| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 217, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.857142857142857e-05, | |
| "loss": 1.536, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.714285714285714e-05, | |
| "loss": 1.525, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.571428571428571e-05, | |
| "loss": 1.2374, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011428571428571428, | |
| "loss": 1.1414, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00014285714285714287, | |
| "loss": 1.0714, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00017142857142857143, | |
| "loss": 1.0713, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 0.97, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019998881018102737, | |
| "loss": 2.257, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019995524322835034, | |
| "loss": 1.001, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00019989930665413147, | |
| "loss": 0.9339, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001998210129767735, | |
| "loss": 0.9392, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00019972037971811802, | |
| "loss": 0.8991, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00019959742939952392, | |
| "loss": 0.9341, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00019945218953682734, | |
| "loss": 0.9036, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00019928469263418374, | |
| "loss": 0.9214, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00019909497617679348, | |
| "loss": 0.9046, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00019888308262251285, | |
| "loss": 0.91, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00019864905939235214, | |
| "loss": 0.971, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00019839295885986296, | |
| "loss": 0.9479, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00019811483833941728, | |
| "loss": 0.9127, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00019781476007338058, | |
| "loss": 0.9677, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00019749279121818235, | |
| "loss": 0.9415, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00019714900382928675, | |
| "loss": 1.0188, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00019678347484506669, | |
| "loss": 1.1678, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00019639628606958533, | |
| "loss": 1.1167, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0001959875241542889, | |
| "loss": 1.0513, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0001955572805786141, | |
| "loss": 0.9545, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00019510565162951537, | |
| "loss": 0.8698, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00019463273837991643, | |
| "loss": 0.8928, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00019413864666609034, | |
| "loss": 0.9672, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00019362348706397373, | |
| "loss": 0.847, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00019308737486442045, | |
| "loss": 0.8751, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00019253043004739968, | |
| "loss": 0.8916, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0001919527772551451, | |
| "loss": 0.8536, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0001913545457642601, | |
| "loss": 0.8425, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0001907358694567865, | |
| "loss": 0.8738, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0001900968867902419, | |
| "loss": 0.9022, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0001894377407666337, | |
| "loss": 0.9337, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.00018875857890045543, | |
| "loss": 0.8617, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0001880595531856738, | |
| "loss": 0.8994, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.00018734082006171299, | |
| "loss": 0.9195, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.00018660254037844388, | |
| "loss": 0.9204, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00018584487936018661, | |
| "loss": 0.8778, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00018506800656873398, | |
| "loss": 0.8408, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0001842720958654039, | |
| "loss": 0.9217, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00018345732537213027, | |
| "loss": 0.871, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0001826238774315995, | |
| "loss": 0.9053, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00018177193856644316, | |
| "loss": 0.9292, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00018090169943749476, | |
| "loss": 0.8279, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00018001335480112064, | |
| "loss": 0.8972, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00017910710346563416, | |
| "loss": 0.8518, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.000178183148246803, | |
| "loss": 0.8356, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00017724169592245995, | |
| "loss": 0.889, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00017628295718622665, | |
| "loss": 0.8499, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00017530714660036112, | |
| "loss": 0.8789, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.00017431448254773944, | |
| "loss": 0.8757, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.00017330518718298264, | |
| "loss": 0.9377, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00017227948638273916, | |
| "loss": 0.8636, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0001712376096951345, | |
| "loss": 0.8761, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00017017979028839916, | |
| "loss": 0.8737, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00016910626489868649, | |
| "loss": 0.8604, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00016801727377709194, | |
| "loss": 0.8085, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00016691306063588583, | |
| "loss": 0.8978, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00016579387259397127, | |
| "loss": 0.9272, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.00016465996012157995, | |
| "loss": 0.976, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0001635115769842179, | |
| "loss": 0.8097, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00016234898018587337, | |
| "loss": 0.8192, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00016117242991150064, | |
| "loss": 0.8579, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.00015998218946879138, | |
| "loss": 0.7813, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.00015877852522924732, | |
| "loss": 0.8448, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.00015756170656856737, | |
| "loss": 0.8173, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0001563320058063622, | |
| "loss": 0.8345, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00015508969814521025, | |
| "loss": 0.8009, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00015383506160906825, | |
| "loss": 0.8763, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00015256837698105047, | |
| "loss": 0.8516, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00015128992774059063, | |
| "loss": 0.8166, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00015000000000000001, | |
| "loss": 0.8473, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.00014869888244043673, | |
| "loss": 0.8136, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.00014738686624729986, | |
| "loss": 0.8285, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00014606424504506324, | |
| "loss": 0.8474, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00014473131483156327, | |
| "loss": 0.8421, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.00014338837391175582, | |
| "loss": 0.7994, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.00014203572283095657, | |
| "loss": 0.865, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00014067366430758004, | |
| "loss": 0.8603, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00013930250316539238, | |
| "loss": 0.8355, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.00013792254626529286, | |
| "loss": 0.8517, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.00013653410243663952, | |
| "loss": 0.8496, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0001351374824081343, | |
| "loss": 0.7882, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.00013373299873828303, | |
| "loss": 0.7927, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.00013232096574544602, | |
| "loss": 0.8434, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.00013090169943749476, | |
| "loss": 0.9031, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.00012947551744109043, | |
| "loss": 0.8491, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00012804273893060028, | |
| "loss": 0.8433, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00012660368455666752, | |
| "loss": 0.8862, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00012515867637445086, | |
| "loss": 0.8081, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00012370803777154977, | |
| "loss": 0.8594, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.00012225209339563145, | |
| "loss": 0.8561, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.00012079116908177593, | |
| "loss": 0.7771, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00011932559177955533, | |
| "loss": 0.8719, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00011785568947986367, | |
| "loss": 0.8254, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00011638179114151377, | |
| "loss": 0.8607, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00011490422661761744, | |
| "loss": 0.8523, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00011342332658176555, | |
| "loss": 0.7828, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00011193942245402443, | |
| "loss": 0.8125, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00011045284632676536, | |
| "loss": 0.8394, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00010896393089034336, | |
| "loss": 0.8442, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00010747300935864243, | |
| "loss": 0.7688, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00010598041539450343, | |
| "loss": 0.7935, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00010448648303505151, | |
| "loss": 0.8522, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.00010299154661693987, | |
| "loss": 0.799, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.00010149594070152638, | |
| "loss": 0.9002, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0001, | |
| "loss": 0.824, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.850405929847366e-05, | |
| "loss": 0.8112, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.700845338306018e-05, | |
| "loss": 0.7934, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.551351696494854e-05, | |
| "loss": 0.757, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.401958460549658e-05, | |
| "loss": 0.8035, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.252699064135758e-05, | |
| "loss": 0.8351, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.103606910965666e-05, | |
| "loss": 0.7876, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.954715367323468e-05, | |
| "loss": 0.8538, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.806057754597558e-05, | |
| "loss": 0.7701, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.657667341823448e-05, | |
| "loss": 0.8407, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.509577338238255e-05, | |
| "loss": 0.7777, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.361820885848624e-05, | |
| "loss": 0.7863, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.214431052013634e-05, | |
| "loss": 0.7934, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.067440822044469e-05, | |
| "loss": 0.8408, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.920883091822408e-05, | |
| "loss": 0.8094, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.774790660436858e-05, | |
| "loss": 0.8085, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.629196222845026e-05, | |
| "loss": 0.8305, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.484132362554915e-05, | |
| "loss": 0.8376, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.339631544333249e-05, | |
| "loss": 0.7636, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.195726106939974e-05, | |
| "loss": 0.8408, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.052448255890957e-05, | |
| "loss": 0.7903, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 6.909830056250527e-05, | |
| "loss": 0.8249, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.767903425455401e-05, | |
| "loss": 0.7758, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.626700126171702e-05, | |
| "loss": 0.841, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.486251759186572e-05, | |
| "loss": 0.7916, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.34658975633605e-05, | |
| "loss": 0.7983, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 6.207745373470716e-05, | |
| "loss": 0.8417, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 6.069749683460765e-05, | |
| "loss": 0.7911, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.9326335692419995e-05, | |
| "loss": 0.8527, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.796427716904347e-05, | |
| "loss": 0.741, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.6611626088244194e-05, | |
| "loss": 0.8258, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.526868516843673e-05, | |
| "loss": 0.7294, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.393575495493679e-05, | |
| "loss": 0.8324, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.261313375270014e-05, | |
| "loss": 0.7886, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.130111755956327e-05, | |
| "loss": 0.7995, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.000000000000002e-05, | |
| "loss": 0.7352, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.87100722594094e-05, | |
| "loss": 0.7491, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.743162301894952e-05, | |
| "loss": 0.8072, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.616493839093179e-05, | |
| "loss": 0.7869, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.491030185478976e-05, | |
| "loss": 0.8552, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.3667994193637796e-05, | |
| "loss": 0.796, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.2438293431432665e-05, | |
| "loss": 0.8093, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.12214747707527e-05, | |
| "loss": 0.7211, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.001781053120863e-05, | |
| "loss": 0.7327, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.8827570088499356e-05, | |
| "loss": 0.7804, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.7651019814126654e-05, | |
| "loss": 0.7857, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.6488423015782125e-05, | |
| "loss": 0.8032, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.534003987842005e-05, | |
| "loss": 0.8019, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.4206127406028745e-05, | |
| "loss": 0.7965, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.308693936411421e-05, | |
| "loss": 0.822, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.198272622290804e-05, | |
| "loss": 0.7934, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.089373510131354e-05, | |
| "loss": 0.7945, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.9820209711600854e-05, | |
| "loss": 0.7876, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.876239030486554e-05, | |
| "loss": 0.8116, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.7720513617260856e-05, | |
| "loss": 0.718, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.669481281701739e-05, | |
| "loss": 0.7309, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.5685517452260567e-05, | |
| "loss": 0.7675, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.4692853399638917e-05, | |
| "loss": 0.7928, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.371704281377335e-05, | |
| "loss": 0.7445, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.275830407754006e-05, | |
| "loss": 0.781, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.181685175319702e-05, | |
| "loss": 0.7934, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.0892896534365904e-05, | |
| "loss": 0.7815, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.9986645198879385e-05, | |
| "loss": 0.7563, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.9098300562505266e-05, | |
| "loss": 0.8097, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.8228061433556866e-05, | |
| "loss": 0.8249, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.7376122568400532e-05, | |
| "loss": 0.8009, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.6542674627869737e-05, | |
| "loss": 0.7568, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.5727904134596083e-05, | |
| "loss": 0.7726, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.4931993431266056e-05, | |
| "loss": 0.7236, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.415512063981339e-05, | |
| "loss": 0.7927, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.339745962155613e-05, | |
| "loss": 0.7514, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.2659179938287035e-05, | |
| "loss": 0.7458, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.19404468143262e-05, | |
| "loss": 0.7872, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.124142109954459e-05, | |
| "loss": 0.7613, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.0562259233366334e-05, | |
| "loss": 0.7444, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 9.903113209758096e-06, | |
| "loss": 0.7934, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.264130543213512e-06, | |
| "loss": 0.7432, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.645454235739903e-06, | |
| "loss": 0.7907, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.047222744854943e-06, | |
| "loss": 0.7454, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.46956995260033e-06, | |
| "loss": 0.7942, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 6.9126251355795864e-06, | |
| "loss": 0.8348, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 6.37651293602628e-06, | |
| "loss": 0.7413, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5.861353333909692e-06, | |
| "loss": 0.8235, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.367261620083575e-06, | |
| "loss": 0.7685, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.8943483704846475e-06, | |
| "loss": 0.7457, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.442719421385922e-06, | |
| "loss": 0.8274, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.012475845711106e-06, | |
| "loss": 0.8045, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.6037139304146762e-06, | |
| "loss": 0.7463, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.2165251549333587e-06, | |
| "loss": 0.7905, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.8509961707132494e-06, | |
| "loss": 0.8179, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.5072087818176382e-06, | |
| "loss": 0.7379, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.1852399266194314e-06, | |
| "loss": 0.7861, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.885161660582746e-06, | |
| "loss": 0.7344, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.6070411401370334e-06, | |
| "loss": 0.7692, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.350940607647866e-06, | |
| "loss": 0.7955, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.1169173774871478e-06, | |
| "loss": 0.8031, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 9.0502382320653e-07, | |
| "loss": 0.7685, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 7.153073658162646e-07, | |
| "loss": 0.8191, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 5.478104631726711e-07, | |
| "loss": 0.7824, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.025706004760932e-07, | |
| "loss": 0.8028, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.7962028188198706e-07, | |
| "loss": 0.7194, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.7898702322648453e-07, | |
| "loss": 0.8275, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.0069334586854107e-07, | |
| "loss": 0.7808, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.475677164966774e-08, | |
| "loss": 0.7297, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.1189818972656696e-08, | |
| "loss": 0.7465, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.7947, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 217, | |
| "total_flos": 3685402214400.0, | |
| "train_loss": 0.8509236571426215, | |
| "train_runtime": 481.0178, | |
| "train_samples_per_second": 7.201, | |
| "train_steps_per_second": 0.451 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 217, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50000, | |
| "total_flos": 3685402214400.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |