| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.005349166867260424, | |
| "global_step": 150, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.061666666666666e-05, | |
| "loss": 4.7122, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 6.123333333333332e-05, | |
| "loss": 4.6911, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.185e-05, | |
| "loss": 4.724, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00012246666666666664, | |
| "loss": 4.6028, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00015308333333333333, | |
| "loss": 4.2087, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001837, | |
| "loss": 3.9055, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018369229239676365, | |
| "loss": 3.6594, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018366917088062342, | |
| "loss": 3.3363, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018363063933206865, | |
| "loss": 3.1973, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018357670421785793, | |
| "loss": 3.0688, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001835073745899338, | |
| "loss": 2.8881, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018342266208390355, | |
| "loss": 2.9587, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018332258091708636, | |
| "loss": 2.844, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018320714788612737, | |
| "loss": 2.7755, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001830763823641785, | |
| "loss": 2.7166, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018293030629764714, | |
| "loss": 2.6935, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018276894420251292, | |
| "loss": 2.7225, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018259232316021308, | |
| "loss": 2.6965, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018240047281309753, | |
| "loss": 2.6583, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018219342535945378, | |
| "loss": 2.6823, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018197121554810325, | |
| "loss": 2.6228, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018173388067256935, | |
| "loss": 2.4769, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018148146056481843, | |
| "loss": 2.437, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001812139975885748, | |
| "loss": 2.4892, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018093153663221082, | |
| "loss": 2.4044, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018063412510121334, | |
| "loss": 2.3818, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00018032181291022746, | |
| "loss": 2.3269, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00017999465247467954, | |
| "loss": 2.348, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00017965269870198018, | |
| "loss": 2.2777, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001792960089823091, | |
| "loss": 2.2388, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00017892464317898344, | |
| "loss": 2.354, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00017853866361841074, | |
| "loss": 2.2417, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001781381350796288, | |
| "loss": 2.2249, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00017772312478343374, | |
| "loss": 2.1922, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00017729370238109832, | |
| "loss": 2.0696, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001768499399426824, | |
| "loss": 2.0623, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001763919119449373, | |
| "loss": 2.1225, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001759196952588065, | |
| "loss": 2.0444, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00017543336913652426, | |
| "loss": 2.1441, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001749330151983147, | |
| "loss": 2.0458, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00017441871741869357, | |
| "loss": 2.109, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001738905621123746, | |
| "loss": 2.1298, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001733486379197835, | |
| "loss": 2.0426, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001727930357921812, | |
| "loss": 2.0356, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00017222384897639967, | |
| "loss": 1.9853, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00017164117299919217, | |
| "loss": 2.0567, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.000171045105651201, | |
| "loss": 1.9754, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00017043574697054526, | |
| "loss": 2.07, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001698131992260315, | |
| "loss": 1.9816, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001691775668999898, | |
| "loss": 2.047, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00016852895667073854, | |
| "loss": 2.0334, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00016786747739468068, | |
| "loss": 2.012, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001671932400880341, | |
| "loss": 1.968, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001665063579082, | |
| "loss": 1.9014, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00016580694613477136, | |
| "loss": 1.9496, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00016509512215018588, | |
| "loss": 1.9224, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00016437100542002542, | |
| "loss": 1.9407, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00016363471747296617, | |
| "loss": 1.9245, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00016288638188038235, | |
| "loss": 1.8933, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00016212612423560733, | |
| "loss": 1.9372, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001613540721328553, | |
| "loss": 1.9722, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001605703551458069, | |
| "loss": 1.92, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00015977510480586304, | |
| "loss": 1.9598, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001589684545800698, | |
| "loss": 1.8803, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00015815053984871886, | |
| "loss": 1.9744, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00015732149788262627, | |
| "loss": 1.8559, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001564814678200945, | |
| "loss": 1.938, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001556305906435608, | |
| "loss": 1.9794, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00015476900915593596, | |
| "loss": 1.9055, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001538968679566378, | |
| "loss": 1.9784, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00015301431341732292, | |
| "loss": 1.9461, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00015212149365732115, | |
| "loss": 1.91, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00015121855851877672, | |
| "loss": 1.927, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00015030565954150013, | |
| "loss": 1.9173, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00014938294993753506, | |
| "loss": 1.9565, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00014845058456544515, | |
| "loss": 1.9549, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00014750871990432368, | |
| "loss": 1.9234, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00014655751402753173, | |
| "loss": 1.9352, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00014559712657616884, | |
| "loss": 1.902, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00014462771873228027, | |
| "loss": 1.9706, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00014364945319180576, | |
| "loss": 1.8017, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001426624941372744, | |
| "loss": 1.8949, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00014166700721024962, | |
| "loss": 1.9567, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00014066315948352973, | |
| "loss": 1.773, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00013965111943310788, | |
| "loss": 1.8803, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001386310569098969, | |
| "loss": 1.9013, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00013760314311122285, | |
| "loss": 1.8125, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001365675505520934, | |
| "loss": 1.8461, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00013552445303624408, | |
| "loss": 1.866, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00013447402562696928, | |
| "loss": 1.8816, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001334164446177409, | |
| "loss": 1.8429, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00013235188750262118, | |
| "loss": 1.8838, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00013128053294647385, | |
| "loss": 1.8784, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00013020256075497866, | |
| "loss": 1.8919, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001291181518444547, | |
| "loss": 1.9255, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.000128027488211497, | |
| "loss": 1.8819, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001269307529024322, | |
| "loss": 1.7187, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00012582812998259752, | |
| "loss": 1.8393, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001247198045054494, | |
| "loss": 1.899, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00012360596248150578, | |
| "loss": 1.8248, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00012248679084712788, | |
| "loss": 1.7758, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00012136247743314659, | |
| "loss": 1.8896, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00012023321093333892, | |
| "loss": 1.8227, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001190991808727594, | |
| "loss": 1.7834, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00011796057757593203, | |
| "loss": 1.9018, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001168175921349081, | |
| "loss": 1.8601, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00011567041637719514, | |
| "loss": 1.77, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00011451924283356254, | |
| "loss": 1.7776, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00011336426470572898, | |
| "loss": 1.7542, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00011220567583393738, | |
| "loss": 1.9201, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00011104367066442263, | |
| "loss": 1.7519, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00010987844421677765, | |
| "loss": 1.7057, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00010871019205122319, | |
| "loss": 1.8886, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00010753911023578701, | |
| "loss": 1.7797, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00010636539531339759, | |
| "loss": 1.8853, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00010518924426889829, | |
| "loss": 1.826, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00010401085449598752, | |
| "loss": 1.7474, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00010283042376408993, | |
| "loss": 1.8552, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00010164815018516475, | |
| "loss": 1.7706, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00010046423218045675, | |
| "loss": 1.7976, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.927886844719498e-05, | |
| "loss": 1.7527, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.80922579252454e-05, | |
| "loss": 1.8078, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.690459976372293e-05, | |
| "loss": 1.8556, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.571609328756794e-05, | |
| "loss": 1.7546, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.452693796409372e-05, | |
| "loss": 1.8492, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.333733336950976e-05, | |
| "loss": 1.8373, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.214747915542679e-05, | |
| "loss": 1.8429, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.095757501534924e-05, | |
| "loss": 1.8388, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.976782065116062e-05, | |
| "loss": 1.7977, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.857841573960755e-05, | |
| "loss": 1.8325, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.738955989878794e-05, | |
| "loss": 1.8479, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.620145265464898e-05, | |
| "loss": 1.65, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.501429340750077e-05, | |
| "loss": 1.8146, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.382828139855082e-05, | |
| "loss": 1.7693, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.264361567646539e-05, | |
| "loss": 1.9136, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.146049506396317e-05, | |
| "loss": 1.8657, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.02791181244466e-05, | |
| "loss": 1.8238, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 7.909968312867728e-05, | |
| "loss": 1.8138, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 7.792238802149986e-05, | |
| "loss": 1.8056, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 7.674743038862096e-05, | |
| "loss": 1.8203, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.557500742344842e-05, | |
| "loss": 1.7885, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.440531589399613e-05, | |
| "loss": 1.737, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.323855210986062e-05, | |
| "loss": 1.796, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.207491188927419e-05, | |
| "loss": 1.7995, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.091459052624075e-05, | |
| "loss": 1.7675, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.97577827577596e-05, | |
| "loss": 1.851, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.860468273114271e-05, | |
| "loss": 1.7414, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.745548397143076e-05, | |
| "loss": 1.819, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.631037934891401e-05, | |
| "loss": 1.8442, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.51695610467628e-05, | |
| "loss": 1.7498, | |
| "step": 150 | |
| } | |
| ], | |
| "max_steps": 200, | |
| "num_train_epochs": 1, | |
| "total_flos": 7.555500537053184e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |