| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 500.0, | |
| "global_step": 31000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 39.53297424316406, | |
| "eval_runtime": 1.5249, | |
| "eval_samples_per_second": 79.35, | |
| "eval_steps_per_second": 10.493, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 39.283233642578125, | |
| "eval_runtime": 1.5209, | |
| "eval_samples_per_second": 79.56, | |
| "eval_steps_per_second": 10.52, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 39.727474212646484, | |
| "eval_runtime": 1.5199, | |
| "eval_samples_per_second": 79.609, | |
| "eval_steps_per_second": 10.527, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 38.73893356323242, | |
| "eval_runtime": 1.5313, | |
| "eval_samples_per_second": 79.02, | |
| "eval_steps_per_second": 10.449, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 38.108787536621094, | |
| "eval_runtime": 1.5538, | |
| "eval_samples_per_second": 77.875, | |
| "eval_steps_per_second": 10.298, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_loss": 37.65414047241211, | |
| "eval_runtime": 1.5218, | |
| "eval_samples_per_second": 79.511, | |
| "eval_steps_per_second": 10.514, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_loss": 36.28599548339844, | |
| "eval_runtime": 1.5198, | |
| "eval_samples_per_second": 79.613, | |
| "eval_steps_per_second": 10.527, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_loss": 36.02519607543945, | |
| "eval_runtime": 1.5195, | |
| "eval_samples_per_second": 79.632, | |
| "eval_steps_per_second": 10.53, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 4.9193548387096775e-08, | |
| "loss": 40.88, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_loss": 35.2044563293457, | |
| "eval_runtime": 1.5286, | |
| "eval_samples_per_second": 79.159, | |
| "eval_steps_per_second": 10.467, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_loss": 34.13139724731445, | |
| "eval_runtime": 1.5277, | |
| "eval_samples_per_second": 79.203, | |
| "eval_steps_per_second": 10.473, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_loss": 32.90681838989258, | |
| "eval_runtime": 1.5353, | |
| "eval_samples_per_second": 78.812, | |
| "eval_steps_per_second": 10.421, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_loss": 32.091732025146484, | |
| "eval_runtime": 1.5361, | |
| "eval_samples_per_second": 78.769, | |
| "eval_steps_per_second": 10.416, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_loss": 31.341379165649414, | |
| "eval_runtime": 1.5306, | |
| "eval_samples_per_second": 79.053, | |
| "eval_steps_per_second": 10.453, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_loss": 30.152177810668945, | |
| "eval_runtime": 1.5194, | |
| "eval_samples_per_second": 79.635, | |
| "eval_steps_per_second": 10.53, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_loss": 29.086437225341797, | |
| "eval_runtime": 1.522, | |
| "eval_samples_per_second": 79.502, | |
| "eval_steps_per_second": 10.513, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_loss": 27.76565933227539, | |
| "eval_runtime": 1.5196, | |
| "eval_samples_per_second": 79.627, | |
| "eval_steps_per_second": 10.529, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 16.13, | |
| "learning_rate": 4.8387096774193546e-08, | |
| "loss": 32.3449, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_loss": 26.679487228393555, | |
| "eval_runtime": 1.5328, | |
| "eval_samples_per_second": 78.942, | |
| "eval_steps_per_second": 10.439, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_loss": 25.430702209472656, | |
| "eval_runtime": 1.5318, | |
| "eval_samples_per_second": 78.994, | |
| "eval_steps_per_second": 10.445, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_loss": 24.457077026367188, | |
| "eval_runtime": 1.5339, | |
| "eval_samples_per_second": 78.882, | |
| "eval_steps_per_second": 10.431, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_loss": 24.03400993347168, | |
| "eval_runtime": 1.5361, | |
| "eval_samples_per_second": 78.771, | |
| "eval_steps_per_second": 10.416, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_loss": 22.688884735107422, | |
| "eval_runtime": 1.5349, | |
| "eval_samples_per_second": 78.834, | |
| "eval_steps_per_second": 10.424, | |
| "step": 1302 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_loss": 21.914871215820312, | |
| "eval_runtime": 1.5268, | |
| "eval_samples_per_second": 79.251, | |
| "eval_steps_per_second": 10.48, | |
| "step": 1364 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_loss": 20.947105407714844, | |
| "eval_runtime": 1.52, | |
| "eval_samples_per_second": 79.607, | |
| "eval_steps_per_second": 10.527, | |
| "step": 1426 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_loss": 20.71515464782715, | |
| "eval_runtime": 1.5285, | |
| "eval_samples_per_second": 79.164, | |
| "eval_steps_per_second": 10.468, | |
| "step": 1488 | |
| }, | |
| { | |
| "epoch": 24.19, | |
| "learning_rate": 4.7580645161290323e-08, | |
| "loss": 24.114, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_loss": 19.957895278930664, | |
| "eval_runtime": 1.5457, | |
| "eval_samples_per_second": 78.279, | |
| "eval_steps_per_second": 10.351, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_loss": 19.421499252319336, | |
| "eval_runtime": 1.5734, | |
| "eval_samples_per_second": 76.905, | |
| "eval_steps_per_second": 10.169, | |
| "step": 1612 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_loss": 18.702451705932617, | |
| "eval_runtime": 1.5338, | |
| "eval_samples_per_second": 78.89, | |
| "eval_steps_per_second": 10.432, | |
| "step": 1674 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_loss": 18.053003311157227, | |
| "eval_runtime": 1.5299, | |
| "eval_samples_per_second": 79.092, | |
| "eval_steps_per_second": 10.458, | |
| "step": 1736 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_loss": 17.729013442993164, | |
| "eval_runtime": 1.5369, | |
| "eval_samples_per_second": 78.732, | |
| "eval_steps_per_second": 10.411, | |
| "step": 1798 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_loss": 17.282075881958008, | |
| "eval_runtime": 1.5244, | |
| "eval_samples_per_second": 79.377, | |
| "eval_steps_per_second": 10.496, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_loss": 16.866586685180664, | |
| "eval_runtime": 1.5191, | |
| "eval_samples_per_second": 79.654, | |
| "eval_steps_per_second": 10.533, | |
| "step": 1922 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_loss": 16.420963287353516, | |
| "eval_runtime": 1.5281, | |
| "eval_samples_per_second": 79.185, | |
| "eval_steps_per_second": 10.471, | |
| "step": 1984 | |
| }, | |
| { | |
| "epoch": 32.26, | |
| "learning_rate": 4.6774193548387094e-08, | |
| "loss": 18.4015, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_loss": 16.233917236328125, | |
| "eval_runtime": 1.5685, | |
| "eval_samples_per_second": 77.146, | |
| "eval_steps_per_second": 10.201, | |
| "step": 2046 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_loss": 15.798954010009766, | |
| "eval_runtime": 1.5347, | |
| "eval_samples_per_second": 78.844, | |
| "eval_steps_per_second": 10.426, | |
| "step": 2108 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_loss": 15.77268123626709, | |
| "eval_runtime": 1.5269, | |
| "eval_samples_per_second": 79.247, | |
| "eval_steps_per_second": 10.479, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_loss": 15.234251976013184, | |
| "eval_runtime": 1.5303, | |
| "eval_samples_per_second": 79.07, | |
| "eval_steps_per_second": 10.456, | |
| "step": 2232 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_loss": 15.142735481262207, | |
| "eval_runtime": 1.5404, | |
| "eval_samples_per_second": 78.549, | |
| "eval_steps_per_second": 10.387, | |
| "step": 2294 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_loss": 14.873641014099121, | |
| "eval_runtime": 1.5199, | |
| "eval_samples_per_second": 79.611, | |
| "eval_steps_per_second": 10.527, | |
| "step": 2356 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_loss": 14.704235076904297, | |
| "eval_runtime": 1.5212, | |
| "eval_samples_per_second": 79.542, | |
| "eval_steps_per_second": 10.518, | |
| "step": 2418 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_loss": 14.539189338684082, | |
| "eval_runtime": 1.5248, | |
| "eval_samples_per_second": 79.353, | |
| "eval_steps_per_second": 10.493, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 40.32, | |
| "learning_rate": 4.5967741935483865e-08, | |
| "loss": 14.946, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_loss": 14.535553932189941, | |
| "eval_runtime": 1.5276, | |
| "eval_samples_per_second": 79.21, | |
| "eval_steps_per_second": 10.474, | |
| "step": 2542 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_loss": 14.1603364944458, | |
| "eval_runtime": 1.5303, | |
| "eval_samples_per_second": 79.072, | |
| "eval_steps_per_second": 10.456, | |
| "step": 2604 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_loss": 14.145159721374512, | |
| "eval_runtime": 1.5312, | |
| "eval_samples_per_second": 79.024, | |
| "eval_steps_per_second": 10.449, | |
| "step": 2666 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_loss": 14.198166847229004, | |
| "eval_runtime": 1.5287, | |
| "eval_samples_per_second": 79.152, | |
| "eval_steps_per_second": 10.466, | |
| "step": 2728 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_loss": 13.976308822631836, | |
| "eval_runtime": 1.5411, | |
| "eval_samples_per_second": 78.517, | |
| "eval_steps_per_second": 10.382, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_loss": 13.921603202819824, | |
| "eval_runtime": 1.5243, | |
| "eval_samples_per_second": 79.382, | |
| "eval_steps_per_second": 10.497, | |
| "step": 2852 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_loss": 13.767107963562012, | |
| "eval_runtime": 1.5235, | |
| "eval_samples_per_second": 79.42, | |
| "eval_steps_per_second": 10.502, | |
| "step": 2914 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_loss": 13.634342193603516, | |
| "eval_runtime": 1.5202, | |
| "eval_samples_per_second": 79.597, | |
| "eval_steps_per_second": 10.525, | |
| "step": 2976 | |
| }, | |
| { | |
| "epoch": 48.39, | |
| "learning_rate": 4.516129032258064e-08, | |
| "loss": 13.1518, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_loss": 13.62414264678955, | |
| "eval_runtime": 1.5563, | |
| "eval_samples_per_second": 77.747, | |
| "eval_steps_per_second": 10.281, | |
| "step": 3038 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_loss": 13.416315078735352, | |
| "eval_runtime": 1.5393, | |
| "eval_samples_per_second": 78.607, | |
| "eval_steps_per_second": 10.394, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_loss": 13.440138816833496, | |
| "eval_runtime": 1.5302, | |
| "eval_samples_per_second": 79.075, | |
| "eval_steps_per_second": 10.456, | |
| "step": 3162 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_loss": 13.560946464538574, | |
| "eval_runtime": 1.5457, | |
| "eval_samples_per_second": 78.282, | |
| "eval_steps_per_second": 10.351, | |
| "step": 3224 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_loss": 13.285831451416016, | |
| "eval_runtime": 1.5455, | |
| "eval_samples_per_second": 78.294, | |
| "eval_steps_per_second": 10.353, | |
| "step": 3286 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_loss": 13.15819263458252, | |
| "eval_runtime": 1.5204, | |
| "eval_samples_per_second": 79.586, | |
| "eval_steps_per_second": 10.524, | |
| "step": 3348 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_loss": 13.245644569396973, | |
| "eval_runtime": 1.5209, | |
| "eval_samples_per_second": 79.557, | |
| "eval_steps_per_second": 10.52, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_loss": 13.234107971191406, | |
| "eval_runtime": 1.5187, | |
| "eval_samples_per_second": 79.673, | |
| "eval_steps_per_second": 10.535, | |
| "step": 3472 | |
| }, | |
| { | |
| "epoch": 56.45, | |
| "learning_rate": 4.435483870967742e-08, | |
| "loss": 12.1359, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_loss": 13.015525817871094, | |
| "eval_runtime": 1.5313, | |
| "eval_samples_per_second": 79.02, | |
| "eval_steps_per_second": 10.449, | |
| "step": 3534 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_loss": 13.029335021972656, | |
| "eval_runtime": 1.5381, | |
| "eval_samples_per_second": 78.667, | |
| "eval_steps_per_second": 10.402, | |
| "step": 3596 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_loss": 12.971243858337402, | |
| "eval_runtime": 1.5525, | |
| "eval_samples_per_second": 77.94, | |
| "eval_steps_per_second": 10.306, | |
| "step": 3658 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_loss": 12.94456672668457, | |
| "eval_runtime": 1.5335, | |
| "eval_samples_per_second": 78.903, | |
| "eval_steps_per_second": 10.433, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_loss": 12.830053329467773, | |
| "eval_runtime": 1.5349, | |
| "eval_samples_per_second": 78.831, | |
| "eval_steps_per_second": 10.424, | |
| "step": 3782 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_loss": 12.75737190246582, | |
| "eval_runtime": 1.5321, | |
| "eval_samples_per_second": 78.978, | |
| "eval_steps_per_second": 10.443, | |
| "step": 3844 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_loss": 12.696720123291016, | |
| "eval_runtime": 1.5195, | |
| "eval_samples_per_second": 79.632, | |
| "eval_steps_per_second": 10.53, | |
| "step": 3906 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_loss": 12.653301239013672, | |
| "eval_runtime": 1.5219, | |
| "eval_samples_per_second": 79.506, | |
| "eval_steps_per_second": 10.513, | |
| "step": 3968 | |
| }, | |
| { | |
| "epoch": 64.52, | |
| "learning_rate": 4.354838709677419e-08, | |
| "loss": 11.5584, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_loss": 12.554749488830566, | |
| "eval_runtime": 1.5433, | |
| "eval_samples_per_second": 78.403, | |
| "eval_steps_per_second": 10.367, | |
| "step": 4030 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_loss": 12.586564064025879, | |
| "eval_runtime": 1.5317, | |
| "eval_samples_per_second": 78.997, | |
| "eval_steps_per_second": 10.446, | |
| "step": 4092 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_loss": 12.756257057189941, | |
| "eval_runtime": 1.5267, | |
| "eval_samples_per_second": 79.256, | |
| "eval_steps_per_second": 10.48, | |
| "step": 4154 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_loss": 12.648726463317871, | |
| "eval_runtime": 1.5304, | |
| "eval_samples_per_second": 79.064, | |
| "eval_steps_per_second": 10.455, | |
| "step": 4216 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_loss": 12.601462364196777, | |
| "eval_runtime": 1.5297, | |
| "eval_samples_per_second": 79.102, | |
| "eval_steps_per_second": 10.46, | |
| "step": 4278 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_loss": 12.509653091430664, | |
| "eval_runtime": 1.5291, | |
| "eval_samples_per_second": 79.132, | |
| "eval_steps_per_second": 10.464, | |
| "step": 4340 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_loss": 12.53116226196289, | |
| "eval_runtime": 1.5214, | |
| "eval_samples_per_second": 79.53, | |
| "eval_steps_per_second": 10.516, | |
| "step": 4402 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_loss": 12.467875480651855, | |
| "eval_runtime": 1.5274, | |
| "eval_samples_per_second": 79.221, | |
| "eval_steps_per_second": 10.475, | |
| "step": 4464 | |
| }, | |
| { | |
| "epoch": 72.58, | |
| "learning_rate": 4.274193548387097e-08, | |
| "loss": 11.224, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_loss": 12.480475425720215, | |
| "eval_runtime": 1.5409, | |
| "eval_samples_per_second": 78.527, | |
| "eval_steps_per_second": 10.384, | |
| "step": 4526 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_loss": 12.44310474395752, | |
| "eval_runtime": 1.5365, | |
| "eval_samples_per_second": 78.752, | |
| "eval_steps_per_second": 10.413, | |
| "step": 4588 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_loss": 12.302526473999023, | |
| "eval_runtime": 1.5302, | |
| "eval_samples_per_second": 79.074, | |
| "eval_steps_per_second": 10.456, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_loss": 12.154280662536621, | |
| "eval_runtime": 1.528, | |
| "eval_samples_per_second": 79.19, | |
| "eval_steps_per_second": 10.471, | |
| "step": 4712 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_loss": 12.35551929473877, | |
| "eval_runtime": 1.5273, | |
| "eval_samples_per_second": 79.225, | |
| "eval_steps_per_second": 10.476, | |
| "step": 4774 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_loss": 12.264195442199707, | |
| "eval_runtime": 1.5333, | |
| "eval_samples_per_second": 78.913, | |
| "eval_steps_per_second": 10.435, | |
| "step": 4836 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_loss": 12.355469703674316, | |
| "eval_runtime": 1.5317, | |
| "eval_samples_per_second": 78.997, | |
| "eval_steps_per_second": 10.446, | |
| "step": 4898 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_loss": 12.291287422180176, | |
| "eval_runtime": 1.5256, | |
| "eval_samples_per_second": 79.315, | |
| "eval_steps_per_second": 10.488, | |
| "step": 4960 | |
| }, | |
| { | |
| "epoch": 80.65, | |
| "learning_rate": 4.193548387096774e-08, | |
| "loss": 10.9897, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 81.0, | |
| "eval_loss": 12.248346328735352, | |
| "eval_runtime": 1.5278, | |
| "eval_samples_per_second": 79.197, | |
| "eval_steps_per_second": 10.472, | |
| "step": 5022 | |
| }, | |
| { | |
| "epoch": 82.0, | |
| "eval_loss": 12.181654930114746, | |
| "eval_runtime": 1.5283, | |
| "eval_samples_per_second": 79.175, | |
| "eval_steps_per_second": 10.469, | |
| "step": 5084 | |
| }, | |
| { | |
| "epoch": 83.0, | |
| "eval_loss": 12.359329223632812, | |
| "eval_runtime": 1.5268, | |
| "eval_samples_per_second": 79.251, | |
| "eval_steps_per_second": 10.479, | |
| "step": 5146 | |
| }, | |
| { | |
| "epoch": 84.0, | |
| "eval_loss": 12.071714401245117, | |
| "eval_runtime": 1.5407, | |
| "eval_samples_per_second": 78.535, | |
| "eval_steps_per_second": 10.385, | |
| "step": 5208 | |
| }, | |
| { | |
| "epoch": 85.0, | |
| "eval_loss": 12.247845649719238, | |
| "eval_runtime": 1.5293, | |
| "eval_samples_per_second": 79.122, | |
| "eval_steps_per_second": 10.462, | |
| "step": 5270 | |
| }, | |
| { | |
| "epoch": 86.0, | |
| "eval_loss": 12.065207481384277, | |
| "eval_runtime": 1.5401, | |
| "eval_samples_per_second": 78.568, | |
| "eval_steps_per_second": 10.389, | |
| "step": 5332 | |
| }, | |
| { | |
| "epoch": 87.0, | |
| "eval_loss": 12.124922752380371, | |
| "eval_runtime": 1.5205, | |
| "eval_samples_per_second": 79.578, | |
| "eval_steps_per_second": 10.523, | |
| "step": 5394 | |
| }, | |
| { | |
| "epoch": 88.0, | |
| "eval_loss": 12.090045928955078, | |
| "eval_runtime": 1.5308, | |
| "eval_samples_per_second": 79.043, | |
| "eval_steps_per_second": 10.452, | |
| "step": 5456 | |
| }, | |
| { | |
| "epoch": 88.71, | |
| "learning_rate": 4.1129032258064516e-08, | |
| "loss": 10.8156, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 89.0, | |
| "eval_loss": 11.984047889709473, | |
| "eval_runtime": 1.5305, | |
| "eval_samples_per_second": 79.061, | |
| "eval_steps_per_second": 10.454, | |
| "step": 5518 | |
| }, | |
| { | |
| "epoch": 90.0, | |
| "eval_loss": 12.120992660522461, | |
| "eval_runtime": 1.53, | |
| "eval_samples_per_second": 79.086, | |
| "eval_steps_per_second": 10.458, | |
| "step": 5580 | |
| }, | |
| { | |
| "epoch": 91.0, | |
| "eval_loss": 12.18094253540039, | |
| "eval_runtime": 1.5426, | |
| "eval_samples_per_second": 78.437, | |
| "eval_steps_per_second": 10.372, | |
| "step": 5642 | |
| }, | |
| { | |
| "epoch": 92.0, | |
| "eval_loss": 12.021756172180176, | |
| "eval_runtime": 1.5405, | |
| "eval_samples_per_second": 78.545, | |
| "eval_steps_per_second": 10.386, | |
| "step": 5704 | |
| }, | |
| { | |
| "epoch": 93.0, | |
| "eval_loss": 12.143865585327148, | |
| "eval_runtime": 1.5315, | |
| "eval_samples_per_second": 79.009, | |
| "eval_steps_per_second": 10.447, | |
| "step": 5766 | |
| }, | |
| { | |
| "epoch": 94.0, | |
| "eval_loss": 11.918268203735352, | |
| "eval_runtime": 1.5285, | |
| "eval_samples_per_second": 79.163, | |
| "eval_steps_per_second": 10.468, | |
| "step": 5828 | |
| }, | |
| { | |
| "epoch": 95.0, | |
| "eval_loss": 12.067418098449707, | |
| "eval_runtime": 1.5204, | |
| "eval_samples_per_second": 79.584, | |
| "eval_steps_per_second": 10.524, | |
| "step": 5890 | |
| }, | |
| { | |
| "epoch": 96.0, | |
| "eval_loss": 12.009296417236328, | |
| "eval_runtime": 1.52, | |
| "eval_samples_per_second": 79.604, | |
| "eval_steps_per_second": 10.526, | |
| "step": 5952 | |
| }, | |
| { | |
| "epoch": 96.77, | |
| "learning_rate": 4.032258064516129e-08, | |
| "loss": 10.6906, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 97.0, | |
| "eval_loss": 12.008870124816895, | |
| "eval_runtime": 1.5307, | |
| "eval_samples_per_second": 79.05, | |
| "eval_steps_per_second": 10.453, | |
| "step": 6014 | |
| }, | |
| { | |
| "epoch": 98.0, | |
| "eval_loss": 11.956018447875977, | |
| "eval_runtime": 1.5309, | |
| "eval_samples_per_second": 79.041, | |
| "eval_steps_per_second": 10.452, | |
| "step": 6076 | |
| }, | |
| { | |
| "epoch": 99.0, | |
| "eval_loss": 11.767322540283203, | |
| "eval_runtime": 1.5351, | |
| "eval_samples_per_second": 78.825, | |
| "eval_steps_per_second": 10.423, | |
| "step": 6138 | |
| }, | |
| { | |
| "epoch": 100.0, | |
| "eval_loss": 12.034104347229004, | |
| "eval_runtime": 1.5321, | |
| "eval_samples_per_second": 78.979, | |
| "eval_steps_per_second": 10.443, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 101.0, | |
| "eval_loss": 12.03824234008789, | |
| "eval_runtime": 1.5289, | |
| "eval_samples_per_second": 79.14, | |
| "eval_steps_per_second": 10.465, | |
| "step": 6262 | |
| }, | |
| { | |
| "epoch": 102.0, | |
| "eval_loss": 11.87580680847168, | |
| "eval_runtime": 1.5338, | |
| "eval_samples_per_second": 78.891, | |
| "eval_steps_per_second": 10.432, | |
| "step": 6324 | |
| }, | |
| { | |
| "epoch": 103.0, | |
| "eval_loss": 11.93928337097168, | |
| "eval_runtime": 1.5258, | |
| "eval_samples_per_second": 79.301, | |
| "eval_steps_per_second": 10.486, | |
| "step": 6386 | |
| }, | |
| { | |
| "epoch": 104.0, | |
| "eval_loss": 11.878522872924805, | |
| "eval_runtime": 1.5273, | |
| "eval_samples_per_second": 79.226, | |
| "eval_steps_per_second": 10.476, | |
| "step": 6448 | |
| }, | |
| { | |
| "epoch": 104.84, | |
| "learning_rate": 3.951612903225806e-08, | |
| "loss": 10.5767, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 105.0, | |
| "eval_loss": 11.81560230255127, | |
| "eval_runtime": 1.5317, | |
| "eval_samples_per_second": 78.998, | |
| "eval_steps_per_second": 10.446, | |
| "step": 6510 | |
| }, | |
| { | |
| "epoch": 106.0, | |
| "eval_loss": 11.750551223754883, | |
| "eval_runtime": 1.5404, | |
| "eval_samples_per_second": 78.549, | |
| "eval_steps_per_second": 10.387, | |
| "step": 6572 | |
| }, | |
| { | |
| "epoch": 107.0, | |
| "eval_loss": 11.90053653717041, | |
| "eval_runtime": 1.5286, | |
| "eval_samples_per_second": 79.156, | |
| "eval_steps_per_second": 10.467, | |
| "step": 6634 | |
| }, | |
| { | |
| "epoch": 108.0, | |
| "eval_loss": 11.880046844482422, | |
| "eval_runtime": 1.5341, | |
| "eval_samples_per_second": 78.874, | |
| "eval_steps_per_second": 10.43, | |
| "step": 6696 | |
| }, | |
| { | |
| "epoch": 109.0, | |
| "eval_loss": 11.9345703125, | |
| "eval_runtime": 1.5463, | |
| "eval_samples_per_second": 78.251, | |
| "eval_steps_per_second": 10.347, | |
| "step": 6758 | |
| }, | |
| { | |
| "epoch": 110.0, | |
| "eval_loss": 11.833414077758789, | |
| "eval_runtime": 1.5284, | |
| "eval_samples_per_second": 79.166, | |
| "eval_steps_per_second": 10.468, | |
| "step": 6820 | |
| }, | |
| { | |
| "epoch": 111.0, | |
| "eval_loss": 11.616347312927246, | |
| "eval_runtime": 1.5189, | |
| "eval_samples_per_second": 79.663, | |
| "eval_steps_per_second": 10.534, | |
| "step": 6882 | |
| }, | |
| { | |
| "epoch": 112.0, | |
| "eval_loss": 11.829395294189453, | |
| "eval_runtime": 1.5288, | |
| "eval_samples_per_second": 79.144, | |
| "eval_steps_per_second": 10.465, | |
| "step": 6944 | |
| }, | |
| { | |
| "epoch": 112.9, | |
| "learning_rate": 3.8709677419354835e-08, | |
| "loss": 10.5013, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 113.0, | |
| "eval_loss": 11.786639213562012, | |
| "eval_runtime": 1.5596, | |
| "eval_samples_per_second": 77.582, | |
| "eval_steps_per_second": 10.259, | |
| "step": 7006 | |
| }, | |
| { | |
| "epoch": 114.0, | |
| "eval_loss": 11.724261283874512, | |
| "eval_runtime": 1.5228, | |
| "eval_samples_per_second": 79.46, | |
| "eval_steps_per_second": 10.507, | |
| "step": 7068 | |
| }, | |
| { | |
| "epoch": 115.0, | |
| "eval_loss": 11.56452751159668, | |
| "eval_runtime": 1.5342, | |
| "eval_samples_per_second": 78.868, | |
| "eval_steps_per_second": 10.429, | |
| "step": 7130 | |
| }, | |
| { | |
| "epoch": 116.0, | |
| "eval_loss": 11.7550687789917, | |
| "eval_runtime": 1.5293, | |
| "eval_samples_per_second": 79.122, | |
| "eval_steps_per_second": 10.462, | |
| "step": 7192 | |
| }, | |
| { | |
| "epoch": 117.0, | |
| "eval_loss": 11.625850677490234, | |
| "eval_runtime": 1.5487, | |
| "eval_samples_per_second": 78.129, | |
| "eval_steps_per_second": 10.331, | |
| "step": 7254 | |
| }, | |
| { | |
| "epoch": 118.0, | |
| "eval_loss": 11.723539352416992, | |
| "eval_runtime": 1.5327, | |
| "eval_samples_per_second": 78.946, | |
| "eval_steps_per_second": 10.439, | |
| "step": 7316 | |
| }, | |
| { | |
| "epoch": 119.0, | |
| "eval_loss": 11.699857711791992, | |
| "eval_runtime": 1.5315, | |
| "eval_samples_per_second": 79.006, | |
| "eval_steps_per_second": 10.447, | |
| "step": 7378 | |
| }, | |
| { | |
| "epoch": 120.0, | |
| "eval_loss": 11.485078811645508, | |
| "eval_runtime": 1.5203, | |
| "eval_samples_per_second": 79.591, | |
| "eval_steps_per_second": 10.524, | |
| "step": 7440 | |
| }, | |
| { | |
| "epoch": 120.97, | |
| "learning_rate": 3.790322580645161e-08, | |
| "loss": 10.4493, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 121.0, | |
| "eval_loss": 11.495078086853027, | |
| "eval_runtime": 1.5271, | |
| "eval_samples_per_second": 79.234, | |
| "eval_steps_per_second": 10.477, | |
| "step": 7502 | |
| }, | |
| { | |
| "epoch": 122.0, | |
| "eval_loss": 11.521384239196777, | |
| "eval_runtime": 1.536, | |
| "eval_samples_per_second": 78.776, | |
| "eval_steps_per_second": 10.417, | |
| "step": 7564 | |
| }, | |
| { | |
| "epoch": 123.0, | |
| "eval_loss": 11.739558219909668, | |
| "eval_runtime": 1.528, | |
| "eval_samples_per_second": 79.186, | |
| "eval_steps_per_second": 10.471, | |
| "step": 7626 | |
| }, | |
| { | |
| "epoch": 124.0, | |
| "eval_loss": 11.696319580078125, | |
| "eval_runtime": 1.5494, | |
| "eval_samples_per_second": 78.094, | |
| "eval_steps_per_second": 10.327, | |
| "step": 7688 | |
| }, | |
| { | |
| "epoch": 125.0, | |
| "eval_loss": 11.530503273010254, | |
| "eval_runtime": 1.5529, | |
| "eval_samples_per_second": 77.921, | |
| "eval_steps_per_second": 10.304, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 126.0, | |
| "eval_loss": 11.771147727966309, | |
| "eval_runtime": 1.5301, | |
| "eval_samples_per_second": 79.081, | |
| "eval_steps_per_second": 10.457, | |
| "step": 7812 | |
| }, | |
| { | |
| "epoch": 127.0, | |
| "eval_loss": 11.80306339263916, | |
| "eval_runtime": 1.5204, | |
| "eval_samples_per_second": 79.583, | |
| "eval_steps_per_second": 10.523, | |
| "step": 7874 | |
| }, | |
| { | |
| "epoch": 128.0, | |
| "eval_loss": 11.595075607299805, | |
| "eval_runtime": 1.5262, | |
| "eval_samples_per_second": 79.282, | |
| "eval_steps_per_second": 10.484, | |
| "step": 7936 | |
| }, | |
| { | |
| "epoch": 129.0, | |
| "eval_loss": 11.47359848022461, | |
| "eval_runtime": 1.5189, | |
| "eval_samples_per_second": 79.663, | |
| "eval_steps_per_second": 10.534, | |
| "step": 7998 | |
| }, | |
| { | |
| "epoch": 129.03, | |
| "learning_rate": 3.7096774193548384e-08, | |
| "loss": 10.3973, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 130.0, | |
| "eval_loss": 11.615568161010742, | |
| "eval_runtime": 1.5324, | |
| "eval_samples_per_second": 78.963, | |
| "eval_steps_per_second": 10.441, | |
| "step": 8060 | |
| }, | |
| { | |
| "epoch": 131.0, | |
| "eval_loss": 11.544720649719238, | |
| "eval_runtime": 1.5295, | |
| "eval_samples_per_second": 79.11, | |
| "eval_steps_per_second": 10.461, | |
| "step": 8122 | |
| }, | |
| { | |
| "epoch": 132.0, | |
| "eval_loss": 11.356687545776367, | |
| "eval_runtime": 1.5299, | |
| "eval_samples_per_second": 79.09, | |
| "eval_steps_per_second": 10.458, | |
| "step": 8184 | |
| }, | |
| { | |
| "epoch": 133.0, | |
| "eval_loss": 11.483879089355469, | |
| "eval_runtime": 1.5293, | |
| "eval_samples_per_second": 79.121, | |
| "eval_steps_per_second": 10.462, | |
| "step": 8246 | |
| }, | |
| { | |
| "epoch": 134.0, | |
| "eval_loss": 11.441631317138672, | |
| "eval_runtime": 1.5309, | |
| "eval_samples_per_second": 79.039, | |
| "eval_steps_per_second": 10.451, | |
| "step": 8308 | |
| }, | |
| { | |
| "epoch": 135.0, | |
| "eval_loss": 11.539510726928711, | |
| "eval_runtime": 1.528, | |
| "eval_samples_per_second": 79.191, | |
| "eval_steps_per_second": 10.471, | |
| "step": 8370 | |
| }, | |
| { | |
| "epoch": 136.0, | |
| "eval_loss": 11.437644004821777, | |
| "eval_runtime": 1.5258, | |
| "eval_samples_per_second": 79.301, | |
| "eval_steps_per_second": 10.486, | |
| "step": 8432 | |
| }, | |
| { | |
| "epoch": 137.0, | |
| "eval_loss": 11.378108978271484, | |
| "eval_runtime": 1.5197, | |
| "eval_samples_per_second": 79.619, | |
| "eval_steps_per_second": 10.528, | |
| "step": 8494 | |
| }, | |
| { | |
| "epoch": 137.1, | |
| "learning_rate": 3.629032258064516e-08, | |
| "loss": 10.3569, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 138.0, | |
| "eval_loss": 11.417948722839355, | |
| "eval_runtime": 1.5364, | |
| "eval_samples_per_second": 78.758, | |
| "eval_steps_per_second": 10.414, | |
| "step": 8556 | |
| }, | |
| { | |
| "epoch": 139.0, | |
| "eval_loss": 11.674735069274902, | |
| "eval_runtime": 1.5314, | |
| "eval_samples_per_second": 79.012, | |
| "eval_steps_per_second": 10.448, | |
| "step": 8618 | |
| }, | |
| { | |
| "epoch": 140.0, | |
| "eval_loss": 11.611095428466797, | |
| "eval_runtime": 1.5272, | |
| "eval_samples_per_second": 79.231, | |
| "eval_steps_per_second": 10.477, | |
| "step": 8680 | |
| }, | |
| { | |
| "epoch": 141.0, | |
| "eval_loss": 11.50928020477295, | |
| "eval_runtime": 1.5304, | |
| "eval_samples_per_second": 79.062, | |
| "eval_steps_per_second": 10.455, | |
| "step": 8742 | |
| }, | |
| { | |
| "epoch": 142.0, | |
| "eval_loss": 11.475595474243164, | |
| "eval_runtime": 1.5278, | |
| "eval_samples_per_second": 79.201, | |
| "eval_steps_per_second": 10.473, | |
| "step": 8804 | |
| }, | |
| { | |
| "epoch": 143.0, | |
| "eval_loss": 11.34980583190918, | |
| "eval_runtime": 1.5199, | |
| "eval_samples_per_second": 79.612, | |
| "eval_steps_per_second": 10.527, | |
| "step": 8866 | |
| }, | |
| { | |
| "epoch": 144.0, | |
| "eval_loss": 11.562585830688477, | |
| "eval_runtime": 1.5213, | |
| "eval_samples_per_second": 79.535, | |
| "eval_steps_per_second": 10.517, | |
| "step": 8928 | |
| }, | |
| { | |
| "epoch": 145.0, | |
| "eval_loss": 11.347404479980469, | |
| "eval_runtime": 1.5204, | |
| "eval_samples_per_second": 79.584, | |
| "eval_steps_per_second": 10.524, | |
| "step": 8990 | |
| }, | |
| { | |
| "epoch": 145.16, | |
| "learning_rate": 3.548387096774194e-08, | |
| "loss": 10.3565, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 146.0, | |
| "eval_loss": 11.377586364746094, | |
| "eval_runtime": 1.5334, | |
| "eval_samples_per_second": 78.91, | |
| "eval_steps_per_second": 10.434, | |
| "step": 9052 | |
| }, | |
| { | |
| "epoch": 147.0, | |
| "eval_loss": 11.362828254699707, | |
| "eval_runtime": 1.5275, | |
| "eval_samples_per_second": 79.217, | |
| "eval_steps_per_second": 10.475, | |
| "step": 9114 | |
| }, | |
| { | |
| "epoch": 148.0, | |
| "eval_loss": 11.485925674438477, | |
| "eval_runtime": 1.5452, | |
| "eval_samples_per_second": 78.308, | |
| "eval_steps_per_second": 10.355, | |
| "step": 9176 | |
| }, | |
| { | |
| "epoch": 149.0, | |
| "eval_loss": 11.422063827514648, | |
| "eval_runtime": 1.5281, | |
| "eval_samples_per_second": 79.184, | |
| "eval_steps_per_second": 10.471, | |
| "step": 9238 | |
| }, | |
| { | |
| "epoch": 150.0, | |
| "eval_loss": 11.419939041137695, | |
| "eval_runtime": 1.54, | |
| "eval_samples_per_second": 78.57, | |
| "eval_steps_per_second": 10.389, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 151.0, | |
| "eval_loss": 11.570083618164062, | |
| "eval_runtime": 1.5233, | |
| "eval_samples_per_second": 79.432, | |
| "eval_steps_per_second": 10.503, | |
| "step": 9362 | |
| }, | |
| { | |
| "epoch": 152.0, | |
| "eval_loss": 11.463680267333984, | |
| "eval_runtime": 1.5197, | |
| "eval_samples_per_second": 79.621, | |
| "eval_steps_per_second": 10.528, | |
| "step": 9424 | |
| }, | |
| { | |
| "epoch": 153.0, | |
| "eval_loss": 11.65864086151123, | |
| "eval_runtime": 1.5194, | |
| "eval_samples_per_second": 79.634, | |
| "eval_steps_per_second": 10.53, | |
| "step": 9486 | |
| }, | |
| { | |
| "epoch": 153.23, | |
| "learning_rate": 3.467741935483871e-08, | |
| "loss": 10.3122, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 154.0, | |
| "eval_loss": 11.487783432006836, | |
| "eval_runtime": 1.5302, | |
| "eval_samples_per_second": 79.074, | |
| "eval_steps_per_second": 10.456, | |
| "step": 9548 | |
| }, | |
| { | |
| "epoch": 155.0, | |
| "eval_loss": 11.590331077575684, | |
| "eval_runtime": 1.5284, | |
| "eval_samples_per_second": 79.169, | |
| "eval_steps_per_second": 10.469, | |
| "step": 9610 | |
| }, | |
| { | |
| "epoch": 156.0, | |
| "eval_loss": 11.381085395812988, | |
| "eval_runtime": 1.5362, | |
| "eval_samples_per_second": 78.767, | |
| "eval_steps_per_second": 10.415, | |
| "step": 9672 | |
| }, | |
| { | |
| "epoch": 157.0, | |
| "eval_loss": 11.454978942871094, | |
| "eval_runtime": 1.5486, | |
| "eval_samples_per_second": 78.136, | |
| "eval_steps_per_second": 10.332, | |
| "step": 9734 | |
| }, | |
| { | |
| "epoch": 158.0, | |
| "eval_loss": 11.569897651672363, | |
| "eval_runtime": 1.5314, | |
| "eval_samples_per_second": 79.011, | |
| "eval_steps_per_second": 10.448, | |
| "step": 9796 | |
| }, | |
| { | |
| "epoch": 159.0, | |
| "eval_loss": 11.515423774719238, | |
| "eval_runtime": 1.52, | |
| "eval_samples_per_second": 79.603, | |
| "eval_steps_per_second": 10.526, | |
| "step": 9858 | |
| }, | |
| { | |
| "epoch": 160.0, | |
| "eval_loss": 11.501609802246094, | |
| "eval_runtime": 1.5198, | |
| "eval_samples_per_second": 79.617, | |
| "eval_steps_per_second": 10.528, | |
| "step": 9920 | |
| }, | |
| { | |
| "epoch": 161.0, | |
| "eval_loss": 11.571191787719727, | |
| "eval_runtime": 1.5313, | |
| "eval_samples_per_second": 79.018, | |
| "eval_steps_per_second": 10.449, | |
| "step": 9982 | |
| }, | |
| { | |
| "epoch": 161.29, | |
| "learning_rate": 3.387096774193548e-08, | |
| "loss": 10.2888, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 162.0, | |
| "eval_loss": 11.289880752563477, | |
| "eval_runtime": 1.5323, | |
| "eval_samples_per_second": 78.967, | |
| "eval_steps_per_second": 10.442, | |
| "step": 10044 | |
| }, | |
| { | |
| "epoch": 163.0, | |
| "eval_loss": 11.250905990600586, | |
| "eval_runtime": 1.5304, | |
| "eval_samples_per_second": 79.065, | |
| "eval_steps_per_second": 10.455, | |
| "step": 10106 | |
| }, | |
| { | |
| "epoch": 164.0, | |
| "eval_loss": 11.452507972717285, | |
| "eval_runtime": 1.5273, | |
| "eval_samples_per_second": 79.227, | |
| "eval_steps_per_second": 10.476, | |
| "step": 10168 | |
| }, | |
| { | |
| "epoch": 165.0, | |
| "eval_loss": 11.363972663879395, | |
| "eval_runtime": 1.5279, | |
| "eval_samples_per_second": 79.192, | |
| "eval_steps_per_second": 10.472, | |
| "step": 10230 | |
| }, | |
| { | |
| "epoch": 166.0, | |
| "eval_loss": 11.337541580200195, | |
| "eval_runtime": 1.5301, | |
| "eval_samples_per_second": 79.077, | |
| "eval_steps_per_second": 10.457, | |
| "step": 10292 | |
| }, | |
| { | |
| "epoch": 167.0, | |
| "eval_loss": 11.396878242492676, | |
| "eval_runtime": 1.5279, | |
| "eval_samples_per_second": 79.193, | |
| "eval_steps_per_second": 10.472, | |
| "step": 10354 | |
| }, | |
| { | |
| "epoch": 168.0, | |
| "eval_loss": 11.478124618530273, | |
| "eval_runtime": 1.5196, | |
| "eval_samples_per_second": 79.628, | |
| "eval_steps_per_second": 10.529, | |
| "step": 10416 | |
| }, | |
| { | |
| "epoch": 169.0, | |
| "eval_loss": 11.397954940795898, | |
| "eval_runtime": 1.5231, | |
| "eval_samples_per_second": 79.443, | |
| "eval_steps_per_second": 10.505, | |
| "step": 10478 | |
| }, | |
| { | |
| "epoch": 169.35, | |
| "learning_rate": 3.306451612903226e-08, | |
| "loss": 10.2642, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 170.0, | |
| "eval_loss": 11.28774356842041, | |
| "eval_runtime": 1.5277, | |
| "eval_samples_per_second": 79.204, | |
| "eval_steps_per_second": 10.473, | |
| "step": 10540 | |
| }, | |
| { | |
| "epoch": 171.0, | |
| "eval_loss": 11.336891174316406, | |
| "eval_runtime": 1.5318, | |
| "eval_samples_per_second": 78.993, | |
| "eval_steps_per_second": 10.445, | |
| "step": 10602 | |
| }, | |
| { | |
| "epoch": 172.0, | |
| "eval_loss": 11.385379791259766, | |
| "eval_runtime": 1.5315, | |
| "eval_samples_per_second": 79.006, | |
| "eval_steps_per_second": 10.447, | |
| "step": 10664 | |
| }, | |
| { | |
| "epoch": 173.0, | |
| "eval_loss": 11.50495433807373, | |
| "eval_runtime": 1.5331, | |
| "eval_samples_per_second": 78.924, | |
| "eval_steps_per_second": 10.436, | |
| "step": 10726 | |
| }, | |
| { | |
| "epoch": 174.0, | |
| "eval_loss": 11.482619285583496, | |
| "eval_runtime": 1.5436, | |
| "eval_samples_per_second": 78.386, | |
| "eval_steps_per_second": 10.365, | |
| "step": 10788 | |
| }, | |
| { | |
| "epoch": 175.0, | |
| "eval_loss": 11.342474937438965, | |
| "eval_runtime": 1.5316, | |
| "eval_samples_per_second": 79.002, | |
| "eval_steps_per_second": 10.447, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 176.0, | |
| "eval_loss": 11.241662979125977, | |
| "eval_runtime": 1.5196, | |
| "eval_samples_per_second": 79.629, | |
| "eval_steps_per_second": 10.529, | |
| "step": 10912 | |
| }, | |
| { | |
| "epoch": 177.0, | |
| "eval_loss": 11.126252174377441, | |
| "eval_runtime": 1.5201, | |
| "eval_samples_per_second": 79.603, | |
| "eval_steps_per_second": 10.526, | |
| "step": 10974 | |
| }, | |
| { | |
| "epoch": 177.42, | |
| "learning_rate": 3.225806451612903e-08, | |
| "loss": 10.2586, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 178.0, | |
| "eval_loss": 11.408693313598633, | |
| "eval_runtime": 1.5266, | |
| "eval_samples_per_second": 79.261, | |
| "eval_steps_per_second": 10.481, | |
| "step": 11036 | |
| }, | |
| { | |
| "epoch": 179.0, | |
| "eval_loss": 11.391864776611328, | |
| "eval_runtime": 1.552, | |
| "eval_samples_per_second": 77.965, | |
| "eval_steps_per_second": 10.309, | |
| "step": 11098 | |
| }, | |
| { | |
| "epoch": 180.0, | |
| "eval_loss": 11.251392364501953, | |
| "eval_runtime": 1.5284, | |
| "eval_samples_per_second": 79.165, | |
| "eval_steps_per_second": 10.468, | |
| "step": 11160 | |
| }, | |
| { | |
| "epoch": 181.0, | |
| "eval_loss": 11.256957054138184, | |
| "eval_runtime": 1.5287, | |
| "eval_samples_per_second": 79.153, | |
| "eval_steps_per_second": 10.467, | |
| "step": 11222 | |
| }, | |
| { | |
| "epoch": 182.0, | |
| "eval_loss": 11.44185733795166, | |
| "eval_runtime": 1.5278, | |
| "eval_samples_per_second": 79.197, | |
| "eval_steps_per_second": 10.472, | |
| "step": 11284 | |
| }, | |
| { | |
| "epoch": 183.0, | |
| "eval_loss": 11.41098690032959, | |
| "eval_runtime": 1.5598, | |
| "eval_samples_per_second": 77.572, | |
| "eval_steps_per_second": 10.257, | |
| "step": 11346 | |
| }, | |
| { | |
| "epoch": 184.0, | |
| "eval_loss": 11.391864776611328, | |
| "eval_runtime": 1.5217, | |
| "eval_samples_per_second": 79.518, | |
| "eval_steps_per_second": 10.515, | |
| "step": 11408 | |
| }, | |
| { | |
| "epoch": 185.0, | |
| "eval_loss": 11.242551803588867, | |
| "eval_runtime": 1.5268, | |
| "eval_samples_per_second": 79.25, | |
| "eval_steps_per_second": 10.479, | |
| "step": 11470 | |
| }, | |
| { | |
| "epoch": 185.48, | |
| "learning_rate": 3.1451612903225806e-08, | |
| "loss": 10.2479, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 186.0, | |
| "eval_loss": 11.386938095092773, | |
| "eval_runtime": 1.5272, | |
| "eval_samples_per_second": 79.228, | |
| "eval_steps_per_second": 10.476, | |
| "step": 11532 | |
| }, | |
| { | |
| "epoch": 187.0, | |
| "eval_loss": 11.295321464538574, | |
| "eval_runtime": 1.5298, | |
| "eval_samples_per_second": 79.094, | |
| "eval_steps_per_second": 10.459, | |
| "step": 11594 | |
| }, | |
| { | |
| "epoch": 188.0, | |
| "eval_loss": 11.33161449432373, | |
| "eval_runtime": 1.5284, | |
| "eval_samples_per_second": 79.166, | |
| "eval_steps_per_second": 10.468, | |
| "step": 11656 | |
| }, | |
| { | |
| "epoch": 189.0, | |
| "eval_loss": 11.137282371520996, | |
| "eval_runtime": 1.5487, | |
| "eval_samples_per_second": 78.13, | |
| "eval_steps_per_second": 10.331, | |
| "step": 11718 | |
| }, | |
| { | |
| "epoch": 190.0, | |
| "eval_loss": 11.165376663208008, | |
| "eval_runtime": 1.5348, | |
| "eval_samples_per_second": 78.838, | |
| "eval_steps_per_second": 10.425, | |
| "step": 11780 | |
| }, | |
| { | |
| "epoch": 191.0, | |
| "eval_loss": 11.320448875427246, | |
| "eval_runtime": 1.5398, | |
| "eval_samples_per_second": 78.582, | |
| "eval_steps_per_second": 10.391, | |
| "step": 11842 | |
| }, | |
| { | |
| "epoch": 192.0, | |
| "eval_loss": 11.312908172607422, | |
| "eval_runtime": 1.5205, | |
| "eval_samples_per_second": 79.582, | |
| "eval_steps_per_second": 10.523, | |
| "step": 11904 | |
| }, | |
| { | |
| "epoch": 193.0, | |
| "eval_loss": 11.271549224853516, | |
| "eval_runtime": 1.5262, | |
| "eval_samples_per_second": 79.283, | |
| "eval_steps_per_second": 10.484, | |
| "step": 11966 | |
| }, | |
| { | |
| "epoch": 193.55, | |
| "learning_rate": 3.064516129032258e-08, | |
| "loss": 10.2172, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 194.0, | |
| "eval_loss": 11.2645845413208, | |
| "eval_runtime": 1.5507, | |
| "eval_samples_per_second": 78.027, | |
| "eval_steps_per_second": 10.318, | |
| "step": 12028 | |
| }, | |
| { | |
| "epoch": 195.0, | |
| "eval_loss": 11.335646629333496, | |
| "eval_runtime": 1.5277, | |
| "eval_samples_per_second": 79.206, | |
| "eval_steps_per_second": 10.474, | |
| "step": 12090 | |
| }, | |
| { | |
| "epoch": 196.0, | |
| "eval_loss": 11.301328659057617, | |
| "eval_runtime": 1.5287, | |
| "eval_samples_per_second": 79.153, | |
| "eval_steps_per_second": 10.466, | |
| "step": 12152 | |
| }, | |
| { | |
| "epoch": 197.0, | |
| "eval_loss": 11.537822723388672, | |
| "eval_runtime": 1.5263, | |
| "eval_samples_per_second": 79.277, | |
| "eval_steps_per_second": 10.483, | |
| "step": 12214 | |
| }, | |
| { | |
| "epoch": 198.0, | |
| "eval_loss": 11.364118576049805, | |
| "eval_runtime": 1.529, | |
| "eval_samples_per_second": 79.138, | |
| "eval_steps_per_second": 10.465, | |
| "step": 12276 | |
| }, | |
| { | |
| "epoch": 199.0, | |
| "eval_loss": 11.369778633117676, | |
| "eval_runtime": 1.5279, | |
| "eval_samples_per_second": 79.195, | |
| "eval_steps_per_second": 10.472, | |
| "step": 12338 | |
| }, | |
| { | |
| "epoch": 200.0, | |
| "eval_loss": 11.267717361450195, | |
| "eval_runtime": 1.5199, | |
| "eval_samples_per_second": 79.612, | |
| "eval_steps_per_second": 10.527, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 201.0, | |
| "eval_loss": 11.297280311584473, | |
| "eval_runtime": 1.521, | |
| "eval_samples_per_second": 79.552, | |
| "eval_steps_per_second": 10.519, | |
| "step": 12462 | |
| }, | |
| { | |
| "epoch": 201.61, | |
| "learning_rate": 2.9838709677419354e-08, | |
| "loss": 10.2023, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 202.0, | |
| "eval_loss": 11.140520095825195, | |
| "eval_runtime": 1.5299, | |
| "eval_samples_per_second": 79.092, | |
| "eval_steps_per_second": 10.458, | |
| "step": 12524 | |
| }, | |
| { | |
| "epoch": 203.0, | |
| "eval_loss": 11.32856559753418, | |
| "eval_runtime": 1.5298, | |
| "eval_samples_per_second": 79.093, | |
| "eval_steps_per_second": 10.459, | |
| "step": 12586 | |
| }, | |
| { | |
| "epoch": 204.0, | |
| "eval_loss": 11.212227821350098, | |
| "eval_runtime": 1.528, | |
| "eval_samples_per_second": 79.19, | |
| "eval_steps_per_second": 10.471, | |
| "step": 12648 | |
| }, | |
| { | |
| "epoch": 205.0, | |
| "eval_loss": 11.232370376586914, | |
| "eval_runtime": 1.5285, | |
| "eval_samples_per_second": 79.16, | |
| "eval_steps_per_second": 10.467, | |
| "step": 12710 | |
| }, | |
| { | |
| "epoch": 206.0, | |
| "eval_loss": 11.35122299194336, | |
| "eval_runtime": 1.5675, | |
| "eval_samples_per_second": 77.193, | |
| "eval_steps_per_second": 10.207, | |
| "step": 12772 | |
| }, | |
| { | |
| "epoch": 207.0, | |
| "eval_loss": 11.088189125061035, | |
| "eval_runtime": 1.5298, | |
| "eval_samples_per_second": 79.096, | |
| "eval_steps_per_second": 10.459, | |
| "step": 12834 | |
| }, | |
| { | |
| "epoch": 208.0, | |
| "eval_loss": 11.35745620727539, | |
| "eval_runtime": 1.5205, | |
| "eval_samples_per_second": 79.581, | |
| "eval_steps_per_second": 10.523, | |
| "step": 12896 | |
| }, | |
| { | |
| "epoch": 209.0, | |
| "eval_loss": 11.250239372253418, | |
| "eval_runtime": 1.5204, | |
| "eval_samples_per_second": 79.584, | |
| "eval_steps_per_second": 10.523, | |
| "step": 12958 | |
| }, | |
| { | |
| "epoch": 209.68, | |
| "learning_rate": 2.9032258064516128e-08, | |
| "loss": 10.1926, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 210.0, | |
| "eval_loss": 11.327762603759766, | |
| "eval_runtime": 1.5376, | |
| "eval_samples_per_second": 78.692, | |
| "eval_steps_per_second": 10.406, | |
| "step": 13020 | |
| }, | |
| { | |
| "epoch": 211.0, | |
| "eval_loss": 11.21988582611084, | |
| "eval_runtime": 1.5336, | |
| "eval_samples_per_second": 78.901, | |
| "eval_steps_per_second": 10.433, | |
| "step": 13082 | |
| }, | |
| { | |
| "epoch": 212.0, | |
| "eval_loss": 11.256662368774414, | |
| "eval_runtime": 1.5406, | |
| "eval_samples_per_second": 78.541, | |
| "eval_steps_per_second": 10.386, | |
| "step": 13144 | |
| }, | |
| { | |
| "epoch": 213.0, | |
| "eval_loss": 11.28549861907959, | |
| "eval_runtime": 1.5261, | |
| "eval_samples_per_second": 79.288, | |
| "eval_steps_per_second": 10.484, | |
| "step": 13206 | |
| }, | |
| { | |
| "epoch": 214.0, | |
| "eval_loss": 11.189271926879883, | |
| "eval_runtime": 1.5291, | |
| "eval_samples_per_second": 79.132, | |
| "eval_steps_per_second": 10.464, | |
| "step": 13268 | |
| }, | |
| { | |
| "epoch": 215.0, | |
| "eval_loss": 11.260034561157227, | |
| "eval_runtime": 1.5268, | |
| "eval_samples_per_second": 79.249, | |
| "eval_steps_per_second": 10.479, | |
| "step": 13330 | |
| }, | |
| { | |
| "epoch": 216.0, | |
| "eval_loss": 11.284974098205566, | |
| "eval_runtime": 1.5308, | |
| "eval_samples_per_second": 79.043, | |
| "eval_steps_per_second": 10.452, | |
| "step": 13392 | |
| }, | |
| { | |
| "epoch": 217.0, | |
| "eval_loss": 11.227448463439941, | |
| "eval_runtime": 1.5233, | |
| "eval_samples_per_second": 79.431, | |
| "eval_steps_per_second": 10.503, | |
| "step": 13454 | |
| }, | |
| { | |
| "epoch": 217.74, | |
| "learning_rate": 2.8225806451612906e-08, | |
| "loss": 10.2075, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 218.0, | |
| "eval_loss": 11.110801696777344, | |
| "eval_runtime": 1.5287, | |
| "eval_samples_per_second": 79.151, | |
| "eval_steps_per_second": 10.466, | |
| "step": 13516 | |
| }, | |
| { | |
| "epoch": 219.0, | |
| "eval_loss": 11.275555610656738, | |
| "eval_runtime": 1.5279, | |
| "eval_samples_per_second": 79.192, | |
| "eval_steps_per_second": 10.472, | |
| "step": 13578 | |
| }, | |
| { | |
| "epoch": 220.0, | |
| "eval_loss": 11.283916473388672, | |
| "eval_runtime": 1.5257, | |
| "eval_samples_per_second": 79.308, | |
| "eval_steps_per_second": 10.487, | |
| "step": 13640 | |
| }, | |
| { | |
| "epoch": 221.0, | |
| "eval_loss": 11.216109275817871, | |
| "eval_runtime": 1.528, | |
| "eval_samples_per_second": 79.188, | |
| "eval_steps_per_second": 10.471, | |
| "step": 13702 | |
| }, | |
| { | |
| "epoch": 222.0, | |
| "eval_loss": 11.186559677124023, | |
| "eval_runtime": 1.5313, | |
| "eval_samples_per_second": 79.017, | |
| "eval_steps_per_second": 10.448, | |
| "step": 13764 | |
| }, | |
| { | |
| "epoch": 223.0, | |
| "eval_loss": 11.12224006652832, | |
| "eval_runtime": 1.5319, | |
| "eval_samples_per_second": 78.989, | |
| "eval_steps_per_second": 10.445, | |
| "step": 13826 | |
| }, | |
| { | |
| "epoch": 224.0, | |
| "eval_loss": 11.264845848083496, | |
| "eval_runtime": 1.5275, | |
| "eval_samples_per_second": 79.216, | |
| "eval_steps_per_second": 10.475, | |
| "step": 13888 | |
| }, | |
| { | |
| "epoch": 225.0, | |
| "eval_loss": 11.293610572814941, | |
| "eval_runtime": 1.5184, | |
| "eval_samples_per_second": 79.688, | |
| "eval_steps_per_second": 10.537, | |
| "step": 13950 | |
| }, | |
| { | |
| "epoch": 225.81, | |
| "learning_rate": 2.7419354838709673e-08, | |
| "loss": 10.1822, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 226.0, | |
| "eval_loss": 11.136199951171875, | |
| "eval_runtime": 1.5304, | |
| "eval_samples_per_second": 79.063, | |
| "eval_steps_per_second": 10.455, | |
| "step": 14012 | |
| }, | |
| { | |
| "epoch": 227.0, | |
| "eval_loss": 11.259844779968262, | |
| "eval_runtime": 1.5312, | |
| "eval_samples_per_second": 79.022, | |
| "eval_steps_per_second": 10.449, | |
| "step": 14074 | |
| }, | |
| { | |
| "epoch": 228.0, | |
| "eval_loss": 11.335241317749023, | |
| "eval_runtime": 1.5268, | |
| "eval_samples_per_second": 79.253, | |
| "eval_steps_per_second": 10.48, | |
| "step": 14136 | |
| }, | |
| { | |
| "epoch": 229.0, | |
| "eval_loss": 11.118280410766602, | |
| "eval_runtime": 1.529, | |
| "eval_samples_per_second": 79.138, | |
| "eval_steps_per_second": 10.465, | |
| "step": 14198 | |
| }, | |
| { | |
| "epoch": 230.0, | |
| "eval_loss": 11.207478523254395, | |
| "eval_runtime": 1.5282, | |
| "eval_samples_per_second": 79.178, | |
| "eval_steps_per_second": 10.47, | |
| "step": 14260 | |
| }, | |
| { | |
| "epoch": 231.0, | |
| "eval_loss": 11.190764427185059, | |
| "eval_runtime": 1.531, | |
| "eval_samples_per_second": 79.032, | |
| "eval_steps_per_second": 10.451, | |
| "step": 14322 | |
| }, | |
| { | |
| "epoch": 232.0, | |
| "eval_loss": 11.133115768432617, | |
| "eval_runtime": 1.5224, | |
| "eval_samples_per_second": 79.482, | |
| "eval_steps_per_second": 10.51, | |
| "step": 14384 | |
| }, | |
| { | |
| "epoch": 233.0, | |
| "eval_loss": 11.092888832092285, | |
| "eval_runtime": 1.521, | |
| "eval_samples_per_second": 79.551, | |
| "eval_steps_per_second": 10.519, | |
| "step": 14446 | |
| }, | |
| { | |
| "epoch": 233.87, | |
| "learning_rate": 2.661290322580645e-08, | |
| "loss": 10.1863, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 234.0, | |
| "eval_loss": 11.106066703796387, | |
| "eval_runtime": 1.5271, | |
| "eval_samples_per_second": 79.233, | |
| "eval_steps_per_second": 10.477, | |
| "step": 14508 | |
| }, | |
| { | |
| "epoch": 235.0, | |
| "eval_loss": 11.054638862609863, | |
| "eval_runtime": 1.5301, | |
| "eval_samples_per_second": 79.078, | |
| "eval_steps_per_second": 10.457, | |
| "step": 14570 | |
| }, | |
| { | |
| "epoch": 236.0, | |
| "eval_loss": 11.176545143127441, | |
| "eval_runtime": 1.528, | |
| "eval_samples_per_second": 79.189, | |
| "eval_steps_per_second": 10.471, | |
| "step": 14632 | |
| }, | |
| { | |
| "epoch": 237.0, | |
| "eval_loss": 11.25126838684082, | |
| "eval_runtime": 1.5393, | |
| "eval_samples_per_second": 78.608, | |
| "eval_steps_per_second": 10.394, | |
| "step": 14694 | |
| }, | |
| { | |
| "epoch": 238.0, | |
| "eval_loss": 11.091902732849121, | |
| "eval_runtime": 1.5287, | |
| "eval_samples_per_second": 79.155, | |
| "eval_steps_per_second": 10.467, | |
| "step": 14756 | |
| }, | |
| { | |
| "epoch": 239.0, | |
| "eval_loss": 11.2738676071167, | |
| "eval_runtime": 1.5269, | |
| "eval_samples_per_second": 79.244, | |
| "eval_steps_per_second": 10.479, | |
| "step": 14818 | |
| }, | |
| { | |
| "epoch": 240.0, | |
| "eval_loss": 11.034156799316406, | |
| "eval_runtime": 1.5197, | |
| "eval_samples_per_second": 79.619, | |
| "eval_steps_per_second": 10.528, | |
| "step": 14880 | |
| }, | |
| { | |
| "epoch": 241.0, | |
| "eval_loss": 11.236822128295898, | |
| "eval_runtime": 1.525, | |
| "eval_samples_per_second": 79.342, | |
| "eval_steps_per_second": 10.492, | |
| "step": 14942 | |
| }, | |
| { | |
| "epoch": 241.94, | |
| "learning_rate": 2.5806451612903225e-08, | |
| "loss": 10.1724, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 242.0, | |
| "eval_loss": 11.25521469116211, | |
| "eval_runtime": 1.5321, | |
| "eval_samples_per_second": 78.976, | |
| "eval_steps_per_second": 10.443, | |
| "step": 15004 | |
| }, | |
| { | |
| "epoch": 243.0, | |
| "eval_loss": 11.151459693908691, | |
| "eval_runtime": 1.5265, | |
| "eval_samples_per_second": 79.267, | |
| "eval_steps_per_second": 10.482, | |
| "step": 15066 | |
| }, | |
| { | |
| "epoch": 244.0, | |
| "eval_loss": 11.049065589904785, | |
| "eval_runtime": 1.525, | |
| "eval_samples_per_second": 79.342, | |
| "eval_steps_per_second": 10.492, | |
| "step": 15128 | |
| }, | |
| { | |
| "epoch": 245.0, | |
| "eval_loss": 11.018394470214844, | |
| "eval_runtime": 1.5424, | |
| "eval_samples_per_second": 78.451, | |
| "eval_steps_per_second": 10.374, | |
| "step": 15190 | |
| }, | |
| { | |
| "epoch": 246.0, | |
| "eval_loss": 11.378674507141113, | |
| "eval_runtime": 1.5254, | |
| "eval_samples_per_second": 79.325, | |
| "eval_steps_per_second": 10.489, | |
| "step": 15252 | |
| }, | |
| { | |
| "epoch": 247.0, | |
| "eval_loss": 11.232932090759277, | |
| "eval_runtime": 1.5294, | |
| "eval_samples_per_second": 79.117, | |
| "eval_steps_per_second": 10.462, | |
| "step": 15314 | |
| }, | |
| { | |
| "epoch": 248.0, | |
| "eval_loss": 11.072883605957031, | |
| "eval_runtime": 1.5274, | |
| "eval_samples_per_second": 79.221, | |
| "eval_steps_per_second": 10.476, | |
| "step": 15376 | |
| }, | |
| { | |
| "epoch": 249.0, | |
| "eval_loss": 11.147173881530762, | |
| "eval_runtime": 1.5246, | |
| "eval_samples_per_second": 79.366, | |
| "eval_steps_per_second": 10.495, | |
| "step": 15438 | |
| }, | |
| { | |
| "epoch": 250.0, | |
| "learning_rate": 2.5e-08, | |
| "loss": 10.1671, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 250.0, | |
| "eval_loss": 11.06186294555664, | |
| "eval_runtime": 1.5501, | |
| "eval_samples_per_second": 78.061, | |
| "eval_steps_per_second": 10.322, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 251.0, | |
| "eval_loss": 11.219667434692383, | |
| "eval_runtime": 1.5267, | |
| "eval_samples_per_second": 79.257, | |
| "eval_steps_per_second": 10.48, | |
| "step": 15562 | |
| }, | |
| { | |
| "epoch": 252.0, | |
| "eval_loss": 11.224080085754395, | |
| "eval_runtime": 1.5293, | |
| "eval_samples_per_second": 79.121, | |
| "eval_steps_per_second": 10.462, | |
| "step": 15624 | |
| }, | |
| { | |
| "epoch": 253.0, | |
| "eval_loss": 11.310192108154297, | |
| "eval_runtime": 1.528, | |
| "eval_samples_per_second": 79.187, | |
| "eval_steps_per_second": 10.471, | |
| "step": 15686 | |
| }, | |
| { | |
| "epoch": 254.0, | |
| "eval_loss": 11.289962768554688, | |
| "eval_runtime": 1.5252, | |
| "eval_samples_per_second": 79.335, | |
| "eval_steps_per_second": 10.491, | |
| "step": 15748 | |
| }, | |
| { | |
| "epoch": 255.0, | |
| "eval_loss": 11.300124168395996, | |
| "eval_runtime": 1.5278, | |
| "eval_samples_per_second": 79.2, | |
| "eval_steps_per_second": 10.473, | |
| "step": 15810 | |
| }, | |
| { | |
| "epoch": 256.0, | |
| "eval_loss": 11.220646858215332, | |
| "eval_runtime": 1.5215, | |
| "eval_samples_per_second": 79.528, | |
| "eval_steps_per_second": 10.516, | |
| "step": 15872 | |
| }, | |
| { | |
| "epoch": 257.0, | |
| "eval_loss": 11.102697372436523, | |
| "eval_runtime": 1.5207, | |
| "eval_samples_per_second": 79.571, | |
| "eval_steps_per_second": 10.522, | |
| "step": 15934 | |
| }, | |
| { | |
| "epoch": 258.0, | |
| "eval_loss": 11.306656837463379, | |
| "eval_runtime": 1.5184, | |
| "eval_samples_per_second": 79.692, | |
| "eval_steps_per_second": 10.538, | |
| "step": 15996 | |
| }, | |
| { | |
| "epoch": 258.06, | |
| "learning_rate": 2.4193548387096773e-08, | |
| "loss": 10.1579, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 259.0, | |
| "eval_loss": 11.30165958404541, | |
| "eval_runtime": 1.5278, | |
| "eval_samples_per_second": 79.201, | |
| "eval_steps_per_second": 10.473, | |
| "step": 16058 | |
| }, | |
| { | |
| "epoch": 260.0, | |
| "eval_loss": 11.098042488098145, | |
| "eval_runtime": 1.5428, | |
| "eval_samples_per_second": 78.43, | |
| "eval_steps_per_second": 10.371, | |
| "step": 16120 | |
| }, | |
| { | |
| "epoch": 261.0, | |
| "eval_loss": 11.215791702270508, | |
| "eval_runtime": 1.5285, | |
| "eval_samples_per_second": 79.163, | |
| "eval_steps_per_second": 10.468, | |
| "step": 16182 | |
| }, | |
| { | |
| "epoch": 262.0, | |
| "eval_loss": 10.940754890441895, | |
| "eval_runtime": 1.5416, | |
| "eval_samples_per_second": 78.488, | |
| "eval_steps_per_second": 10.379, | |
| "step": 16244 | |
| }, | |
| { | |
| "epoch": 263.0, | |
| "eval_loss": 11.108644485473633, | |
| "eval_runtime": 1.5277, | |
| "eval_samples_per_second": 79.205, | |
| "eval_steps_per_second": 10.473, | |
| "step": 16306 | |
| }, | |
| { | |
| "epoch": 264.0, | |
| "eval_loss": 11.28579330444336, | |
| "eval_runtime": 1.5188, | |
| "eval_samples_per_second": 79.67, | |
| "eval_steps_per_second": 10.535, | |
| "step": 16368 | |
| }, | |
| { | |
| "epoch": 265.0, | |
| "eval_loss": 11.280501365661621, | |
| "eval_runtime": 1.518, | |
| "eval_samples_per_second": 79.708, | |
| "eval_steps_per_second": 10.54, | |
| "step": 16430 | |
| }, | |
| { | |
| "epoch": 266.0, | |
| "eval_loss": 11.08869457244873, | |
| "eval_runtime": 1.5206, | |
| "eval_samples_per_second": 79.573, | |
| "eval_steps_per_second": 10.522, | |
| "step": 16492 | |
| }, | |
| { | |
| "epoch": 266.13, | |
| "learning_rate": 2.3387096774193547e-08, | |
| "loss": 10.1552, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 267.0, | |
| "eval_loss": 11.171466827392578, | |
| "eval_runtime": 1.5404, | |
| "eval_samples_per_second": 78.554, | |
| "eval_steps_per_second": 10.387, | |
| "step": 16554 | |
| }, | |
| { | |
| "epoch": 268.0, | |
| "eval_loss": 11.1470365524292, | |
| "eval_runtime": 1.5289, | |
| "eval_samples_per_second": 79.141, | |
| "eval_steps_per_second": 10.465, | |
| "step": 16616 | |
| }, | |
| { | |
| "epoch": 269.0, | |
| "eval_loss": 11.106681823730469, | |
| "eval_runtime": 1.5263, | |
| "eval_samples_per_second": 79.276, | |
| "eval_steps_per_second": 10.483, | |
| "step": 16678 | |
| }, | |
| { | |
| "epoch": 270.0, | |
| "eval_loss": 11.163908958435059, | |
| "eval_runtime": 1.5252, | |
| "eval_samples_per_second": 79.336, | |
| "eval_steps_per_second": 10.491, | |
| "step": 16740 | |
| }, | |
| { | |
| "epoch": 271.0, | |
| "eval_loss": 11.146965980529785, | |
| "eval_runtime": 1.5255, | |
| "eval_samples_per_second": 79.316, | |
| "eval_steps_per_second": 10.488, | |
| "step": 16802 | |
| }, | |
| { | |
| "epoch": 272.0, | |
| "eval_loss": 11.206341743469238, | |
| "eval_runtime": 1.5228, | |
| "eval_samples_per_second": 79.461, | |
| "eval_steps_per_second": 10.507, | |
| "step": 16864 | |
| }, | |
| { | |
| "epoch": 273.0, | |
| "eval_loss": 11.16470718383789, | |
| "eval_runtime": 1.5197, | |
| "eval_samples_per_second": 79.62, | |
| "eval_steps_per_second": 10.528, | |
| "step": 16926 | |
| }, | |
| { | |
| "epoch": 274.0, | |
| "eval_loss": 11.006061553955078, | |
| "eval_runtime": 1.5209, | |
| "eval_samples_per_second": 79.559, | |
| "eval_steps_per_second": 10.52, | |
| "step": 16988 | |
| }, | |
| { | |
| "epoch": 274.19, | |
| "learning_rate": 2.258064516129032e-08, | |
| "loss": 10.1468, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 275.0, | |
| "eval_loss": 11.382197380065918, | |
| "eval_runtime": 1.5267, | |
| "eval_samples_per_second": 79.254, | |
| "eval_steps_per_second": 10.48, | |
| "step": 17050 | |
| }, | |
| { | |
| "epoch": 276.0, | |
| "eval_loss": 10.946673393249512, | |
| "eval_runtime": 1.5278, | |
| "eval_samples_per_second": 79.2, | |
| "eval_steps_per_second": 10.473, | |
| "step": 17112 | |
| }, | |
| { | |
| "epoch": 277.0, | |
| "eval_loss": 11.2734956741333, | |
| "eval_runtime": 1.5371, | |
| "eval_samples_per_second": 78.717, | |
| "eval_steps_per_second": 10.409, | |
| "step": 17174 | |
| }, | |
| { | |
| "epoch": 278.0, | |
| "eval_loss": 11.013189315795898, | |
| "eval_runtime": 1.5409, | |
| "eval_samples_per_second": 78.528, | |
| "eval_steps_per_second": 10.384, | |
| "step": 17236 | |
| }, | |
| { | |
| "epoch": 279.0, | |
| "eval_loss": 11.160365104675293, | |
| "eval_runtime": 1.5281, | |
| "eval_samples_per_second": 79.181, | |
| "eval_steps_per_second": 10.47, | |
| "step": 17298 | |
| }, | |
| { | |
| "epoch": 280.0, | |
| "eval_loss": 11.16196060180664, | |
| "eval_runtime": 1.5194, | |
| "eval_samples_per_second": 79.636, | |
| "eval_steps_per_second": 10.53, | |
| "step": 17360 | |
| }, | |
| { | |
| "epoch": 281.0, | |
| "eval_loss": 11.064959526062012, | |
| "eval_runtime": 1.5184, | |
| "eval_samples_per_second": 79.689, | |
| "eval_steps_per_second": 10.537, | |
| "step": 17422 | |
| }, | |
| { | |
| "epoch": 282.0, | |
| "eval_loss": 10.991950035095215, | |
| "eval_runtime": 1.5262, | |
| "eval_samples_per_second": 79.284, | |
| "eval_steps_per_second": 10.484, | |
| "step": 17484 | |
| }, | |
| { | |
| "epoch": 282.26, | |
| "learning_rate": 2.1774193548387095e-08, | |
| "loss": 10.1402, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 283.0, | |
| "eval_loss": 11.044986724853516, | |
| "eval_runtime": 1.5294, | |
| "eval_samples_per_second": 79.118, | |
| "eval_steps_per_second": 10.462, | |
| "step": 17546 | |
| }, | |
| { | |
| "epoch": 284.0, | |
| "eval_loss": 10.982977867126465, | |
| "eval_runtime": 1.5353, | |
| "eval_samples_per_second": 78.811, | |
| "eval_steps_per_second": 10.421, | |
| "step": 17608 | |
| }, | |
| { | |
| "epoch": 285.0, | |
| "eval_loss": 11.013529777526855, | |
| "eval_runtime": 1.5385, | |
| "eval_samples_per_second": 78.649, | |
| "eval_steps_per_second": 10.4, | |
| "step": 17670 | |
| }, | |
| { | |
| "epoch": 286.0, | |
| "eval_loss": 11.264179229736328, | |
| "eval_runtime": 1.5281, | |
| "eval_samples_per_second": 79.183, | |
| "eval_steps_per_second": 10.47, | |
| "step": 17732 | |
| }, | |
| { | |
| "epoch": 287.0, | |
| "eval_loss": 11.11242389678955, | |
| "eval_runtime": 1.5289, | |
| "eval_samples_per_second": 79.142, | |
| "eval_steps_per_second": 10.465, | |
| "step": 17794 | |
| }, | |
| { | |
| "epoch": 288.0, | |
| "eval_loss": 11.192839622497559, | |
| "eval_runtime": 1.5192, | |
| "eval_samples_per_second": 79.646, | |
| "eval_steps_per_second": 10.532, | |
| "step": 17856 | |
| }, | |
| { | |
| "epoch": 289.0, | |
| "eval_loss": 11.128388404846191, | |
| "eval_runtime": 1.5284, | |
| "eval_samples_per_second": 79.167, | |
| "eval_steps_per_second": 10.468, | |
| "step": 17918 | |
| }, | |
| { | |
| "epoch": 290.0, | |
| "eval_loss": 11.208813667297363, | |
| "eval_runtime": 1.5227, | |
| "eval_samples_per_second": 79.463, | |
| "eval_steps_per_second": 10.507, | |
| "step": 17980 | |
| }, | |
| { | |
| "epoch": 290.32, | |
| "learning_rate": 2.096774193548387e-08, | |
| "loss": 10.147, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 291.0, | |
| "eval_loss": 11.075602531433105, | |
| "eval_runtime": 1.5304, | |
| "eval_samples_per_second": 79.062, | |
| "eval_steps_per_second": 10.454, | |
| "step": 18042 | |
| }, | |
| { | |
| "epoch": 292.0, | |
| "eval_loss": 11.051429748535156, | |
| "eval_runtime": 1.5297, | |
| "eval_samples_per_second": 79.103, | |
| "eval_steps_per_second": 10.46, | |
| "step": 18104 | |
| }, | |
| { | |
| "epoch": 293.0, | |
| "eval_loss": 11.14745044708252, | |
| "eval_runtime": 1.528, | |
| "eval_samples_per_second": 79.191, | |
| "eval_steps_per_second": 10.471, | |
| "step": 18166 | |
| }, | |
| { | |
| "epoch": 294.0, | |
| "eval_loss": 11.044754981994629, | |
| "eval_runtime": 1.5371, | |
| "eval_samples_per_second": 78.718, | |
| "eval_steps_per_second": 10.409, | |
| "step": 18228 | |
| }, | |
| { | |
| "epoch": 295.0, | |
| "eval_loss": 11.241650581359863, | |
| "eval_runtime": 1.5413, | |
| "eval_samples_per_second": 78.505, | |
| "eval_steps_per_second": 10.381, | |
| "step": 18290 | |
| }, | |
| { | |
| "epoch": 296.0, | |
| "eval_loss": 11.107006072998047, | |
| "eval_runtime": 1.5336, | |
| "eval_samples_per_second": 78.901, | |
| "eval_steps_per_second": 10.433, | |
| "step": 18352 | |
| }, | |
| { | |
| "epoch": 297.0, | |
| "eval_loss": 11.014945030212402, | |
| "eval_runtime": 1.5335, | |
| "eval_samples_per_second": 78.906, | |
| "eval_steps_per_second": 10.434, | |
| "step": 18414 | |
| }, | |
| { | |
| "epoch": 298.0, | |
| "eval_loss": 11.168006896972656, | |
| "eval_runtime": 1.5218, | |
| "eval_samples_per_second": 79.51, | |
| "eval_steps_per_second": 10.514, | |
| "step": 18476 | |
| }, | |
| { | |
| "epoch": 298.39, | |
| "learning_rate": 2.0161290322580644e-08, | |
| "loss": 10.1331, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 299.0, | |
| "eval_loss": 11.100564956665039, | |
| "eval_runtime": 1.5329, | |
| "eval_samples_per_second": 78.935, | |
| "eval_steps_per_second": 10.438, | |
| "step": 18538 | |
| }, | |
| { | |
| "epoch": 300.0, | |
| "eval_loss": 10.944679260253906, | |
| "eval_runtime": 1.5398, | |
| "eval_samples_per_second": 78.581, | |
| "eval_steps_per_second": 10.391, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 301.0, | |
| "eval_loss": 10.988174438476562, | |
| "eval_runtime": 1.54, | |
| "eval_samples_per_second": 78.569, | |
| "eval_steps_per_second": 10.389, | |
| "step": 18662 | |
| }, | |
| { | |
| "epoch": 302.0, | |
| "eval_loss": 11.02690315246582, | |
| "eval_runtime": 1.5425, | |
| "eval_samples_per_second": 78.446, | |
| "eval_steps_per_second": 10.373, | |
| "step": 18724 | |
| }, | |
| { | |
| "epoch": 303.0, | |
| "eval_loss": 10.976366996765137, | |
| "eval_runtime": 1.5362, | |
| "eval_samples_per_second": 78.765, | |
| "eval_steps_per_second": 10.415, | |
| "step": 18786 | |
| }, | |
| { | |
| "epoch": 304.0, | |
| "eval_loss": 11.088923454284668, | |
| "eval_runtime": 1.5246, | |
| "eval_samples_per_second": 79.365, | |
| "eval_steps_per_second": 10.495, | |
| "step": 18848 | |
| }, | |
| { | |
| "epoch": 305.0, | |
| "eval_loss": 11.024333953857422, | |
| "eval_runtime": 1.5195, | |
| "eval_samples_per_second": 79.629, | |
| "eval_steps_per_second": 10.529, | |
| "step": 18910 | |
| }, | |
| { | |
| "epoch": 306.0, | |
| "eval_loss": 11.025029182434082, | |
| "eval_runtime": 1.524, | |
| "eval_samples_per_second": 79.394, | |
| "eval_steps_per_second": 10.498, | |
| "step": 18972 | |
| }, | |
| { | |
| "epoch": 306.45, | |
| "learning_rate": 1.9354838709677418e-08, | |
| "loss": 10.1244, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 307.0, | |
| "eval_loss": 11.113330841064453, | |
| "eval_runtime": 1.54, | |
| "eval_samples_per_second": 78.569, | |
| "eval_steps_per_second": 10.389, | |
| "step": 19034 | |
| }, | |
| { | |
| "epoch": 308.0, | |
| "eval_loss": 11.013413429260254, | |
| "eval_runtime": 1.5273, | |
| "eval_samples_per_second": 79.226, | |
| "eval_steps_per_second": 10.476, | |
| "step": 19096 | |
| }, | |
| { | |
| "epoch": 309.0, | |
| "eval_loss": 10.995118141174316, | |
| "eval_runtime": 1.5332, | |
| "eval_samples_per_second": 78.921, | |
| "eval_steps_per_second": 10.436, | |
| "step": 19158 | |
| }, | |
| { | |
| "epoch": 310.0, | |
| "eval_loss": 11.261984825134277, | |
| "eval_runtime": 1.5238, | |
| "eval_samples_per_second": 79.407, | |
| "eval_steps_per_second": 10.5, | |
| "step": 19220 | |
| }, | |
| { | |
| "epoch": 311.0, | |
| "eval_loss": 11.110681533813477, | |
| "eval_runtime": 1.5588, | |
| "eval_samples_per_second": 77.623, | |
| "eval_steps_per_second": 10.264, | |
| "step": 19282 | |
| }, | |
| { | |
| "epoch": 312.0, | |
| "eval_loss": 11.05971622467041, | |
| "eval_runtime": 1.5228, | |
| "eval_samples_per_second": 79.459, | |
| "eval_steps_per_second": 10.507, | |
| "step": 19344 | |
| }, | |
| { | |
| "epoch": 313.0, | |
| "eval_loss": 10.960829734802246, | |
| "eval_runtime": 1.5312, | |
| "eval_samples_per_second": 79.024, | |
| "eval_steps_per_second": 10.449, | |
| "step": 19406 | |
| }, | |
| { | |
| "epoch": 314.0, | |
| "eval_loss": 10.997540473937988, | |
| "eval_runtime": 1.5275, | |
| "eval_samples_per_second": 79.216, | |
| "eval_steps_per_second": 10.475, | |
| "step": 19468 | |
| }, | |
| { | |
| "epoch": 314.52, | |
| "learning_rate": 1.8548387096774192e-08, | |
| "loss": 10.1251, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 315.0, | |
| "eval_loss": 11.017022132873535, | |
| "eval_runtime": 1.5516, | |
| "eval_samples_per_second": 77.982, | |
| "eval_steps_per_second": 10.312, | |
| "step": 19530 | |
| }, | |
| { | |
| "epoch": 316.0, | |
| "eval_loss": 11.152543067932129, | |
| "eval_runtime": 1.5321, | |
| "eval_samples_per_second": 78.977, | |
| "eval_steps_per_second": 10.443, | |
| "step": 19592 | |
| }, | |
| { | |
| "epoch": 317.0, | |
| "eval_loss": 11.0108003616333, | |
| "eval_runtime": 1.521, | |
| "eval_samples_per_second": 79.554, | |
| "eval_steps_per_second": 10.52, | |
| "step": 19654 | |
| }, | |
| { | |
| "epoch": 318.0, | |
| "eval_loss": 11.024676322937012, | |
| "eval_runtime": 1.5196, | |
| "eval_samples_per_second": 79.626, | |
| "eval_steps_per_second": 10.529, | |
| "step": 19716 | |
| }, | |
| { | |
| "epoch": 319.0, | |
| "eval_loss": 11.113091468811035, | |
| "eval_runtime": 1.5205, | |
| "eval_samples_per_second": 79.581, | |
| "eval_steps_per_second": 10.523, | |
| "step": 19778 | |
| }, | |
| { | |
| "epoch": 320.0, | |
| "eval_loss": 11.04623794555664, | |
| "eval_runtime": 1.5226, | |
| "eval_samples_per_second": 79.468, | |
| "eval_steps_per_second": 10.508, | |
| "step": 19840 | |
| }, | |
| { | |
| "epoch": 321.0, | |
| "eval_loss": 11.06059741973877, | |
| "eval_runtime": 1.5198, | |
| "eval_samples_per_second": 79.613, | |
| "eval_steps_per_second": 10.527, | |
| "step": 19902 | |
| }, | |
| { | |
| "epoch": 322.0, | |
| "eval_loss": 11.004892349243164, | |
| "eval_runtime": 1.5295, | |
| "eval_samples_per_second": 79.109, | |
| "eval_steps_per_second": 10.461, | |
| "step": 19964 | |
| }, | |
| { | |
| "epoch": 322.58, | |
| "learning_rate": 1.774193548387097e-08, | |
| "loss": 10.1178, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 323.0, | |
| "eval_loss": 11.110086441040039, | |
| "eval_runtime": 1.5366, | |
| "eval_samples_per_second": 78.746, | |
| "eval_steps_per_second": 10.413, | |
| "step": 20026 | |
| }, | |
| { | |
| "epoch": 324.0, | |
| "eval_loss": 11.034819602966309, | |
| "eval_runtime": 1.5318, | |
| "eval_samples_per_second": 78.992, | |
| "eval_steps_per_second": 10.445, | |
| "step": 20088 | |
| }, | |
| { | |
| "epoch": 325.0, | |
| "eval_loss": 11.286378860473633, | |
| "eval_runtime": 1.5319, | |
| "eval_samples_per_second": 78.988, | |
| "eval_steps_per_second": 10.445, | |
| "step": 20150 | |
| }, | |
| { | |
| "epoch": 326.0, | |
| "eval_loss": 10.950843811035156, | |
| "eval_runtime": 1.5282, | |
| "eval_samples_per_second": 79.177, | |
| "eval_steps_per_second": 10.47, | |
| "step": 20212 | |
| }, | |
| { | |
| "epoch": 327.0, | |
| "eval_loss": 10.995577812194824, | |
| "eval_runtime": 1.5301, | |
| "eval_samples_per_second": 79.078, | |
| "eval_steps_per_second": 10.457, | |
| "step": 20274 | |
| }, | |
| { | |
| "epoch": 328.0, | |
| "eval_loss": 10.956055641174316, | |
| "eval_runtime": 1.5197, | |
| "eval_samples_per_second": 79.62, | |
| "eval_steps_per_second": 10.528, | |
| "step": 20336 | |
| }, | |
| { | |
| "epoch": 329.0, | |
| "eval_loss": 10.953376770019531, | |
| "eval_runtime": 1.522, | |
| "eval_samples_per_second": 79.498, | |
| "eval_steps_per_second": 10.512, | |
| "step": 20398 | |
| }, | |
| { | |
| "epoch": 330.0, | |
| "eval_loss": 11.365254402160645, | |
| "eval_runtime": 1.5263, | |
| "eval_samples_per_second": 79.275, | |
| "eval_steps_per_second": 10.483, | |
| "step": 20460 | |
| }, | |
| { | |
| "epoch": 330.65, | |
| "learning_rate": 1.693548387096774e-08, | |
| "loss": 10.1215, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 331.0, | |
| "eval_loss": 11.118794441223145, | |
| "eval_runtime": 1.5308, | |
| "eval_samples_per_second": 79.044, | |
| "eval_steps_per_second": 10.452, | |
| "step": 20522 | |
| }, | |
| { | |
| "epoch": 332.0, | |
| "eval_loss": 11.090803146362305, | |
| "eval_runtime": 1.5271, | |
| "eval_samples_per_second": 79.237, | |
| "eval_steps_per_second": 10.478, | |
| "step": 20584 | |
| }, | |
| { | |
| "epoch": 333.0, | |
| "eval_loss": 10.98558235168457, | |
| "eval_runtime": 1.5289, | |
| "eval_samples_per_second": 79.144, | |
| "eval_steps_per_second": 10.465, | |
| "step": 20646 | |
| }, | |
| { | |
| "epoch": 334.0, | |
| "eval_loss": 10.987250328063965, | |
| "eval_runtime": 1.5588, | |
| "eval_samples_per_second": 77.622, | |
| "eval_steps_per_second": 10.264, | |
| "step": 20708 | |
| }, | |
| { | |
| "epoch": 335.0, | |
| "eval_loss": 11.048944473266602, | |
| "eval_runtime": 1.5348, | |
| "eval_samples_per_second": 78.84, | |
| "eval_steps_per_second": 10.425, | |
| "step": 20770 | |
| }, | |
| { | |
| "epoch": 336.0, | |
| "eval_loss": 11.112990379333496, | |
| "eval_runtime": 1.5248, | |
| "eval_samples_per_second": 79.355, | |
| "eval_steps_per_second": 10.493, | |
| "step": 20832 | |
| }, | |
| { | |
| "epoch": 337.0, | |
| "eval_loss": 10.877412796020508, | |
| "eval_runtime": 1.5203, | |
| "eval_samples_per_second": 79.589, | |
| "eval_steps_per_second": 10.524, | |
| "step": 20894 | |
| }, | |
| { | |
| "epoch": 338.0, | |
| "eval_loss": 10.787759780883789, | |
| "eval_runtime": 1.5247, | |
| "eval_samples_per_second": 79.361, | |
| "eval_steps_per_second": 10.494, | |
| "step": 20956 | |
| }, | |
| { | |
| "epoch": 338.71, | |
| "learning_rate": 1.6129032258064514e-08, | |
| "loss": 10.1267, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 339.0, | |
| "eval_loss": 11.173955917358398, | |
| "eval_runtime": 1.535, | |
| "eval_samples_per_second": 78.827, | |
| "eval_steps_per_second": 10.423, | |
| "step": 21018 | |
| }, | |
| { | |
| "epoch": 340.0, | |
| "eval_loss": 11.00390911102295, | |
| "eval_runtime": 1.5311, | |
| "eval_samples_per_second": 79.03, | |
| "eval_steps_per_second": 10.45, | |
| "step": 21080 | |
| }, | |
| { | |
| "epoch": 341.0, | |
| "eval_loss": 11.032459259033203, | |
| "eval_runtime": 1.5355, | |
| "eval_samples_per_second": 78.804, | |
| "eval_steps_per_second": 10.42, | |
| "step": 21142 | |
| }, | |
| { | |
| "epoch": 342.0, | |
| "eval_loss": 11.04984188079834, | |
| "eval_runtime": 1.5396, | |
| "eval_samples_per_second": 78.591, | |
| "eval_steps_per_second": 10.392, | |
| "step": 21204 | |
| }, | |
| { | |
| "epoch": 343.0, | |
| "eval_loss": 11.023720741271973, | |
| "eval_runtime": 1.5305, | |
| "eval_samples_per_second": 79.057, | |
| "eval_steps_per_second": 10.454, | |
| "step": 21266 | |
| }, | |
| { | |
| "epoch": 344.0, | |
| "eval_loss": 11.093502044677734, | |
| "eval_runtime": 1.5353, | |
| "eval_samples_per_second": 78.813, | |
| "eval_steps_per_second": 10.422, | |
| "step": 21328 | |
| }, | |
| { | |
| "epoch": 345.0, | |
| "eval_loss": 10.816137313842773, | |
| "eval_runtime": 1.5197, | |
| "eval_samples_per_second": 79.62, | |
| "eval_steps_per_second": 10.528, | |
| "step": 21390 | |
| }, | |
| { | |
| "epoch": 346.0, | |
| "eval_loss": 11.070144653320312, | |
| "eval_runtime": 1.5197, | |
| "eval_samples_per_second": 79.623, | |
| "eval_steps_per_second": 10.529, | |
| "step": 21452 | |
| }, | |
| { | |
| "epoch": 346.77, | |
| "learning_rate": 1.532258064516129e-08, | |
| "loss": 10.1105, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 347.0, | |
| "eval_loss": 10.87633991241455, | |
| "eval_runtime": 1.5347, | |
| "eval_samples_per_second": 78.841, | |
| "eval_steps_per_second": 10.425, | |
| "step": 21514 | |
| }, | |
| { | |
| "epoch": 348.0, | |
| "eval_loss": 11.06225299835205, | |
| "eval_runtime": 1.5319, | |
| "eval_samples_per_second": 78.984, | |
| "eval_steps_per_second": 10.444, | |
| "step": 21576 | |
| }, | |
| { | |
| "epoch": 349.0, | |
| "eval_loss": 10.950231552124023, | |
| "eval_runtime": 1.5483, | |
| "eval_samples_per_second": 78.15, | |
| "eval_steps_per_second": 10.334, | |
| "step": 21638 | |
| }, | |
| { | |
| "epoch": 350.0, | |
| "eval_loss": 11.076042175292969, | |
| "eval_runtime": 1.5401, | |
| "eval_samples_per_second": 78.567, | |
| "eval_steps_per_second": 10.389, | |
| "step": 21700 | |
| }, | |
| { | |
| "epoch": 351.0, | |
| "eval_loss": 11.147188186645508, | |
| "eval_runtime": 1.5351, | |
| "eval_samples_per_second": 78.823, | |
| "eval_steps_per_second": 10.423, | |
| "step": 21762 | |
| }, | |
| { | |
| "epoch": 352.0, | |
| "eval_loss": 11.15673828125, | |
| "eval_runtime": 1.5292, | |
| "eval_samples_per_second": 79.126, | |
| "eval_steps_per_second": 10.463, | |
| "step": 21824 | |
| }, | |
| { | |
| "epoch": 353.0, | |
| "eval_loss": 11.01186466217041, | |
| "eval_runtime": 1.5267, | |
| "eval_samples_per_second": 79.255, | |
| "eval_steps_per_second": 10.48, | |
| "step": 21886 | |
| }, | |
| { | |
| "epoch": 354.0, | |
| "eval_loss": 11.07396125793457, | |
| "eval_runtime": 1.5223, | |
| "eval_samples_per_second": 79.486, | |
| "eval_steps_per_second": 10.511, | |
| "step": 21948 | |
| }, | |
| { | |
| "epoch": 354.84, | |
| "learning_rate": 1.4516129032258064e-08, | |
| "loss": 10.1124, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 355.0, | |
| "eval_loss": 11.035076141357422, | |
| "eval_runtime": 1.5453, | |
| "eval_samples_per_second": 78.302, | |
| "eval_steps_per_second": 10.354, | |
| "step": 22010 | |
| }, | |
| { | |
| "epoch": 356.0, | |
| "eval_loss": 10.995853424072266, | |
| "eval_runtime": 1.5492, | |
| "eval_samples_per_second": 78.107, | |
| "eval_steps_per_second": 10.328, | |
| "step": 22072 | |
| }, | |
| { | |
| "epoch": 357.0, | |
| "eval_loss": 10.982865333557129, | |
| "eval_runtime": 1.5378, | |
| "eval_samples_per_second": 78.686, | |
| "eval_steps_per_second": 10.405, | |
| "step": 22134 | |
| }, | |
| { | |
| "epoch": 358.0, | |
| "eval_loss": 11.162384986877441, | |
| "eval_runtime": 1.5367, | |
| "eval_samples_per_second": 78.739, | |
| "eval_steps_per_second": 10.412, | |
| "step": 22196 | |
| }, | |
| { | |
| "epoch": 359.0, | |
| "eval_loss": 11.10819149017334, | |
| "eval_runtime": 1.5353, | |
| "eval_samples_per_second": 78.812, | |
| "eval_steps_per_second": 10.421, | |
| "step": 22258 | |
| }, | |
| { | |
| "epoch": 360.0, | |
| "eval_loss": 11.049622535705566, | |
| "eval_runtime": 1.5377, | |
| "eval_samples_per_second": 78.689, | |
| "eval_steps_per_second": 10.405, | |
| "step": 22320 | |
| }, | |
| { | |
| "epoch": 361.0, | |
| "eval_loss": 11.111467361450195, | |
| "eval_runtime": 1.52, | |
| "eval_samples_per_second": 79.604, | |
| "eval_steps_per_second": 10.526, | |
| "step": 22382 | |
| }, | |
| { | |
| "epoch": 362.0, | |
| "eval_loss": 10.887182235717773, | |
| "eval_runtime": 1.5238, | |
| "eval_samples_per_second": 79.408, | |
| "eval_steps_per_second": 10.5, | |
| "step": 22444 | |
| }, | |
| { | |
| "epoch": 362.9, | |
| "learning_rate": 1.3709677419354837e-08, | |
| "loss": 10.1057, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 363.0, | |
| "eval_loss": 10.899175643920898, | |
| "eval_runtime": 1.5381, | |
| "eval_samples_per_second": 78.671, | |
| "eval_steps_per_second": 10.403, | |
| "step": 22506 | |
| }, | |
| { | |
| "epoch": 364.0, | |
| "eval_loss": 11.026792526245117, | |
| "eval_runtime": 1.5471, | |
| "eval_samples_per_second": 78.21, | |
| "eval_steps_per_second": 10.342, | |
| "step": 22568 | |
| }, | |
| { | |
| "epoch": 365.0, | |
| "eval_loss": 10.96151065826416, | |
| "eval_runtime": 1.5277, | |
| "eval_samples_per_second": 79.204, | |
| "eval_steps_per_second": 10.473, | |
| "step": 22630 | |
| }, | |
| { | |
| "epoch": 366.0, | |
| "eval_loss": 10.967521667480469, | |
| "eval_runtime": 1.5284, | |
| "eval_samples_per_second": 79.165, | |
| "eval_steps_per_second": 10.468, | |
| "step": 22692 | |
| }, | |
| { | |
| "epoch": 367.0, | |
| "eval_loss": 11.17754077911377, | |
| "eval_runtime": 1.5345, | |
| "eval_samples_per_second": 78.853, | |
| "eval_steps_per_second": 10.427, | |
| "step": 22754 | |
| }, | |
| { | |
| "epoch": 368.0, | |
| "eval_loss": 10.994891166687012, | |
| "eval_runtime": 1.5285, | |
| "eval_samples_per_second": 79.161, | |
| "eval_steps_per_second": 10.468, | |
| "step": 22816 | |
| }, | |
| { | |
| "epoch": 369.0, | |
| "eval_loss": 11.081649780273438, | |
| "eval_runtime": 1.5214, | |
| "eval_samples_per_second": 79.533, | |
| "eval_steps_per_second": 10.517, | |
| "step": 22878 | |
| }, | |
| { | |
| "epoch": 370.0, | |
| "eval_loss": 11.051599502563477, | |
| "eval_runtime": 1.5213, | |
| "eval_samples_per_second": 79.538, | |
| "eval_steps_per_second": 10.517, | |
| "step": 22940 | |
| }, | |
| { | |
| "epoch": 370.97, | |
| "learning_rate": 1.2903225806451612e-08, | |
| "loss": 10.1217, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 371.0, | |
| "eval_loss": 11.067154884338379, | |
| "eval_runtime": 1.5291, | |
| "eval_samples_per_second": 79.131, | |
| "eval_steps_per_second": 10.464, | |
| "step": 23002 | |
| }, | |
| { | |
| "epoch": 372.0, | |
| "eval_loss": 10.966989517211914, | |
| "eval_runtime": 1.5272, | |
| "eval_samples_per_second": 79.228, | |
| "eval_steps_per_second": 10.476, | |
| "step": 23064 | |
| }, | |
| { | |
| "epoch": 373.0, | |
| "eval_loss": 11.200106620788574, | |
| "eval_runtime": 1.5333, | |
| "eval_samples_per_second": 78.916, | |
| "eval_steps_per_second": 10.435, | |
| "step": 23126 | |
| }, | |
| { | |
| "epoch": 374.0, | |
| "eval_loss": 11.017847061157227, | |
| "eval_runtime": 1.534, | |
| "eval_samples_per_second": 78.88, | |
| "eval_steps_per_second": 10.43, | |
| "step": 23188 | |
| }, | |
| { | |
| "epoch": 375.0, | |
| "eval_loss": 10.895892143249512, | |
| "eval_runtime": 1.5401, | |
| "eval_samples_per_second": 78.565, | |
| "eval_steps_per_second": 10.389, | |
| "step": 23250 | |
| }, | |
| { | |
| "epoch": 376.0, | |
| "eval_loss": 11.167061805725098, | |
| "eval_runtime": 1.528, | |
| "eval_samples_per_second": 79.188, | |
| "eval_steps_per_second": 10.471, | |
| "step": 23312 | |
| }, | |
| { | |
| "epoch": 377.0, | |
| "eval_loss": 11.18134593963623, | |
| "eval_runtime": 1.5229, | |
| "eval_samples_per_second": 79.454, | |
| "eval_steps_per_second": 10.506, | |
| "step": 23374 | |
| }, | |
| { | |
| "epoch": 378.0, | |
| "eval_loss": 11.156011581420898, | |
| "eval_runtime": 1.5299, | |
| "eval_samples_per_second": 79.09, | |
| "eval_steps_per_second": 10.458, | |
| "step": 23436 | |
| }, | |
| { | |
| "epoch": 379.0, | |
| "eval_loss": 11.084783554077148, | |
| "eval_runtime": 1.5211, | |
| "eval_samples_per_second": 79.545, | |
| "eval_steps_per_second": 10.518, | |
| "step": 23498 | |
| }, | |
| { | |
| "epoch": 379.03, | |
| "learning_rate": 1.2096774193548386e-08, | |
| "loss": 10.1098, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 380.0, | |
| "eval_loss": 11.04666805267334, | |
| "eval_runtime": 1.5221, | |
| "eval_samples_per_second": 79.493, | |
| "eval_steps_per_second": 10.512, | |
| "step": 23560 | |
| }, | |
| { | |
| "epoch": 381.0, | |
| "eval_loss": 11.148889541625977, | |
| "eval_runtime": 1.548, | |
| "eval_samples_per_second": 78.166, | |
| "eval_steps_per_second": 10.336, | |
| "step": 23622 | |
| }, | |
| { | |
| "epoch": 382.0, | |
| "eval_loss": 11.006622314453125, | |
| "eval_runtime": 1.5259, | |
| "eval_samples_per_second": 79.298, | |
| "eval_steps_per_second": 10.486, | |
| "step": 23684 | |
| }, | |
| { | |
| "epoch": 383.0, | |
| "eval_loss": 10.977095603942871, | |
| "eval_runtime": 1.5282, | |
| "eval_samples_per_second": 79.178, | |
| "eval_steps_per_second": 10.47, | |
| "step": 23746 | |
| }, | |
| { | |
| "epoch": 384.0, | |
| "eval_loss": 10.895329475402832, | |
| "eval_runtime": 1.5304, | |
| "eval_samples_per_second": 79.066, | |
| "eval_steps_per_second": 10.455, | |
| "step": 23808 | |
| }, | |
| { | |
| "epoch": 385.0, | |
| "eval_loss": 10.903176307678223, | |
| "eval_runtime": 1.5397, | |
| "eval_samples_per_second": 78.586, | |
| "eval_steps_per_second": 10.392, | |
| "step": 23870 | |
| }, | |
| { | |
| "epoch": 386.0, | |
| "eval_loss": 10.892215728759766, | |
| "eval_runtime": 1.5258, | |
| "eval_samples_per_second": 79.302, | |
| "eval_steps_per_second": 10.486, | |
| "step": 23932 | |
| }, | |
| { | |
| "epoch": 387.0, | |
| "eval_loss": 10.993136405944824, | |
| "eval_runtime": 1.5252, | |
| "eval_samples_per_second": 79.332, | |
| "eval_steps_per_second": 10.49, | |
| "step": 23994 | |
| }, | |
| { | |
| "epoch": 387.1, | |
| "learning_rate": 1.129032258064516e-08, | |
| "loss": 10.1262, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 388.0, | |
| "eval_loss": 11.127774238586426, | |
| "eval_runtime": 1.561, | |
| "eval_samples_per_second": 77.516, | |
| "eval_steps_per_second": 10.25, | |
| "step": 24056 | |
| }, | |
| { | |
| "epoch": 389.0, | |
| "eval_loss": 11.101140022277832, | |
| "eval_runtime": 1.534, | |
| "eval_samples_per_second": 78.877, | |
| "eval_steps_per_second": 10.43, | |
| "step": 24118 | |
| }, | |
| { | |
| "epoch": 390.0, | |
| "eval_loss": 11.042156219482422, | |
| "eval_runtime": 1.5371, | |
| "eval_samples_per_second": 78.721, | |
| "eval_steps_per_second": 10.409, | |
| "step": 24180 | |
| }, | |
| { | |
| "epoch": 391.0, | |
| "eval_loss": 11.063082695007324, | |
| "eval_runtime": 1.5258, | |
| "eval_samples_per_second": 79.304, | |
| "eval_steps_per_second": 10.487, | |
| "step": 24242 | |
| }, | |
| { | |
| "epoch": 392.0, | |
| "eval_loss": 10.970271110534668, | |
| "eval_runtime": 1.5279, | |
| "eval_samples_per_second": 79.192, | |
| "eval_steps_per_second": 10.472, | |
| "step": 24304 | |
| }, | |
| { | |
| "epoch": 393.0, | |
| "eval_loss": 10.953568458557129, | |
| "eval_runtime": 1.5296, | |
| "eval_samples_per_second": 79.105, | |
| "eval_steps_per_second": 10.46, | |
| "step": 24366 | |
| }, | |
| { | |
| "epoch": 394.0, | |
| "eval_loss": 11.076667785644531, | |
| "eval_runtime": 1.5221, | |
| "eval_samples_per_second": 79.496, | |
| "eval_steps_per_second": 10.512, | |
| "step": 24428 | |
| }, | |
| { | |
| "epoch": 395.0, | |
| "eval_loss": 10.922811508178711, | |
| "eval_runtime": 1.5248, | |
| "eval_samples_per_second": 79.353, | |
| "eval_steps_per_second": 10.493, | |
| "step": 24490 | |
| }, | |
| { | |
| "epoch": 395.16, | |
| "learning_rate": 1.0483870967741935e-08, | |
| "loss": 10.1038, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 396.0, | |
| "eval_loss": 11.142001152038574, | |
| "eval_runtime": 1.5327, | |
| "eval_samples_per_second": 78.944, | |
| "eval_steps_per_second": 10.439, | |
| "step": 24552 | |
| }, | |
| { | |
| "epoch": 397.0, | |
| "eval_loss": 11.145098686218262, | |
| "eval_runtime": 1.5305, | |
| "eval_samples_per_second": 79.06, | |
| "eval_steps_per_second": 10.454, | |
| "step": 24614 | |
| }, | |
| { | |
| "epoch": 398.0, | |
| "eval_loss": 11.166299819946289, | |
| "eval_runtime": 1.5557, | |
| "eval_samples_per_second": 77.777, | |
| "eval_steps_per_second": 10.285, | |
| "step": 24676 | |
| }, | |
| { | |
| "epoch": 399.0, | |
| "eval_loss": 11.075632095336914, | |
| "eval_runtime": 1.5363, | |
| "eval_samples_per_second": 78.761, | |
| "eval_steps_per_second": 10.415, | |
| "step": 24738 | |
| }, | |
| { | |
| "epoch": 400.0, | |
| "eval_loss": 11.114908218383789, | |
| "eval_runtime": 1.5291, | |
| "eval_samples_per_second": 79.134, | |
| "eval_steps_per_second": 10.464, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 401.0, | |
| "eval_loss": 11.065619468688965, | |
| "eval_runtime": 1.5197, | |
| "eval_samples_per_second": 79.621, | |
| "eval_steps_per_second": 10.528, | |
| "step": 24862 | |
| }, | |
| { | |
| "epoch": 402.0, | |
| "eval_loss": 11.075223922729492, | |
| "eval_runtime": 1.5218, | |
| "eval_samples_per_second": 79.51, | |
| "eval_steps_per_second": 10.514, | |
| "step": 24924 | |
| }, | |
| { | |
| "epoch": 403.0, | |
| "eval_loss": 11.141654968261719, | |
| "eval_runtime": 1.5194, | |
| "eval_samples_per_second": 79.637, | |
| "eval_steps_per_second": 10.53, | |
| "step": 24986 | |
| }, | |
| { | |
| "epoch": 403.23, | |
| "learning_rate": 9.677419354838709e-09, | |
| "loss": 10.1159, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 404.0, | |
| "eval_loss": 10.96292495727539, | |
| "eval_runtime": 1.5308, | |
| "eval_samples_per_second": 79.043, | |
| "eval_steps_per_second": 10.452, | |
| "step": 25048 | |
| }, | |
| { | |
| "epoch": 405.0, | |
| "eval_loss": 10.93803596496582, | |
| "eval_runtime": 1.541, | |
| "eval_samples_per_second": 78.519, | |
| "eval_steps_per_second": 10.383, | |
| "step": 25110 | |
| }, | |
| { | |
| "epoch": 406.0, | |
| "eval_loss": 11.00986385345459, | |
| "eval_runtime": 1.5347, | |
| "eval_samples_per_second": 78.841, | |
| "eval_steps_per_second": 10.425, | |
| "step": 25172 | |
| }, | |
| { | |
| "epoch": 407.0, | |
| "eval_loss": 11.137813568115234, | |
| "eval_runtime": 1.5284, | |
| "eval_samples_per_second": 79.17, | |
| "eval_steps_per_second": 10.469, | |
| "step": 25234 | |
| }, | |
| { | |
| "epoch": 408.0, | |
| "eval_loss": 10.851067543029785, | |
| "eval_runtime": 1.5273, | |
| "eval_samples_per_second": 79.223, | |
| "eval_steps_per_second": 10.476, | |
| "step": 25296 | |
| }, | |
| { | |
| "epoch": 409.0, | |
| "eval_loss": 11.01217269897461, | |
| "eval_runtime": 1.5232, | |
| "eval_samples_per_second": 79.439, | |
| "eval_steps_per_second": 10.504, | |
| "step": 25358 | |
| }, | |
| { | |
| "epoch": 410.0, | |
| "eval_loss": 11.01009750366211, | |
| "eval_runtime": 1.5196, | |
| "eval_samples_per_second": 79.627, | |
| "eval_steps_per_second": 10.529, | |
| "step": 25420 | |
| }, | |
| { | |
| "epoch": 411.0, | |
| "eval_loss": 10.873472213745117, | |
| "eval_runtime": 1.5201, | |
| "eval_samples_per_second": 79.6, | |
| "eval_steps_per_second": 10.526, | |
| "step": 25482 | |
| }, | |
| { | |
| "epoch": 411.29, | |
| "learning_rate": 8.870967741935485e-09, | |
| "loss": 10.0848, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 412.0, | |
| "eval_loss": 11.008174896240234, | |
| "eval_runtime": 1.5272, | |
| "eval_samples_per_second": 79.232, | |
| "eval_steps_per_second": 10.477, | |
| "step": 25544 | |
| }, | |
| { | |
| "epoch": 413.0, | |
| "eval_loss": 11.012767791748047, | |
| "eval_runtime": 1.5353, | |
| "eval_samples_per_second": 78.81, | |
| "eval_steps_per_second": 10.421, | |
| "step": 25606 | |
| }, | |
| { | |
| "epoch": 414.0, | |
| "eval_loss": 11.137340545654297, | |
| "eval_runtime": 1.5293, | |
| "eval_samples_per_second": 79.119, | |
| "eval_steps_per_second": 10.462, | |
| "step": 25668 | |
| }, | |
| { | |
| "epoch": 415.0, | |
| "eval_loss": 11.137984275817871, | |
| "eval_runtime": 1.5285, | |
| "eval_samples_per_second": 79.163, | |
| "eval_steps_per_second": 10.468, | |
| "step": 25730 | |
| }, | |
| { | |
| "epoch": 416.0, | |
| "eval_loss": 10.93875503540039, | |
| "eval_runtime": 1.5284, | |
| "eval_samples_per_second": 79.167, | |
| "eval_steps_per_second": 10.468, | |
| "step": 25792 | |
| }, | |
| { | |
| "epoch": 417.0, | |
| "eval_loss": 10.93083381652832, | |
| "eval_runtime": 1.524, | |
| "eval_samples_per_second": 79.396, | |
| "eval_steps_per_second": 10.499, | |
| "step": 25854 | |
| }, | |
| { | |
| "epoch": 418.0, | |
| "eval_loss": 11.082795143127441, | |
| "eval_runtime": 1.5216, | |
| "eval_samples_per_second": 79.523, | |
| "eval_steps_per_second": 10.515, | |
| "step": 25916 | |
| }, | |
| { | |
| "epoch": 419.0, | |
| "eval_loss": 11.256653785705566, | |
| "eval_runtime": 1.5222, | |
| "eval_samples_per_second": 79.488, | |
| "eval_steps_per_second": 10.511, | |
| "step": 25978 | |
| }, | |
| { | |
| "epoch": 419.35, | |
| "learning_rate": 8.064516129032257e-09, | |
| "loss": 10.0923, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 420.0, | |
| "eval_loss": 11.040205955505371, | |
| "eval_runtime": 1.5329, | |
| "eval_samples_per_second": 78.935, | |
| "eval_steps_per_second": 10.438, | |
| "step": 26040 | |
| }, | |
| { | |
| "epoch": 421.0, | |
| "eval_loss": 11.028487205505371, | |
| "eval_runtime": 1.541, | |
| "eval_samples_per_second": 78.518, | |
| "eval_steps_per_second": 10.383, | |
| "step": 26102 | |
| }, | |
| { | |
| "epoch": 422.0, | |
| "eval_loss": 11.002765655517578, | |
| "eval_runtime": 1.5314, | |
| "eval_samples_per_second": 79.012, | |
| "eval_steps_per_second": 10.448, | |
| "step": 26164 | |
| }, | |
| { | |
| "epoch": 423.0, | |
| "eval_loss": 10.978453636169434, | |
| "eval_runtime": 1.5289, | |
| "eval_samples_per_second": 79.144, | |
| "eval_steps_per_second": 10.465, | |
| "step": 26226 | |
| }, | |
| { | |
| "epoch": 424.0, | |
| "eval_loss": 11.046844482421875, | |
| "eval_runtime": 1.5378, | |
| "eval_samples_per_second": 78.685, | |
| "eval_steps_per_second": 10.405, | |
| "step": 26288 | |
| }, | |
| { | |
| "epoch": 425.0, | |
| "eval_loss": 11.059452056884766, | |
| "eval_runtime": 1.523, | |
| "eval_samples_per_second": 79.446, | |
| "eval_steps_per_second": 10.505, | |
| "step": 26350 | |
| }, | |
| { | |
| "epoch": 426.0, | |
| "eval_loss": 11.098678588867188, | |
| "eval_runtime": 1.521, | |
| "eval_samples_per_second": 79.551, | |
| "eval_steps_per_second": 10.519, | |
| "step": 26412 | |
| }, | |
| { | |
| "epoch": 427.0, | |
| "eval_loss": 10.876445770263672, | |
| "eval_runtime": 1.5209, | |
| "eval_samples_per_second": 79.556, | |
| "eval_steps_per_second": 10.52, | |
| "step": 26474 | |
| }, | |
| { | |
| "epoch": 427.42, | |
| "learning_rate": 7.258064516129032e-09, | |
| "loss": 10.1, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 428.0, | |
| "eval_loss": 10.964273452758789, | |
| "eval_runtime": 1.5385, | |
| "eval_samples_per_second": 78.65, | |
| "eval_steps_per_second": 10.4, | |
| "step": 26536 | |
| }, | |
| { | |
| "epoch": 429.0, | |
| "eval_loss": 10.986409187316895, | |
| "eval_runtime": 1.5468, | |
| "eval_samples_per_second": 78.225, | |
| "eval_steps_per_second": 10.344, | |
| "step": 26598 | |
| }, | |
| { | |
| "epoch": 430.0, | |
| "eval_loss": 11.001900672912598, | |
| "eval_runtime": 1.529, | |
| "eval_samples_per_second": 79.135, | |
| "eval_steps_per_second": 10.464, | |
| "step": 26660 | |
| }, | |
| { | |
| "epoch": 431.0, | |
| "eval_loss": 11.103708267211914, | |
| "eval_runtime": 1.5277, | |
| "eval_samples_per_second": 79.205, | |
| "eval_steps_per_second": 10.473, | |
| "step": 26722 | |
| }, | |
| { | |
| "epoch": 432.0, | |
| "eval_loss": 10.956040382385254, | |
| "eval_runtime": 1.5283, | |
| "eval_samples_per_second": 79.174, | |
| "eval_steps_per_second": 10.469, | |
| "step": 26784 | |
| }, | |
| { | |
| "epoch": 433.0, | |
| "eval_loss": 10.98078727722168, | |
| "eval_runtime": 1.545, | |
| "eval_samples_per_second": 78.316, | |
| "eval_steps_per_second": 10.356, | |
| "step": 26846 | |
| }, | |
| { | |
| "epoch": 434.0, | |
| "eval_loss": 11.044650077819824, | |
| "eval_runtime": 1.5194, | |
| "eval_samples_per_second": 79.639, | |
| "eval_steps_per_second": 10.531, | |
| "step": 26908 | |
| }, | |
| { | |
| "epoch": 435.0, | |
| "eval_loss": 10.943286895751953, | |
| "eval_runtime": 1.5208, | |
| "eval_samples_per_second": 79.563, | |
| "eval_steps_per_second": 10.521, | |
| "step": 26970 | |
| }, | |
| { | |
| "epoch": 435.48, | |
| "learning_rate": 6.451612903225806e-09, | |
| "loss": 10.0944, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 436.0, | |
| "eval_loss": 11.040830612182617, | |
| "eval_runtime": 1.5264, | |
| "eval_samples_per_second": 79.272, | |
| "eval_steps_per_second": 10.482, | |
| "step": 27032 | |
| }, | |
| { | |
| "epoch": 437.0, | |
| "eval_loss": 10.829809188842773, | |
| "eval_runtime": 1.531, | |
| "eval_samples_per_second": 79.032, | |
| "eval_steps_per_second": 10.45, | |
| "step": 27094 | |
| }, | |
| { | |
| "epoch": 438.0, | |
| "eval_loss": 11.169089317321777, | |
| "eval_runtime": 1.5289, | |
| "eval_samples_per_second": 79.142, | |
| "eval_steps_per_second": 10.465, | |
| "step": 27156 | |
| }, | |
| { | |
| "epoch": 439.0, | |
| "eval_loss": 11.029060363769531, | |
| "eval_runtime": 1.5418, | |
| "eval_samples_per_second": 78.481, | |
| "eval_steps_per_second": 10.378, | |
| "step": 27218 | |
| }, | |
| { | |
| "epoch": 440.0, | |
| "eval_loss": 10.99075984954834, | |
| "eval_runtime": 1.5372, | |
| "eval_samples_per_second": 78.712, | |
| "eval_steps_per_second": 10.408, | |
| "step": 27280 | |
| }, | |
| { | |
| "epoch": 441.0, | |
| "eval_loss": 10.889039039611816, | |
| "eval_runtime": 1.5276, | |
| "eval_samples_per_second": 79.211, | |
| "eval_steps_per_second": 10.474, | |
| "step": 27342 | |
| }, | |
| { | |
| "epoch": 442.0, | |
| "eval_loss": 11.024713516235352, | |
| "eval_runtime": 1.5184, | |
| "eval_samples_per_second": 79.688, | |
| "eval_steps_per_second": 10.537, | |
| "step": 27404 | |
| }, | |
| { | |
| "epoch": 443.0, | |
| "eval_loss": 11.10232925415039, | |
| "eval_runtime": 1.5225, | |
| "eval_samples_per_second": 79.473, | |
| "eval_steps_per_second": 10.509, | |
| "step": 27466 | |
| }, | |
| { | |
| "epoch": 443.55, | |
| "learning_rate": 5.64516129032258e-09, | |
| "loss": 10.0965, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 444.0, | |
| "eval_loss": 11.125219345092773, | |
| "eval_runtime": 1.5507, | |
| "eval_samples_per_second": 78.027, | |
| "eval_steps_per_second": 10.318, | |
| "step": 27528 | |
| }, | |
| { | |
| "epoch": 445.0, | |
| "eval_loss": 11.037354469299316, | |
| "eval_runtime": 1.5307, | |
| "eval_samples_per_second": 79.048, | |
| "eval_steps_per_second": 10.453, | |
| "step": 27590 | |
| }, | |
| { | |
| "epoch": 446.0, | |
| "eval_loss": 11.032657623291016, | |
| "eval_runtime": 1.5272, | |
| "eval_samples_per_second": 79.231, | |
| "eval_steps_per_second": 10.477, | |
| "step": 27652 | |
| }, | |
| { | |
| "epoch": 447.0, | |
| "eval_loss": 10.83928394317627, | |
| "eval_runtime": 1.5299, | |
| "eval_samples_per_second": 79.09, | |
| "eval_steps_per_second": 10.458, | |
| "step": 27714 | |
| }, | |
| { | |
| "epoch": 448.0, | |
| "eval_loss": 10.952762603759766, | |
| "eval_runtime": 1.5436, | |
| "eval_samples_per_second": 78.39, | |
| "eval_steps_per_second": 10.366, | |
| "step": 27776 | |
| }, | |
| { | |
| "epoch": 449.0, | |
| "eval_loss": 11.012396812438965, | |
| "eval_runtime": 1.5272, | |
| "eval_samples_per_second": 79.231, | |
| "eval_steps_per_second": 10.477, | |
| "step": 27838 | |
| }, | |
| { | |
| "epoch": 450.0, | |
| "eval_loss": 10.84825611114502, | |
| "eval_runtime": 1.5273, | |
| "eval_samples_per_second": 79.226, | |
| "eval_steps_per_second": 10.476, | |
| "step": 27900 | |
| }, | |
| { | |
| "epoch": 451.0, | |
| "eval_loss": 10.897711753845215, | |
| "eval_runtime": 1.527, | |
| "eval_samples_per_second": 79.241, | |
| "eval_steps_per_second": 10.478, | |
| "step": 27962 | |
| }, | |
| { | |
| "epoch": 451.61, | |
| "learning_rate": 4.8387096774193544e-09, | |
| "loss": 10.1023, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 452.0, | |
| "eval_loss": 11.033035278320312, | |
| "eval_runtime": 1.5393, | |
| "eval_samples_per_second": 78.607, | |
| "eval_steps_per_second": 10.394, | |
| "step": 28024 | |
| }, | |
| { | |
| "epoch": 453.0, | |
| "eval_loss": 10.92820930480957, | |
| "eval_runtime": 1.5378, | |
| "eval_samples_per_second": 78.682, | |
| "eval_steps_per_second": 10.404, | |
| "step": 28086 | |
| }, | |
| { | |
| "epoch": 454.0, | |
| "eval_loss": 11.028227806091309, | |
| "eval_runtime": 1.5335, | |
| "eval_samples_per_second": 78.905, | |
| "eval_steps_per_second": 10.434, | |
| "step": 28148 | |
| }, | |
| { | |
| "epoch": 455.0, | |
| "eval_loss": 10.94653034210205, | |
| "eval_runtime": 1.5323, | |
| "eval_samples_per_second": 78.968, | |
| "eval_steps_per_second": 10.442, | |
| "step": 28210 | |
| }, | |
| { | |
| "epoch": 456.0, | |
| "eval_loss": 10.933752059936523, | |
| "eval_runtime": 1.542, | |
| "eval_samples_per_second": 78.468, | |
| "eval_steps_per_second": 10.376, | |
| "step": 28272 | |
| }, | |
| { | |
| "epoch": 457.0, | |
| "eval_loss": 10.998141288757324, | |
| "eval_runtime": 1.5243, | |
| "eval_samples_per_second": 79.38, | |
| "eval_steps_per_second": 10.496, | |
| "step": 28334 | |
| }, | |
| { | |
| "epoch": 458.0, | |
| "eval_loss": 11.004565238952637, | |
| "eval_runtime": 1.5184, | |
| "eval_samples_per_second": 79.69, | |
| "eval_steps_per_second": 10.537, | |
| "step": 28396 | |
| }, | |
| { | |
| "epoch": 459.0, | |
| "eval_loss": 11.058143615722656, | |
| "eval_runtime": 1.5198, | |
| "eval_samples_per_second": 79.618, | |
| "eval_steps_per_second": 10.528, | |
| "step": 28458 | |
| }, | |
| { | |
| "epoch": 459.68, | |
| "learning_rate": 4.0322580645161286e-09, | |
| "loss": 10.0928, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 460.0, | |
| "eval_loss": 11.069917678833008, | |
| "eval_runtime": 1.5309, | |
| "eval_samples_per_second": 79.038, | |
| "eval_steps_per_second": 10.451, | |
| "step": 28520 | |
| }, | |
| { | |
| "epoch": 461.0, | |
| "eval_loss": 11.008392333984375, | |
| "eval_runtime": 1.5319, | |
| "eval_samples_per_second": 78.986, | |
| "eval_steps_per_second": 10.444, | |
| "step": 28582 | |
| }, | |
| { | |
| "epoch": 462.0, | |
| "eval_loss": 10.968953132629395, | |
| "eval_runtime": 1.5279, | |
| "eval_samples_per_second": 79.192, | |
| "eval_steps_per_second": 10.472, | |
| "step": 28644 | |
| }, | |
| { | |
| "epoch": 463.0, | |
| "eval_loss": 11.05852222442627, | |
| "eval_runtime": 1.5388, | |
| "eval_samples_per_second": 78.635, | |
| "eval_steps_per_second": 10.398, | |
| "step": 28706 | |
| }, | |
| { | |
| "epoch": 464.0, | |
| "eval_loss": 11.158599853515625, | |
| "eval_runtime": 1.5307, | |
| "eval_samples_per_second": 79.048, | |
| "eval_steps_per_second": 10.453, | |
| "step": 28768 | |
| }, | |
| { | |
| "epoch": 465.0, | |
| "eval_loss": 11.008247375488281, | |
| "eval_runtime": 1.5274, | |
| "eval_samples_per_second": 79.218, | |
| "eval_steps_per_second": 10.475, | |
| "step": 28830 | |
| }, | |
| { | |
| "epoch": 466.0, | |
| "eval_loss": 11.049484252929688, | |
| "eval_runtime": 1.5195, | |
| "eval_samples_per_second": 79.634, | |
| "eval_steps_per_second": 10.53, | |
| "step": 28892 | |
| }, | |
| { | |
| "epoch": 467.0, | |
| "eval_loss": 10.853099822998047, | |
| "eval_runtime": 1.5203, | |
| "eval_samples_per_second": 79.589, | |
| "eval_steps_per_second": 10.524, | |
| "step": 28954 | |
| }, | |
| { | |
| "epoch": 467.74, | |
| "learning_rate": 3.225806451612903e-09, | |
| "loss": 10.0925, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 468.0, | |
| "eval_loss": 10.858702659606934, | |
| "eval_runtime": 1.5294, | |
| "eval_samples_per_second": 79.118, | |
| "eval_steps_per_second": 10.462, | |
| "step": 29016 | |
| }, | |
| { | |
| "epoch": 469.0, | |
| "eval_loss": 11.159988403320312, | |
| "eval_runtime": 1.5283, | |
| "eval_samples_per_second": 79.173, | |
| "eval_steps_per_second": 10.469, | |
| "step": 29078 | |
| }, | |
| { | |
| "epoch": 470.0, | |
| "eval_loss": 10.992026329040527, | |
| "eval_runtime": 1.5307, | |
| "eval_samples_per_second": 79.05, | |
| "eval_steps_per_second": 10.453, | |
| "step": 29140 | |
| }, | |
| { | |
| "epoch": 471.0, | |
| "eval_loss": 11.020038604736328, | |
| "eval_runtime": 1.5472, | |
| "eval_samples_per_second": 78.208, | |
| "eval_steps_per_second": 10.342, | |
| "step": 29202 | |
| }, | |
| { | |
| "epoch": 472.0, | |
| "eval_loss": 11.056954383850098, | |
| "eval_runtime": 1.5255, | |
| "eval_samples_per_second": 79.316, | |
| "eval_steps_per_second": 10.488, | |
| "step": 29264 | |
| }, | |
| { | |
| "epoch": 473.0, | |
| "eval_loss": 10.931620597839355, | |
| "eval_runtime": 1.5261, | |
| "eval_samples_per_second": 79.287, | |
| "eval_steps_per_second": 10.484, | |
| "step": 29326 | |
| }, | |
| { | |
| "epoch": 474.0, | |
| "eval_loss": 11.059667587280273, | |
| "eval_runtime": 1.5202, | |
| "eval_samples_per_second": 79.597, | |
| "eval_steps_per_second": 10.525, | |
| "step": 29388 | |
| }, | |
| { | |
| "epoch": 475.0, | |
| "eval_loss": 10.994057655334473, | |
| "eval_runtime": 1.5275, | |
| "eval_samples_per_second": 79.216, | |
| "eval_steps_per_second": 10.475, | |
| "step": 29450 | |
| }, | |
| { | |
| "epoch": 475.81, | |
| "learning_rate": 2.4193548387096772e-09, | |
| "loss": 10.0956, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 476.0, | |
| "eval_loss": 11.07702350616455, | |
| "eval_runtime": 1.5276, | |
| "eval_samples_per_second": 79.21, | |
| "eval_steps_per_second": 10.474, | |
| "step": 29512 | |
| }, | |
| { | |
| "epoch": 477.0, | |
| "eval_loss": 11.006840705871582, | |
| "eval_runtime": 1.5298, | |
| "eval_samples_per_second": 79.098, | |
| "eval_steps_per_second": 10.459, | |
| "step": 29574 | |
| }, | |
| { | |
| "epoch": 478.0, | |
| "eval_loss": 10.907955169677734, | |
| "eval_runtime": 1.5329, | |
| "eval_samples_per_second": 78.936, | |
| "eval_steps_per_second": 10.438, | |
| "step": 29636 | |
| }, | |
| { | |
| "epoch": 479.0, | |
| "eval_loss": 10.865551948547363, | |
| "eval_runtime": 1.5272, | |
| "eval_samples_per_second": 79.23, | |
| "eval_steps_per_second": 10.477, | |
| "step": 29698 | |
| }, | |
| { | |
| "epoch": 480.0, | |
| "eval_loss": 11.035009384155273, | |
| "eval_runtime": 1.5396, | |
| "eval_samples_per_second": 78.594, | |
| "eval_steps_per_second": 10.393, | |
| "step": 29760 | |
| }, | |
| { | |
| "epoch": 481.0, | |
| "eval_loss": 11.06725788116455, | |
| "eval_runtime": 1.5306, | |
| "eval_samples_per_second": 79.056, | |
| "eval_steps_per_second": 10.454, | |
| "step": 29822 | |
| }, | |
| { | |
| "epoch": 482.0, | |
| "eval_loss": 11.034406661987305, | |
| "eval_runtime": 1.5223, | |
| "eval_samples_per_second": 79.487, | |
| "eval_steps_per_second": 10.511, | |
| "step": 29884 | |
| }, | |
| { | |
| "epoch": 483.0, | |
| "eval_loss": 10.981928825378418, | |
| "eval_runtime": 1.5189, | |
| "eval_samples_per_second": 79.661, | |
| "eval_steps_per_second": 10.534, | |
| "step": 29946 | |
| }, | |
| { | |
| "epoch": 483.87, | |
| "learning_rate": 1.6129032258064515e-09, | |
| "loss": 10.112, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 484.0, | |
| "eval_loss": 11.066652297973633, | |
| "eval_runtime": 1.5282, | |
| "eval_samples_per_second": 79.179, | |
| "eval_steps_per_second": 10.47, | |
| "step": 30008 | |
| }, | |
| { | |
| "epoch": 485.0, | |
| "eval_loss": 10.937297821044922, | |
| "eval_runtime": 1.5264, | |
| "eval_samples_per_second": 79.27, | |
| "eval_steps_per_second": 10.482, | |
| "step": 30070 | |
| }, | |
| { | |
| "epoch": 486.0, | |
| "eval_loss": 10.977715492248535, | |
| "eval_runtime": 1.537, | |
| "eval_samples_per_second": 78.724, | |
| "eval_steps_per_second": 10.41, | |
| "step": 30132 | |
| }, | |
| { | |
| "epoch": 487.0, | |
| "eval_loss": 10.983244895935059, | |
| "eval_runtime": 1.5322, | |
| "eval_samples_per_second": 78.969, | |
| "eval_steps_per_second": 10.442, | |
| "step": 30194 | |
| }, | |
| { | |
| "epoch": 488.0, | |
| "eval_loss": 11.142578125, | |
| "eval_runtime": 1.532, | |
| "eval_samples_per_second": 78.984, | |
| "eval_steps_per_second": 10.444, | |
| "step": 30256 | |
| }, | |
| { | |
| "epoch": 489.0, | |
| "eval_loss": 11.0770845413208, | |
| "eval_runtime": 1.527, | |
| "eval_samples_per_second": 79.241, | |
| "eval_steps_per_second": 10.478, | |
| "step": 30318 | |
| }, | |
| { | |
| "epoch": 490.0, | |
| "eval_loss": 10.951501846313477, | |
| "eval_runtime": 1.5214, | |
| "eval_samples_per_second": 79.534, | |
| "eval_steps_per_second": 10.517, | |
| "step": 30380 | |
| }, | |
| { | |
| "epoch": 491.0, | |
| "eval_loss": 10.961722373962402, | |
| "eval_runtime": 1.5203, | |
| "eval_samples_per_second": 79.59, | |
| "eval_steps_per_second": 10.524, | |
| "step": 30442 | |
| }, | |
| { | |
| "epoch": 491.94, | |
| "learning_rate": 8.064516129032258e-10, | |
| "loss": 10.1007, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 492.0, | |
| "eval_loss": 10.952271461486816, | |
| "eval_runtime": 1.528, | |
| "eval_samples_per_second": 79.187, | |
| "eval_steps_per_second": 10.471, | |
| "step": 30504 | |
| }, | |
| { | |
| "epoch": 493.0, | |
| "eval_loss": 10.87966537475586, | |
| "eval_runtime": 1.5563, | |
| "eval_samples_per_second": 77.748, | |
| "eval_steps_per_second": 10.281, | |
| "step": 30566 | |
| }, | |
| { | |
| "epoch": 494.0, | |
| "eval_loss": 10.981633186340332, | |
| "eval_runtime": 1.533, | |
| "eval_samples_per_second": 78.933, | |
| "eval_steps_per_second": 10.437, | |
| "step": 30628 | |
| }, | |
| { | |
| "epoch": 495.0, | |
| "eval_loss": 10.952598571777344, | |
| "eval_runtime": 1.5283, | |
| "eval_samples_per_second": 79.171, | |
| "eval_steps_per_second": 10.469, | |
| "step": 30690 | |
| }, | |
| { | |
| "epoch": 496.0, | |
| "eval_loss": 11.00783920288086, | |
| "eval_runtime": 1.5275, | |
| "eval_samples_per_second": 79.212, | |
| "eval_steps_per_second": 10.474, | |
| "step": 30752 | |
| }, | |
| { | |
| "epoch": 497.0, | |
| "eval_loss": 11.156755447387695, | |
| "eval_runtime": 1.536, | |
| "eval_samples_per_second": 78.778, | |
| "eval_steps_per_second": 10.417, | |
| "step": 30814 | |
| }, | |
| { | |
| "epoch": 498.0, | |
| "eval_loss": 10.986204147338867, | |
| "eval_runtime": 1.521, | |
| "eval_samples_per_second": 79.554, | |
| "eval_steps_per_second": 10.519, | |
| "step": 30876 | |
| }, | |
| { | |
| "epoch": 499.0, | |
| "eval_loss": 11.053728103637695, | |
| "eval_runtime": 1.5217, | |
| "eval_samples_per_second": 79.518, | |
| "eval_steps_per_second": 10.515, | |
| "step": 30938 | |
| }, | |
| { | |
| "epoch": 500.0, | |
| "learning_rate": 0.0, | |
| "loss": 10.0953, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 500.0, | |
| "eval_loss": 11.041763305664062, | |
| "eval_runtime": 1.5553, | |
| "eval_samples_per_second": 77.797, | |
| "eval_steps_per_second": 10.287, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 500.0, | |
| "step": 31000, | |
| "total_flos": 1.6202616878592e+16, | |
| "train_loss": 11.618560483870969, | |
| "train_runtime": 17098.8381, | |
| "train_samples_per_second": 14.358, | |
| "train_steps_per_second": 1.813 | |
| } | |
| ], | |
| "max_steps": 31000, | |
| "num_train_epochs": 500, | |
| "total_flos": 1.6202616878592e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |