| { | |
| "best_metric": 0.974936306476593, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-100", | |
| "epoch": 0.3714020427112349, | |
| "eval_steps": 50, | |
| "global_step": 100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.003714020427112349, | |
| "grad_norm": 1.807255744934082, | |
| "learning_rate": 1.6666666666666668e-07, | |
| "loss": 1.3609, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.003714020427112349, | |
| "eval_loss": 3.732529640197754, | |
| "eval_runtime": 33.8777, | |
| "eval_samples_per_second": 13.401, | |
| "eval_steps_per_second": 1.683, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.007428040854224698, | |
| "grad_norm": 3.0510687828063965, | |
| "learning_rate": 3.3333333333333335e-07, | |
| "loss": 2.1135, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.011142061281337047, | |
| "grad_norm": 3.5649664402008057, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 2.2581, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.014856081708449397, | |
| "grad_norm": 3.335415840148926, | |
| "learning_rate": 6.666666666666667e-07, | |
| "loss": 2.13, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.018570102135561744, | |
| "grad_norm": 4.153068542480469, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 2.5975, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.022284122562674095, | |
| "grad_norm": 3.5235085487365723, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 2.4077, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.025998142989786442, | |
| "grad_norm": 3.9450502395629883, | |
| "learning_rate": 1.1666666666666668e-06, | |
| "loss": 2.7531, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.029712163416898793, | |
| "grad_norm": 4.102875232696533, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 2.7811, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.033426183844011144, | |
| "grad_norm": 4.623213768005371, | |
| "learning_rate": 1.5e-06, | |
| "loss": 2.9448, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.03714020427112349, | |
| "grad_norm": 4.344907760620117, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 2.8671, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04085422469823584, | |
| "grad_norm": 4.495368480682373, | |
| "learning_rate": 1.8333333333333333e-06, | |
| "loss": 3.0411, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.04456824512534819, | |
| "grad_norm": 4.336016654968262, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 2.9452, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.04828226555246054, | |
| "grad_norm": 4.537540435791016, | |
| "learning_rate": 2.166666666666667e-06, | |
| "loss": 3.0757, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.051996285979572884, | |
| "grad_norm": 4.8718953132629395, | |
| "learning_rate": 2.3333333333333336e-06, | |
| "loss": 3.119, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.055710306406685235, | |
| "grad_norm": 4.760570049285889, | |
| "learning_rate": 2.5e-06, | |
| "loss": 3.0015, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.059424326833797586, | |
| "grad_norm": 4.955251216888428, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 3.2608, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.06313834726090993, | |
| "grad_norm": 4.898932456970215, | |
| "learning_rate": 2.8333333333333335e-06, | |
| "loss": 3.1958, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.06685236768802229, | |
| "grad_norm": 4.579758644104004, | |
| "learning_rate": 3e-06, | |
| "loss": 2.9159, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.07056638811513463, | |
| "grad_norm": 4.835488796234131, | |
| "learning_rate": 3.1666666666666667e-06, | |
| "loss": 3.0749, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.07428040854224698, | |
| "grad_norm": 4.509119510650635, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 3.0182, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.07799442896935933, | |
| "grad_norm": 4.78074312210083, | |
| "learning_rate": 3.5e-06, | |
| "loss": 3.3061, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.08170844939647168, | |
| "grad_norm": 4.850345611572266, | |
| "learning_rate": 3.6666666666666666e-06, | |
| "loss": 3.4289, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.08542246982358404, | |
| "grad_norm": 4.934477806091309, | |
| "learning_rate": 3.833333333333334e-06, | |
| "loss": 3.0821, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.08913649025069638, | |
| "grad_norm": 4.942277908325195, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 3.1767, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.09285051067780872, | |
| "grad_norm": 5.097426891326904, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 3.5164, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.09656453110492108, | |
| "grad_norm": 4.779514312744141, | |
| "learning_rate": 4.333333333333334e-06, | |
| "loss": 3.1813, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.10027855153203342, | |
| "grad_norm": 4.577581405639648, | |
| "learning_rate": 4.5e-06, | |
| "loss": 2.941, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.10399257195914577, | |
| "grad_norm": 4.432027339935303, | |
| "learning_rate": 4.666666666666667e-06, | |
| "loss": 3.0727, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.10770659238625813, | |
| "grad_norm": 4.62089729309082, | |
| "learning_rate": 4.833333333333333e-06, | |
| "loss": 3.1745, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.11142061281337047, | |
| "grad_norm": 5.194108486175537, | |
| "learning_rate": 5e-06, | |
| "loss": 3.2936, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.11513463324048283, | |
| "grad_norm": 4.469546794891357, | |
| "learning_rate": 4.997482666353287e-06, | |
| "loss": 2.911, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.11884865366759517, | |
| "grad_norm": 4.538417339324951, | |
| "learning_rate": 4.989935734988098e-06, | |
| "loss": 3.1691, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.12256267409470752, | |
| "grad_norm": 4.686911582946777, | |
| "learning_rate": 4.977374404419838e-06, | |
| "loss": 2.6434, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.12627669452181986, | |
| "grad_norm": 4.282558441162109, | |
| "learning_rate": 4.959823971496575e-06, | |
| "loss": 2.6201, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.12999071494893222, | |
| "grad_norm": 4.451693534851074, | |
| "learning_rate": 4.937319780454559e-06, | |
| "loss": 2.816, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.13370473537604458, | |
| "grad_norm": 4.864418029785156, | |
| "learning_rate": 4.909907151739634e-06, | |
| "loss": 2.6204, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.1374187558031569, | |
| "grad_norm": 4.817498207092285, | |
| "learning_rate": 4.8776412907378845e-06, | |
| "loss": 2.3239, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.14113277623026926, | |
| "grad_norm": 4.677555084228516, | |
| "learning_rate": 4.8405871765993435e-06, | |
| "loss": 2.3435, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.14484679665738162, | |
| "grad_norm": 4.646546363830566, | |
| "learning_rate": 4.7988194313786275e-06, | |
| "loss": 2.5372, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.14856081708449395, | |
| "grad_norm": 4.66044807434082, | |
| "learning_rate": 4.752422169756048e-06, | |
| "loss": 2.5341, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1522748375116063, | |
| "grad_norm": 4.495102882385254, | |
| "learning_rate": 4.701488829641845e-06, | |
| "loss": 2.2117, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.15598885793871867, | |
| "grad_norm": 4.308347702026367, | |
| "learning_rate": 4.646121984004666e-06, | |
| "loss": 1.9536, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.15970287836583102, | |
| "grad_norm": 4.461782455444336, | |
| "learning_rate": 4.586433134303257e-06, | |
| "loss": 2.3499, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.16341689879294335, | |
| "grad_norm": 4.448497295379639, | |
| "learning_rate": 4.522542485937369e-06, | |
| "loss": 1.6131, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.1671309192200557, | |
| "grad_norm": 4.729413032531738, | |
| "learning_rate": 4.454578706170075e-06, | |
| "loss": 1.7194, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.17084493964716807, | |
| "grad_norm": 4.8837995529174805, | |
| "learning_rate": 4.382678665009028e-06, | |
| "loss": 2.0311, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.1745589600742804, | |
| "grad_norm": 5.577359199523926, | |
| "learning_rate": 4.3069871595684795e-06, | |
| "loss": 2.2559, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.17827298050139276, | |
| "grad_norm": 6.341835021972656, | |
| "learning_rate": 4.227656622467162e-06, | |
| "loss": 1.9221, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.18198700092850512, | |
| "grad_norm": 7.828035354614258, | |
| "learning_rate": 4.144846814849282e-06, | |
| "loss": 2.4759, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.18570102135561745, | |
| "grad_norm": 8.434853553771973, | |
| "learning_rate": 4.058724504646834e-06, | |
| "loss": 3.2325, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.18570102135561745, | |
| "eval_loss": 1.611772060394287, | |
| "eval_runtime": 33.8579, | |
| "eval_samples_per_second": 13.409, | |
| "eval_steps_per_second": 1.684, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.1894150417827298, | |
| "grad_norm": 1.2142438888549805, | |
| "learning_rate": 3.969463130731183e-06, | |
| "loss": 0.8321, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.19312906220984216, | |
| "grad_norm": 2.127845048904419, | |
| "learning_rate": 3.8772424536302565e-06, | |
| "loss": 1.1498, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.1968430826369545, | |
| "grad_norm": 2.3324849605560303, | |
| "learning_rate": 3.782248193514766e-06, | |
| "loss": 1.2167, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.20055710306406685, | |
| "grad_norm": 2.875474691390991, | |
| "learning_rate": 3.684671656182497e-06, | |
| "loss": 1.3003, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.2042711234911792, | |
| "grad_norm": 2.1670145988464355, | |
| "learning_rate": 3.5847093477938955e-06, | |
| "loss": 1.3164, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.20798514391829154, | |
| "grad_norm": 2.674121379852295, | |
| "learning_rate": 3.4825625791348093e-06, | |
| "loss": 1.2361, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.2116991643454039, | |
| "grad_norm": 2.671863555908203, | |
| "learning_rate": 3.3784370602033572e-06, | |
| "loss": 1.2381, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.21541318477251625, | |
| "grad_norm": 2.4426980018615723, | |
| "learning_rate": 3.272542485937369e-06, | |
| "loss": 1.3495, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.2191272051996286, | |
| "grad_norm": 2.743527412414551, | |
| "learning_rate": 3.165092113916688e-06, | |
| "loss": 1.3194, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.22284122562674094, | |
| "grad_norm": 2.485083818435669, | |
| "learning_rate": 3.056302334890786e-06, | |
| "loss": 1.3947, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.2265552460538533, | |
| "grad_norm": 2.3887181282043457, | |
| "learning_rate": 2.946392236996592e-06, | |
| "loss": 1.2135, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.23026926648096566, | |
| "grad_norm": 2.7468771934509277, | |
| "learning_rate": 2.835583164544139e-06, | |
| "loss": 1.2523, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.233983286908078, | |
| "grad_norm": 2.5111334323883057, | |
| "learning_rate": 2.724098272258584e-06, | |
| "loss": 1.2471, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.23769730733519034, | |
| "grad_norm": 2.4568049907684326, | |
| "learning_rate": 2.6121620758762877e-06, | |
| "loss": 1.1975, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.2414113277623027, | |
| "grad_norm": 2.5363759994506836, | |
| "learning_rate": 2.5e-06, | |
| "loss": 1.1698, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.24512534818941503, | |
| "grad_norm": 2.9774973392486572, | |
| "learning_rate": 2.3878379241237136e-06, | |
| "loss": 0.9025, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.2488393686165274, | |
| "grad_norm": 3.5985605716705322, | |
| "learning_rate": 2.2759017277414165e-06, | |
| "loss": 1.22, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.2525533890436397, | |
| "grad_norm": 2.659881830215454, | |
| "learning_rate": 2.1644168354558623e-06, | |
| "loss": 1.1345, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.2562674094707521, | |
| "grad_norm": 2.756131172180176, | |
| "learning_rate": 2.053607763003409e-06, | |
| "loss": 1.1833, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.25998142989786444, | |
| "grad_norm": 2.8494410514831543, | |
| "learning_rate": 1.9436976651092143e-06, | |
| "loss": 1.3538, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.26369545032497677, | |
| "grad_norm": 2.196274518966675, | |
| "learning_rate": 1.8349078860833125e-06, | |
| "loss": 0.8695, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.26740947075208915, | |
| "grad_norm": 2.421745538711548, | |
| "learning_rate": 1.7274575140626318e-06, | |
| "loss": 1.1275, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.2711234911792015, | |
| "grad_norm": 3.1137969493865967, | |
| "learning_rate": 1.6215629397966432e-06, | |
| "loss": 1.2171, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.2748375116063138, | |
| "grad_norm": 2.646501064300537, | |
| "learning_rate": 1.5174374208651913e-06, | |
| "loss": 1.234, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.2785515320334262, | |
| "grad_norm": 2.5165648460388184, | |
| "learning_rate": 1.415290652206105e-06, | |
| "loss": 1.1426, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.2822655524605385, | |
| "grad_norm": 2.521559476852417, | |
| "learning_rate": 1.3153283438175036e-06, | |
| "loss": 1.0068, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.28597957288765086, | |
| "grad_norm": 2.283813238143921, | |
| "learning_rate": 1.217751806485235e-06, | |
| "loss": 1.1615, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.28969359331476324, | |
| "grad_norm": 3.4305078983306885, | |
| "learning_rate": 1.122757546369744e-06, | |
| "loss": 1.1356, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.2934076137418756, | |
| "grad_norm": 2.714026927947998, | |
| "learning_rate": 1.0305368692688175e-06, | |
| "loss": 1.086, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.2971216341689879, | |
| "grad_norm": 2.8863515853881836, | |
| "learning_rate": 9.412754953531664e-07, | |
| "loss": 0.854, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.3008356545961003, | |
| "grad_norm": 2.6695327758789062, | |
| "learning_rate": 8.551531851507186e-07, | |
| "loss": 1.1173, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.3045496750232126, | |
| "grad_norm": 2.6120901107788086, | |
| "learning_rate": 7.723433775328385e-07, | |
| "loss": 0.9034, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.308263695450325, | |
| "grad_norm": 2.606943368911743, | |
| "learning_rate": 6.930128404315214e-07, | |
| "loss": 1.0242, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.31197771587743733, | |
| "grad_norm": 2.955063819885254, | |
| "learning_rate": 6.17321334990973e-07, | |
| "loss": 0.9499, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.31569173630454966, | |
| "grad_norm": 2.4737842082977295, | |
| "learning_rate": 5.454212938299256e-07, | |
| "loss": 1.0119, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.31940575673166205, | |
| "grad_norm": 3.265795946121216, | |
| "learning_rate": 4.774575140626317e-07, | |
| "loss": 0.855, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.3231197771587744, | |
| "grad_norm": 3.479175090789795, | |
| "learning_rate": 4.1356686569674344e-07, | |
| "loss": 0.8147, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.3268337975858867, | |
| "grad_norm": 3.1530840396881104, | |
| "learning_rate": 3.538780159953348e-07, | |
| "loss": 1.1186, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.3305478180129991, | |
| "grad_norm": 4.578067779541016, | |
| "learning_rate": 2.98511170358155e-07, | |
| "loss": 0.9307, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.3342618384401114, | |
| "grad_norm": 3.7903590202331543, | |
| "learning_rate": 2.4757783024395244e-07, | |
| "loss": 0.8859, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.33797585886722376, | |
| "grad_norm": 4.344192028045654, | |
| "learning_rate": 2.0118056862137358e-07, | |
| "loss": 0.9719, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.34168987929433614, | |
| "grad_norm": 6.286164283752441, | |
| "learning_rate": 1.59412823400657e-07, | |
| "loss": 1.1164, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.34540389972144847, | |
| "grad_norm": 4.787858963012695, | |
| "learning_rate": 1.223587092621162e-07, | |
| "loss": 1.0283, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.3491179201485608, | |
| "grad_norm": 7.135058403015137, | |
| "learning_rate": 9.00928482603669e-08, | |
| "loss": 0.9365, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.3528319405756732, | |
| "grad_norm": 7.054929256439209, | |
| "learning_rate": 6.268021954544095e-08, | |
| "loss": 1.0428, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.3565459610027855, | |
| "grad_norm": 8.841050148010254, | |
| "learning_rate": 4.017602850342584e-08, | |
| "loss": 0.9227, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.36025998142989785, | |
| "grad_norm": 9.116321563720703, | |
| "learning_rate": 2.262559558016325e-08, | |
| "loss": 1.0278, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.36397400185701023, | |
| "grad_norm": 8.069734573364258, | |
| "learning_rate": 1.006426501190233e-08, | |
| "loss": 1.2367, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.36768802228412256, | |
| "grad_norm": 10.388524055480957, | |
| "learning_rate": 2.5173336467135266e-09, | |
| "loss": 1.0777, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.3714020427112349, | |
| "grad_norm": 9.496840476989746, | |
| "learning_rate": 0.0, | |
| "loss": 1.8257, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.3714020427112349, | |
| "eval_loss": 0.974936306476593, | |
| "eval_runtime": 33.8821, | |
| "eval_samples_per_second": 13.399, | |
| "eval_steps_per_second": 1.682, | |
| "step": 100 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 100, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.35249644224512e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |