{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 263, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0038022813688212928, "grad_norm": 5.838922807698357, "learning_rate": 1.25e-06, "loss": 0.4415, "step": 1 }, { "epoch": 0.0076045627376425855, "grad_norm": 6.450188921175257, "learning_rate": 2.5e-06, "loss": 0.2714, "step": 2 }, { "epoch": 0.011406844106463879, "grad_norm": 6.328807457221606, "learning_rate": 3.7500000000000005e-06, "loss": 0.2392, "step": 3 }, { "epoch": 0.015209125475285171, "grad_norm": 3.1900170578862403, "learning_rate": 5e-06, "loss": 0.1902, "step": 4 }, { "epoch": 0.019011406844106463, "grad_norm": 4.396787628400217, "learning_rate": 6.25e-06, "loss": 0.2287, "step": 5 }, { "epoch": 0.022813688212927757, "grad_norm": 2.328876214447226, "learning_rate": 7.500000000000001e-06, "loss": 0.1073, "step": 6 }, { "epoch": 0.026615969581749048, "grad_norm": 1.9893066003363478, "learning_rate": 8.750000000000001e-06, "loss": 0.0892, "step": 7 }, { "epoch": 0.030418250950570342, "grad_norm": 3.4491162158646227, "learning_rate": 1e-05, "loss": 0.1859, "step": 8 }, { "epoch": 0.034220532319391636, "grad_norm": 3.391007553648945, "learning_rate": 9.999620550574155e-06, "loss": 0.1162, "step": 9 }, { "epoch": 0.03802281368821293, "grad_norm": 5.243929471727456, "learning_rate": 9.99848225988936e-06, "loss": 0.2052, "step": 10 }, { "epoch": 0.04182509505703422, "grad_norm": 2.556605157640159, "learning_rate": 9.996585300715117e-06, "loss": 0.1279, "step": 11 }, { "epoch": 0.045627376425855515, "grad_norm": 2.3229950077008232, "learning_rate": 9.99392996097145e-06, "loss": 0.2329, "step": 12 }, { "epoch": 0.049429657794676805, "grad_norm": 3.4538715709146754, "learning_rate": 9.990516643685222e-06, "loss": 0.0859, "step": 13 }, { "epoch": 0.053231939163498096, "grad_norm": 2.144930306306445, "learning_rate": 9.98634586692894e-06, "loss": 0.1541, "step": 14 }, { "epoch": 0.057034220532319393, "grad_norm": 1.6344989568819797, "learning_rate": 9.981418263742148e-06, "loss": 0.1433, "step": 15 }, { "epoch": 0.060836501901140684, "grad_norm": 1.2978801538360458, "learning_rate": 9.975734582035323e-06, "loss": 0.0885, "step": 16 }, { "epoch": 0.06463878326996197, "grad_norm": 2.0950541376792366, "learning_rate": 9.96929568447637e-06, "loss": 0.1319, "step": 17 }, { "epoch": 0.06844106463878327, "grad_norm": 1.0425316398679343, "learning_rate": 9.96210254835968e-06, "loss": 0.1279, "step": 18 }, { "epoch": 0.07224334600760456, "grad_norm": 1.7860823823073846, "learning_rate": 9.954156265457801e-06, "loss": 0.1857, "step": 19 }, { "epoch": 0.07604562737642585, "grad_norm": 1.7415782346194582, "learning_rate": 9.945458041855732e-06, "loss": 0.0717, "step": 20 }, { "epoch": 0.07984790874524715, "grad_norm": 1.0879866640913596, "learning_rate": 9.936009197767847e-06, "loss": 0.1016, "step": 21 }, { "epoch": 0.08365019011406843, "grad_norm": 2.2439360478661645, "learning_rate": 9.925811167337533e-06, "loss": 0.0935, "step": 22 }, { "epoch": 0.08745247148288973, "grad_norm": 1.8496647469351228, "learning_rate": 9.91486549841951e-06, "loss": 0.1117, "step": 23 }, { "epoch": 0.09125475285171103, "grad_norm": 1.1499446513004061, "learning_rate": 9.903173852344889e-06, "loss": 0.0786, "step": 24 }, { "epoch": 0.09505703422053231, "grad_norm": 1.0813157058702787, "learning_rate": 9.890738003669029e-06, "loss": 0.0466, "step": 25 }, { "epoch": 0.09885931558935361, "grad_norm": 1.1206071268561641, "learning_rate": 9.877559839902185e-06, "loss": 0.0473, "step": 26 }, { "epoch": 0.10266159695817491, "grad_norm": 1.2183071418887779, "learning_rate": 9.863641361223025e-06, "loss": 0.0625, "step": 27 }, { "epoch": 0.10646387832699619, "grad_norm": 1.7316764015595658, "learning_rate": 9.848984680175049e-06, "loss": 0.0485, "step": 28 }, { "epoch": 0.11026615969581749, "grad_norm": 0.6990239085195359, "learning_rate": 9.833592021345938e-06, "loss": 0.0573, "step": 29 }, { "epoch": 0.11406844106463879, "grad_norm": 2.2785738442682413, "learning_rate": 9.817465721029916e-06, "loss": 0.0796, "step": 30 }, { "epoch": 0.11787072243346007, "grad_norm": 1.0584228821124018, "learning_rate": 9.800608226873143e-06, "loss": 0.0403, "step": 31 }, { "epoch": 0.12167300380228137, "grad_norm": 2.5879877206601636, "learning_rate": 9.783022097502204e-06, "loss": 0.1062, "step": 32 }, { "epoch": 0.12547528517110265, "grad_norm": 2.5381277344322433, "learning_rate": 9.764710002135784e-06, "loss": 0.0769, "step": 33 }, { "epoch": 0.12927756653992395, "grad_norm": 1.0747071356178863, "learning_rate": 9.745674720179507e-06, "loss": 0.0544, "step": 34 }, { "epoch": 0.13307984790874525, "grad_norm": 3.102732036616969, "learning_rate": 9.7259191408041e-06, "loss": 0.0926, "step": 35 }, { "epoch": 0.13688212927756654, "grad_norm": 1.0789665373993582, "learning_rate": 9.705446262506858e-06, "loss": 0.041, "step": 36 }, { "epoch": 0.14068441064638784, "grad_norm": 1.872380809657841, "learning_rate": 9.684259192656554e-06, "loss": 0.0987, "step": 37 }, { "epoch": 0.1444866920152091, "grad_norm": 1.6725384234341496, "learning_rate": 9.66236114702178e-06, "loss": 0.0664, "step": 38 }, { "epoch": 0.1482889733840304, "grad_norm": 1.3155529867243292, "learning_rate": 9.639755449282874e-06, "loss": 0.0471, "step": 39 }, { "epoch": 0.1520912547528517, "grad_norm": 0.7375858425847438, "learning_rate": 9.616445530527448e-06, "loss": 0.0542, "step": 40 }, { "epoch": 0.155893536121673, "grad_norm": 2.5798050001278865, "learning_rate": 9.592434928729617e-06, "loss": 0.0851, "step": 41 }, { "epoch": 0.1596958174904943, "grad_norm": 0.9563326747806046, "learning_rate": 9.567727288213005e-06, "loss": 0.0701, "step": 42 }, { "epoch": 0.1634980988593156, "grad_norm": 1.2201713738820383, "learning_rate": 9.542326359097619e-06, "loss": 0.0483, "step": 43 }, { "epoch": 0.16730038022813687, "grad_norm": 1.5349602348665532, "learning_rate": 9.516235996730645e-06, "loss": 0.0493, "step": 44 }, { "epoch": 0.17110266159695817, "grad_norm": 1.2129366778348925, "learning_rate": 9.489460161101291e-06, "loss": 0.0423, "step": 45 }, { "epoch": 0.17490494296577946, "grad_norm": 1.5701039195877315, "learning_rate": 9.46200291623974e-06, "loss": 0.0594, "step": 46 }, { "epoch": 0.17870722433460076, "grad_norm": 1.7001361570347544, "learning_rate": 9.43386842960031e-06, "loss": 0.056, "step": 47 }, { "epoch": 0.18250950570342206, "grad_norm": 0.8564787838977387, "learning_rate": 9.405060971428924e-06, "loss": 0.0498, "step": 48 }, { "epoch": 0.18631178707224336, "grad_norm": 2.4376315388128056, "learning_rate": 9.375584914114963e-06, "loss": 0.0717, "step": 49 }, { "epoch": 0.19011406844106463, "grad_norm": 3.0710870034344704, "learning_rate": 9.345444731527642e-06, "loss": 0.0597, "step": 50 }, { "epoch": 0.19391634980988592, "grad_norm": 1.1595171258313646, "learning_rate": 9.31464499833695e-06, "loss": 0.0484, "step": 51 }, { "epoch": 0.19771863117870722, "grad_norm": 2.036685842781906, "learning_rate": 9.283190389319315e-06, "loss": 0.0533, "step": 52 }, { "epoch": 0.20152091254752852, "grad_norm": 1.6358538836833496, "learning_rate": 9.251085678648072e-06, "loss": 0.063, "step": 53 }, { "epoch": 0.20532319391634982, "grad_norm": 2.341827886552907, "learning_rate": 9.218335739168833e-06, "loss": 0.0614, "step": 54 }, { "epoch": 0.20912547528517111, "grad_norm": 1.6561642430586991, "learning_rate": 9.18494554165989e-06, "loss": 0.0335, "step": 55 }, { "epoch": 0.21292775665399238, "grad_norm": 1.395413976072295, "learning_rate": 9.150920154077753e-06, "loss": 0.0536, "step": 56 }, { "epoch": 0.21673003802281368, "grad_norm": 2.657218435113838, "learning_rate": 9.116264740787937e-06, "loss": 0.0697, "step": 57 }, { "epoch": 0.22053231939163498, "grad_norm": 3.3032959131412634, "learning_rate": 9.08098456178111e-06, "loss": 0.0833, "step": 58 }, { "epoch": 0.22433460076045628, "grad_norm": 1.8964099123844205, "learning_rate": 9.045084971874738e-06, "loss": 0.0568, "step": 59 }, { "epoch": 0.22813688212927757, "grad_norm": 1.5300857159970949, "learning_rate": 9.008571419900334e-06, "loss": 0.0599, "step": 60 }, { "epoch": 0.23193916349809887, "grad_norm": 1.5891594516335554, "learning_rate": 8.97144944787643e-06, "loss": 0.0454, "step": 61 }, { "epoch": 0.23574144486692014, "grad_norm": 1.3527045916224263, "learning_rate": 8.933724690167417e-06, "loss": 0.0429, "step": 62 }, { "epoch": 0.23954372623574144, "grad_norm": 3.4176684375694446, "learning_rate": 8.895402872628352e-06, "loss": 0.0535, "step": 63 }, { "epoch": 0.24334600760456274, "grad_norm": 0.9708842431312176, "learning_rate": 8.856489811735904e-06, "loss": 0.05, "step": 64 }, { "epoch": 0.24714828897338403, "grad_norm": 1.4542670605196113, "learning_rate": 8.816991413705515e-06, "loss": 0.0553, "step": 65 }, { "epoch": 0.2509505703422053, "grad_norm": 1.3407992355297549, "learning_rate": 8.776913673594968e-06, "loss": 0.0337, "step": 66 }, { "epoch": 0.25475285171102663, "grad_norm": 2.282768165416241, "learning_rate": 8.736262674394455e-06, "loss": 0.054, "step": 67 }, { "epoch": 0.2585551330798479, "grad_norm": 2.120541919990877, "learning_rate": 8.695044586103297e-06, "loss": 0.0528, "step": 68 }, { "epoch": 0.2623574144486692, "grad_norm": 2.340243548218491, "learning_rate": 8.653265664793466e-06, "loss": 0.0559, "step": 69 }, { "epoch": 0.2661596958174905, "grad_norm": 2.6043724991846915, "learning_rate": 8.610932251660046e-06, "loss": 0.0788, "step": 70 }, { "epoch": 0.26996197718631176, "grad_norm": 1.981130327340551, "learning_rate": 8.568050772058763e-06, "loss": 0.0631, "step": 71 }, { "epoch": 0.2737642585551331, "grad_norm": 3.0833275950321535, "learning_rate": 8.524627734530738e-06, "loss": 0.066, "step": 72 }, { "epoch": 0.27756653992395436, "grad_norm": 2.2997642624470354, "learning_rate": 8.480669729814635e-06, "loss": 0.056, "step": 73 }, { "epoch": 0.2813688212927757, "grad_norm": 1.4420430072448926, "learning_rate": 8.436183429846314e-06, "loss": 0.0425, "step": 74 }, { "epoch": 0.28517110266159695, "grad_norm": 1.6013511726718204, "learning_rate": 8.39117558674617e-06, "loss": 0.0501, "step": 75 }, { "epoch": 0.2889733840304182, "grad_norm": 1.4338725585293788, "learning_rate": 8.345653031794292e-06, "loss": 0.0422, "step": 76 }, { "epoch": 0.29277566539923955, "grad_norm": 3.890683595146634, "learning_rate": 8.299622674393615e-06, "loss": 0.1042, "step": 77 }, { "epoch": 0.2965779467680608, "grad_norm": 2.8479412113795406, "learning_rate": 8.25309150102121e-06, "loss": 0.065, "step": 78 }, { "epoch": 0.30038022813688214, "grad_norm": 1.431054324105267, "learning_rate": 8.206066574167893e-06, "loss": 0.0458, "step": 79 }, { "epoch": 0.3041825095057034, "grad_norm": 0.8730610840802391, "learning_rate": 8.158555031266255e-06, "loss": 0.04, "step": 80 }, { "epoch": 0.30798479087452474, "grad_norm": 2.448239293063525, "learning_rate": 8.110564083607371e-06, "loss": 0.0759, "step": 81 }, { "epoch": 0.311787072243346, "grad_norm": 1.8456070372447848, "learning_rate": 8.06210101524625e-06, "loss": 0.0576, "step": 82 }, { "epoch": 0.3155893536121673, "grad_norm": 1.834734351428776, "learning_rate": 8.013173181896283e-06, "loss": 0.06, "step": 83 }, { "epoch": 0.3193916349809886, "grad_norm": 1.563283754929812, "learning_rate": 7.963788009812775e-06, "loss": 0.0513, "step": 84 }, { "epoch": 0.3231939163498099, "grad_norm": 1.1475605231367754, "learning_rate": 7.913952994665805e-06, "loss": 0.0543, "step": 85 }, { "epoch": 0.3269961977186312, "grad_norm": 1.4206944686043848, "learning_rate": 7.863675700402527e-06, "loss": 0.0374, "step": 86 }, { "epoch": 0.33079847908745247, "grad_norm": 40.37566470088141, "learning_rate": 7.812963758099118e-06, "loss": 0.2393, "step": 87 }, { "epoch": 0.33460076045627374, "grad_norm": 0.781171019162742, "learning_rate": 7.76182486480253e-06, "loss": 0.0439, "step": 88 }, { "epoch": 0.33840304182509506, "grad_norm": 1.2238520077458774, "learning_rate": 7.710266782362248e-06, "loss": 0.0363, "step": 89 }, { "epoch": 0.34220532319391633, "grad_norm": 3.0641582178280307, "learning_rate": 7.658297336252181e-06, "loss": 0.0662, "step": 90 }, { "epoch": 0.34600760456273766, "grad_norm": 1.965932071811284, "learning_rate": 7.605924414382926e-06, "loss": 0.0472, "step": 91 }, { "epoch": 0.34980988593155893, "grad_norm": 1.755513579015423, "learning_rate": 7.553155965904535e-06, "loss": 0.0355, "step": 92 }, { "epoch": 0.35361216730038025, "grad_norm": 1.7758852391959217, "learning_rate": 7.500000000000001e-06, "loss": 0.0439, "step": 93 }, { "epoch": 0.3574144486692015, "grad_norm": 1.3555739799872628, "learning_rate": 7.4464645846696186e-06, "loss": 0.0503, "step": 94 }, { "epoch": 0.3612167300380228, "grad_norm": 1.4562802703339102, "learning_rate": 7.392557845506433e-06, "loss": 0.0451, "step": 95 }, { "epoch": 0.3650190114068441, "grad_norm": 1.242339996611598, "learning_rate": 7.3382879644629345e-06, "loss": 0.0372, "step": 96 }, { "epoch": 0.3688212927756654, "grad_norm": 1.7658082693457402, "learning_rate": 7.283663178609204e-06, "loss": 0.0516, "step": 97 }, { "epoch": 0.3726235741444867, "grad_norm": 2.308009029051408, "learning_rate": 7.2286917788826926e-06, "loss": 0.049, "step": 98 }, { "epoch": 0.376425855513308, "grad_norm": 1.5974409986244866, "learning_rate": 7.173382108829826e-06, "loss": 0.052, "step": 99 }, { "epoch": 0.38022813688212925, "grad_norm": 0.8994670069496103, "learning_rate": 7.117742563339622e-06, "loss": 0.0256, "step": 100 }, { "epoch": 0.3840304182509506, "grad_norm": 2.4678061096370465, "learning_rate": 7.061781587369518e-06, "loss": 0.0666, "step": 101 }, { "epoch": 0.38783269961977185, "grad_norm": 0.866063867737264, "learning_rate": 7.005507674663594e-06, "loss": 0.0404, "step": 102 }, { "epoch": 0.3916349809885932, "grad_norm": 1.2929123997330905, "learning_rate": 6.948929366463397e-06, "loss": 0.0416, "step": 103 }, { "epoch": 0.39543726235741444, "grad_norm": 1.6302913701612327, "learning_rate": 6.892055250211552e-06, "loss": 0.0444, "step": 104 }, { "epoch": 0.39923954372623577, "grad_norm": 2.728101647559586, "learning_rate": 6.834893958248361e-06, "loss": 0.0475, "step": 105 }, { "epoch": 0.40304182509505704, "grad_norm": 2.2488039758111915, "learning_rate": 6.77745416650159e-06, "loss": 0.0659, "step": 106 }, { "epoch": 0.4068441064638783, "grad_norm": 1.977313537391127, "learning_rate": 6.719744593169642e-06, "loss": 0.052, "step": 107 }, { "epoch": 0.41064638783269963, "grad_norm": 5.706465703956973, "learning_rate": 6.6617739973982985e-06, "loss": 0.2998, "step": 108 }, { "epoch": 0.4144486692015209, "grad_norm": 3.8000626899961825, "learning_rate": 6.6035511779512764e-06, "loss": 0.1164, "step": 109 }, { "epoch": 0.41825095057034223, "grad_norm": 10.502568472372902, "learning_rate": 6.545084971874738e-06, "loss": 0.123, "step": 110 }, { "epoch": 0.4220532319391635, "grad_norm": 1.8313837197728655, "learning_rate": 6.486384253156014e-06, "loss": 0.1134, "step": 111 }, { "epoch": 0.42585551330798477, "grad_norm": 2.0327190996550177, "learning_rate": 6.427457931376712e-06, "loss": 0.0888, "step": 112 }, { "epoch": 0.4296577946768061, "grad_norm": 1.7161432735601356, "learning_rate": 6.368314950360416e-06, "loss": 0.0902, "step": 113 }, { "epoch": 0.43346007604562736, "grad_norm": 1.074310172184068, "learning_rate": 6.308964286815203e-06, "loss": 0.0581, "step": 114 }, { "epoch": 0.4372623574144487, "grad_norm": 1.6451467195844793, "learning_rate": 6.249414948971154e-06, "loss": 0.0503, "step": 115 }, { "epoch": 0.44106463878326996, "grad_norm": 4.814077108982163, "learning_rate": 6.189675975213094e-06, "loss": 0.1629, "step": 116 }, { "epoch": 0.4448669201520912, "grad_norm": 1.4504882892841855, "learning_rate": 6.129756432708739e-06, "loss": 0.077, "step": 117 }, { "epoch": 0.44866920152091255, "grad_norm": 2.5745840097932517, "learning_rate": 6.0696654160324875e-06, "loss": 0.0785, "step": 118 }, { "epoch": 0.4524714828897338, "grad_norm": 1.2717081467380555, "learning_rate": 6.009412045785051e-06, "loss": 0.0721, "step": 119 }, { "epoch": 0.45627376425855515, "grad_norm": 1.3500755816952386, "learning_rate": 5.9490054672091305e-06, "loss": 0.0585, "step": 120 }, { "epoch": 0.4600760456273764, "grad_norm": 0.6582503184547626, "learning_rate": 5.888454848801345e-06, "loss": 0.0647, "step": 121 }, { "epoch": 0.46387832699619774, "grad_norm": 2.057266551706436, "learning_rate": 5.82776938092065e-06, "loss": 0.1195, "step": 122 }, { "epoch": 0.467680608365019, "grad_norm": 1.717649125812939, "learning_rate": 5.766958274393428e-06, "loss": 0.0762, "step": 123 }, { "epoch": 0.4714828897338403, "grad_norm": 1.0979465765609746, "learning_rate": 5.706030759115458e-06, "loss": 0.074, "step": 124 }, { "epoch": 0.4752851711026616, "grad_norm": 2.4576710255834184, "learning_rate": 5.644996082651018e-06, "loss": 0.09, "step": 125 }, { "epoch": 0.4790874524714829, "grad_norm": 1.4141031874389902, "learning_rate": 5.583863508829281e-06, "loss": 0.0712, "step": 126 }, { "epoch": 0.4828897338403042, "grad_norm": 1.3722960078778967, "learning_rate": 5.522642316338268e-06, "loss": 0.0774, "step": 127 }, { "epoch": 0.4866920152091255, "grad_norm": 1.9899235244969848, "learning_rate": 5.46134179731651e-06, "loss": 0.0727, "step": 128 }, { "epoch": 0.49049429657794674, "grad_norm": 0.9857281384665181, "learning_rate": 5.399971255942708e-06, "loss": 0.0709, "step": 129 }, { "epoch": 0.49429657794676807, "grad_norm": 1.873178689272728, "learning_rate": 5.338540007023538e-06, "loss": 0.086, "step": 130 }, { "epoch": 0.49809885931558934, "grad_norm": 1.2985642993407172, "learning_rate": 5.27705737457985e-06, "loss": 0.0729, "step": 131 }, { "epoch": 0.5019011406844106, "grad_norm": 2.00845296448955, "learning_rate": 5.2155326904314795e-06, "loss": 0.0796, "step": 132 }, { "epoch": 0.5057034220532319, "grad_norm": 1.3854160082227878, "learning_rate": 5.153975292780852e-06, "loss": 0.0655, "step": 133 }, { "epoch": 0.5095057034220533, "grad_norm": 0.8110311967262005, "learning_rate": 5.09239452479565e-06, "loss": 0.0832, "step": 134 }, { "epoch": 0.5133079847908745, "grad_norm": 1.349215941618463, "learning_rate": 5.030799733190694e-06, "loss": 0.0734, "step": 135 }, { "epoch": 0.5171102661596958, "grad_norm": 1.7298488968449564, "learning_rate": 4.9692002668093075e-06, "loss": 0.0854, "step": 136 }, { "epoch": 0.5209125475285171, "grad_norm": 1.9286164063763356, "learning_rate": 4.907605475204352e-06, "loss": 0.0967, "step": 137 }, { "epoch": 0.5247148288973384, "grad_norm": 1.5511875987133978, "learning_rate": 4.846024707219149e-06, "loss": 0.0895, "step": 138 }, { "epoch": 0.5285171102661597, "grad_norm": 1.8297476277504392, "learning_rate": 4.784467309568524e-06, "loss": 0.0861, "step": 139 }, { "epoch": 0.532319391634981, "grad_norm": 1.4103839066534454, "learning_rate": 4.7229426254201504e-06, "loss": 0.0783, "step": 140 }, { "epoch": 0.5361216730038023, "grad_norm": 2.1074351624344136, "learning_rate": 4.661459992976463e-06, "loss": 0.0822, "step": 141 }, { "epoch": 0.5399239543726235, "grad_norm": 2.0662466538476827, "learning_rate": 4.6000287440572925e-06, "loss": 0.0923, "step": 142 }, { "epoch": 0.5437262357414449, "grad_norm": 1.9436401310479368, "learning_rate": 4.53865820268349e-06, "loss": 0.0973, "step": 143 }, { "epoch": 0.5475285171102662, "grad_norm": 1.6678903520906458, "learning_rate": 4.477357683661734e-06, "loss": 0.0877, "step": 144 }, { "epoch": 0.5513307984790875, "grad_norm": 0.809599618210663, "learning_rate": 4.41613649117072e-06, "loss": 0.0786, "step": 145 }, { "epoch": 0.5551330798479087, "grad_norm": 1.0977795425796129, "learning_rate": 4.355003917348985e-06, "loss": 0.0831, "step": 146 }, { "epoch": 0.55893536121673, "grad_norm": 1.8349191027751834, "learning_rate": 4.293969240884545e-06, "loss": 0.0546, "step": 147 }, { "epoch": 0.5627376425855514, "grad_norm": 1.2008364879121194, "learning_rate": 4.233041725606573e-06, "loss": 0.0614, "step": 148 }, { "epoch": 0.5665399239543726, "grad_norm": 0.732201424861496, "learning_rate": 4.17223061907935e-06, "loss": 0.0734, "step": 149 }, { "epoch": 0.5703422053231939, "grad_norm": 1.225050397564832, "learning_rate": 4.111545151198657e-06, "loss": 0.0651, "step": 150 }, { "epoch": 0.5741444866920152, "grad_norm": 1.2302003045619856, "learning_rate": 4.050994532790871e-06, "loss": 0.0837, "step": 151 }, { "epoch": 0.5779467680608364, "grad_norm": 1.8397382613360544, "learning_rate": 3.99058795421495e-06, "loss": 0.0607, "step": 152 }, { "epoch": 0.5817490494296578, "grad_norm": 1.4093018088954288, "learning_rate": 3.930334583967514e-06, "loss": 0.0575, "step": 153 }, { "epoch": 0.5855513307984791, "grad_norm": 1.4975102213218823, "learning_rate": 3.870243567291263e-06, "loss": 0.0634, "step": 154 }, { "epoch": 0.5893536121673004, "grad_norm": 1.8275993786721776, "learning_rate": 3.8103240247869077e-06, "loss": 0.0477, "step": 155 }, { "epoch": 0.5931558935361216, "grad_norm": 1.5246149628230938, "learning_rate": 3.7505850510288455e-06, "loss": 0.0927, "step": 156 }, { "epoch": 0.596958174904943, "grad_norm": 1.07378251359328, "learning_rate": 3.6910357131847986e-06, "loss": 0.0798, "step": 157 }, { "epoch": 0.6007604562737643, "grad_norm": 1.2099308034216252, "learning_rate": 3.6316850496395863e-06, "loss": 0.0724, "step": 158 }, { "epoch": 0.6045627376425855, "grad_norm": 1.0732935946375615, "learning_rate": 3.5725420686232903e-06, "loss": 0.0773, "step": 159 }, { "epoch": 0.6083650190114068, "grad_norm": 0.9225394975341887, "learning_rate": 3.513615746843987e-06, "loss": 0.0635, "step": 160 }, { "epoch": 0.6121673003802282, "grad_norm": 2.6158603947076577, "learning_rate": 3.4549150281252635e-06, "loss": 0.1059, "step": 161 }, { "epoch": 0.6159695817490495, "grad_norm": 0.9294415514571752, "learning_rate": 3.3964488220487252e-06, "loss": 0.0709, "step": 162 }, { "epoch": 0.6197718631178707, "grad_norm": 1.9222737047405196, "learning_rate": 3.3382260026017027e-06, "loss": 0.062, "step": 163 }, { "epoch": 0.623574144486692, "grad_norm": 1.906035535415949, "learning_rate": 3.2802554068303595e-06, "loss": 0.0967, "step": 164 }, { "epoch": 0.6273764258555133, "grad_norm": 1.7676981737731525, "learning_rate": 3.22254583349841e-06, "loss": 0.0961, "step": 165 }, { "epoch": 0.6311787072243346, "grad_norm": 1.605378292764449, "learning_rate": 3.16510604175164e-06, "loss": 0.0507, "step": 166 }, { "epoch": 0.6349809885931559, "grad_norm": 1.1486017436180254, "learning_rate": 3.107944749788449e-06, "loss": 0.0653, "step": 167 }, { "epoch": 0.6387832699619772, "grad_norm": 1.9815488934112822, "learning_rate": 3.0510706335366034e-06, "loss": 0.0687, "step": 168 }, { "epoch": 0.6425855513307985, "grad_norm": 1.747055284984615, "learning_rate": 2.9944923253364066e-06, "loss": 0.0753, "step": 169 }, { "epoch": 0.6463878326996197, "grad_norm": 1.5044113934612662, "learning_rate": 2.9382184126304834e-06, "loss": 0.0751, "step": 170 }, { "epoch": 0.6501901140684411, "grad_norm": 1.4456809835756477, "learning_rate": 2.8822574366603804e-06, "loss": 0.0857, "step": 171 }, { "epoch": 0.6539923954372624, "grad_norm": 0.8421771448658065, "learning_rate": 2.8266178911701757e-06, "loss": 0.0754, "step": 172 }, { "epoch": 0.6577946768060836, "grad_norm": 1.6606796114651394, "learning_rate": 2.771308221117309e-06, "loss": 0.0633, "step": 173 }, { "epoch": 0.6615969581749049, "grad_norm": 1.0518360600901147, "learning_rate": 2.7163368213907975e-06, "loss": 0.0595, "step": 174 }, { "epoch": 0.6653992395437263, "grad_norm": 0.8118249216729664, "learning_rate": 2.6617120355370667e-06, "loss": 0.0721, "step": 175 }, { "epoch": 0.6692015209125475, "grad_norm": 2.735849675959911, "learning_rate": 2.607442154493568e-06, "loss": 0.0981, "step": 176 }, { "epoch": 0.6730038022813688, "grad_norm": 1.5818010804365756, "learning_rate": 2.5535354153303827e-06, "loss": 0.0944, "step": 177 }, { "epoch": 0.6768060836501901, "grad_norm": 2.106035867055531, "learning_rate": 2.5000000000000015e-06, "loss": 0.0563, "step": 178 }, { "epoch": 0.6806083650190115, "grad_norm": 1.1144497943115816, "learning_rate": 2.4468440340954664e-06, "loss": 0.08, "step": 179 }, { "epoch": 0.6844106463878327, "grad_norm": 0.6703641595119413, "learning_rate": 2.3940755856170744e-06, "loss": 0.0681, "step": 180 }, { "epoch": 0.688212927756654, "grad_norm": 1.5423718248019078, "learning_rate": 2.341702663747819e-06, "loss": 0.0934, "step": 181 }, { "epoch": 0.6920152091254753, "grad_norm": 2.5974586850504173, "learning_rate": 2.289733217637753e-06, "loss": 0.0727, "step": 182 }, { "epoch": 0.6958174904942965, "grad_norm": 1.6582686566551548, "learning_rate": 2.238175135197471e-06, "loss": 0.0708, "step": 183 }, { "epoch": 0.6996197718631179, "grad_norm": 1.5364491430874354, "learning_rate": 2.1870362419008844e-06, "loss": 0.0896, "step": 184 }, { "epoch": 0.7034220532319392, "grad_norm": 0.8883156220926425, "learning_rate": 2.136324299597474e-06, "loss": 0.0805, "step": 185 }, { "epoch": 0.7072243346007605, "grad_norm": 0.722419818120629, "learning_rate": 2.0860470053341957e-06, "loss": 0.0677, "step": 186 }, { "epoch": 0.7110266159695817, "grad_norm": 0.8687543599479524, "learning_rate": 2.0362119901872262e-06, "loss": 0.0743, "step": 187 }, { "epoch": 0.714828897338403, "grad_norm": 1.1765271610095978, "learning_rate": 1.9868268181037186e-06, "loss": 0.073, "step": 188 }, { "epoch": 0.7186311787072244, "grad_norm": 1.57171034016372, "learning_rate": 1.937898984753751e-06, "loss": 0.0774, "step": 189 }, { "epoch": 0.7224334600760456, "grad_norm": 2.5653255183794843, "learning_rate": 1.8894359163926312e-06, "loss": 0.0845, "step": 190 }, { "epoch": 0.7262357414448669, "grad_norm": 1.8207532400192823, "learning_rate": 1.8414449687337467e-06, "loss": 0.0866, "step": 191 }, { "epoch": 0.7300380228136882, "grad_norm": 2.494001462793861, "learning_rate": 1.7939334258321094e-06, "loss": 0.0745, "step": 192 }, { "epoch": 0.7338403041825095, "grad_norm": 1.2902772325556104, "learning_rate": 1.746908498978791e-06, "loss": 0.0755, "step": 193 }, { "epoch": 0.7376425855513308, "grad_norm": 1.6350962042870758, "learning_rate": 1.7003773256063882e-06, "loss": 0.0734, "step": 194 }, { "epoch": 0.7414448669201521, "grad_norm": 0.9676259463093386, "learning_rate": 1.6543469682057105e-06, "loss": 0.0816, "step": 195 }, { "epoch": 0.7452471482889734, "grad_norm": 0.5610374914180468, "learning_rate": 1.60882441325383e-06, "loss": 0.0754, "step": 196 }, { "epoch": 0.7490494296577946, "grad_norm": 0.7804394462668183, "learning_rate": 1.5638165701536866e-06, "loss": 0.0747, "step": 197 }, { "epoch": 0.752851711026616, "grad_norm": 1.550496983980984, "learning_rate": 1.5193302701853674e-06, "loss": 0.0791, "step": 198 }, { "epoch": 0.7566539923954373, "grad_norm": 1.9010955783137817, "learning_rate": 1.475372265469265e-06, "loss": 0.0732, "step": 199 }, { "epoch": 0.7604562737642585, "grad_norm": 1.6079174758382067, "learning_rate": 1.4319492279412388e-06, "loss": 0.0708, "step": 200 }, { "epoch": 0.7642585551330798, "grad_norm": 0.6607128276682519, "learning_rate": 1.389067748339954e-06, "loss": 0.0715, "step": 201 }, { "epoch": 0.7680608365019012, "grad_norm": 1.3595868708508696, "learning_rate": 1.3467343352065349e-06, "loss": 0.0781, "step": 202 }, { "epoch": 0.7718631178707225, "grad_norm": 2.419173076809107, "learning_rate": 1.3049554138967052e-06, "loss": 0.0739, "step": 203 }, { "epoch": 0.7756653992395437, "grad_norm": 1.8044702216287307, "learning_rate": 1.2637373256055445e-06, "loss": 0.0825, "step": 204 }, { "epoch": 0.779467680608365, "grad_norm": 0.9717656143179705, "learning_rate": 1.2230863264050308e-06, "loss": 0.0777, "step": 205 }, { "epoch": 0.7832699619771863, "grad_norm": 1.181299656152716, "learning_rate": 1.1830085862944851e-06, "loss": 0.0726, "step": 206 }, { "epoch": 0.7870722433460076, "grad_norm": 0.9219123710812074, "learning_rate": 1.1435101882640964e-06, "loss": 0.0832, "step": 207 }, { "epoch": 0.7908745247148289, "grad_norm": 0.6914642768758692, "learning_rate": 1.1045971273716476e-06, "loss": 0.0793, "step": 208 }, { "epoch": 0.7946768060836502, "grad_norm": 2.273763477220486, "learning_rate": 1.066275309832584e-06, "loss": 0.0715, "step": 209 }, { "epoch": 0.7984790874524715, "grad_norm": 1.6498287914408991, "learning_rate": 1.02855055212357e-06, "loss": 0.0669, "step": 210 }, { "epoch": 0.8022813688212928, "grad_norm": 1.102115306685056, "learning_rate": 9.91428580099667e-07, "loss": 0.0775, "step": 211 }, { "epoch": 0.8060836501901141, "grad_norm": 2.4914475016820967, "learning_rate": 9.549150281252633e-07, "loss": 0.0677, "step": 212 }, { "epoch": 0.8098859315589354, "grad_norm": 0.6217527292902375, "learning_rate": 9.190154382188921e-07, "loss": 0.0704, "step": 213 }, { "epoch": 0.8136882129277566, "grad_norm": 0.776872411148349, "learning_rate": 8.837352592120646e-07, "loss": 0.0683, "step": 214 }, { "epoch": 0.8174904942965779, "grad_norm": 1.3830390856962926, "learning_rate": 8.490798459222477e-07, "loss": 0.0719, "step": 215 }, { "epoch": 0.8212927756653993, "grad_norm": 0.9016535949900154, "learning_rate": 8.150544583401116e-07, "loss": 0.0706, "step": 216 }, { "epoch": 0.8250950570342205, "grad_norm": 1.2565895090887536, "learning_rate": 7.816642608311692e-07, "loss": 0.0691, "step": 217 }, { "epoch": 0.8288973384030418, "grad_norm": 0.4685806197849472, "learning_rate": 7.489143213519301e-07, "loss": 0.0749, "step": 218 }, { "epoch": 0.8326996197718631, "grad_norm": 1.8324078968655844, "learning_rate": 7.168096106806871e-07, "loss": 0.0842, "step": 219 }, { "epoch": 0.8365019011406845, "grad_norm": 0.615073356028593, "learning_rate": 6.853550016630517e-07, "loss": 0.0697, "step": 220 }, { "epoch": 0.8403041825095057, "grad_norm": 2.4222883575922154, "learning_rate": 6.545552684723583e-07, "loss": 0.0803, "step": 221 }, { "epoch": 0.844106463878327, "grad_norm": 2.325809289721842, "learning_rate": 6.244150858850368e-07, "loss": 0.0969, "step": 222 }, { "epoch": 0.8479087452471483, "grad_norm": 1.176358839322873, "learning_rate": 5.949390285710777e-07, "loss": 0.0599, "step": 223 }, { "epoch": 0.8517110266159695, "grad_norm": 0.7047528759218481, "learning_rate": 5.661315703996905e-07, "loss": 0.0807, "step": 224 }, { "epoch": 0.8555133079847909, "grad_norm": 1.1938640871812451, "learning_rate": 5.379970837602611e-07, "loss": 0.0712, "step": 225 }, { "epoch": 0.8593155893536122, "grad_norm": 2.0073147773750635, "learning_rate": 5.105398388987098e-07, "loss": 0.0723, "step": 226 }, { "epoch": 0.8631178707224335, "grad_norm": 1.910913263898643, "learning_rate": 4.837640032693558e-07, "loss": 0.0795, "step": 227 }, { "epoch": 0.8669201520912547, "grad_norm": 1.8337713325079474, "learning_rate": 4.576736409023813e-07, "loss": 0.068, "step": 228 }, { "epoch": 0.870722433460076, "grad_norm": 1.2290049663038816, "learning_rate": 4.322727117869951e-07, "loss": 0.0768, "step": 229 }, { "epoch": 0.8745247148288974, "grad_norm": 1.4789730966126406, "learning_rate": 4.0756507127038494e-07, "loss": 0.0767, "step": 230 }, { "epoch": 0.8783269961977186, "grad_norm": 1.6910881638069963, "learning_rate": 3.8355446947255293e-07, "loss": 0.0665, "step": 231 }, { "epoch": 0.8821292775665399, "grad_norm": 1.033604422841365, "learning_rate": 3.602445507171276e-07, "loss": 0.0778, "step": 232 }, { "epoch": 0.8859315589353612, "grad_norm": 0.8013760374315584, "learning_rate": 3.3763885297822153e-07, "loss": 0.0657, "step": 233 }, { "epoch": 0.8897338403041825, "grad_norm": 2.429431976291132, "learning_rate": 3.1574080734344757e-07, "loss": 0.0644, "step": 234 }, { "epoch": 0.8935361216730038, "grad_norm": 0.9711502835652512, "learning_rate": 2.9455373749314285e-07, "loss": 0.0706, "step": 235 }, { "epoch": 0.8973384030418251, "grad_norm": 1.275326544398461, "learning_rate": 2.7408085919590265e-07, "loss": 0.0679, "step": 236 }, { "epoch": 0.9011406844106464, "grad_norm": 1.2439206544006953, "learning_rate": 2.5432527982049424e-07, "loss": 0.0686, "step": 237 }, { "epoch": 0.9049429657794676, "grad_norm": 1.1151619016488201, "learning_rate": 2.3528999786421758e-07, "loss": 0.067, "step": 238 }, { "epoch": 0.908745247148289, "grad_norm": 1.5637158385664465, "learning_rate": 2.1697790249779638e-07, "loss": 0.0653, "step": 239 }, { "epoch": 0.9125475285171103, "grad_norm": 0.4492873956448081, "learning_rate": 1.9939177312685963e-07, "loss": 0.0782, "step": 240 }, { "epoch": 0.9163498098859315, "grad_norm": 1.1564997677392477, "learning_rate": 1.825342789700846e-07, "loss": 0.0608, "step": 241 }, { "epoch": 0.9201520912547528, "grad_norm": 1.8601658350989132, "learning_rate": 1.664079786540629e-07, "loss": 0.0853, "step": 242 }, { "epoch": 0.9239543726235742, "grad_norm": 1.227671794720867, "learning_rate": 1.510153198249531e-07, "loss": 0.0681, "step": 243 }, { "epoch": 0.9277566539923955, "grad_norm": 1.318746672359122, "learning_rate": 1.363586387769761e-07, "loss": 0.0716, "step": 244 }, { "epoch": 0.9315589353612167, "grad_norm": 1.6242713478117103, "learning_rate": 1.22440160097817e-07, "loss": 0.0585, "step": 245 }, { "epoch": 0.935361216730038, "grad_norm": 2.4521843943377455, "learning_rate": 1.0926199633097156e-07, "loss": 0.0748, "step": 246 }, { "epoch": 0.9391634980988594, "grad_norm": 1.624621392645855, "learning_rate": 9.682614765511134e-08, "loss": 0.0778, "step": 247 }, { "epoch": 0.9429657794676806, "grad_norm": 0.4946954920558412, "learning_rate": 8.513450158049109e-08, "loss": 0.0787, "step": 248 }, { "epoch": 0.9467680608365019, "grad_norm": 0.5126439315912313, "learning_rate": 7.418883266246734e-08, "loss": 0.0779, "step": 249 }, { "epoch": 0.9505703422053232, "grad_norm": 1.9296492595340622, "learning_rate": 6.399080223215503e-08, "loss": 0.071, "step": 250 }, { "epoch": 0.9543726235741445, "grad_norm": 2.381450276806091, "learning_rate": 5.454195814427021e-08, "loss": 0.0571, "step": 251 }, { "epoch": 0.9581749049429658, "grad_norm": 1.6084565235987887, "learning_rate": 4.584373454219859e-08, "loss": 0.0895, "step": 252 }, { "epoch": 0.9619771863117871, "grad_norm": 0.8689683317083515, "learning_rate": 3.7897451640321326e-08, "loss": 0.0666, "step": 253 }, { "epoch": 0.9657794676806084, "grad_norm": 1.3522488546526146, "learning_rate": 3.0704315523631956e-08, "loss": 0.0611, "step": 254 }, { "epoch": 0.9695817490494296, "grad_norm": 2.2795049559749754, "learning_rate": 2.426541796467785e-08, "loss": 0.0536, "step": 255 }, { "epoch": 0.973384030418251, "grad_norm": 1.087776806550556, "learning_rate": 1.8581736257852756e-08, "loss": 0.0602, "step": 256 }, { "epoch": 0.9771863117870723, "grad_norm": 1.1862726730320716, "learning_rate": 1.3654133071059894e-08, "loss": 0.0722, "step": 257 }, { "epoch": 0.9809885931558935, "grad_norm": 0.7704198670078842, "learning_rate": 9.48335631477948e-09, "loss": 0.0639, "step": 258 }, { "epoch": 0.9847908745247148, "grad_norm": 1.0439502260084565, "learning_rate": 6.070039028550634e-09, "loss": 0.0623, "step": 259 }, { "epoch": 0.9885931558935361, "grad_norm": 0.9401759231345554, "learning_rate": 3.41469928488547e-09, "loss": 0.0735, "step": 260 }, { "epoch": 0.9923954372623575, "grad_norm": 1.8274467631316753, "learning_rate": 1.5177401106419853e-09, "loss": 0.0714, "step": 261 }, { "epoch": 0.9961977186311787, "grad_norm": 1.347810152487735, "learning_rate": 3.7944942584688947e-10, "loss": 0.0698, "step": 262 }, { "epoch": 1.0, "grad_norm": 1.4596446787851405, "learning_rate": 0.0, "loss": 0.0874, "step": 263 }, { "epoch": 1.0, "step": 263, "total_flos": 8314083559424.0, "train_loss": 0.04440210763889121, "train_runtime": 1286.9653, "train_samples_per_second": 1.632, "train_steps_per_second": 0.204 } ], "logging_steps": 1.0, "max_steps": 263, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 100, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 8314083559424.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }