| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9989281886387996, | |
| "global_step": 233, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.4285714285714286e-06, | |
| "loss": 0.8496, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.8571428571428573e-06, | |
| "loss": 0.8464, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2857142857142855e-06, | |
| "loss": 0.8613, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5.7142857142857145e-06, | |
| "loss": 0.8911, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.1428571428571436e-06, | |
| "loss": 0.8857, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 8.571428571428571e-06, | |
| "loss": 0.8173, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8239, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.999516923531906e-06, | |
| "loss": 0.8904, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.998067787472772e-06, | |
| "loss": 0.8792, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.995652871840006e-06, | |
| "loss": 0.8376, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.992272643269181e-06, | |
| "loss": 0.8633, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.987927754923844e-06, | |
| "loss": 0.8431, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.982619046369321e-06, | |
| "loss": 0.8543, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.976347543410487e-06, | |
| "loss": 0.8458, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.96911445789354e-06, | |
| "loss": 0.8829, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.960921187471841e-06, | |
| "loss": 0.8638, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.951769315335843e-06, | |
| "loss": 0.8933, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.94166060990718e-06, | |
| "loss": 0.8132, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.930597024496933e-06, | |
| "loss": 0.8528, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.918580696928206e-06, | |
| "loss": 0.8643, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.905613949123036e-06, | |
| "loss": 0.8508, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.891699286653714e-06, | |
| "loss": 0.8491, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.87683939825864e-06, | |
| "loss": 0.8541, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.861037155322777e-06, | |
| "loss": 0.8565, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.844295611322804e-06, | |
| "loss": 0.8383, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.826618001237101e-06, | |
| "loss": 0.8842, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.808007740920647e-06, | |
| "loss": 0.8439, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.788468426444968e-06, | |
| "loss": 0.8613, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.768003833403278e-06, | |
| "loss": 0.8536, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.746617916180906e-06, | |
| "loss": 0.8499, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.724314807191197e-06, | |
| "loss": 0.9052, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.701098816076995e-06, | |
| "loss": 0.8328, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.6769744288779e-06, | |
| "loss": 0.8637, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.651946307163417e-06, | |
| "loss": 0.8685, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.626019287132202e-06, | |
| "loss": 0.8849, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.599198378677559e-06, | |
| "loss": 0.876, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.571488764419381e-06, | |
| "loss": 0.8354, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.542895798702702e-06, | |
| "loss": 0.8596, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.51342500656308e-06, | |
| "loss": 0.8291, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.483082082658984e-06, | |
| "loss": 0.8565, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.451872890171419e-06, | |
| "loss": 0.8549, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.41980345967098e-06, | |
| "loss": 0.8533, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.386879987952549e-06, | |
| "loss": 0.8696, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.353108836837907e-06, | |
| "loss": 0.8344, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.318496531946411e-06, | |
| "loss": 0.8221, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.283049761434059e-06, | |
| "loss": 0.8148, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.246775374701139e-06, | |
| "loss": 0.8571, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.209680381068698e-06, | |
| "loss": 0.8727, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.171771948424138e-06, | |
| "loss": 0.8737, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.13305740183616e-06, | |
| "loss": 0.8395, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.093544222139338e-06, | |
| "loss": 0.8227, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.053240044488587e-06, | |
| "loss": 0.9094, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.012152656883824e-06, | |
| "loss": 0.832, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.970289998665083e-06, | |
| "loss": 0.8383, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.927660158978392e-06, | |
| "loss": 0.8391, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.884271375212714e-06, | |
| "loss": 0.8271, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.84013203140821e-06, | |
| "loss": 0.8405, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.795250656636207e-06, | |
| "loss": 0.8517, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.749635923351108e-06, | |
| "loss": 0.8577, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.70329664571461e-06, | |
| "loss": 0.8626, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.656241777892544e-06, | |
| "loss": 0.8635, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.608480412324652e-06, | |
| "loss": 0.8613, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.56002177796765e-06, | |
| "loss": 0.8493, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.510875238511911e-06, | |
| "loss": 0.8266, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.461050290572114e-06, | |
| "loss": 0.8619, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.410556561852212e-06, | |
| "loss": 0.8376, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.359403809285054e-06, | |
| "loss": 0.8016, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.30760191714706e-06, | |
| "loss": 0.8304, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.255160895148263e-06, | |
| "loss": 0.8297, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.202090876498144e-06, | |
| "loss": 0.8511, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.14840211594757e-06, | |
| "loss": 0.8369, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.09410498780727e-06, | |
| "loss": 0.8646, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.039209983943201e-06, | |
| "loss": 0.8434, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.983727711749194e-06, | |
| "loss": 0.8706, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.927668892097288e-06, | |
| "loss": 0.8386, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.871044357266124e-06, | |
| "loss": 0.8591, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.81386504884782e-06, | |
| "loss": 0.8636, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.75614201563372e-06, | |
| "loss": 0.8124, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.697886411479422e-06, | |
| "loss": 0.8588, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.639109493149537e-06, | |
| "loss": 0.8437, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.579822618142505e-06, | |
| "loss": 0.8422, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.520037242496e-06, | |
| "loss": 0.8224, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.459764918573264e-06, | |
| "loss": 0.8511, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.399017292830848e-06, | |
| "loss": 0.8975, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.3378061035681415e-06, | |
| "loss": 0.8138, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 7.276143178659195e-06, | |
| "loss": 0.8309, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 7.2140404332671986e-06, | |
| "loss": 0.8347, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 7.1515098675421125e-06, | |
| "loss": 0.8287, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 7.088563564301874e-06, | |
| "loss": 0.8544, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 7.0252136866976205e-06, | |
| "loss": 0.8427, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.961472475863406e-06, | |
| "loss": 0.8082, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.897352248550828e-06, | |
| "loss": 0.8291, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.832865394749065e-06, | |
| "loss": 0.8369, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.768024375290747e-06, | |
| "loss": 0.8196, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.702841719444141e-06, | |
| "loss": 0.8267, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.637330022492112e-06, | |
| "loss": 0.8428, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.571501943298335e-06, | |
| "loss": 0.851, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.50537020186121e-06, | |
| "loss": 0.8255, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.4389475768559675e-06, | |
| "loss": 0.8296, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.372246903165445e-06, | |
| "loss": 0.8578, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.305281069399989e-06, | |
| "loss": 0.8744, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 6.238063015406982e-06, | |
| "loss": 0.8358, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 6.17060572977047e-06, | |
| "loss": 0.8287, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 6.1029222473013705e-06, | |
| "loss": 0.8464, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 6.035025646518747e-06, | |
| "loss": 0.8443, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.966929047122641e-06, | |
| "loss": 0.8269, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.898645607458941e-06, | |
| "loss": 0.8334, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.830188521976794e-06, | |
| "loss": 0.9108, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.761571018679025e-06, | |
| "loss": 0.8164, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.6928063565660955e-06, | |
| "loss": 0.8439, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.623907823074044e-06, | |
| "loss": 0.8513, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.5548887315069575e-06, | |
| "loss": 0.8198, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.48576241846443e-06, | |
| "loss": 0.8724, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.416542241264524e-06, | |
| "loss": 0.8116, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.347241575362729e-06, | |
| "loss": 0.8932, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.277873811767415e-06, | |
| "loss": 0.8415, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.208452354452275e-06, | |
| "loss": 0.8137, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.1389906177662705e-06, | |
| "loss": 0.7899, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.069502023841576e-06, | |
| "loss": 0.8527, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5e-06, | |
| "loss": 0.8108, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.9304979761584256e-06, | |
| "loss": 0.8387, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.86100938223373e-06, | |
| "loss": 0.8395, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.791547645547727e-06, | |
| "loss": 0.8428, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.722126188232586e-06, | |
| "loss": 0.8528, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.652758424637271e-06, | |
| "loss": 0.8132, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.583457758735477e-06, | |
| "loss": 0.8055, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.514237581535571e-06, | |
| "loss": 0.8801, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.4451112684930424e-06, | |
| "loss": 0.8476, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.3760921769259585e-06, | |
| "loss": 0.8304, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.307193643433907e-06, | |
| "loss": 0.848, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.2384289813209754e-06, | |
| "loss": 0.815, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.1698114780232085e-06, | |
| "loss": 0.8726, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.101354392541061e-06, | |
| "loss": 0.8188, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.033070952877362e-06, | |
| "loss": 0.847, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.964974353481254e-06, | |
| "loss": 0.8031, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.89707775269863e-06, | |
| "loss": 0.8712, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.829394270229531e-06, | |
| "loss": 0.8314, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.7619369845930195e-06, | |
| "loss": 0.8546, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.694718930600012e-06, | |
| "loss": 0.8424, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.6277530968345552e-06, | |
| "loss": 0.8531, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.5610524231440324e-06, | |
| "loss": 0.8085, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.4946297981387913e-06, | |
| "loss": 0.8608, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.428498056701665e-06, | |
| "loss": 0.836, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.3626699775078884e-06, | |
| "loss": 0.8483, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.2971582805558622e-06, | |
| "loss": 0.8269, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.2319756247092552e-06, | |
| "loss": 0.8339, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.167134605250938e-06, | |
| "loss": 0.8445, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.102647751449174e-06, | |
| "loss": 0.8458, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.0385275241365965e-06, | |
| "loss": 0.8105, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.9747863133023803e-06, | |
| "loss": 0.8312, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.9114364356981274e-06, | |
| "loss": 0.8052, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.8484901324578883e-06, | |
| "loss": 0.8428, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.7859595667328027e-06, | |
| "loss": 0.8171, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.723856821340806e-06, | |
| "loss": 0.8212, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.6621938964318593e-06, | |
| "loss": 0.8481, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.600982707169154e-06, | |
| "loss": 0.836, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.5402350814267364e-06, | |
| "loss": 0.8358, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.4799627575040014e-06, | |
| "loss": 0.8139, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.4201773818574956e-06, | |
| "loss": 0.8381, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.360890506850464e-06, | |
| "loss": 0.8482, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.302113588520578e-06, | |
| "loss": 0.8348, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.243857984366284e-06, | |
| "loss": 0.8305, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.1861349511521817e-06, | |
| "loss": 0.8395, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.128955642733877e-06, | |
| "loss": 0.8171, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.072331107902713e-06, | |
| "loss": 0.8389, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.0162722882508072e-06, | |
| "loss": 0.8271, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.960790016056801e-06, | |
| "loss": 0.8552, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.90589501219273e-06, | |
| "loss": 0.8264, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.8515978840524302e-06, | |
| "loss": 0.8391, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.7979091235018564e-06, | |
| "loss": 0.8331, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.7448391048517378e-06, | |
| "loss": 0.8261, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.6923980828529424e-06, | |
| "loss": 0.8287, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.640596190714947e-06, | |
| "loss": 0.8568, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.589443438147789e-06, | |
| "loss": 0.8773, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.5389497094278861e-06, | |
| "loss": 0.8582, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.4891247614880904e-06, | |
| "loss": 0.8434, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.4399782220323515e-06, | |
| "loss": 0.8256, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.3915195876753495e-06, | |
| "loss": 0.8252, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.3437582221074574e-06, | |
| "loss": 0.8074, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.2967033542853918e-06, | |
| "loss": 0.8385, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.250364076648894e-06, | |
| "loss": 0.8102, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.2047493433637935e-06, | |
| "loss": 0.8655, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.1598679685917901e-06, | |
| "loss": 0.8143, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.1157286247872873e-06, | |
| "loss": 0.8163, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.0723398410216085e-06, | |
| "loss": 0.8289, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.0297100013349181e-06, | |
| "loss": 0.8151, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.878473431161767e-07, | |
| "loss": 0.8052, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.467599555114137e-07, | |
| "loss": 0.7995, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.064557778606631e-07, | |
| "loss": 0.8933, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 8.669425981638413e-07, | |
| "loss": 0.847, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.282280515758639e-07, | |
| "loss": 0.8323, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.903196189313039e-07, | |
| "loss": 0.8677, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.532246252988617e-07, | |
| "loss": 0.8319, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.16950238565941e-07, | |
| "loss": 0.8074, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.815034680535915e-07, | |
| "loss": 0.781, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.46891163162095e-07, | |
| "loss": 0.8195, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.131200120474512e-07, | |
| "loss": 0.8607, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5.801965403290221e-07, | |
| "loss": 0.8323, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5.481271098285818e-07, | |
| "loss": 0.8281, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5.169179173410178e-07, | |
| "loss": 0.8104, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.865749934369224e-07, | |
| "loss": 0.8208, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.571042012972993e-07, | |
| "loss": 0.828, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.2851123558061927e-07, | |
| "loss": 0.81, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.008016213224408e-07, | |
| "loss": 0.8513, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.739807128677986e-07, | |
| "loss": 0.8184, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.480536928365824e-07, | |
| "loss": 0.814, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.230255711220992e-07, | |
| "loss": 0.8325, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.9890118392300493e-07, | |
| "loss": 0.8353, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.756851928088056e-07, | |
| "loss": 0.8306, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.533820838190959e-07, | |
| "loss": 0.8459, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.3199616659672352e-07, | |
| "loss": 0.8497, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.1153157355503274e-07, | |
| "loss": 0.8122, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.9199225907935492e-07, | |
| "loss": 0.8373, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.7338199876289984e-07, | |
| "loss": 0.8326, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.5570438867719695e-07, | |
| "loss": 0.8571, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.3896284467722398e-07, | |
| "loss": 0.8512, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.2316060174136e-07, | |
| "loss": 0.8574, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.0830071334628655e-07, | |
| "loss": 0.8303, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 9.43860508769645e-08, | |
| "loss": 0.8163, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 8.141930307179468e-08, | |
| "loss": 0.838, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.940297550306895e-08, | |
| "loss": 0.8343, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.833939009282086e-08, | |
| "loss": 0.8288, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.823068466415615e-08, | |
| "loss": 0.8798, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.907881252816048e-08, | |
| "loss": 0.8127, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.088554210646133e-08, | |
| "loss": 0.8718, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.3652456589512983e-08, | |
| "loss": 0.8641, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.7380953630678488e-08, | |
| "loss": 0.8025, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.2072245076156786e-08, | |
| "loss": 0.8217, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 7.727356730820035e-09, | |
| "loss": 0.8422, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.347128159993829e-09, | |
| "loss": 0.8371, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.9322125272297488e-09, | |
| "loss": 0.8539, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.830764680946453e-10, | |
| "loss": 0.8061, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.8302, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 233, | |
| "total_flos": 4.1801555584619643e+18, | |
| "train_loss": 0.8414768988482431, | |
| "train_runtime": 12374.6376, | |
| "train_samples_per_second": 4.824, | |
| "train_steps_per_second": 0.019 | |
| } | |
| ], | |
| "max_steps": 233, | |
| "num_train_epochs": 1, | |
| "total_flos": 4.1801555584619643e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |