{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.3326133909287257, "eval_steps": 500, "global_step": 308, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "grad_norm": 2.1186674794018874, "learning_rate": 0.0, "loss": 0.762, "step": 1 }, { "epoch": 0.0, "grad_norm": 1.5049784219658717, "learning_rate": 6.020599913279623e-06, "loss": 0.6777, "step": 2 }, { "epoch": 0.0, "grad_norm": 1.5560748940122215, "learning_rate": 9.542425094393249e-06, "loss": 0.6918, "step": 3 }, { "epoch": 0.0, "grad_norm": 1.2511342624485708, "learning_rate": 1.2041199826559246e-05, "loss": 0.6375, "step": 4 }, { "epoch": 0.01, "grad_norm": 0.8125037046494756, "learning_rate": 1.3979400086720374e-05, "loss": 0.5701, "step": 5 }, { "epoch": 0.01, "grad_norm": 1.8434809472279226, "learning_rate": 1.5563025007672873e-05, "loss": 0.6514, "step": 6 }, { "epoch": 0.01, "grad_norm": 1.5763574654037487, "learning_rate": 1.6901960800285137e-05, "loss": 0.6088, "step": 7 }, { "epoch": 0.01, "grad_norm": 0.9845853308357277, "learning_rate": 1.806179973983887e-05, "loss": 0.5696, "step": 8 }, { "epoch": 0.01, "grad_norm": 0.9825649271082681, "learning_rate": 1.9084850188786497e-05, "loss": 0.5759, "step": 9 }, { "epoch": 0.01, "grad_norm": 0.8132097372225459, "learning_rate": 1.9999999999999998e-05, "loss": 0.5595, "step": 10 }, { "epoch": 0.01, "grad_norm": 1.319819296186598, "learning_rate": 2e-05, "loss": 0.5612, "step": 11 }, { "epoch": 0.01, "grad_norm": 0.8912280533965539, "learning_rate": 1.997816593886463e-05, "loss": 0.5426, "step": 12 }, { "epoch": 0.01, "grad_norm": 0.5696945500526134, "learning_rate": 1.995633187772926e-05, "loss": 0.5468, "step": 13 }, { "epoch": 0.02, "grad_norm": 1.292370522394383, "learning_rate": 1.993449781659389e-05, "loss": 0.5656, "step": 14 }, { "epoch": 0.02, "grad_norm": 0.5804040423506179, "learning_rate": 1.9912663755458517e-05, "loss": 0.5344, "step": 15 }, { "epoch": 0.02, "grad_norm": 1.0611806419542331, "learning_rate": 1.9890829694323144e-05, "loss": 0.5549, "step": 16 }, { "epoch": 0.02, "grad_norm": 0.5684069321913675, "learning_rate": 1.9868995633187776e-05, "loss": 0.5271, "step": 17 }, { "epoch": 0.02, "grad_norm": 0.8373449011138275, "learning_rate": 1.9847161572052404e-05, "loss": 0.5529, "step": 18 }, { "epoch": 0.02, "grad_norm": 0.6233456890603677, "learning_rate": 1.982532751091703e-05, "loss": 0.5498, "step": 19 }, { "epoch": 0.02, "grad_norm": 0.5892227390005575, "learning_rate": 1.980349344978166e-05, "loss": 0.5275, "step": 20 }, { "epoch": 0.02, "grad_norm": 0.5360881738197603, "learning_rate": 1.978165938864629e-05, "loss": 0.5459, "step": 21 }, { "epoch": 0.02, "grad_norm": 0.5577347292500204, "learning_rate": 1.975982532751092e-05, "loss": 0.533, "step": 22 }, { "epoch": 0.02, "grad_norm": 0.6845128012101741, "learning_rate": 1.9737991266375546e-05, "loss": 0.5496, "step": 23 }, { "epoch": 0.03, "grad_norm": 0.42643509104346145, "learning_rate": 1.9716157205240178e-05, "loss": 0.5112, "step": 24 }, { "epoch": 0.03, "grad_norm": 0.5149003630149541, "learning_rate": 1.9694323144104806e-05, "loss": 0.4802, "step": 25 }, { "epoch": 0.03, "grad_norm": 0.5163933779775761, "learning_rate": 1.9672489082969434e-05, "loss": 0.4877, "step": 26 }, { "epoch": 0.03, "grad_norm": 0.5207228225450767, "learning_rate": 1.965065502183406e-05, "loss": 0.5513, "step": 27 }, { "epoch": 0.03, "grad_norm": 0.4777497059069117, "learning_rate": 1.9628820960698693e-05, "loss": 0.532, "step": 28 }, { "epoch": 0.03, "grad_norm": 0.4794666570948325, "learning_rate": 1.960698689956332e-05, "loss": 0.5131, "step": 29 }, { "epoch": 0.03, "grad_norm": 0.4763149650448877, "learning_rate": 1.958515283842795e-05, "loss": 0.5193, "step": 30 }, { "epoch": 0.03, "grad_norm": 0.47104986565242496, "learning_rate": 1.9563318777292576e-05, "loss": 0.486, "step": 31 }, { "epoch": 0.03, "grad_norm": 0.44186554703050224, "learning_rate": 1.9541484716157208e-05, "loss": 0.5062, "step": 32 }, { "epoch": 0.04, "grad_norm": 0.5577375612757014, "learning_rate": 1.9519650655021836e-05, "loss": 0.5142, "step": 33 }, { "epoch": 0.04, "grad_norm": 0.4466514535111405, "learning_rate": 1.9497816593886463e-05, "loss": 0.5105, "step": 34 }, { "epoch": 0.04, "grad_norm": 0.4599752241697936, "learning_rate": 1.9475982532751095e-05, "loss": 0.5162, "step": 35 }, { "epoch": 0.04, "grad_norm": 0.44857709087715164, "learning_rate": 1.9454148471615723e-05, "loss": 0.5367, "step": 36 }, { "epoch": 0.04, "grad_norm": 0.41263712857832235, "learning_rate": 1.943231441048035e-05, "loss": 0.5027, "step": 37 }, { "epoch": 0.04, "grad_norm": 0.4521119533509463, "learning_rate": 1.941048034934498e-05, "loss": 0.5118, "step": 38 }, { "epoch": 0.04, "grad_norm": 0.4662594115199691, "learning_rate": 1.938864628820961e-05, "loss": 0.5002, "step": 39 }, { "epoch": 0.04, "grad_norm": 0.4446567090191832, "learning_rate": 1.9366812227074238e-05, "loss": 0.4839, "step": 40 }, { "epoch": 0.04, "grad_norm": 0.48251939475160455, "learning_rate": 1.9344978165938865e-05, "loss": 0.4929, "step": 41 }, { "epoch": 0.05, "grad_norm": 0.5540231836852683, "learning_rate": 1.9323144104803497e-05, "loss": 0.5149, "step": 42 }, { "epoch": 0.05, "grad_norm": 0.43090837602909066, "learning_rate": 1.9301310043668125e-05, "loss": 0.4932, "step": 43 }, { "epoch": 0.05, "grad_norm": 0.5198842620428126, "learning_rate": 1.9279475982532753e-05, "loss": 0.5107, "step": 44 }, { "epoch": 0.05, "grad_norm": 0.45405522592051506, "learning_rate": 1.925764192139738e-05, "loss": 0.5104, "step": 45 }, { "epoch": 0.05, "grad_norm": 0.4563593041116948, "learning_rate": 1.9235807860262012e-05, "loss": 0.4915, "step": 46 }, { "epoch": 0.05, "grad_norm": 0.4652171198209756, "learning_rate": 1.921397379912664e-05, "loss": 0.4773, "step": 47 }, { "epoch": 0.05, "grad_norm": 0.5075515663557973, "learning_rate": 1.9192139737991267e-05, "loss": 0.4918, "step": 48 }, { "epoch": 0.05, "grad_norm": 0.4876926842169163, "learning_rate": 1.9170305676855895e-05, "loss": 0.5052, "step": 49 }, { "epoch": 0.05, "grad_norm": 0.5657178444298022, "learning_rate": 1.9148471615720527e-05, "loss": 0.4877, "step": 50 }, { "epoch": 0.06, "grad_norm": 0.4224792144955699, "learning_rate": 1.9126637554585155e-05, "loss": 0.4844, "step": 51 }, { "epoch": 0.06, "grad_norm": 0.47793247734973493, "learning_rate": 1.9104803493449782e-05, "loss": 0.4946, "step": 52 }, { "epoch": 0.06, "grad_norm": 0.47163318223121553, "learning_rate": 1.908296943231441e-05, "loss": 0.4777, "step": 53 }, { "epoch": 0.06, "grad_norm": 0.4468196827989752, "learning_rate": 1.906113537117904e-05, "loss": 0.4912, "step": 54 }, { "epoch": 0.06, "grad_norm": 0.42476515311951113, "learning_rate": 1.903930131004367e-05, "loss": 0.4904, "step": 55 }, { "epoch": 0.06, "grad_norm": 0.4301592924354783, "learning_rate": 1.9017467248908297e-05, "loss": 0.4843, "step": 56 }, { "epoch": 0.06, "grad_norm": 0.45155900802054094, "learning_rate": 1.899563318777293e-05, "loss": 0.4794, "step": 57 }, { "epoch": 0.06, "grad_norm": 0.47557594241815077, "learning_rate": 1.8973799126637557e-05, "loss": 0.4915, "step": 58 }, { "epoch": 0.06, "grad_norm": 0.41055652523585845, "learning_rate": 1.8951965065502184e-05, "loss": 0.4761, "step": 59 }, { "epoch": 0.06, "grad_norm": 0.49673065092092256, "learning_rate": 1.8930131004366816e-05, "loss": 0.4876, "step": 60 }, { "epoch": 0.07, "grad_norm": 0.4658799523135649, "learning_rate": 1.8908296943231444e-05, "loss": 0.4953, "step": 61 }, { "epoch": 0.07, "grad_norm": 0.4184265481713602, "learning_rate": 1.888646288209607e-05, "loss": 0.4734, "step": 62 }, { "epoch": 0.07, "grad_norm": 0.5665612896476564, "learning_rate": 1.88646288209607e-05, "loss": 0.5221, "step": 63 }, { "epoch": 0.07, "grad_norm": 0.40809681069198844, "learning_rate": 1.884279475982533e-05, "loss": 0.4803, "step": 64 }, { "epoch": 0.07, "grad_norm": 0.5235211035164702, "learning_rate": 1.882096069868996e-05, "loss": 0.4944, "step": 65 }, { "epoch": 0.07, "grad_norm": 0.4966721954560684, "learning_rate": 1.8799126637554586e-05, "loss": 0.5025, "step": 66 }, { "epoch": 0.07, "grad_norm": 0.43399894957700963, "learning_rate": 1.8777292576419214e-05, "loss": 0.4861, "step": 67 }, { "epoch": 0.07, "grad_norm": 0.6261012627094004, "learning_rate": 1.8755458515283846e-05, "loss": 0.491, "step": 68 }, { "epoch": 0.07, "grad_norm": 0.5504631314192578, "learning_rate": 1.8733624454148474e-05, "loss": 0.4788, "step": 69 }, { "epoch": 0.08, "grad_norm": 0.5063285566533242, "learning_rate": 1.87117903930131e-05, "loss": 0.4802, "step": 70 }, { "epoch": 0.08, "grad_norm": 0.5795384217921828, "learning_rate": 1.868995633187773e-05, "loss": 0.4838, "step": 71 }, { "epoch": 0.08, "grad_norm": 0.39411140991455096, "learning_rate": 1.866812227074236e-05, "loss": 0.4636, "step": 72 }, { "epoch": 0.08, "grad_norm": 0.4230648674489383, "learning_rate": 1.864628820960699e-05, "loss": 0.4694, "step": 73 }, { "epoch": 0.08, "grad_norm": 0.5138620814929178, "learning_rate": 1.8624454148471616e-05, "loss": 0.4865, "step": 74 }, { "epoch": 0.08, "grad_norm": 0.41965115714343726, "learning_rate": 1.8602620087336244e-05, "loss": 0.4833, "step": 75 }, { "epoch": 0.08, "grad_norm": 0.4149053560192593, "learning_rate": 1.8580786026200876e-05, "loss": 0.4786, "step": 76 }, { "epoch": 0.08, "grad_norm": 0.46599988055432306, "learning_rate": 1.8558951965065503e-05, "loss": 0.497, "step": 77 }, { "epoch": 0.08, "grad_norm": 0.4811313526242602, "learning_rate": 1.853711790393013e-05, "loss": 0.4485, "step": 78 }, { "epoch": 0.09, "grad_norm": 0.4392560728996164, "learning_rate": 1.8515283842794763e-05, "loss": 0.434, "step": 79 }, { "epoch": 0.09, "grad_norm": 0.45146186393597715, "learning_rate": 1.849344978165939e-05, "loss": 0.4857, "step": 80 }, { "epoch": 0.09, "grad_norm": 0.4048483216039733, "learning_rate": 1.847161572052402e-05, "loss": 0.463, "step": 81 }, { "epoch": 0.09, "grad_norm": 0.39375794039394735, "learning_rate": 1.844978165938865e-05, "loss": 0.4777, "step": 82 }, { "epoch": 0.09, "grad_norm": 0.49842165141484956, "learning_rate": 1.8427947598253278e-05, "loss": 0.4743, "step": 83 }, { "epoch": 0.09, "grad_norm": 0.45279844618971316, "learning_rate": 1.8406113537117905e-05, "loss": 0.4606, "step": 84 }, { "epoch": 0.09, "grad_norm": 0.4186717031357162, "learning_rate": 1.8384279475982533e-05, "loss": 0.4869, "step": 85 }, { "epoch": 0.09, "grad_norm": 0.41969330355451234, "learning_rate": 1.8362445414847165e-05, "loss": 0.4617, "step": 86 }, { "epoch": 0.09, "grad_norm": 0.4496593802158418, "learning_rate": 1.8340611353711792e-05, "loss": 0.473, "step": 87 }, { "epoch": 0.1, "grad_norm": 0.41451974130263514, "learning_rate": 1.831877729257642e-05, "loss": 0.4561, "step": 88 }, { "epoch": 0.1, "grad_norm": 0.43177829462958645, "learning_rate": 1.8296943231441048e-05, "loss": 0.4473, "step": 89 }, { "epoch": 0.1, "grad_norm": 0.5142765330070325, "learning_rate": 1.827510917030568e-05, "loss": 0.4651, "step": 90 }, { "epoch": 0.1, "grad_norm": 0.5298650303411407, "learning_rate": 1.8253275109170307e-05, "loss": 0.4324, "step": 91 }, { "epoch": 0.1, "grad_norm": 0.45728734983986014, "learning_rate": 1.8231441048034935e-05, "loss": 0.4609, "step": 92 }, { "epoch": 0.1, "grad_norm": 0.476340414110513, "learning_rate": 1.8209606986899563e-05, "loss": 0.4864, "step": 93 }, { "epoch": 0.1, "grad_norm": 0.5424768428870163, "learning_rate": 1.8187772925764194e-05, "loss": 0.4481, "step": 94 }, { "epoch": 0.1, "grad_norm": 0.3967853775356532, "learning_rate": 1.8165938864628822e-05, "loss": 0.443, "step": 95 }, { "epoch": 0.1, "grad_norm": 0.37481758528836623, "learning_rate": 1.814410480349345e-05, "loss": 0.4484, "step": 96 }, { "epoch": 0.1, "grad_norm": 0.4467615010105185, "learning_rate": 1.8122270742358078e-05, "loss": 0.4644, "step": 97 }, { "epoch": 0.11, "grad_norm": 0.3864694041939627, "learning_rate": 1.810043668122271e-05, "loss": 0.4442, "step": 98 }, { "epoch": 0.11, "grad_norm": 0.44938445973951147, "learning_rate": 1.8078602620087337e-05, "loss": 0.4633, "step": 99 }, { "epoch": 0.11, "grad_norm": 0.4153749251398598, "learning_rate": 1.805676855895197e-05, "loss": 0.4622, "step": 100 }, { "epoch": 0.11, "grad_norm": 0.37908567525993214, "learning_rate": 1.8034934497816597e-05, "loss": 0.4557, "step": 101 }, { "epoch": 0.11, "grad_norm": 0.43344971266237664, "learning_rate": 1.8013100436681224e-05, "loss": 0.4861, "step": 102 }, { "epoch": 0.11, "grad_norm": 0.3897708901183335, "learning_rate": 1.7991266375545852e-05, "loss": 0.4314, "step": 103 }, { "epoch": 0.11, "grad_norm": 0.38897087826119375, "learning_rate": 1.7969432314410484e-05, "loss": 0.4478, "step": 104 }, { "epoch": 0.11, "grad_norm": 0.38883284489467035, "learning_rate": 1.794759825327511e-05, "loss": 0.4518, "step": 105 }, { "epoch": 0.11, "grad_norm": 0.43206300394588104, "learning_rate": 1.792576419213974e-05, "loss": 0.4617, "step": 106 }, { "epoch": 0.12, "grad_norm": 0.40125737231009045, "learning_rate": 1.7903930131004367e-05, "loss": 0.4502, "step": 107 }, { "epoch": 0.12, "grad_norm": 0.435314612619678, "learning_rate": 1.7882096069869e-05, "loss": 0.4766, "step": 108 }, { "epoch": 0.12, "grad_norm": 0.4144017071697519, "learning_rate": 1.7860262008733626e-05, "loss": 0.4665, "step": 109 }, { "epoch": 0.12, "grad_norm": 0.4192066755628224, "learning_rate": 1.7838427947598254e-05, "loss": 0.4871, "step": 110 }, { "epoch": 0.12, "grad_norm": 0.45215784643941886, "learning_rate": 1.7816593886462882e-05, "loss": 0.4745, "step": 111 }, { "epoch": 0.12, "grad_norm": 0.3946414831281833, "learning_rate": 1.7794759825327513e-05, "loss": 0.4692, "step": 112 }, { "epoch": 0.12, "grad_norm": 0.361997860030177, "learning_rate": 1.777292576419214e-05, "loss": 0.4554, "step": 113 }, { "epoch": 0.12, "grad_norm": 0.40175416951250204, "learning_rate": 1.775109170305677e-05, "loss": 0.4536, "step": 114 }, { "epoch": 0.12, "grad_norm": 0.47114297094100405, "learning_rate": 1.7729257641921397e-05, "loss": 0.4708, "step": 115 }, { "epoch": 0.13, "grad_norm": 0.41885478146398897, "learning_rate": 1.770742358078603e-05, "loss": 0.451, "step": 116 }, { "epoch": 0.13, "grad_norm": 0.4552034524733042, "learning_rate": 1.7685589519650656e-05, "loss": 0.4495, "step": 117 }, { "epoch": 0.13, "grad_norm": 0.4294672314730056, "learning_rate": 1.7663755458515288e-05, "loss": 0.4377, "step": 118 }, { "epoch": 0.13, "grad_norm": 0.41158434793213744, "learning_rate": 1.7641921397379912e-05, "loss": 0.4603, "step": 119 }, { "epoch": 0.13, "grad_norm": 0.40828284256365754, "learning_rate": 1.7620087336244543e-05, "loss": 0.452, "step": 120 }, { "epoch": 0.13, "grad_norm": 0.4116486419111368, "learning_rate": 1.759825327510917e-05, "loss": 0.4503, "step": 121 }, { "epoch": 0.13, "grad_norm": 0.3981270422052198, "learning_rate": 1.7576419213973803e-05, "loss": 0.4606, "step": 122 }, { "epoch": 0.13, "grad_norm": 0.4493126314995839, "learning_rate": 1.755458515283843e-05, "loss": 0.4667, "step": 123 }, { "epoch": 0.13, "grad_norm": 0.424114780787477, "learning_rate": 1.753275109170306e-05, "loss": 0.4422, "step": 124 }, { "epoch": 0.13, "grad_norm": 0.3852850414308942, "learning_rate": 1.7510917030567686e-05, "loss": 0.4429, "step": 125 }, { "epoch": 0.14, "grad_norm": 0.4282509458412285, "learning_rate": 1.7489082969432317e-05, "loss": 0.4432, "step": 126 }, { "epoch": 0.14, "grad_norm": 0.41473375647446675, "learning_rate": 1.7467248908296945e-05, "loss": 0.4681, "step": 127 }, { "epoch": 0.14, "grad_norm": 0.42524655917736237, "learning_rate": 1.7445414847161573e-05, "loss": 0.4666, "step": 128 }, { "epoch": 0.14, "grad_norm": 0.4577002723023361, "learning_rate": 1.74235807860262e-05, "loss": 0.4386, "step": 129 }, { "epoch": 0.14, "grad_norm": 0.4495477549652464, "learning_rate": 1.7401746724890832e-05, "loss": 0.4456, "step": 130 }, { "epoch": 0.14, "grad_norm": 0.4589218421705617, "learning_rate": 1.737991266375546e-05, "loss": 0.447, "step": 131 }, { "epoch": 0.14, "grad_norm": 0.41129567959332086, "learning_rate": 1.7358078602620088e-05, "loss": 0.4714, "step": 132 }, { "epoch": 0.14, "grad_norm": 0.4091662311389778, "learning_rate": 1.7336244541484716e-05, "loss": 0.4325, "step": 133 }, { "epoch": 0.14, "grad_norm": 0.43554863694180995, "learning_rate": 1.7314410480349347e-05, "loss": 0.428, "step": 134 }, { "epoch": 0.15, "grad_norm": 0.48782967094292995, "learning_rate": 1.7292576419213975e-05, "loss": 0.4614, "step": 135 }, { "epoch": 0.15, "grad_norm": 0.410345941910886, "learning_rate": 1.7270742358078607e-05, "loss": 0.4358, "step": 136 }, { "epoch": 0.15, "grad_norm": 0.4658797763961389, "learning_rate": 1.724890829694323e-05, "loss": 0.4228, "step": 137 }, { "epoch": 0.15, "grad_norm": 0.46687534901658256, "learning_rate": 1.7227074235807862e-05, "loss": 0.4363, "step": 138 }, { "epoch": 0.15, "grad_norm": 0.41452146680368457, "learning_rate": 1.720524017467249e-05, "loss": 0.4451, "step": 139 }, { "epoch": 0.15, "grad_norm": 0.435633971396934, "learning_rate": 1.718340611353712e-05, "loss": 0.4234, "step": 140 }, { "epoch": 0.15, "grad_norm": 0.3833328913948718, "learning_rate": 1.7161572052401746e-05, "loss": 0.4457, "step": 141 }, { "epoch": 0.15, "grad_norm": 0.4362215880238739, "learning_rate": 1.7139737991266377e-05, "loss": 0.433, "step": 142 }, { "epoch": 0.15, "grad_norm": 0.4495745203741214, "learning_rate": 1.7117903930131005e-05, "loss": 0.4275, "step": 143 }, { "epoch": 0.16, "grad_norm": 0.4364392500220487, "learning_rate": 1.7096069868995636e-05, "loss": 0.4448, "step": 144 }, { "epoch": 0.16, "grad_norm": 0.43234675029655206, "learning_rate": 1.7074235807860264e-05, "loss": 0.4333, "step": 145 }, { "epoch": 0.16, "grad_norm": 0.38155397507175676, "learning_rate": 1.7052401746724892e-05, "loss": 0.4151, "step": 146 }, { "epoch": 0.16, "grad_norm": 0.41784216844651545, "learning_rate": 1.703056768558952e-05, "loss": 0.4604, "step": 147 }, { "epoch": 0.16, "grad_norm": 0.4282599403566064, "learning_rate": 1.700873362445415e-05, "loss": 0.4031, "step": 148 }, { "epoch": 0.16, "grad_norm": 0.40487465599518896, "learning_rate": 1.698689956331878e-05, "loss": 0.4409, "step": 149 }, { "epoch": 0.16, "grad_norm": 0.4694993863584187, "learning_rate": 1.6965065502183407e-05, "loss": 0.4351, "step": 150 }, { "epoch": 0.16, "grad_norm": 0.40097074583417713, "learning_rate": 1.6943231441048035e-05, "loss": 0.4454, "step": 151 }, { "epoch": 0.16, "grad_norm": 0.41500384012996383, "learning_rate": 1.6921397379912666e-05, "loss": 0.4133, "step": 152 }, { "epoch": 0.17, "grad_norm": 0.4091240382988871, "learning_rate": 1.6899563318777294e-05, "loss": 0.4387, "step": 153 }, { "epoch": 0.17, "grad_norm": 0.38499409045601, "learning_rate": 1.6877729257641922e-05, "loss": 0.4176, "step": 154 }, { "epoch": 0.17, "grad_norm": 0.3923298254546311, "learning_rate": 1.685589519650655e-05, "loss": 0.4394, "step": 155 }, { "epoch": 0.17, "grad_norm": 0.39382094470718404, "learning_rate": 1.683406113537118e-05, "loss": 0.4203, "step": 156 }, { "epoch": 0.17, "grad_norm": 0.3947764097635848, "learning_rate": 1.681222707423581e-05, "loss": 0.4079, "step": 157 }, { "epoch": 0.17, "grad_norm": 0.3842228774523482, "learning_rate": 1.679039301310044e-05, "loss": 0.4203, "step": 158 }, { "epoch": 0.17, "grad_norm": 0.4362604428547085, "learning_rate": 1.6768558951965065e-05, "loss": 0.4214, "step": 159 }, { "epoch": 0.17, "grad_norm": 0.3897244945668934, "learning_rate": 1.6746724890829696e-05, "loss": 0.4135, "step": 160 }, { "epoch": 0.17, "grad_norm": 0.39055816078553107, "learning_rate": 1.6724890829694324e-05, "loss": 0.4264, "step": 161 }, { "epoch": 0.17, "grad_norm": 0.4059738357585979, "learning_rate": 1.6703056768558955e-05, "loss": 0.4426, "step": 162 }, { "epoch": 0.18, "grad_norm": 0.3719736409193062, "learning_rate": 1.668122270742358e-05, "loss": 0.3741, "step": 163 }, { "epoch": 0.18, "grad_norm": 0.3994391137360452, "learning_rate": 1.665938864628821e-05, "loss": 0.3903, "step": 164 }, { "epoch": 0.18, "grad_norm": 0.4102321554518753, "learning_rate": 1.663755458515284e-05, "loss": 0.3947, "step": 165 }, { "epoch": 0.18, "grad_norm": 0.40605248638251906, "learning_rate": 1.661572052401747e-05, "loss": 0.374, "step": 166 }, { "epoch": 0.18, "grad_norm": 0.39829795862216355, "learning_rate": 1.6593886462882098e-05, "loss": 0.3558, "step": 167 }, { "epoch": 0.18, "grad_norm": 0.5668042067428868, "learning_rate": 1.6572052401746726e-05, "loss": 0.3966, "step": 168 }, { "epoch": 0.18, "grad_norm": 0.37012620577733346, "learning_rate": 1.6550218340611354e-05, "loss": 0.3696, "step": 169 }, { "epoch": 0.18, "grad_norm": 0.5230789664738862, "learning_rate": 1.6528384279475985e-05, "loss": 0.3746, "step": 170 }, { "epoch": 0.18, "grad_norm": 0.38417092465063496, "learning_rate": 1.6506550218340613e-05, "loss": 0.3643, "step": 171 }, { "epoch": 0.19, "grad_norm": 0.39278262301667943, "learning_rate": 1.648471615720524e-05, "loss": 0.3707, "step": 172 }, { "epoch": 0.19, "grad_norm": 0.4445691345644359, "learning_rate": 1.646288209606987e-05, "loss": 0.3825, "step": 173 }, { "epoch": 0.19, "grad_norm": 0.6480928907101309, "learning_rate": 1.64410480349345e-05, "loss": 0.3854, "step": 174 }, { "epoch": 0.19, "grad_norm": 0.4272433035623177, "learning_rate": 1.6419213973799128e-05, "loss": 0.3703, "step": 175 }, { "epoch": 0.19, "grad_norm": 0.3739993134928334, "learning_rate": 1.639737991266376e-05, "loss": 0.3701, "step": 176 }, { "epoch": 0.19, "grad_norm": 0.4010063647136101, "learning_rate": 1.6375545851528384e-05, "loss": 0.3723, "step": 177 }, { "epoch": 0.19, "grad_norm": 0.3790230325277792, "learning_rate": 1.6353711790393015e-05, "loss": 0.3676, "step": 178 }, { "epoch": 0.19, "grad_norm": 0.3755499066206077, "learning_rate": 1.6331877729257643e-05, "loss": 0.3642, "step": 179 }, { "epoch": 0.19, "grad_norm": 0.4241683578769766, "learning_rate": 1.6310043668122274e-05, "loss": 0.3899, "step": 180 }, { "epoch": 0.2, "grad_norm": 0.38321915261506356, "learning_rate": 1.62882096069869e-05, "loss": 0.3614, "step": 181 }, { "epoch": 0.2, "grad_norm": 0.3761870154387026, "learning_rate": 1.626637554585153e-05, "loss": 0.3655, "step": 182 }, { "epoch": 0.2, "grad_norm": 0.39129047927836064, "learning_rate": 1.6244541484716158e-05, "loss": 0.383, "step": 183 }, { "epoch": 0.2, "grad_norm": 0.4724242531647335, "learning_rate": 1.622270742358079e-05, "loss": 0.3654, "step": 184 }, { "epoch": 0.2, "grad_norm": 0.4319860707229489, "learning_rate": 1.6200873362445414e-05, "loss": 0.3906, "step": 185 }, { "epoch": 0.2, "grad_norm": 0.45729422543214604, "learning_rate": 1.6179039301310045e-05, "loss": 0.3765, "step": 186 }, { "epoch": 0.2, "grad_norm": 0.45529560900836535, "learning_rate": 1.6157205240174673e-05, "loss": 0.3751, "step": 187 }, { "epoch": 0.2, "grad_norm": 0.4503043781159598, "learning_rate": 1.6135371179039304e-05, "loss": 0.3687, "step": 188 }, { "epoch": 0.2, "grad_norm": 0.45425963573071876, "learning_rate": 1.6113537117903932e-05, "loss": 0.3551, "step": 189 }, { "epoch": 0.21, "grad_norm": 0.35503621037592853, "learning_rate": 1.609170305676856e-05, "loss": 0.349, "step": 190 }, { "epoch": 0.21, "grad_norm": 0.3796166907322132, "learning_rate": 1.6069868995633188e-05, "loss": 0.3716, "step": 191 }, { "epoch": 0.21, "grad_norm": 0.4173119812888326, "learning_rate": 1.604803493449782e-05, "loss": 0.3784, "step": 192 }, { "epoch": 0.21, "grad_norm": 0.38778404930691235, "learning_rate": 1.6026200873362447e-05, "loss": 0.3828, "step": 193 }, { "epoch": 0.21, "grad_norm": 0.376339388079517, "learning_rate": 1.6004366812227075e-05, "loss": 0.3787, "step": 194 }, { "epoch": 0.21, "grad_norm": 0.375246423976208, "learning_rate": 1.5982532751091703e-05, "loss": 0.3528, "step": 195 }, { "epoch": 0.21, "grad_norm": 0.37393557958350465, "learning_rate": 1.5960698689956334e-05, "loss": 0.3508, "step": 196 }, { "epoch": 0.21, "grad_norm": 0.38427011175269715, "learning_rate": 1.5938864628820962e-05, "loss": 0.3537, "step": 197 }, { "epoch": 0.21, "grad_norm": 0.38025086526439966, "learning_rate": 1.5917030567685593e-05, "loss": 0.3594, "step": 198 }, { "epoch": 0.21, "grad_norm": 0.46667328798614865, "learning_rate": 1.5895196506550218e-05, "loss": 0.3601, "step": 199 }, { "epoch": 0.22, "grad_norm": 0.4141132395623408, "learning_rate": 1.587336244541485e-05, "loss": 0.3586, "step": 200 }, { "epoch": 0.22, "grad_norm": 0.3803714659154048, "learning_rate": 1.5851528384279477e-05, "loss": 0.363, "step": 201 }, { "epoch": 0.22, "grad_norm": 0.4605657889095618, "learning_rate": 1.582969432314411e-05, "loss": 0.3538, "step": 202 }, { "epoch": 0.22, "grad_norm": 0.4058624399636699, "learning_rate": 1.5807860262008733e-05, "loss": 0.3601, "step": 203 }, { "epoch": 0.22, "grad_norm": 0.5379500589660902, "learning_rate": 1.5786026200873364e-05, "loss": 0.3669, "step": 204 }, { "epoch": 0.22, "grad_norm": 0.4219898314844134, "learning_rate": 1.5764192139737992e-05, "loss": 0.353, "step": 205 }, { "epoch": 0.22, "grad_norm": 0.38152400004537884, "learning_rate": 1.5742358078602623e-05, "loss": 0.3604, "step": 206 }, { "epoch": 0.22, "grad_norm": 0.40015810285717957, "learning_rate": 1.5720524017467248e-05, "loss": 0.3619, "step": 207 }, { "epoch": 0.22, "grad_norm": 0.41681499423940854, "learning_rate": 1.569868995633188e-05, "loss": 0.3357, "step": 208 }, { "epoch": 0.23, "grad_norm": 0.4976226945041277, "learning_rate": 1.5676855895196507e-05, "loss": 0.3691, "step": 209 }, { "epoch": 0.23, "grad_norm": 0.43800093718650573, "learning_rate": 1.5655021834061138e-05, "loss": 0.3606, "step": 210 }, { "epoch": 0.23, "grad_norm": 0.3919587444642411, "learning_rate": 1.5633187772925766e-05, "loss": 0.3597, "step": 211 }, { "epoch": 0.23, "grad_norm": 0.49836512850382286, "learning_rate": 1.5611353711790394e-05, "loss": 0.3596, "step": 212 }, { "epoch": 0.23, "grad_norm": 0.4395052391516801, "learning_rate": 1.5589519650655022e-05, "loss": 0.352, "step": 213 }, { "epoch": 0.23, "grad_norm": 0.45220282868066214, "learning_rate": 1.5567685589519653e-05, "loss": 0.3589, "step": 214 }, { "epoch": 0.23, "grad_norm": 0.43696638330335513, "learning_rate": 1.554585152838428e-05, "loss": 0.352, "step": 215 }, { "epoch": 0.23, "grad_norm": 0.4493622923687038, "learning_rate": 1.552401746724891e-05, "loss": 0.3452, "step": 216 }, { "epoch": 0.23, "grad_norm": 0.4368637942785743, "learning_rate": 1.5502183406113537e-05, "loss": 0.3606, "step": 217 }, { "epoch": 0.24, "grad_norm": 0.39254783399638093, "learning_rate": 1.5480349344978168e-05, "loss": 0.3446, "step": 218 }, { "epoch": 0.24, "grad_norm": 0.39267930582092203, "learning_rate": 1.5458515283842796e-05, "loss": 0.3463, "step": 219 }, { "epoch": 0.24, "grad_norm": 0.49336143547588307, "learning_rate": 1.5436681222707427e-05, "loss": 0.3545, "step": 220 }, { "epoch": 0.24, "grad_norm": 0.3861913649178407, "learning_rate": 1.5414847161572052e-05, "loss": 0.3463, "step": 221 }, { "epoch": 0.24, "grad_norm": 0.39706584967316105, "learning_rate": 1.5393013100436683e-05, "loss": 0.3492, "step": 222 }, { "epoch": 0.24, "grad_norm": 0.5278921313657406, "learning_rate": 1.537117903930131e-05, "loss": 0.3629, "step": 223 }, { "epoch": 0.24, "grad_norm": 0.3925508897738005, "learning_rate": 1.5349344978165942e-05, "loss": 0.3303, "step": 224 }, { "epoch": 0.24, "grad_norm": 0.3833632691587687, "learning_rate": 1.5327510917030567e-05, "loss": 0.3542, "step": 225 }, { "epoch": 0.24, "grad_norm": 0.4143435226325874, "learning_rate": 1.5305676855895198e-05, "loss": 0.3557, "step": 226 }, { "epoch": 0.25, "grad_norm": 0.5459803483958254, "learning_rate": 1.5283842794759826e-05, "loss": 0.3431, "step": 227 }, { "epoch": 0.25, "grad_norm": 0.4130510897175159, "learning_rate": 1.5262008733624457e-05, "loss": 0.3463, "step": 228 }, { "epoch": 0.25, "grad_norm": 0.4749086480386068, "learning_rate": 1.5240174672489083e-05, "loss": 0.3596, "step": 229 }, { "epoch": 0.25, "grad_norm": 0.4727236644494208, "learning_rate": 1.5218340611353713e-05, "loss": 0.3384, "step": 230 }, { "epoch": 0.25, "grad_norm": 0.4264208163593211, "learning_rate": 1.5196506550218343e-05, "loss": 0.3633, "step": 231 }, { "epoch": 0.25, "grad_norm": 0.45316939300908593, "learning_rate": 1.517467248908297e-05, "loss": 0.3532, "step": 232 }, { "epoch": 0.25, "grad_norm": 0.5014932507396431, "learning_rate": 1.51528384279476e-05, "loss": 0.3581, "step": 233 }, { "epoch": 0.25, "grad_norm": 0.3740888692728483, "learning_rate": 1.5131004366812228e-05, "loss": 0.3591, "step": 234 }, { "epoch": 0.25, "grad_norm": 0.41532342481631906, "learning_rate": 1.5109170305676858e-05, "loss": 0.3459, "step": 235 }, { "epoch": 0.25, "grad_norm": 0.3887632636355805, "learning_rate": 1.5087336244541485e-05, "loss": 0.3447, "step": 236 }, { "epoch": 0.26, "grad_norm": 0.40230721242978856, "learning_rate": 1.5065502183406115e-05, "loss": 0.3672, "step": 237 }, { "epoch": 0.26, "grad_norm": 0.39422442525497225, "learning_rate": 1.5043668122270743e-05, "loss": 0.35, "step": 238 }, { "epoch": 0.26, "grad_norm": 0.42685260200105113, "learning_rate": 1.5021834061135372e-05, "loss": 0.3614, "step": 239 }, { "epoch": 0.26, "grad_norm": 0.39862400252659125, "learning_rate": 1.5000000000000002e-05, "loss": 0.3485, "step": 240 }, { "epoch": 0.26, "grad_norm": 0.4428714302171062, "learning_rate": 1.497816593886463e-05, "loss": 0.36, "step": 241 }, { "epoch": 0.26, "grad_norm": 0.4185235272235272, "learning_rate": 1.495633187772926e-05, "loss": 0.3492, "step": 242 }, { "epoch": 0.26, "grad_norm": 0.39847525249468174, "learning_rate": 1.4934497816593887e-05, "loss": 0.3567, "step": 243 }, { "epoch": 0.26, "grad_norm": 0.3855301479432999, "learning_rate": 1.4912663755458517e-05, "loss": 0.3571, "step": 244 }, { "epoch": 0.26, "grad_norm": 0.393234899196987, "learning_rate": 1.4890829694323145e-05, "loss": 0.3566, "step": 245 }, { "epoch": 0.27, "grad_norm": 0.45253412592224085, "learning_rate": 1.4868995633187775e-05, "loss": 0.3527, "step": 246 }, { "epoch": 0.27, "grad_norm": 0.4218417613746097, "learning_rate": 1.4847161572052402e-05, "loss": 0.3391, "step": 247 }, { "epoch": 0.27, "grad_norm": 0.447603156651112, "learning_rate": 1.4825327510917032e-05, "loss": 0.3711, "step": 248 }, { "epoch": 0.27, "grad_norm": 0.39097790988732256, "learning_rate": 1.480349344978166e-05, "loss": 0.3376, "step": 249 }, { "epoch": 0.27, "grad_norm": 0.4318473286465606, "learning_rate": 1.478165938864629e-05, "loss": 0.3506, "step": 250 }, { "epoch": 0.27, "grad_norm": 0.45407014143329794, "learning_rate": 1.4759825327510919e-05, "loss": 0.3658, "step": 251 }, { "epoch": 0.27, "grad_norm": 0.4126763263167815, "learning_rate": 1.4737991266375547e-05, "loss": 0.3383, "step": 252 }, { "epoch": 0.27, "grad_norm": 0.3962187981580303, "learning_rate": 1.4716157205240177e-05, "loss": 0.3449, "step": 253 }, { "epoch": 0.27, "grad_norm": 0.44411484803724655, "learning_rate": 1.4694323144104804e-05, "loss": 0.3329, "step": 254 }, { "epoch": 0.28, "grad_norm": 0.40937179753455943, "learning_rate": 1.4672489082969434e-05, "loss": 0.3473, "step": 255 }, { "epoch": 0.28, "grad_norm": 0.4170433030971853, "learning_rate": 1.4650655021834062e-05, "loss": 0.3537, "step": 256 }, { "epoch": 0.28, "grad_norm": 0.5088525538449871, "learning_rate": 1.4628820960698691e-05, "loss": 0.3464, "step": 257 }, { "epoch": 0.28, "grad_norm": 0.42919291729255854, "learning_rate": 1.460698689956332e-05, "loss": 0.3598, "step": 258 }, { "epoch": 0.28, "grad_norm": 0.3968573821655448, "learning_rate": 1.4585152838427949e-05, "loss": 0.3297, "step": 259 }, { "epoch": 0.28, "grad_norm": 0.41865791119556794, "learning_rate": 1.4563318777292577e-05, "loss": 0.3592, "step": 260 }, { "epoch": 0.28, "grad_norm": 0.42353970958609694, "learning_rate": 1.4541484716157206e-05, "loss": 0.3548, "step": 261 }, { "epoch": 0.28, "grad_norm": 0.3742329381749764, "learning_rate": 1.4519650655021836e-05, "loss": 0.3218, "step": 262 }, { "epoch": 0.28, "grad_norm": 0.4626643436578089, "learning_rate": 1.4497816593886464e-05, "loss": 0.3384, "step": 263 }, { "epoch": 0.29, "grad_norm": 0.45856052063838026, "learning_rate": 1.4475982532751093e-05, "loss": 0.3615, "step": 264 }, { "epoch": 0.29, "grad_norm": 0.41283923418455426, "learning_rate": 1.4454148471615721e-05, "loss": 0.3524, "step": 265 }, { "epoch": 0.29, "grad_norm": 0.4074545937674161, "learning_rate": 1.4432314410480351e-05, "loss": 0.3409, "step": 266 }, { "epoch": 0.29, "grad_norm": 0.41807996749714293, "learning_rate": 1.4410480349344979e-05, "loss": 0.3452, "step": 267 }, { "epoch": 0.29, "grad_norm": 0.38052190844013845, "learning_rate": 1.4388646288209608e-05, "loss": 0.3434, "step": 268 }, { "epoch": 0.29, "grad_norm": 0.44469063822750765, "learning_rate": 1.4366812227074236e-05, "loss": 0.35, "step": 269 }, { "epoch": 0.29, "grad_norm": 0.3829776252082372, "learning_rate": 1.4344978165938866e-05, "loss": 0.3263, "step": 270 }, { "epoch": 0.29, "grad_norm": 0.4056196273835837, "learning_rate": 1.4323144104803495e-05, "loss": 0.3539, "step": 271 }, { "epoch": 0.29, "grad_norm": 0.40412139770148303, "learning_rate": 1.4301310043668123e-05, "loss": 0.3598, "step": 272 }, { "epoch": 0.29, "grad_norm": 0.4202242804763555, "learning_rate": 1.4279475982532753e-05, "loss": 0.3337, "step": 273 }, { "epoch": 0.3, "grad_norm": 0.4047872913704621, "learning_rate": 1.4257641921397381e-05, "loss": 0.3511, "step": 274 }, { "epoch": 0.3, "grad_norm": 0.37472436629683387, "learning_rate": 1.423580786026201e-05, "loss": 0.3307, "step": 275 }, { "epoch": 0.3, "grad_norm": 0.41684890179246353, "learning_rate": 1.4213973799126638e-05, "loss": 0.3379, "step": 276 }, { "epoch": 0.3, "grad_norm": 0.4174816823127464, "learning_rate": 1.4192139737991268e-05, "loss": 0.3611, "step": 277 }, { "epoch": 0.3, "grad_norm": 0.4487870671522349, "learning_rate": 1.4170305676855896e-05, "loss": 0.3392, "step": 278 }, { "epoch": 0.3, "grad_norm": 0.41458956443794104, "learning_rate": 1.4148471615720525e-05, "loss": 0.3467, "step": 279 }, { "epoch": 0.3, "grad_norm": 0.46610708626195474, "learning_rate": 1.4126637554585155e-05, "loss": 0.3353, "step": 280 }, { "epoch": 0.3, "grad_norm": 0.4512003769892443, "learning_rate": 1.4104803493449783e-05, "loss": 0.3513, "step": 281 }, { "epoch": 0.3, "grad_norm": 0.3858475360185647, "learning_rate": 1.408296943231441e-05, "loss": 0.3177, "step": 282 }, { "epoch": 0.31, "grad_norm": 0.3971417244165327, "learning_rate": 1.406113537117904e-05, "loss": 0.3517, "step": 283 }, { "epoch": 0.31, "grad_norm": 0.4318791589108847, "learning_rate": 1.403930131004367e-05, "loss": 0.3369, "step": 284 }, { "epoch": 0.31, "grad_norm": 0.39746635489556087, "learning_rate": 1.4017467248908298e-05, "loss": 0.3212, "step": 285 }, { "epoch": 0.31, "grad_norm": 0.43224599512736944, "learning_rate": 1.3995633187772927e-05, "loss": 0.3451, "step": 286 }, { "epoch": 0.31, "grad_norm": 0.4169628481404007, "learning_rate": 1.3973799126637555e-05, "loss": 0.3429, "step": 287 }, { "epoch": 0.31, "grad_norm": 0.38109401419780187, "learning_rate": 1.3951965065502185e-05, "loss": 0.3444, "step": 288 }, { "epoch": 0.31, "grad_norm": 0.4174351360849819, "learning_rate": 1.3930131004366814e-05, "loss": 0.3395, "step": 289 }, { "epoch": 0.31, "grad_norm": 0.4089628933545283, "learning_rate": 1.3908296943231442e-05, "loss": 0.3534, "step": 290 }, { "epoch": 0.31, "grad_norm": 0.45790289224080694, "learning_rate": 1.388646288209607e-05, "loss": 0.3365, "step": 291 }, { "epoch": 0.32, "grad_norm": 0.37659122705694703, "learning_rate": 1.38646288209607e-05, "loss": 0.3331, "step": 292 }, { "epoch": 0.32, "grad_norm": 0.3837943816847714, "learning_rate": 1.384279475982533e-05, "loss": 0.3182, "step": 293 }, { "epoch": 0.32, "grad_norm": 0.3822006474102691, "learning_rate": 1.3820960698689957e-05, "loss": 0.3074, "step": 294 }, { "epoch": 0.32, "grad_norm": 0.3796325094330029, "learning_rate": 1.3799126637554587e-05, "loss": 0.311, "step": 295 }, { "epoch": 0.32, "grad_norm": 0.3910995271433598, "learning_rate": 1.3777292576419215e-05, "loss": 0.3151, "step": 296 }, { "epoch": 0.32, "grad_norm": 0.4162711670508305, "learning_rate": 1.3755458515283844e-05, "loss": 0.318, "step": 297 }, { "epoch": 0.32, "grad_norm": 0.39243348147197954, "learning_rate": 1.3733624454148474e-05, "loss": 0.315, "step": 298 }, { "epoch": 0.32, "grad_norm": 0.3718609158068742, "learning_rate": 1.3711790393013102e-05, "loss": 0.3068, "step": 299 }, { "epoch": 0.32, "grad_norm": 0.4114227903281369, "learning_rate": 1.368995633187773e-05, "loss": 0.3086, "step": 300 }, { "epoch": 0.33, "grad_norm": 0.37117152152007493, "learning_rate": 1.366812227074236e-05, "loss": 0.3082, "step": 301 }, { "epoch": 0.33, "grad_norm": 0.37955301673885294, "learning_rate": 1.3646288209606989e-05, "loss": 0.3048, "step": 302 }, { "epoch": 0.33, "grad_norm": 0.7145676209823392, "learning_rate": 1.3624454148471617e-05, "loss": 0.3151, "step": 303 }, { "epoch": 0.33, "grad_norm": 0.4155052349515932, "learning_rate": 1.3602620087336245e-05, "loss": 0.322, "step": 304 }, { "epoch": 0.33, "grad_norm": 0.41405357495171785, "learning_rate": 1.3580786026200874e-05, "loss": 0.3246, "step": 305 }, { "epoch": 0.33, "grad_norm": 0.4272800105120246, "learning_rate": 1.3558951965065504e-05, "loss": 0.3218, "step": 306 }, { "epoch": 0.33, "grad_norm": 0.39827328458103745, "learning_rate": 1.3537117903930132e-05, "loss": 0.3177, "step": 307 }, { "epoch": 0.33, "grad_norm": 0.42606470314677275, "learning_rate": 1.3515283842794761e-05, "loss": 0.3158, "step": 308 } ], "logging_steps": 1, "max_steps": 926, "num_input_tokens_seen": 0, "num_train_epochs": 9223372036854775807, "save_steps": 154, "total_flos": 329999752101888.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }