": 2.138, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.8674748516894506, | |
| "grad_norm": 0.307925820350647, | |
| "learning_rate": 4.77958776425022e-06, | |
| "loss": 2.1345, | |
| "step": 2102 | |
| }, | |
| { | |
| "epoch": 0.8683002321382512, | |
| "grad_norm": 0.309372216463089, | |
| "learning_rate": 4.721527887563659e-06, | |
| "loss": 2.1448, | |
| "step": 2104 | |
| }, | |
| { | |
| "epoch": 0.8691256125870519, | |
| "grad_norm": 0.31039103865623474, | |
| "learning_rate": 4.663805329712318e-06, | |
| "loss": 2.1158, | |
| "step": 2106 | |
| }, | |
| { | |
| "epoch": 0.8699509930358524, | |
| "grad_norm": 0.3017139136791229, | |
| "learning_rate": 4.6064205207218546e-06, | |
| "loss": 2.1628, | |
| "step": 2108 | |
| }, | |
| { | |
| "epoch": 0.8707763734846531, | |
| "grad_norm": 0.30297982692718506, | |
| "learning_rate": 4.549373888101793e-06, | |
| "loss": 2.1455, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.8716017539334537, | |
| "grad_norm": 0.31339317560195923, | |
| "learning_rate": 4.492665856842249e-06, | |
| "loss": 2.1317, | |
| "step": 2112 | |
| }, | |
| { | |
| "epoch": 0.8724271343822543, | |
| "grad_norm": 0.30679646134376526, | |
| "learning_rate": 4.4362968494108145e-06, | |
| "loss": 2.1319, | |
| "step": 2114 | |
| }, | |
| { | |
| "epoch": 0.8732525148310549, | |
| "grad_norm": 0.31017109751701355, | |
| "learning_rate": 4.3802672857494006e-06, | |
| "loss": 2.1495, | |
| "step": 2116 | |
| }, | |
| { | |
| "epoch": 0.8740778952798556, | |
| "grad_norm": 0.31378743052482605, | |
| "learning_rate": 4.324577583271089e-06, | |
| "loss": 2.1022, | |
| "step": 2118 | |
| }, | |
| { | |
| "epoch": 0.8749032757286562, | |
| "grad_norm": 0.30802688002586365, | |
| "learning_rate": 4.269228156857069e-06, | |
| "loss": 2.1201, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.8757286561774568, | |
| "grad_norm": 0.3153627812862396, | |
| "learning_rate": 4.2142194188534934e-06, | |
| "loss": 2.1629, | |
| "step": 2122 | |
| }, | |
| { | |
| "epoch": 0.8765540366262574, | |
| "grad_norm": 0.3161097466945648, | |
| "learning_rate": 4.159551779068438e-06, | |
| "loss": 2.1497, | |
| "step": 2124 | |
| }, | |
| { | |
| "epoch": 0.8773794170750581, | |
| "grad_norm": 0.310384064912796, | |
| "learning_rate": 4.1052256447688285e-06, | |
| "loss": 2.135, | |
| "step": 2126 | |
| }, | |
| { | |
| "epoch": 0.8782047975238586, | |
| "grad_norm": 0.332768976688385, | |
| "learning_rate": 4.051241420677427e-06, | |
| "loss": 2.1394, | |
| "step": 2128 | |
| }, | |
| { | |
| "epoch": 0.8790301779726593, | |
| "grad_norm": 0.3080803155899048, | |
| "learning_rate": 3.997599508969829e-06, | |
| "loss": 2.1377, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.8798555584214599, | |
| "grad_norm": 0.314807653427124, | |
| "learning_rate": 3.9443003092714095e-06, | |
| "loss": 2.1421, | |
| "step": 2132 | |
| }, | |
| { | |
| "epoch": 0.8806809388702606, | |
| "grad_norm": 0.31590813398361206, | |
| "learning_rate": 3.891344218654403e-06, | |
| "loss": 2.1268, | |
| "step": 2134 | |
| }, | |
| { | |
| "epoch": 0.8815063193190611, | |
| "grad_norm": 0.3047875463962555, | |
| "learning_rate": 3.8387316316349285e-06, | |
| "loss": 2.1248, | |
| "step": 2136 | |
| }, | |
| { | |
| "epoch": 0.8823316997678617, | |
| "grad_norm": 0.3278840482234955, | |
| "learning_rate": 3.7864629401700214e-06, | |
| "loss": 2.1316, | |
| "step": 2138 | |
| }, | |
| { | |
| "epoch": 0.8831570802166624, | |
| "grad_norm": 0.3061862587928772, | |
| "learning_rate": 3.7345385336547856e-06, | |
| "loss": 2.1468, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.8839824606654629, | |
| "grad_norm": 0.31418314576148987, | |
| "learning_rate": 3.682958798919406e-06, | |
| "loss": 2.1235, | |
| "step": 2142 | |
| }, | |
| { | |
| "epoch": 0.8848078411142636, | |
| "grad_norm": 0.31270700693130493, | |
| "learning_rate": 3.631724120226321e-06, | |
| "loss": 2.1345, | |
| "step": 2144 | |
| }, | |
| { | |
| "epoch": 0.8856332215630642, | |
| "grad_norm": 0.3076173961162567, | |
| "learning_rate": 3.5808348792673364e-06, | |
| "loss": 2.1069, | |
| "step": 2146 | |
| }, | |
| { | |
| "epoch": 0.8864586020118649, | |
| "grad_norm": 0.3199022114276886, | |
| "learning_rate": 3.5302914551608112e-06, | |
| "loss": 2.1474, | |
| "step": 2148 | |
| }, | |
| { | |
| "epoch": 0.8872839824606654, | |
| "grad_norm": 0.30249086022377014, | |
| "learning_rate": 3.480094224448788e-06, | |
| "loss": 2.1373, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.8881093629094661, | |
| "grad_norm": 0.3049810528755188, | |
| "learning_rate": 3.4302435610942372e-06, | |
| "loss": 2.1069, | |
| "step": 2152 | |
| }, | |
| { | |
| "epoch": 0.8889347433582667, | |
| "grad_norm": 0.310183048248291, | |
| "learning_rate": 3.3807398364782307e-06, | |
| "loss": 2.154, | |
| "step": 2154 | |
| }, | |
| { | |
| "epoch": 0.8897601238070674, | |
| "grad_norm": 0.30666887760162354, | |
| "learning_rate": 3.331583419397194e-06, | |
| "loss": 2.101, | |
| "step": 2156 | |
| }, | |
| { | |
| "epoch": 0.8905855042558679, | |
| "grad_norm": 0.30306148529052734, | |
| "learning_rate": 3.2827746760601573e-06, | |
| "loss": 2.1107, | |
| "step": 2158 | |
| }, | |
| { | |
| "epoch": 0.8914108847046686, | |
| "grad_norm": 0.30855709314346313, | |
| "learning_rate": 3.2343139700860168e-06, | |
| "loss": 2.1451, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.8922362651534692, | |
| "grad_norm": 0.31602942943573, | |
| "learning_rate": 3.186201662500865e-06, | |
| "loss": 2.1116, | |
| "step": 2162 | |
| }, | |
| { | |
| "epoch": 0.8930616456022697, | |
| "grad_norm": 0.3141464293003082, | |
| "learning_rate": 3.138438111735231e-06, | |
| "loss": 2.1339, | |
| "step": 2164 | |
| }, | |
| { | |
| "epoch": 0.8938870260510704, | |
| "grad_norm": 0.3070449233055115, | |
| "learning_rate": 3.0910236736214794e-06, | |
| "loss": 2.1485, | |
| "step": 2166 | |
| }, | |
| { | |
| "epoch": 0.894712406499871, | |
| "grad_norm": 0.29712599515914917, | |
| "learning_rate": 3.043958701391114e-06, | |
| "loss": 2.1151, | |
| "step": 2168 | |
| }, | |
| { | |
| "epoch": 0.8955377869486717, | |
| "grad_norm": 0.30952054262161255, | |
| "learning_rate": 2.9972435456721627e-06, | |
| "loss": 2.1224, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.8963631673974722, | |
| "grad_norm": 0.3056076169013977, | |
| "learning_rate": 2.9508785544865856e-06, | |
| "loss": 2.1457, | |
| "step": 2172 | |
| }, | |
| { | |
| "epoch": 0.8971885478462729, | |
| "grad_norm": 0.2979845702648163, | |
| "learning_rate": 2.9048640732476317e-06, | |
| "loss": 2.1149, | |
| "step": 2174 | |
| }, | |
| { | |
| "epoch": 0.8980139282950735, | |
| "grad_norm": 0.30862563848495483, | |
| "learning_rate": 2.8592004447573207e-06, | |
| "loss": 2.1382, | |
| "step": 2176 | |
| }, | |
| { | |
| "epoch": 0.8988393087438741, | |
| "grad_norm": 0.3080708682537079, | |
| "learning_rate": 2.813888009203841e-06, | |
| "loss": 2.1036, | |
| "step": 2178 | |
| }, | |
| { | |
| "epoch": 0.8996646891926747, | |
| "grad_norm": 0.3009863793849945, | |
| "learning_rate": 2.768927104159058e-06, | |
| "loss": 2.1386, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.9004900696414754, | |
| "grad_norm": 0.3038215637207031, | |
| "learning_rate": 2.724318064575976e-06, | |
| "loss": 2.1344, | |
| "step": 2182 | |
| }, | |
| { | |
| "epoch": 0.901315450090276, | |
| "grad_norm": 0.3082687556743622, | |
| "learning_rate": 2.6800612227862453e-06, | |
| "loss": 2.1287, | |
| "step": 2184 | |
| }, | |
| { | |
| "epoch": 0.9021408305390766, | |
| "grad_norm": 0.29829367995262146, | |
| "learning_rate": 2.6361569084976723e-06, | |
| "loss": 2.1251, | |
| "step": 2186 | |
| }, | |
| { | |
| "epoch": 0.9029662109878772, | |
| "grad_norm": 0.30370181798934937, | |
| "learning_rate": 2.592605448791807e-06, | |
| "loss": 2.1409, | |
| "step": 2188 | |
| }, | |
| { | |
| "epoch": 0.9037915914366779, | |
| "grad_norm": 0.30194738507270813, | |
| "learning_rate": 2.5494071681214483e-06, | |
| "loss": 2.0957, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.9046169718854785, | |
| "grad_norm": 0.3110347092151642, | |
| "learning_rate": 2.5065623883082867e-06, | |
| "loss": 2.1373, | |
| "step": 2192 | |
| }, | |
| { | |
| "epoch": 0.905442352334279, | |
| "grad_norm": 0.3016178607940674, | |
| "learning_rate": 2.4640714285404544e-06, | |
| "loss": 2.1553, | |
| "step": 2194 | |
| }, | |
| { | |
| "epoch": 0.9062677327830797, | |
| "grad_norm": 0.30545875430107117, | |
| "learning_rate": 2.421934605370163e-06, | |
| "loss": 2.1334, | |
| "step": 2196 | |
| }, | |
| { | |
| "epoch": 0.9070931132318804, | |
| "grad_norm": 0.29015982151031494, | |
| "learning_rate": 2.3801522327113802e-06, | |
| "loss": 2.1405, | |
| "step": 2198 | |
| }, | |
| { | |
| "epoch": 0.9079184936806809, | |
| "grad_norm": 0.29557955265045166, | |
| "learning_rate": 2.338724621837435e-06, | |
| "loss": 2.1282, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.9087438741294815, | |
| "grad_norm": 0.3093981444835663, | |
| "learning_rate": 2.2976520813787594e-06, | |
| "loss": 2.1162, | |
| "step": 2202 | |
| }, | |
| { | |
| "epoch": 0.9095692545782822, | |
| "grad_norm": 0.3128198981285095, | |
| "learning_rate": 2.2569349173205133e-06, | |
| "loss": 2.1056, | |
| "step": 2204 | |
| }, | |
| { | |
| "epoch": 0.9103946350270828, | |
| "grad_norm": 0.2990127205848694, | |
| "learning_rate": 2.2165734330003686e-06, | |
| "loss": 2.1289, | |
| "step": 2206 | |
| }, | |
| { | |
| "epoch": 0.9112200154758834, | |
| "grad_norm": 0.3216819167137146, | |
| "learning_rate": 2.176567929106249e-06, | |
| "loss": 2.1068, | |
| "step": 2208 | |
| }, | |
| { | |
| "epoch": 0.912045395924684, | |
| "grad_norm": 0.3107984960079193, | |
| "learning_rate": 2.1369187036740235e-06, | |
| "loss": 2.1332, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.9128707763734847, | |
| "grad_norm": 0.2950679063796997, | |
| "learning_rate": 2.0976260520853886e-06, | |
| "loss": 2.1573, | |
| "step": 2212 | |
| }, | |
| { | |
| "epoch": 0.9136961568222852, | |
| "grad_norm": 0.3093186914920807, | |
| "learning_rate": 2.0586902670655606e-06, | |
| "loss": 2.0812, | |
| "step": 2214 | |
| }, | |
| { | |
| "epoch": 0.9145215372710859, | |
| "grad_norm": 0.2955740690231323, | |
| "learning_rate": 2.02011163868116e-06, | |
| "loss": 2.1069, | |
| "step": 2216 | |
| }, | |
| { | |
| "epoch": 0.9153469177198865, | |
| "grad_norm": 0.3023936450481415, | |
| "learning_rate": 1.9818904543380643e-06, | |
| "loss": 2.1372, | |
| "step": 2218 | |
| }, | |
| { | |
| "epoch": 0.9161722981686872, | |
| "grad_norm": 0.3137916922569275, | |
| "learning_rate": 1.9440269987791914e-06, | |
| "loss": 2.1333, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.9169976786174877, | |
| "grad_norm": 0.3078385591506958, | |
| "learning_rate": 1.90652155408248e-06, | |
| "loss": 2.1218, | |
| "step": 2222 | |
| }, | |
| { | |
| "epoch": 0.9178230590662884, | |
| "grad_norm": 0.29822996258735657, | |
| "learning_rate": 1.8693743996586742e-06, | |
| "loss": 2.148, | |
| "step": 2224 | |
| }, | |
| { | |
| "epoch": 0.918648439515089, | |
| "grad_norm": 0.30666473507881165, | |
| "learning_rate": 1.8325858122493432e-06, | |
| "loss": 2.1107, | |
| "step": 2226 | |
| }, | |
| { | |
| "epoch": 0.9194738199638897, | |
| "grad_norm": 0.3042726516723633, | |
| "learning_rate": 1.7961560659247646e-06, | |
| "loss": 2.1354, | |
| "step": 2228 | |
| }, | |
| { | |
| "epoch": 0.9202992004126902, | |
| "grad_norm": 0.306774765253067, | |
| "learning_rate": 1.7600854320819038e-06, | |
| "loss": 2.1636, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.9211245808614908, | |
| "grad_norm": 0.29949676990509033, | |
| "learning_rate": 1.7243741794423619e-06, | |
| "loss": 2.1297, | |
| "step": 2232 | |
| }, | |
| { | |
| "epoch": 0.9219499613102915, | |
| "grad_norm": 0.3098452687263489, | |
| "learning_rate": 1.6890225740504251e-06, | |
| "loss": 2.1298, | |
| "step": 2234 | |
| }, | |
| { | |
| "epoch": 0.922775341759092, | |
| "grad_norm": 0.3078421354293823, | |
| "learning_rate": 1.6540308792710235e-06, | |
| "loss": 2.1078, | |
| "step": 2236 | |
| }, | |
| { | |
| "epoch": 0.9236007222078927, | |
| "grad_norm": 0.31042400002479553, | |
| "learning_rate": 1.6193993557878317e-06, | |
| "loss": 2.1276, | |
| "step": 2238 | |
| }, | |
| { | |
| "epoch": 0.9244261026566933, | |
| "grad_norm": 0.30934637784957886, | |
| "learning_rate": 1.5851282616012653e-06, | |
| "loss": 2.151, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.925251483105494, | |
| "grad_norm": 0.3095110356807709, | |
| "learning_rate": 1.5512178520266096e-06, | |
| "loss": 2.1373, | |
| "step": 2242 | |
| }, | |
| { | |
| "epoch": 0.9260768635542945, | |
| "grad_norm": 0.2958972454071045, | |
| "learning_rate": 1.5176683796920721e-06, | |
| "loss": 2.1429, | |
| "step": 2244 | |
| }, | |
| { | |
| "epoch": 0.9269022440030952, | |
| "grad_norm": 0.29824674129486084, | |
| "learning_rate": 1.4844800945369498e-06, | |
| "loss": 2.1453, | |
| "step": 2246 | |
| }, | |
| { | |
| "epoch": 0.9277276244518958, | |
| "grad_norm": 0.29917067289352417, | |
| "learning_rate": 1.4516532438097196e-06, | |
| "loss": 2.1218, | |
| "step": 2248 | |
| }, | |
| { | |
| "epoch": 0.9285530049006964, | |
| "grad_norm": 0.2987802028656006, | |
| "learning_rate": 1.419188072066241e-06, | |
| "loss": 2.1268, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.929378385349497, | |
| "grad_norm": 0.29865482449531555, | |
| "learning_rate": 1.3870848211678833e-06, | |
| "loss": 2.1087, | |
| "step": 2252 | |
| }, | |
| { | |
| "epoch": 0.9302037657982977, | |
| "grad_norm": 0.28756940364837646, | |
| "learning_rate": 1.3553437302797733e-06, | |
| "loss": 2.1003, | |
| "step": 2254 | |
| }, | |
| { | |
| "epoch": 0.9310291462470983, | |
| "grad_norm": 0.307478129863739, | |
| "learning_rate": 1.3239650358689737e-06, | |
| "loss": 2.1079, | |
| "step": 2256 | |
| }, | |
| { | |
| "epoch": 0.9318545266958989, | |
| "grad_norm": 0.298652708530426, | |
| "learning_rate": 1.292948971702773e-06, | |
| "loss": 2.1487, | |
| "step": 2258 | |
| }, | |
| { | |
| "epoch": 0.9326799071446995, | |
| "grad_norm": 0.2891993224620819, | |
| "learning_rate": 1.2622957688468717e-06, | |
| "loss": 2.1359, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.9335052875935002, | |
| "grad_norm": 0.30007168650627136, | |
| "learning_rate": 1.2320056556637205e-06, | |
| "loss": 2.1583, | |
| "step": 2262 | |
| }, | |
| { | |
| "epoch": 0.9343306680423007, | |
| "grad_norm": 0.3034486472606659, | |
| "learning_rate": 1.2020788578107956e-06, | |
| "loss": 2.1427, | |
| "step": 2264 | |
| }, | |
| { | |
| "epoch": 0.9351560484911013, | |
| "grad_norm": 0.30218076705932617, | |
| "learning_rate": 1.1725155982389158e-06, | |
| "loss": 2.1206, | |
| "step": 2266 | |
| }, | |
| { | |
| "epoch": 0.935981428939902, | |
| "grad_norm": 0.305792897939682, | |
| "learning_rate": 1.1433160971905942e-06, | |
| "loss": 2.1063, | |
| "step": 2268 | |
| }, | |
| { | |
| "epoch": 0.9368068093887026, | |
| "grad_norm": 0.2913326919078827, | |
| "learning_rate": 1.1144805721983841e-06, | |
| "loss": 2.1368, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.9376321898375032, | |
| "grad_norm": 0.2973470985889435, | |
| "learning_rate": 1.0860092380832632e-06, | |
| "loss": 2.106, | |
| "step": 2272 | |
| }, | |
| { | |
| "epoch": 0.9384575702863038, | |
| "grad_norm": 0.29396718740463257, | |
| "learning_rate": 1.0579023069530346e-06, | |
| "loss": 2.1007, | |
| "step": 2274 | |
| }, | |
| { | |
| "epoch": 0.9392829507351045, | |
| "grad_norm": 0.29689350724220276, | |
| "learning_rate": 1.0301599882007462e-06, | |
| "loss": 2.1312, | |
| "step": 2276 | |
| }, | |
| { | |
| "epoch": 0.9401083311839051, | |
| "grad_norm": 0.2933015823364258, | |
| "learning_rate": 1.0027824885031288e-06, | |
| "loss": 2.1144, | |
| "step": 2278 | |
| }, | |
| { | |
| "epoch": 0.9409337116327057, | |
| "grad_norm": 0.30149662494659424, | |
| "learning_rate": 9.75770011819066e-07, | |
| "loss": 2.1526, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.9417590920815063, | |
| "grad_norm": 0.3010120689868927, | |
| "learning_rate": 9.491227593880492e-07, | |
| "loss": 2.1455, | |
| "step": 2282 | |
| }, | |
| { | |
| "epoch": 0.942584472530307, | |
| "grad_norm": 0.3034345209598541, | |
| "learning_rate": 9.228409297287132e-07, | |
| "loss": 2.1181, | |
| "step": 2284 | |
| }, | |
| { | |
| "epoch": 0.9434098529791075, | |
| "grad_norm": 0.3068033754825592, | |
| "learning_rate": 8.96924718637332e-07, | |
| "loss": 2.1583, | |
| "step": 2286 | |
| }, | |
| { | |
| "epoch": 0.9442352334279082, | |
| "grad_norm": 0.29393401741981506, | |
| "learning_rate": 8.713743191863633e-07, | |
| "loss": 2.157, | |
| "step": 2288 | |
| }, | |
| { | |
| "epoch": 0.9450606138767088, | |
| "grad_norm": 0.2999117970466614, | |
| "learning_rate": 8.46189921723034e-07, | |
| "loss": 2.1656, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.9458859943255095, | |
| "grad_norm": 0.2952830195426941, | |
| "learning_rate": 8.21371713867869e-07, | |
| "loss": 2.1328, | |
| "step": 2292 | |
| }, | |
| { | |
| "epoch": 0.94671137477431, | |
| "grad_norm": 0.29716044664382935, | |
| "learning_rate": 7.969198805133638e-07, | |
| "loss": 2.1334, | |
| "step": 2294 | |
| }, | |
| { | |
| "epoch": 0.9475367552231106, | |
| "grad_norm": 0.298360675573349, | |
| "learning_rate": 7.728346038225475e-07, | |
| "loss": 2.127, | |
| "step": 2296 | |
| }, | |
| { | |
| "epoch": 0.9483621356719113, | |
| "grad_norm": 0.3036721348762512, | |
| "learning_rate": 7.491160632276562e-07, | |
| "loss": 2.1199, | |
| "step": 2298 | |
| }, | |
| { | |
| "epoch": 0.9491875161207118, | |
| "grad_norm": 0.2898666262626648, | |
| "learning_rate": 7.25764435428794e-07, | |
| "loss": 2.0981, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.9500128965695125, | |
| "grad_norm": 0.29367583990097046, | |
| "learning_rate": 7.027798943925967e-07, | |
| "loss": 2.1074, | |
| "step": 2302 | |
| }, | |
| { | |
| "epoch": 0.9508382770183131, | |
| "grad_norm": 0.2900161147117615, | |
| "learning_rate": 6.801626113509651e-07, | |
| "loss": 2.1371, | |
| "step": 2304 | |
| }, | |
| { | |
| "epoch": 0.9516636574671138, | |
| "grad_norm": 0.31588950753211975, | |
| "learning_rate": 6.579127547997721e-07, | |
| "loss": 2.1346, | |
| "step": 2306 | |
| }, | |
| { | |
| "epoch": 0.9524890379159143, | |
| "grad_norm": 0.30221107602119446, | |
| "learning_rate": 6.360304904976022e-07, | |
| "loss": 2.1732, | |
| "step": 2308 | |
| }, | |
| { | |
| "epoch": 0.953314418364715, | |
| "grad_norm": 0.2927485406398773, | |
| "learning_rate": 6.145159814645362e-07, | |
| "loss": 2.0866, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 0.9541397988135156, | |
| "grad_norm": 0.29365116357803345, | |
| "learning_rate": 5.933693879809132e-07, | |
| "loss": 2.1252, | |
| "step": 2312 | |
| }, | |
| { | |
| "epoch": 0.9549651792623163, | |
| "grad_norm": 0.29774293303489685, | |
| "learning_rate": 5.725908675861535e-07, | |
| "loss": 2.134, | |
| "step": 2314 | |
| }, | |
| { | |
| "epoch": 0.9557905597111168, | |
| "grad_norm": 0.30118051171302795, | |
| "learning_rate": 5.521805750775877e-07, | |
| "loss": 2.1385, | |
| "step": 2316 | |
| }, | |
| { | |
| "epoch": 0.9566159401599175, | |
| "grad_norm": 0.2982546389102936, | |
| "learning_rate": 5.32138662509285e-07, | |
| "loss": 2.1316, | |
| "step": 2318 | |
| }, | |
| { | |
| "epoch": 0.9574413206087181, | |
| "grad_norm": 0.29799342155456543, | |
| "learning_rate": 5.124652791909324e-07, | |
| "loss": 2.14, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.9582667010575187, | |
| "grad_norm": 0.29669389128685, | |
| "learning_rate": 4.931605716867293e-07, | |
| "loss": 2.1077, | |
| "step": 2322 | |
| }, | |
| { | |
| "epoch": 0.9590920815063193, | |
| "grad_norm": 0.29356813430786133, | |
| "learning_rate": 4.742246838142672e-07, | |
| "loss": 2.123, | |
| "step": 2324 | |
| }, | |
| { | |
| "epoch": 0.95991746195512, | |
| "grad_norm": 0.3008041977882385, | |
| "learning_rate": 4.5565775664351275e-07, | |
| "loss": 2.1124, | |
| "step": 2326 | |
| }, | |
| { | |
| "epoch": 0.9607428424039206, | |
| "grad_norm": 0.3006245195865631, | |
| "learning_rate": 4.3745992849568707e-07, | |
| "loss": 2.1207, | |
| "step": 2328 | |
| }, | |
| { | |
| "epoch": 0.9615682228527211, | |
| "grad_norm": 0.28674232959747314, | |
| "learning_rate": 4.196313349422942e-07, | |
| "loss": 2.1394, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 0.9623936033015218, | |
| "grad_norm": 0.301516056060791, | |
| "learning_rate": 4.021721088040775e-07, | |
| "loss": 2.1235, | |
| "step": 2332 | |
| }, | |
| { | |
| "epoch": 0.9632189837503224, | |
| "grad_norm": 0.30167555809020996, | |
| "learning_rate": 3.8508238015003697e-07, | |
| "loss": 2.1171, | |
| "step": 2334 | |
| }, | |
| { | |
| "epoch": 0.964044364199123, | |
| "grad_norm": 0.2917619049549103, | |
| "learning_rate": 3.6836227629648e-07, | |
| "loss": 2.1139, | |
| "step": 2336 | |
| }, | |
| { | |
| "epoch": 0.9648697446479236, | |
| "grad_norm": 0.28927844762802124, | |
| "learning_rate": 3.520119218060336e-07, | |
| "loss": 2.1377, | |
| "step": 2338 | |
| }, | |
| { | |
| "epoch": 0.9656951250967243, | |
| "grad_norm": 0.3015216290950775, | |
| "learning_rate": 3.360314384867558e-07, | |
| "loss": 2.1335, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.9665205055455249, | |
| "grad_norm": 0.2863229215145111, | |
| "learning_rate": 3.2042094539120883e-07, | |
| "loss": 2.1282, | |
| "step": 2342 | |
| }, | |
| { | |
| "epoch": 0.9673458859943255, | |
| "grad_norm": 0.28870850801467896, | |
| "learning_rate": 3.051805588155654e-07, | |
| "loss": 2.133, | |
| "step": 2344 | |
| }, | |
| { | |
| "epoch": 0.9681712664431261, | |
| "grad_norm": 0.29704856872558594, | |
| "learning_rate": 2.903103922987649e-07, | |
| "loss": 2.1267, | |
| "step": 2346 | |
| }, | |
| { | |
| "epoch": 0.9689966468919268, | |
| "grad_norm": 0.2941645681858063, | |
| "learning_rate": 2.7581055662164736e-07, | |
| "loss": 2.1464, | |
| "step": 2348 | |
| }, | |
| { | |
| "epoch": 0.9698220273407274, | |
| "grad_norm": 0.30156826972961426, | |
| "learning_rate": 2.6168115980614303e-07, | |
| "loss": 2.1208, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.970647407789528, | |
| "grad_norm": 0.2918868958950043, | |
| "learning_rate": 2.4792230711444544e-07, | |
| "loss": 2.1258, | |
| "step": 2352 | |
| }, | |
| { | |
| "epoch": 0.9714727882383286, | |
| "grad_norm": 0.2962958514690399, | |
| "learning_rate": 2.3453410104825046e-07, | |
| "loss": 2.1517, | |
| "step": 2354 | |
| }, | |
| { | |
| "epoch": 0.9722981686871293, | |
| "grad_norm": 0.3073914051055908, | |
| "learning_rate": 2.2151664134799076e-07, | |
| "loss": 2.0907, | |
| "step": 2356 | |
| }, | |
| { | |
| "epoch": 0.9731235491359298, | |
| "grad_norm": 0.292449414730072, | |
| "learning_rate": 2.0887002499207498e-07, | |
| "loss": 2.1287, | |
| "step": 2358 | |
| }, | |
| { | |
| "epoch": 0.9739489295847304, | |
| "grad_norm": 0.2905406355857849, | |
| "learning_rate": 1.9659434619617723e-07, | |
| "loss": 2.0952, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.9747743100335311, | |
| "grad_norm": 0.2932960093021393, | |
| "learning_rate": 1.846896964125433e-07, | |
| "loss": 2.1425, | |
| "step": 2362 | |
| }, | |
| { | |
| "epoch": 0.9755996904823317, | |
| "grad_norm": 0.30320852994918823, | |
| "learning_rate": 1.731561643292856e-07, | |
| "loss": 2.1331, | |
| "step": 2364 | |
| }, | |
| { | |
| "epoch": 0.9764250709311323, | |
| "grad_norm": 0.28756237030029297, | |
| "learning_rate": 1.6199383586975037e-07, | |
| "loss": 2.1249, | |
| "step": 2366 | |
| }, | |
| { | |
| "epoch": 0.9772504513799329, | |
| "grad_norm": 0.2928750514984131, | |
| "learning_rate": 1.5120279419185701e-07, | |
| "loss": 2.1222, | |
| "step": 2368 | |
| }, | |
| { | |
| "epoch": 0.9780758318287336, | |
| "grad_norm": 0.2939901649951935, | |
| "learning_rate": 1.4078311968749313e-07, | |
| "loss": 2.1391, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 0.9789012122775341, | |
| "grad_norm": 0.2927398681640625, | |
| "learning_rate": 1.307348899818983e-07, | |
| "loss": 2.1125, | |
| "step": 2372 | |
| }, | |
| { | |
| "epoch": 0.9797265927263348, | |
| "grad_norm": 0.294331431388855, | |
| "learning_rate": 1.2105817993309786e-07, | |
| "loss": 2.0947, | |
| "step": 2374 | |
| }, | |
| { | |
| "epoch": 0.9805519731751354, | |
| "grad_norm": 0.29420092701911926, | |
| "learning_rate": 1.1175306163135335e-07, | |
| "loss": 2.1467, | |
| "step": 2376 | |
| }, | |
| { | |
| "epoch": 0.9813773536239361, | |
| "grad_norm": 0.29944416880607605, | |
| "learning_rate": 1.0281960439860739e-07, | |
| "loss": 2.146, | |
| "step": 2378 | |
| }, | |
| { | |
| "epoch": 0.9822027340727366, | |
| "grad_norm": 0.2895830273628235, | |
| "learning_rate": 9.425787478796744e-08, | |
| "loss": 2.1096, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.9830281145215373, | |
| "grad_norm": 0.3013817071914673, | |
| "learning_rate": 8.606793658323398e-08, | |
| "loss": 2.1204, | |
| "step": 2382 | |
| }, | |
| { | |
| "epoch": 0.9838534949703379, | |
| "grad_norm": 0.2919696569442749, | |
| "learning_rate": 7.824985079839531e-08, | |
| "loss": 2.137, | |
| "step": 2384 | |
| }, | |
| { | |
| "epoch": 0.9846788754191385, | |
| "grad_norm": 0.2889643907546997, | |
| "learning_rate": 7.08036756771946e-08, | |
| "loss": 2.0994, | |
| "step": 2386 | |
| }, | |
| { | |
| "epoch": 0.9855042558679391, | |
| "grad_norm": 0.29254478216171265, | |
| "learning_rate": 6.372946669269131e-08, | |
| "loss": 2.1083, | |
| "step": 2388 | |
| }, | |
| { | |
| "epoch": 0.9863296363167398, | |
| "grad_norm": 0.3087567389011383, | |
| "learning_rate": 5.702727654682272e-08, | |
| "loss": 2.1284, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 0.9871550167655404, | |
| "grad_norm": 0.29949310421943665, | |
| "learning_rate": 5.069715517007079e-08, | |
| "loss": 2.1422, | |
| "step": 2392 | |
| }, | |
| { | |
| "epoch": 0.9879803972143409, | |
| "grad_norm": 0.3776766061782837, | |
| "learning_rate": 4.473914972101256e-08, | |
| "loss": 2.1214, | |
| "step": 2394 | |
| }, | |
| { | |
| "epoch": 0.9888057776631416, | |
| "grad_norm": 0.29243144392967224, | |
| "learning_rate": 3.9153304586042605e-08, | |
| "loss": 2.1213, | |
| "step": 2396 | |
| }, | |
| { | |
| "epoch": 0.9896311581119422, | |
| "grad_norm": 0.2868538498878479, | |
| "learning_rate": 3.393966137898441e-08, | |
| "loss": 2.1366, | |
| "step": 2398 | |
| }, | |
| { | |
| "epoch": 0.9904565385607429, | |
| "grad_norm": 0.3015158176422119, | |
| "learning_rate": 2.9098258940818414e-08, | |
| "loss": 2.1389, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.9912819190095434, | |
| "grad_norm": 0.28750908374786377, | |
| "learning_rate": 2.4629133339371113e-08, | |
| "loss": 2.1257, | |
| "step": 2402 | |
| }, | |
| { | |
| "epoch": 0.9921072994583441, | |
| "grad_norm": 0.29580312967300415, | |
| "learning_rate": 2.0532317869059735e-08, | |
| "loss": 2.1176, | |
| "step": 2404 | |
| }, | |
| { | |
| "epoch": 0.9929326799071447, | |
| "grad_norm": 0.29322829842567444, | |
| "learning_rate": 1.6807843050636874e-08, | |
| "loss": 2.1088, | |
| "step": 2406 | |
| }, | |
| { | |
| "epoch": 0.9937580603559453, | |
| "grad_norm": 0.2918194830417633, | |
| "learning_rate": 1.345573663096289e-08, | |
| "loss": 2.1349, | |
| "step": 2408 | |
| }, | |
| { | |
| "epoch": 0.9945834408047459, | |
| "grad_norm": 0.28616565465927124, | |
| "learning_rate": 1.0476023582806083e-08, | |
| "loss": 2.098, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 0.9954088212535466, | |
| "grad_norm": 0.29948458075523376, | |
| "learning_rate": 7.868726104642844e-09, | |
| "loss": 2.1149, | |
| "step": 2412 | |
| }, | |
| { | |
| "epoch": 0.9962342017023472, | |
| "grad_norm": 0.29504823684692383, | |
| "learning_rate": 5.633863620507773e-09, | |
| "loss": 2.122, | |
| "step": 2414 | |
| }, | |
| { | |
| "epoch": 0.9970595821511478, | |
| "grad_norm": 0.29400694370269775, | |
| "learning_rate": 3.7714527798438095e-09, | |
| "loss": 2.1245, | |
| "step": 2416 | |
| }, | |
| { | |
| "epoch": 0.9978849625999484, | |
| "grad_norm": 0.2938084602355957, | |
| "learning_rate": 2.2815074573745434e-09, | |
| "loss": 2.1174, | |
| "step": 2418 | |
| }, | |
| { | |
| "epoch": 0.9987103430487491, | |
| "grad_norm": 0.2996973991394043, | |
| "learning_rate": 1.164038752998753e-09, | |
| "loss": 2.1225, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.9995357234975496, | |
| "grad_norm": 0.3032398223876953, | |
| "learning_rate": 4.1905499171268626e-10, | |
| "loss": 2.1085, | |
| "step": 2422 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.8356114625930786, | |
| "learning_rate": 4.6561723560101245e-11, | |
| "loss": 2.1522, | |
| "step": 2424 | |
| } | |
| ], | |
| "logging_steps": 2, | |
| "max_steps": 2424, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 300, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.7280755528491336e+18, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |